JP2017010307A - Measuring device, measuring system, measuring result collection method, and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a remote natural information measuring system making good use of an approximation curve process.SOLUTION: A measuring device comprises: an approximation curve generation unit that obtains an observation result measured by a sensor, and generates information representing an approximation curve that approximates the aggregation of points indicated by the time-series observation result; and a communication unit for transmitting the information representing the approximation curve generated by the approximation curve generation unit.SELECTED DRAWING: Figure 1

Description

  The present invention relates to a measurement apparatus, a measurement system, a measurement result collection method, and a program.

  In recent years, in the field of agriculture, there has been a movement to conduct more efficient farming activities by collecting information on the natural world related to the cultivation status of crops such as soil, air, and temperature using a measuring device for growing crops. is there. For example, it is conceivable that the amount of carbon dioxide contained in the air in the green house is measured using a measuring device, and carbon dioxide is added as necessary. Such an act of collecting information in the natural world that is particularly useful in the agricultural field using a measuring device is called “sensing for agriculture”.

  Farmers ultimately benefit from sensing for agriculture. However, the information collected by sensing for agriculture is simply a list of numerical values, and it is very difficult for farmers to use as it is. Accordingly, there are service providers that produce and operate devices for processing these information and provide them to farmers in the form of agricultural ICT services. There are also businesses that predict crop yields based on sensed information and provide them as information to be considered when determining the price of crops in a trading market. In this way, there are a wide range of users other than farmers who mediate and indirectly use information collected by sensing for agriculture. Therefore, the demand also varies widely.

  Demand for sensing for agriculture focuses on the freshness of information, that is, the time it takes for the farmer to finally reach the farmer. It can be divided into two categories: acquisition. First, there is a need to realize a service that constantly measures the current room temperature and communicates a warning to a farmer about a sudden rise in room temperature. Therefore, the measuring device needs to inform the user as soon as possible about the current situation. Also, it is often used for important decisions in emergency situations, so accuracy is often required. However, at present, the amount of information is often small in order to make an important decision. Furthermore, there is a need to acquire knowledge such as yield prediction by observing the trend of temperature based on accumulated past temperature information, for example, to acquire knowledge by accumulation and analysis. In order to observe the overall trend in this need, the accuracy of individual information and the comprehensiveness of collected information are not so important. Immediacy is not required. In this case, the more information in the past, the easier the analysis becomes as the granularity of the information becomes finer, and the amount of information tends to increase.

  Information to meet these needs is originally measured by the same measuring device. Therefore, it can be said that the difference in information to satisfy each need arises in the information transmission means and the processing process. In an extreme example, if you want to know the current temperature when you focus on the temperature in a certain city, you can check the value of the measurement device connected online on the Web page. On the other hand, if you want to know the change in temperature throughout the year, you can purchase a weather yearbook provided by an organization such as the Japan Meteorological Agency, which is issued on a storage medium such as a CD. As described above, the structure of the transmission means and the processing process has a great influence on the characteristics of information, and determines whether or not the target service can be realized.

  It should be noted that the observations from the original measuring device have the highest accuracy and granularity in the process. As long as only this information is used, the accuracy and granularity of the information will not increase during transmission or processing. So-called corrections and calibrations are carried out using “external information” that is knowledge obtained from previous observations. In other words, the information amount of the measured information is increased by adding information called previous information.

  From another point of view, classification can be based on the need for preparation for information utilization. For example, there is a need for information in a unified and formatted format. When executing a service by computer processing, it is desirable to collect information in a format that can be processed by the computer. When monitoring and automatic analysis of information by a computer, information is often read repeatedly with the same module, so the format of the information needs to be unified. Even if it is not completely unified, the original information must be within the range that can be shaped by a computer. There is also a need for an interface for users who want to view information. There is a need (particularly a need for visualization) for simpler information shaping and presentation for human information analysis. At present, there are many scenes where human beings consider information based on information in order to obtain knowledge specific to the business field from the collected information. In this case, the observation result is rarely presented as it is to the user who wants to browse the information. In many cases, information is presented in the form of, for example, visualization by graphs, total aggregation by tables, and extraction of values indicating characteristics such as average values. By refining these presentation methods, the user can obtain more effective knowledge more easily.

  Here are some examples of needs and services for agricultural sensing. One is an email service that alerts the temperature inside the greenhouse. When the temperature rises, the crops burnt down in about half a day and become useless. In order to prevent this, it is a service form in which the farmer is notified of a temperature rise warning by e-mail and encourages the farmer to improve the situation. Next, there is a need to know the integrated amount of sunlight in the field. It is known that the degree of crop growth is related to the time of sun exposure. Therefore, knowing the integrated value of the amount of sunlight at a certain point, the remaining amount of sunlight necessary for the completion of the crop can be obtained. There is a need to combine this information with past year-round observations to predict crop yield. There is also a need for humidity measurement for irrigation control. Equipment for automatic irrigation is remotely controlled using a computer. This creates a need for remote information on humidity information that triggers device activation.

  Now, there are some essential issues in implementing sensing for agriculture. First, communication costs are avoided when a line is used for information transmission. In particular, transmission power is suppressed when information communication is performed by wireless communication from a measuring device. Next, it is a countermeasure for mixing abnormal values and illegal values into information derived from the communication path and processing. It also avoids the limit of the amount of information stored. In particular, there is a problem of the amount of buffering information on the measuring device side. There is also room for improvement in improving the method of presenting information to the user.

As for avoidance of communication cost burden, most of the sensings for agriculture have a wide range of fields to be observed outdoors. Therefore, mobile phone communication is often used. When the usage cost of the communication network is a pay-per-use charge related to the communication amount, measures such as reducing the amount of transmission information by thinning the observation results by thinning the observation results to reduce the communication cost. On the other hand, the accuracy of information is lost by reducing the amount of information. Therefore, in this case, it is necessary to take measures to reduce the communication cost as long as the service can be maintained.
For this problem, for example, in Patent Document 1, in a data transmission device that transmits operation data and weather data provided in a photovoltaic power generation system, information indicating importance is given to the transmission data, Describes a data transmission device that preferentially transmits high data.

Regarding the transmission power suppression, there are cases where the agricultural measurement apparatus is powered by solar cells for the same reason. Solar cells supply enough power to measure ambient information and to operate an internal computer. However, wireless communication requires a large amount of power compared to computer operation. In particular, when used outdoors over a long distance and in a wide area, power consumption often increases in order to obtain a sufficient gain of radio waves.
To deal with this problem, for example, in Patent Document 2, among a large number of sensors provided on a farm or the like, the measured values of a plurality of sensors provided within a predetermined range are compared, and the measured values are the same. When it can be considered, a data collection control method is described in which a representative sensor within the range is selected and the other sensors are stopped to obtain a power saving effect.

  There are several factors regarding the mixing of illegal values and abnormal values into information. First, there is an abnormality in the surrounding environment of the measuring device. For example, this is a state in which the measuring device is temporarily hidden by a person's shadow for farm work, and the amount of sunlight is drastically reduced for that short time. This is certainly accurate ambient information, but not useful for agriculture. The next possible factor is an abnormality in the measurement terminal of the measurement device. The measurement device's measurement terminal may be set to a value that can be handled by a computer directly in the analog-to-digital conversion circuit of the measurement device, with respect to changes in analog values such as resistance and capacitance of the electronic structure at the tip of the measurement device Many. In this case, the value may change instantaneously due to electrical noise mixed due to disconnection of the measurement terminal or the loop portion of the circuit operating as a kind of antenna. Such information is often accompanied by rapid changes. This information is completely unrelated to agricultural use.

Another factor may be an abnormality of the measuring device main body. For example, the power supply of the measuring device is temporarily lost. This can be a situation in which, when the measuring device is driven by a solar cell or the like, a fallen leaf is applied to the solar cell. In this case, the information is suddenly lost from a certain point, and when the situation recovers, the information is transmitted as if nothing happened. Although the change in the environment is certainly a change in the situation where the measuring device is placed, it is not the purpose to detect such a change in the environment. Another factor is temporary suspension of information transmission due to deterioration of radio wave conditions. This is a situation that can occur particularly in measuring devices that rely on radio for information transmission and reception. There are various causes of the deterioration of radio wave conditions, and in addition to simple changes in radio wave reachability, it is conceivable that the performance of communication stations deteriorates due to excess of terminals accommodated in mobile communication stations as communication destinations.
In general, measures for this problem include selection of measured values based on threshold values and removal of singularities by a feedback circuit such as a low-pass filter.

  Regarding the problem of the amount of information stored on the measuring device side, at least one piece of digital information is buffered on the measuring device side because of the recent characteristics of measuring devices that convert values from analog to digital. It will be. Actually, in order to improve the efficiency of wireless transmission, a plurality of these are buffered and transmitted. For the transmission of information, for example, preliminary communication for establishing a communication session is required. Such spare communication increases the amount of communication per time as overhead. When the amount of communication information is very small, an increase in communication amount due to this overhead is a value that cannot be ignored. Therefore, in most measuring apparatuses, as much information as possible is buffered within a range that does not affect the service, and the information is transmitted at a time. Naturally, this requires a memory for temporarily storing a large amount of information. An increase in memory leads to an increase in manufacturing cost of the measuring device. In the case of a configuration that requires power to drive the memory, the contents of the memory must be maintained until the next transmission is successful, and information that is input one after another must be accumulated during that time. . In cases where the measurement device is unlikely to be restored, such as when the communication path is interrupted due to the deterioration of the radio wave conditions mentioned above, the measurement device accumulates information up to the maximum capacity it can hold, and then Because one has to be destroyed, information loss occurs. The mainstream approach to this problem is to reduce the amount of information itself by lossy compression. For example, a method is often used in which several irreversible compression patterns are prepared and a pattern suitable for information characteristics is applied.

  Regarding the method of presenting information to the user, there are various methods for obtaining some knowledge from the information. First, a method of finding a representative value. There are systematic studies such as statistics. On the other hand, in recent years, there is a method for assisting a user in acquiring knowledge from information by visualizing and presenting information as it is in a graph or bitmap. In particular, the latter is often used by displaying a graph on a display device or the like. There are advantages and disadvantages regarding the shape of the graph, and they are used according to the purpose. A graph often used for displaying natural environment information in the agricultural field is a line graph, which matches the feature that the natural environment continuously changes. Many methods of presenting information to the user have been proposed, and for example, a line graph or a candle chart has been devised for time series information. Many interactive interfaces have also been devised due to the recent development of computers.

Japanese Patent No. 5554399 Japanese Patent No. 5365587

In addition to the problems described above, there are the following problems. For example, in the field of field monitoring in the field of agriculture, the measuring device is installed in an outdoor field and may easily break down or be destroyed. Therefore, the price of the measuring device needs to be suppressed as much as possible. An effective method for controlling the price of the measuring device is to reduce the price of the members constituting the measuring device. For example, the number of recording parts for temporarily recording observation results before the information is transmitted to the central control unit is reduced. Further, for example, the price of the independent power source can be reduced by reducing the quality of the capacity and electrical characteristics of the independent power source provided in the measuring apparatus. However, when these methods are selected, there is a problem that the measurement apparatus needs to transmit observation results to the central control apparatus more frequently.
Another issue is that it is difficult to improve the observation results. An information transmission path that connects the measuring device and the central control device in the remote natural information measuring device group is often provided by a third party for a fee. An example of such an information transmission path is a cellular phone communication network by a cellular phone operator. In most cases, the price for transmitting information is determined by the amount of information transmitted. High-definition observation results can be realized by observing at shorter time intervals. As a result, the amount of recorded observation results increases. As a result, in order to transmit the increased observation results, more information transmission is required. As measures against these problems, there is a reduction in recording amount using an information compression technique such as a Huffman code. However, even in such compression, the expected recording amount reduction effect may not be obtained depending on the structure of the observation result.

  Therefore, an object of the present invention is to provide a measurement apparatus, a measurement system, a measurement result collection method, and a program that can solve the above-described problems.

  The present invention has been made to solve the above-described problems, obtains observation results measured by a sensor, and generates information representing an approximate curve that is a curve that approximates a set of points indicated by the time-series observation results. And a communication unit that transmits information representing the approximate curve generated by the approximate curve generation unit.

  The present invention also includes a central control device including one or a plurality of the above-described measurement devices and an image generation unit that generates a graph of an approximate curve based on information representing the approximate curve received from the measurement device. It is a measurement system provided.

  Further, the present invention acquires observation results measured by the sensor, generates information representing an approximate curve that is a curve that approximates a set of points indicated by the time-series observation results, and generates information representing the generated approximate curve. This is a method for collecting measurement results to be transmitted.

  According to another aspect of the present invention, the computer of the measuring device acquires the observation result measured by the sensor, and generates information representing an approximate curve that is a curve that approximates a set of points indicated by the observation result in time series, the generation It is a program for functioning as means for transmitting information representing the approximated curve.

  ADVANTAGE OF THE INVENTION According to this invention, the communication amount of the observation result which a measuring apparatus transmits can be suppressed, and it can respond also to high definition of a measurement. In particular, with regard to telegrams during information communication, accurate and high-definition information can be provided to the user regarding some abnormal values while reducing data included in the telegram.

It is a figure which shows the minimum structure of the measuring apparatus by one Embodiment of this invention. 1 is a schematic block diagram of a remote natural information measurement system according to an embodiment of the present invention. 1 is a diagram illustrating a remote natural information measurement system according to an embodiment of the present invention. 1 is a block diagram of a remote natural information measurement system according to an embodiment of the present invention. It is a block diagram which shows an example of the measuring apparatus by one Embodiment of this invention. It is a figure which shows the example of mounting of the measuring apparatus by one Embodiment of this invention. It is a block diagram of the CU part with which the measuring device by one embodiment of the present invention is provided. It is a block diagram which shows an example of the central control apparatus by one Embodiment of this invention. It is a 1st figure explaining the thinning-out process of the observation result by one Embodiment of this invention. It is a 2nd figure explaining the thinning-out process of the observation result by one Embodiment of this invention. It is an example of the flowchart of the thinning-out process of the observation result by one Embodiment of this invention. It is a figure explaining the influence of the thinning-out process of the observation result by one Embodiment of this invention. It is a 1st figure explaining the term by one Embodiment of this invention. It is a 2nd figure explaining the term by one Embodiment of this invention. It is a 3rd figure explaining the term by one Embodiment of this invention. It is a 4th figure explaining the term by one Embodiment of this invention. It is a 5th figure explaining the term by one Embodiment of this invention. It is a 6th figure explaining the term by one Embodiment of this invention. It is a 7th figure explaining the term by one Embodiment of this invention. It is a 1st flowchart of the approximate curve calculation process by one Embodiment of this invention. It is a 2nd flowchart of the approximate curve calculation process by one Embodiment of this invention. It is a figure which shows an example of the screen image by one Embodiment of this invention. It is a figure explaining the information loss confirmation processing by one Embodiment of this invention. It is a flowchart of the confirmation process of the information defect by one Embodiment of this invention. It is a flowchart of the spike removal process by one Embodiment of this invention. It is a figure explaining the example of the difference of the spike and table by one Embodiment of this invention. It is a figure explaining division | segmentation before and behind the edge by one Embodiment of this invention. It is a flowchart of the division process of the table by one Embodiment of this invention. It is a 1st figure explaining the table discovery process by one Embodiment of this invention. It is a 2nd figure explaining the table discovery process by one Embodiment of this invention. It is a flowchart of the section approximation process by one Embodiment of this invention. 5 is a flowchart of pattern matching according to an embodiment of the present invention. It is a figure explaining the compression process in the pattern matching by one Embodiment of this invention. It is a figure explaining the discovery process of the approximated curve by one Embodiment of this invention. It is a 1st figure explaining the pattern file by one Embodiment of this invention. It is a 2nd figure explaining the pattern file by one Embodiment of this invention. It is a 1st figure explaining the kind of pattern file by one Embodiment of this invention. It is a 2nd figure explaining the kind of pattern file by one Embodiment of this invention. It is a 3rd figure explaining the kind of pattern file by one Embodiment of this invention. It is a 4th figure explaining the kind of pattern file by one Embodiment of this invention. It is a 1st figure explaining calculation of the feature-value of the point by one Embodiment of this invention. It is a 2nd figure explaining calculation of the feature-value of the point by one Embodiment of this invention. It is a 3rd figure explaining calculation of the feature-value of the point by one Embodiment of this invention. It is a 1st flowchart of the feature-value calculation of the point by one Embodiment of this invention. It is a 1st flowchart of the feature-value calculation of the point by one Embodiment of this invention. It is a first diagram illustrating a point probe according to an embodiment of the present invention. It is a 2nd figure explaining the probe of the point by one Embodiment of this invention. It is a flowchart of the process of the abnormal point sequence by one Embodiment of this invention. It is a 1st figure explaining the process of the abnormal point sequence by one Embodiment of this invention. It is a 2nd figure explaining the process of the abnormal point sequence by one Embodiment of this invention. It is a 3rd figure explaining the process of the abnormal point sequence by one Embodiment of this invention. It is a 4th figure explaining the process of the abnormal point sequence by one Embodiment of this invention. It is a figure which shows the example of a display of the observation result by one Embodiment of this invention.

<One Embodiment>
Hereinafter, a remote natural information measuring system according to an embodiment of the present invention will be described with reference to FIGS.
FIG. 1 is a diagram showing a minimum configuration of a measuring apparatus according to an embodiment of the present invention.
As shown in FIG. 1, the measuring apparatus 10 includes at least an approximate curve generation unit 104 and a communication unit 107.
The approximate curve generation unit 104 acquires observation results measured by the sensor over time, and generates information representing a curve that approximates a graph indicated by the time-series observation results.
The communication unit 107 transmits information representing the curve generated by the approximate curve generation unit 104.

FIG. 2 is a schematic block diagram of the remote natural information measurement system 1 according to an embodiment of the present invention.
As shown in FIG. 2, the natural information measuring system 1 includes a measuring device 10 and a central control device 20. In FIG. 2, only one measuring device 10 is shown, but normally, a plurality of measuring devices 10 are connected to the central control device 20 via a network.
As shown in FIG. 2, the measuring apparatus 10 includes a sensor 101, a sensor information acquisition unit 102, an immediate transmission determination unit 103, an approximate curve generation unit 104, an instruction information analysis unit 105, and an abnormal value transmission control unit 106. , A communication unit 107, a timer 108, and a storage unit 109.
The sensor 101 is one or more sensors that measure information about the environment around the measurement apparatus 10.
The sensor information acquisition unit 102 performs processing such as acquiring the observation result measured by the sensor 101 from the sensor 101 and converting it to a digital value.
The immediate transmission determination unit 103 determines whether the observation result acquired by the sensor information acquisition unit 102 needs to be immediately transmitted to the central control device 20.
The approximate curve generation unit 104 acquires the observation result acquired by the sensor information acquisition unit 102, and calculates a curve that approximates the graph information indicated by the time-series observation result. In addition, when there is a section including an abnormal value in the observation result, the approximate curve generation unit 104 calculates a curve that includes the abnormal value included in the section.
The instruction information analysis unit 105 analyzes the instruction information transmitted by the central controller 20. For example, the instruction information analysis unit 105 analyzes the instruction information transmitted by the central control device 20 and extracts information that requests transmission of an abnormal value.
When the instruction information analysis unit 105 extracts an abnormal value transmission request from the instruction information, the abnormal value transmission control unit 106 transmits the true observation result of the abnormal value in the section indicated in the transmission request to the central controller 20. Control to do.
The communication unit 107 transmits / receives information to / from the central control device 20. For example, the instruction information analysis unit 105 acquires instruction information from the central control device 20 via the communication unit 107. Further, the abnormal value transmission control unit 106 transmits the true observation result of the abnormal value to the central control device 20 via the communication unit 107. Further, the approximate curve generation unit 104 transmits information representing the approximate curve to the central control device 20 via the communication unit 107. In addition, the immediate transmission determination unit 103 transmits an urgent observation result to the central control device 20 via the communication unit 107.
The timer 108 measures time.
The storage unit 109 stores various information such as observation results acquired by the sensor information acquisition unit 102 and information representing the approximate curve generated by the approximate curve generation unit 104.

As shown in FIG. 2, the central control device 20 includes an input receiving unit 201, an acquisition information identification unit 202, an image generation unit 203, an instruction information generation unit 204, a storage unit 205, and a communication unit 206. ing.
The input receiving unit 201 receives input of instruction information from the user.
The acquisition information identification unit 202 acquires the information transmitted by the measurement device 10 via the communication unit 206. For example, whether the information is an observation result, information indicating an approximate curve, or a true observation result of an abnormal value And the identified information is stored in the storage unit 205 for each type.
The image generation unit 203 generates various images to be browsed by the user and displays them on a display device connected to the central control device 20.
The instruction information generation unit 204 generates a message including instruction information for the measurement apparatus 10 based on an instruction operation received from the user by the input reception unit 201.
The storage unit 205 stores information received from the measurement apparatus 10.
The communication unit 206 communicates with one or a plurality of measurement devices 10.
Hereinafter, a more detailed configuration and operation of the present embodiment will be described with reference to the drawings.

FIG. 3 is a diagram illustrating a remote natural information measurement system according to an embodiment of the present invention.
In this figure, the natural information measuring system 1 includes measuring devices 10A to 10F, a network 3, and a central control device 20. The measuring devices 10A to 10F are collectively referred to as the measuring device 10.
The measuring device 10 is a remote-type natural information measuring device including an approximate curve calculation processing component. The measuring apparatus 10 is installed in a farm field in agriculture. The plurality of measuring devices 10 are connected to the central control device 20 via the network 3. The network 3 is, for example, a mobile phone network provided by a mobile phone company. Further, the central control device 20 and the network 3 are connected via wireless communication or the Internet. The measuring device 10 is a device that transmits an observation result measured by a sensor. The measurement apparatus 10 includes a microcomputer and a measurement terminal connected to, for example, an electric resistance or a capacitor. The central controller 20 is a server terminal device using a general-purpose computer. The central controller 20 may be stored, for example, in a remote data center.

FIG. 4 is a block diagram of a remote natural information measurement system according to an embodiment of the present invention.
As shown in FIG. 4, the measurement apparatus 10 includes at least one measurement terminal 11 (11a, 11b, 11c) and an interface Pr12 (12a, 12b, 12c) for each measurement terminal. Pr12 is one or more sets of electric wires and connections, and is connected to an AD converter included in the measuring apparatus 10. Further, the measuring apparatus 10 is connected to the network 3 via the interface Sr31. In order to transmit the measured information to the central control device 20, the measurement device 10 holds an identifier of the central control device 20 expressed by, for example, a URL (Uniform Resource Locator). The central controller 20 is connected to the network 3 through an interface Sw32. Sw32 is, for example, a dedicated line laid in the data center. The central control device 20 provides data received from the measurement device 10 to the user through a GUI (graphical user interface). One form of provision to the user is a graph display on a site by a Web browser.

FIG. 5 is a block diagram showing an example of a measuring apparatus according to an embodiment of the present invention.
As shown in FIG. 5, the measuring apparatus 10 includes a measurement terminal 11 (11a, 11b, 11c), an interface Pr12 (12a, 12b, 12c), an AD_CONV unit 13 (13a, 13b, 13c), and an interface Di14 (14a, 14b, 14c), TIMING section 15, interface Tm151, CU section 16 (16a, 16b, 16c), interface Do161 (161a, 161b, 161c), MESSAGE_AGGREGATOR section 17, interface Am171a, RU section 18, interface Mc181, CP section 19, interface Cm191 is provided.

In addition, the measurement apparatus 10 periodically transmits the observation result input from the Pr 12 to the central control apparatus 20 via the SrDT 31, and transmits the abnormal value in the observation result received from the central control apparatus 20 via the SrCM 31. In response to the request, it has a function of transmitting the true value of the abnormal value in the observation result. The Sr31 described above includes an SrCM 31a that is an input interface to the measuring apparatus 10 and an SrDT 31b that is an interface for transmitting the observation result and the abnormal value in the true observation result to the central control apparatus 20.
The interface Pr12 is expressed as Pr12. The same applies to the other interfaces Di14, Tm151, Do161, Am171a, Mc181, Cm191, SrCM31, SrCM31a, and SrDT31b.

Each component of the measuring apparatus 10 will be described. The measurement terminals 11 and Pr12 are as described with reference to FIG.
The AD_CONV unit 13 is an analog-digital converter. The AD_CONV unit 13 acquires the electrical signal output from the measurement terminal 11 via Pr12 and converts it into a meaningful format as an observation result. For example, when the measurement terminal 11 is an electrical resistance functioning as a thermometer, the AD_CONV unit 13 replaces the acquired electrical resistance value with an absolute temperature. The AD_CONV unit 13 is a component in which hardware and software are mixed together. The Pr 12 that terminates the signal on the measurement terminal 11 side is a physical device. When the signal is transmitted to the CU unit 16 side, the converter 12 It has an aspect as an electronic device for obtaining information. The AD_CONV unit 13 is mounted, for example, as a one-chip chip set, and a driver of the chip set operates inside an embedded OS (operating system) that operates the microcomputer of the measurement apparatus 10, and is installed in software on the OS. An observation result from the measurement terminal 11 is provided. Many products are provided as integrated circuits constituting a chip set on which the AD_CONV unit 13 is mounted. Such an integrated circuit outputs a bit string, but since the bit string contains only a minimum amount of information, in order to make this significant observation data, the bit string information can be read into observation results, It is necessary to perform conversion and to add additional information such as observation success / failure. The AD_CONV unit 13 also performs these processes.
Di 14 is an interface from the AD_CONV unit 13 to the CU unit 16. Di14 is realized by operating software on the OS.

  The TIMING unit 15 is a timer. The TIMING unit 15 measures time, and outputs time information for periodic information transmission to the CU unit 16 via the Tm 151. When the measurement apparatus 10 is realized by a microcomputer board that operates on an advanced OS such as Linux (registered trademark), the OS often provides a function for obtaining time. The TIMING unit 15 is realized by such a function, for example. The CU unit 16 periodically transmits the observation result to the central control device 20 based on this time information. Tm 151 is an interface between the TIMING unit 15 and the CU unit 16. The Tm 151 may output simple time information to the CU 16 or may output instruction information instructing execution at the time when the CU unit 16 operates. As will be described later, in order to determine the transmission timing, the CU unit 16 always outputs the current time.

  The CU unit 16 controls to transmit the observation result converted by the AD_CONV unit 13 to the central control device 20. The CU unit 16 first selects information to be transmitted immediately from the observation results and performs transmission control to the central control device 20. Further, the CU unit 16 temporarily accumulates observation results in a storage unit (not shown) for a certain period, and then generates an approximate curve therefrom. Further, the CU unit 16 controls to store an abnormal value in the observation result that could not be approximated last in a storage unit (not shown) and to transmit it according to an instruction from the central control device 20. The CU unit 16 has a central function of the present embodiment. The CU unit 16 will be described in more detail later with reference to FIG. The CU unit 16 outputs the observation result to the MESSAGE_AGGREGATOR unit 17 via the Do 161. The Do 161 is an interface between the CU unit 16 and the MESSAGE_AGGREGATOR unit 17.

  The MESSAGE_AGGREGATOR unit 17 obtains observation results for transmission to the central control device 20 via the interface Do161, and generates a set of transmission messages together. Thereby, a plurality of observation results output by each CU unit 16 provided for each type of measurement terminal 11 and a plurality of observation results when the times of the observation results output from each measurement terminal 11 overlap are 1 Information can be transmitted efficiently because it can be combined into one message. The MESSAGE_AGGREGATOR unit 17 outputs a message generated via the interface Am 171 to the RU unit 18. When the MESSAGE_AGGREGATOR unit 17 acquires information to be transmitted from each CU unit 16 and aggregates the information into one electronic message, in addition to the information acquired via the Do 161, information indicating the identifier of the measurement device 10 and the type of the measurement terminal 11 The amount of information to be insufficient when a message does not reach the specified amount of information when adding, unifying the format of information to be stored in an aggregated message, converting a character string into a binary signal, or sending a message is predetermined. Processing that adjusts the amount of information in a message to a specified amount of information by adding a meaningless value, compression of the message, conversion to a transmission format, and the like are performed.

  The RU unit 18 is a component that transmits and receives data to and from the network 3. The RU unit 18 includes hardware for directly transmitting / receiving a message and a driver program for operating the hardware. The RU unit 18 is realized by a 3G communication module connected to a microcomputer board via a PCI-e bus. The RU unit 18 outputs the received information to the CP unit 19 via the interface Mc181.

  The CP unit 19 interprets the information received from the central controller 20 by the RU unit 18 and extracts abnormal value transmission request information. The CP unit 19 outputs abnormal value transmission request information to the CU unit 16 via the interface Cm191. The abnormal value transmission request information is instruction information for requesting transmission of a true measured value for an abnormal value in an observation result. The CP unit 19 instructs the CU unit 16 to transmit a true measurement value of an abnormal value in the observation result. The CP unit 19 does not output an instruction to the CU unit 16 immediately after acquiring information including an abnormal value transmission request, but determines the timing of outputting the instruction according to the state of the measuring apparatus 10. For example, the CP 19 monitors the remaining capacity of an independent power source (not shown) included in the measurement apparatus 10 and determines that there is not enough power left to transmit a message. Holds output of instructions until recovery. As another example, when the solar cell is provided as an independent power source (not shown), for example, the CP unit 19 determines whether the power generation amount can cover transmission power, whether the radio wave condition is suitable for transmission, transmission Whether the information amount of the scheduled message is within the size that can be transmitted, whether the remaining power amount of the independent power source (not shown) is sufficient for the continuous operation of the measuring device after the transmission process, whether the storage unit (not shown) Whether there is room in the free area or not. If it is determined that the transmission possible condition is satisfied, the CP unit 19 transmits the true measurement value of the abnormal value in the observation result to the CU unit 16 having the true observation result of the abnormal value in the corresponding observation result. Instruct.

FIG. 6 is a diagram illustrating an implementation example of a measurement apparatus according to an embodiment of the present invention.
As shown in FIG. 6, the measurement apparatus 10 includes an FPGA (field-programmable gate array) 30 and a 3G communication module 31 as an example. The FPGA 30 and the 3G communication module 31 are connected by a PCI-e bus 32. The 3G communication module 31 is connected to the antenna 34 via the coaxial cable 33. The 3G communication module 31 performs wireless communication with the network 3 using the antenna 34. The AD_CONV unit 13, Di14, TIMING unit 15, Tm151, CU unit 16, Do161, MESSAGE_AGGRREGATOR unit 17, Am171a, CP unit 19, and Cm191 described with reference to FIG. 4 are realized by the FPGA 30 operating. Further, the CP 19 unit and the Mc 181 are realized by the 3G communication module 31.

FIG. 7 is a block diagram of a CU unit included in the measurement apparatus according to the embodiment of the present invention.
As shown in the figure, the CU unit 16 includes a SELECTOR unit 600, a BUFFER unit 601, a CREATE_MESSAGE unit 602, an APF unit 603, an APF_POOL unit 604, a DET_LOG unit 605, a DETAIL_SENDER unit 606, an interface Dd611, an interface As 613, an interface As 613, and an interface As 613. Each component includes an interface As615, an interface Ap616, and an interface As617.
The interface Dd611 is expressed as Dd611. The same applies to the other interfaces Dd611, Di612, As613, Ap614, As615, Ap616, As617.

  The SELECTOR unit 600 acquires the observation result from Di14, and determines whether or not to immediately transmit the observation result to the central controller based on the time information acquired from Tm151 and the predetermined signal shape information. When it is determined to transmit, the SELECTOR unit 600 outputs the observation result to the CREATE_MESSAGE unit 602 via the interface Dd611. The predetermined signal shape information is information for determining the shape of a signal for identifying an emergency situation such as an abnormal rise in temperature. This is given by, for example, a threshold value or rate of change, and is recorded in a storage unit (not shown). The observation result sent to the CREATE_MESSAGE unit 602 via the Dd 611 is immediately converted into a message for transmission to the central controller 20 and output from the interface Do 161 to the MESSAGE_AGGREGATOR unit 17.

  The BUFFER unit 601 acquires observation results from the Di 14 and accumulates the information, and waits for processing to be executed by the APF unit 603. Since the APF unit 603 cannot receive information while the APF unit 603 is executing processing, the BUFFER unit 601 buffers the observation result information. In addition, since the process in the APF unit 603 does not always meet the scheduled time for transmitting the next information to the central controller 20, the BUFFER unit 601 buffers observation result information during that time. For example, when the interval of the scheduled time is not constant, even if the process of the APF unit 603 is not in time for the interval of the short time interval, the process is completed by the next scheduled time and information representing the approximate curve as the processing result is transmitted. There is a possibility. The BUFFER unit 601 temporarily accumulates observation results acquired from the Di 14 in preparation for such a case.

  The CREATE_MESSAGE unit 602 shapes a message when generating information from the CU unit 16 and generates a transmission message. When creating the transmission message, if the information representing the approximate curve can be transmitted at the same time, the CREATE_MESSAGE unit 602 reads the information representing the approximate curve from the APF_POOL unit 604 and includes the information representing the approximate curve in the transmission message.

  The APF unit 603 generates an approximate curve from the observation result with time. The APF unit 603 reads the observation results stored in the BUFFER unit 601 for a certain interval, and converts the observation results into an approximate curve. When the conversion is successful, the APF unit 603 deletes the observation result information corresponding to the approximate curve from the BUFFER unit 601, and records information representing the converted approximate curve in the APF_POOL unit 604. When the conversion fails, the APF unit 603 records the observation result of the section in which the conversion has failed as the abnormal value in the observation result in the DET_LOG unit 605, while the information on the observation result of the section in which the conversion has failed is the BUFFER unit 601. Delete from.

The APF_POOL unit 604 stores information representing the approximate curve generated by the APF unit 603.
The DET_LOG unit 605 stores the observation result (abnormal value) of the section in which the conversion of the approximate curve has failed.
The DETAIL_SENDER unit 606 acquires the abnormal value transmission request information from the central control device 20 via the Cm 191 and makes preparations for transmitting the true measured value of the abnormal value in the observation result to the central control device 20. The DETAIL_SENDER unit 606 reads the true measured value of the abnormal value in the observation result from the DET_LOG unit 605 via the As 615 and instructs the CREATE_MESSAGE unit 602 to transmit the true measured value of the abnormal value.

Dd611 is an interface between the SELECTOR unit 600 and the CREATE_MESSAGE unit 602. Di 612 is an interface between the BUFFER unit 601 and the APF unit 603. As 613 is an interface between the APF unit 603 and the DET_LOG unit 605. Ap 614 is an interface between the APF unit 603 and the APF_POOL unit 604. As 615 is an interface between the DET_LOG unit 605 and the DETAIL_SENDER unit 606. Ap 616 is an interface between the APF_POOL unit 604 and the CREATE_MESSAGE unit 602. As 617 is an interface between the DETAIL_SENDER unit 606 and the CREATE_MESSAGE unit 602.
Note that the BUFFER unit 601, the APF_POOL unit 604, and the DET_LOG unit 605 can be realized by a computer memory, a hard disk, or the like.

FIG. 8 is a block diagram illustrating an example of a central control device according to an embodiment of the present invention.
As shown in the figure, the central control unit 20 includes a MESSAGE_RECOGNIZER unit 21, an APF_POOL unit 22, a DET_LOG unit 23, a GG unit 24, a COMMAND_GENEATOR unit 25, an interface Ap211, an interface As212, an interface Ap213, an interface As214, an interface Mc215, and an interface Ui216. It has parts.

For the same reason as the measurement apparatus 10, the interface Sw32 includes an interface SwCM32a and an interface SwDT32b.
The interface Ap211 is referred to as Ap211. The same applies to the other interfaces As212, Ap213, As214, Mc215, Ui216, Sr32, SwCM32a, and SwDT32b.
The central controller 20 accumulates information such as the approximate curve received from the measuring apparatus 10 and displays it on the GUI. In addition, the central control device 20 collects the true observation results of the portion of the abnormal value range designated by the user via the GUI from the measurement device 10.

The MESSAGE_RECOGNIZER unit 21 acquires the observation result, the approximate curve, and the true measurement value of the abnormal value immediately transmitted via the SwDT 32b. The MESSAGE_RECOGNIZER unit 21 outputs information representing the approximate curve to the APF_POOL unit 22 via the Ap 211. The MESSAGE_RECOGNIZER unit 21 outputs the true measured value of the abnormal value to the DET_LOG unit 23 via the As 212.
The APF_POOL unit 22 stores information representing an approximate curve.
The DET_LOG unit 23 stores the true observation result of the abnormal value in the section in which the APF unit 603 failed to convert the approximate curve.

The GG unit 24 reads the information representing the approximate curve stored in the APF_POOL unit 22 and the true measured value of the abnormal value stored in the DET_LOG unit 23, generates a graph or the like using the information, and sends the information via the Ui 216. Display with GUI. Initially, the graph shows only the range of outliers in the observation results. The user designates an arbitrary range on the GUI. Then, the GG unit 24 requests a true measurement value to the COMMAND_GENEATOR unit 25 via the Mc 215.
The COMMAND_GENEATOR unit 25 transmits the transmission request information of the true measurement value of the abnormal value to the measurement device 10 specified by the user via the GG unit 24 via the SwCM 32a. Since it is necessary to designate the measuring device 10, the central control device 20 stores the identifier of each measuring device 10. The identifier of the measuring device 10 is, for example, a URL list.
Ap 211 is an interface between the MESSAGE_RECOGNIZer unit 21 and the APF_POOL unit 22. As 212 is an interface between the MESSAGE_RECOGNIZer unit 21 and the DET_LOG unit 23. Ap 213 is an interface between the APF_POOL unit 22 and the GG unit 24. As 214 is an interface between the DET_LOG unit 23 and the GG unit 24. Mc 215 is an interface between the GG unit 24 and the COMMAND_GENEATOR unit 25.

Next, the operation of the natural information measuring system 1 of the present embodiment will be described. As described above, the CU unit 16 has a central function of the present embodiment. Hereinafter, the operation of the CU unit 16 will be described in more detail.
The CU unit 16 is started at two different timings. The first timing is when the CU unit 16 acquires the observation result collected from Di14. The second timing is when the CU unit 16 acquires transmission request information of the true observation result of the abnormal value in the observation result from Cm191. When the information is acquired from the Di 14 and started (first timing), the CU unit 16 performs the immediate observation result (the observation result that should be transmitted immediately) and the untransmitted calculation that has been completed up to the present time. The approximate curve for the minute is output from the interface Do161. When the information is acquired from the Cm 191 and started (second timing), the CU unit 16 includes the true measurement value of the abnormal value in the observation result accumulated in the internal storage unit (DET_LOG unit 605). The observation results in the specified range are output from Do 161.

In the present embodiment, it is assumed that the natural information measurement interval in the measurement device 10 is much shorter than the interval at which the immediate observation result is transmitted to the central control device 20. For example, the observation results collected once per second are transmitted once per hour, and the observation results compressed into an approximate curve are transmitted. Therefore, every time the APF unit 603 is started at the first timing, the APF unit 603 obtains time information from the interface Tm 151 and determines whether or not those observation results or compressed observation results should be transmitted to the central controller 20. ing. In the case of an observation result that should not be transmitted or a compressed observation result, message transmission from the Do 161 is not performed, and a recording component (APF_POOL unit 604) prepared for an untransmitted observation result inside the APF unit 603 is used. store.
In addition, when the APF unit 603 matches a predetermined condition, the unprocessed observation result stored in the internal BUFFER unit 601, the untransmitted compressed observation result, and the true value of the abnormal value in the observation result Delete observation results. One such condition is the elimination of observation results when approximate curve calculation processing is not possible in time. This is performed when the situation in which the compression of the observation result by the internal approximate curve calculation is not in time until the next message transmission occurs continuously. As a result, the approximate curve calculation process is not performed on the observation result to be erased, but the approximate curve calculation process of the observation results collected after that is tried again. Processing in this case will be described with reference to FIG.

FIG. 9 is a first diagram illustrating the thinning-out process of observation results according to an embodiment of the present invention.
FIG. 9 shows an operation example when the approximate curve calculation takes time and the approximate curve calculation timing is delayed with respect to the observation result acquisition timing. Time T1 is a timing at which the observation result observed in the observation result section 1 is transmitted to the central controller 20. Time T2 is a timing at which the observation result observed in the observation result section 2 is transmitted to the central controller 20. The same applies to times T3, T4, T5, T6, and T7. That is, at time T1 to T7, the CU unit 16 outputs not only the observation result but also information representing the approximate curve if there is information representing the approximate curve that can be transmitted at that time. In the case of the example in this figure, the approximate curve calculation processing is performed on the observation result observed in the section two before the currently observed section. For example, the CU unit 16 starts the approximate calculation of the observation result section 1 simultaneously with the start of the observation result section 3. If the calculation of the approximate curve for the observation result section 1 is completed within the observation result section 3, the CU unit 16 can start the approximate calculation for the next observation result section 2 at time T3. Further, the CU unit 16 outputs the calculation result of the approximate curve for the observation result section 1 to the MESSAGE_AGGREGATOR unit 17 at time T3, and transmits the information representing the observation result and the approximate curve to the central controller 20 via the RU unit 18 and the like. can do. However, if the time required for calculating the approximate curve for the observation result section 1 (approximate curve calculation section 1) is longer than the observation result section 3, as shown in the figure, the start of the calculation of the approximate curve for the observation result section 2 is delayed. Become. Also, information representing an approximate curve for the observation result section 1 cannot be transmitted at time T3. In such a case, the CU unit 16 deletes the information in the BUFFER unit 601.

  As a case where the processing tends to be delayed in this way, for example, there is a problem in the design of the information processing capability of the measuring apparatus 10, and even if the transmission timing is adjusted, the approximate curve calculation processing is not in time, and the transmission timing continues. Adjustment processing may be required. In order to deal with such a case, in the present embodiment, before resuming the approximate curve calculation process after adjusting the transmission timing, the observation result is thinned to reduce the overall calculation amount. In this case, the observation results stored in the BUFFER unit 601 are evenly thinned using the time interval as an index. For example, when observation results are collected at intervals of one second and all the observation results are objects of approximate curve calculation processing, the calculation processing is not in time. In such a case, in the present embodiment, for example, after that, thinning is performed so that only the observation results at intervals of 5 seconds are targeted for calculation. In order to execute this thinning out at high speed, when the BUFFER unit 601 is configured with a computer memory, a continuous memory space is used when the observation result is stored in the memory, or a ring buffer is produced. . An example of the ring buffer is shown in FIG.

FIG. 10 is a second diagram illustrating the thinning-out process of observation results according to an embodiment of the present invention.
FIG. 10 shows an example of a ring buffer having a special shape used in this embodiment.
In the example of FIG. 10, twelve ring buffer areas B1 to B12 are provided in the storage area. In each of the ring buffer areas B1 to B12, in addition to a normal ring pointer, a series pointer used for thinning is provided. A normal ring-shaped pointer extends in the vertical direction in the figure and from the left column to the right column. As a result, during normal processing, data is processed in the order of the ring buffer areas B1, B2, B3, B4, B5, B6, B7, B8, B9, B10, B11, and B12. On the other hand, the pointer used at the time of thinning is extended for each series, and is processed in the direction in which the series in which the observation result currently processed exists exists. For example, the pointer corresponding to the series 1 is set to be B5 for the ring buffer area B1 and B9 for the area B5. As a result, if the current processing is the ring buffer area B1, the data processing is performed in the order of the ring buffer area B1 → B5 → B9 during the thinning process.

FIG. 11 is an example of a flowchart of the thinning process of observation results according to an embodiment of the present invention.
A determination process for determining whether or not to perform the thinning process will be described with reference to FIG.
As a premise, a decimation process flag indicating whether or not to perform a decimation process, a success number flag that stores the number of times the approximate curve calculation process has been completed at the transmission time, and a number of times that the approximate curve calculation process has not been completed at the transmission time It is assumed that a failure number flag to be stored is provided, the initial value of the thinning-out processing flag is False, and the initial value of the success number flag and the failure number flag is 0.
First, the APF unit 603 acquires an observation result (step S1). Next, the APF unit 603 determines whether or not to perform the thinning process based on the thinning process flag (step S2). If the value of the thinning process flag is False, the APF unit 603 determines that the normal process is performed without performing the thinning process. If the value of the thinning process flag is True, the APF unit 603 determines to perform the thinning process. When it is determined that the normal process is performed, the APF unit 603 performs the normal process (step S3). That is, the APF unit 603 calculates approximate curves for all observation results. When it is determined that the thinning process is performed, the APF unit 603 performs the thinning process (step S4). For example, the APF unit 603 performs approximate curve calculation for only the data stored in a certain series of ring buffers in the ring buffer described with reference to FIG. Next, the APF unit 603 determines whether or not the approximate curve calculation process is successful (step S5). When the approximate curve calculation process is not delayed and the approximate curve calculation result can be transmitted, the APF unit 603 determines that the approximate curve calculation process is successful. When the approximate curve calculation process is delayed and the approximate curve calculation result cannot be transmitted, the APF unit 603 determines that the approximate curve calculation process has failed.

  When it is determined that the approximate curve calculation is successful, the APF unit 603 determines whether or not the value of the failure number flag is greater than 0 (step S6). If the value of the failure count flag is 0, the processing from step S1 is repeated. On the other hand, when the value of the failure number flag is greater than 0, the APF unit 603 increments the value of the success number flag by 1 (step S7). Subsequently, the APF unit 603 compares the value of the success number flag with a predetermined threshold value (step S8). If the value of the success number flag is smaller than the predetermined threshold value, the processing from step S1 is repeated. If the value of the success number flag is equal to or greater than the predetermined threshold, the APF unit 603 sets False to the value of the thinning process flag (step S9), and repeats the process from step S1.

  When it is determined that the approximate curve calculation has failed, the CU unit 16 increments the value of the failure number flag by 1 (step S10). Subsequently, the CU unit 16 compares the value of the failure count flag with a predetermined threshold value (step S11). If the value of the failure number flag is smaller than the predetermined threshold value, the processing from step S1 is repeated. If the value of the failure count flag is equal to or greater than the predetermined threshold, the CU unit 16 sets True to the value of the thinning process flag (step S12), and repeats the processing from step S1.

Next, operation of internal functions provided in the CU unit 16 will be described.
The SELECTOR unit 600 determines whether or not the observation result acquired from the Di 14 is an immediate observation result that is immediately transmitted to the central controller 20. Immediate observation results are observation results that require immediate transmission. For example, the SELECTOR unit 600 compares the observation result with a predetermined threshold (such as temperature), and determines that the observation result is an immediate observation result, for example, if the observation result exceeds the threshold. The SELECTOR unit 600 compares the time when the acquired observation result is observed with the time information acquired from the Tm 151. Then, if the current time acquired from Tm 151 has passed a predetermined time or more from the observation time, the SELECTOR unit 600 determines that the result is an immediate observation result.
If the SELECTOR unit 600 determines that the result is an immediate observation result, the SELECTOR unit 600 outputs information on the observation result to the CREATE_MESSAGE unit 602 via the Dd 611. The CREATE_MESSAGE unit 602 generates a message for transmitting the observation result, and the MESSAGE_AGGREGATOR unit 17 transmits the message together with other messages to the central control device 20 via the RU unit 18. Thereby, the information of the observation result determined as the immediate observation result is immediately transmitted to the central control device 20.
As one of the implementation forms of the SELECTOR unit 600, a form of reading external information in which a threshold value that the SELECTOR unit 600 determines to be emergency information at the time of activation is read. With this implementation, if the external information is realized by a file on a computer, the contents are rewritten, and the SELECTOR unit 600 is restarted, another threshold can be used even after the measurement apparatus 10 is constructed. It becomes.

  The BUFFER unit 601 has a data structure called a queue that stores observation results measured by the measurement terminal 11 acquired from the Di 14 as they are in chronological order. The BUFFER unit 601 stores these observation results until the approximate curve calculation processing in the APF unit 603 is completed. As described above, the observation result acquired by the APF unit 603 and succeeded in the approximate curve calculation process is deleted from the BUFFER unit 601. The BUFFER unit 601 has the structure illustrated in FIG. 10, for example. If the amount of information stored in the BUFFER unit 601 exceeds the storage capacity, the observation results of other sequences are discarded, leaving only one series of data. May be. As a result, observation results in chronological order can be thinned out and stored, and information loss due to excess storage capacity can be minimized. The influence of the thinning process in which the observation result stored in the BUFFER unit 601 is limited to a certain series will be described with reference to FIG.

FIG. 12 is a diagram for explaining the influence of thinning processing of observation results according to an embodiment of the present invention.
In FIG. 12, the values of the observation results stored in the buffers of all series (series 1 to 4) are reflected in the approximate curve until time Tα. After time Tα, only the observation result of series 1 is reflected in the approximate curve. It can be seen that the outline of the approximate curve can be maintained to some extent even if only a part of the data of the series is used if the series is selected well so that the values of the observation results are dispersed.

  The APF unit 603 generates an approximate curve. The APF unit 603 operates using the time information acquired from the Tm 151 as a starting point. It is assumed that the operation start timing is determined in advance. For example, the operation start timing is designated as 0.5 second per second. This operation start timing is just the timing at which the APF unit 603 starts to operate, and may be different from the timing at which the result is output from the Ap 614 or the As 613. Before entering into a detailed description of the operation of the APF unit 603, terms necessary for describing the process of processing the observation result will be described.

  “Point”: A point corresponds to one observation result. The information held by the point has information that is added for the internal processing of this device group, the observation time, the observation result, and the observation result.

  “Point cloud”: A point cloud is a set of one or more points. Unless there is any abnormal situation, there are no observation results measured at the same time, so all points belonging to the point cloud can be uniquely identified. Therefore, a unique point number can be given to a point belonging to the point group for identification.

  “Point sequence”: A point sequence is a sequence of points based on arbitrary information held by the points. Since the points are unique, there is an order relationship between all the points belonging to the point sequence.

  “Section”: In the present embodiment, any continuous part of a sequence of points arranged in particular by time is called a section. If the order relation of points is taken into consideration, only the start point and end point need be specified in order to represent a section.

“Frame”: A sufficiently long point sequence (referred to as an observation point sequence) having observation results for a certain period as a point. It is a partial section of the observation point sequence and is subject to calculation during one approximate curve calculation process. This is a sequence of points. There may be a plurality of frames included in a certain observation point sequence, but the elements do not overlap each other. That is, the observation point sequence can be composed of a series of adjacent frames. That is, the observation result is processed by a plurality of approximate curve calculation processes, but in this case, adjacent frames are always processed in order. An example of the order of processing through frames will be described with reference to FIG.
FIG. 13 is a first diagram illustrating terms according to an embodiment of the present invention.
Frames K1 to K6 are shown in a series of sections in FIG. The APF unit 603 calculates approximate curves in the order of the frame K1 and the frame K2 in order from the left. As shown in the figure, the immediate observation result is immediately transmitted to the central controller 20, and the approximate curve calculation is transmitted to the central controller 20 with a delay of two frames.

“Target frame”: A target frame refers to a frame that is currently the target of approximate curve calculation.
“Peripheral factor”: Although the details of the approximate curve calculation process will be described later, in addition to the target frame, the surrounding points are required for the approximate curve calculation process. These points are used only for intermediate processing in the approximate curve calculation. In this embodiment, a set of these points is called a peripheral factor. Peripheral factors exist separately on the left and right sides of the target frame. The relationship between the peripheral factor and the target frame will be described with reference to FIG.

FIG. 14 is a second diagram illustrating terms according to an embodiment of the present invention.
In the figure, the point sequence in the section described as “target frame” is the target frame. In the drawing, the point sequence excluding the target frame in the section described as “target frame + peripheral factor” is the peripheral factor. As shown in the figure, the peripheral factors exist on the left and right of the target frame, respectively. The linear calculation range in the figure will be described later.

  “Approximate curve”: Approximate curve best represents the tendency of the arrangement of points in the target frame, that is, the shape of a point mapped to a two-dimensional space when the time and observation results are considered as basis vectors. It is a line segment represented by a relatively low-dimensional function. The best expression refers to a line segment expressed in as low a dimension as possible so that the approximate curve passes as close as possible to as many points in the target frame as possible. An example of generating an approximate curve is shown in FIG.

FIG. 15 is a third diagram illustrating terms according to an embodiment of the present invention.
FIG. 15A shows a target frame. A graph Val in FIG. 15B shows an approximate curve of the target frame calculated by the APF unit 603. FIG. 15C is an example of information representing an approximate curve transmitted from the measurement apparatus 10 to the central control apparatus 20.
The information representing the approximate curve transmitted to the central controller 20 includes a function indicating the approximate curve, information defining a time range in which the approximate curve is effective, and the like.

  “Angle change rate”: The angle change rate is the value of the second derivative at a certain point when the point group of the original data can be viewed as an infinitely fine continuous line segment. Actually, the point group is discrete, and here, the angle change rate of a certain point is defined using the difference between the front and rear.

FIG. 16 is a fourth diagram illustrating terms according to an embodiment of the present invention.
FIG. 16 is a diagram for explaining the angle of the graph at a certain point including the time and the observation result, and the angle change rate at the certain point. The angle Deg _t of the graph at the point corresponding to time t can be expressed by the difference between the observation result Val _t at time t and the observation result Val _t−1 at time t−1. Further, the angle change rate Diff _{t at} the point corresponding to the time t can be expressed by the difference between the graph angle Deg _t and the graph angle Deg _t−1 . The angle change rate Diff _t can be expressed by the following equation.
Diff _t = Deg _t -Deg _t-1 =
(Val _t -Val _t-1 )-(Val _t-1 -Val _t-2 )

“Feature point”: A feature point refers to a point having a great influence on the shape of an approximate curve. There may be a plurality of feature points in the target frame. A target frame approximated to a quadratic function and characteristic points therein will be described with reference to FIG.
“Point feature value”: The point feature value is a numerical value of the influence exerted by a point that determines the shape of the approximate curve. In this embodiment, the feature amount of a point is obtained through calculation of an approximate line.

FIG. 17 is a fifth diagram illustrating terms according to an embodiment of the present invention.
The feature points and the feature amounts of the points will be described with reference to FIG.
FIG. 17A is a graph showing a target frame (points 1 to 10) approximated by a quadratic function and an approximate curve thereof. FIG. 17B is a diagram illustrating feature amounts of points 1 to 10. As shown in the figure, among the feature quantities of points 1 to 10, points 1 to 6 and 10 exceed the threshold value V1. However, since the points 7 to 9 can be interpolated by the points 6 and 10, they do not have a great influence on determining the shape of the approximate curve, and the feature amount is low. For example, a point where the feature amount exceeds the threshold value V1 can be set as the feature point. The calculation of the point feature amount will be described later.

“Approximate line”: A line necessary for calculating the feature amount of each point of the target frame. This is temporary information used only for internal processing of the apparatus.
“Straight line calculation range”: a range of time required for calculating an approximate straight line at a certain point. Points included in this time width are used in the calculation. The straight line calculation range is shown in FIG. In FIG. 13, the sections described as “the straight line calculation range at the left end” and “the straight line calculation range at the right end” are the straight line calculation ranges. For example, the straight line calculation range at the left end is a time width necessary for calculating an approximate straight line for the left end point in the target frame.

“Edge”: An edge is a point where the value of an observation result changes abruptly in a given point sequence as compared to either the previous or next point. The change in value ranges from a high value to a low value and vice versa. This is called the edge direction. A high value to a low value is called a right edge, and a low value to a high value is called a left edge.
“Table”: A table is a partial point sequence in a point sequence surrounded by two edges having different directions. Such a partial point sequence protrudes like a table when the observation result is visualized. A table starting from the left edge and ending at the right edge is called an upward convex table, and the opposite is called a downward convex table.
“Spike”: A spike is a table with very few points. Sometimes it consists of a single point.

FIG. 18 is a sixth diagram illustrating terms according to an embodiment of the present invention.
Edges, tables, and spikes will be described with reference to FIG.
In FIG. 18, a point P1 is a spike. Points P2 and P3 are edges. The point P2 is a left edge because it changes from a low value to a high value. P3 is a right edge because it changes from a high value to a low value. Further, the partial point sequence surrounded by the points P2 and P3 is a table. In this example, the table is convex upward because it starts at the left edge and ends at the right edge.

“Pattern file”: A pattern file is a recording of representative arrangements among arrangements of feature points.
“Ui graph”: A Ui graph is a shape of a point mapped to a two-dimensional space when two of time and observation result are considered as basis vectors. In the information display component of this device group, for example, it is expressed in a form close to the waveform displayed on the display device.

FIG. 19 is a seventh diagram illustrating terms according to an embodiment of the present invention.
FIG. 19A shows a point sequence of observation results. FIG. 19B shows a Ui graph corresponding to the point sequence of FIG. The user can know the transition of the observation result by looking at the Ui graph. The display surrounded by the dotted line in the Ui graph illustrated in FIG. 19B is a highlighted display indicating that the observation result in the range includes an abnormal value. By this highlighting, the user can grasp a section in which an abnormal value is included. The upper and lower approximate curves of the convex hull are displayed on the Ui graph surrounded by the dotted line. The upper and lower convex hull approximate curves indicate that there are abnormal values in the range enclosed by the upper and lower convex hull approximate curves. As described above, the measuring apparatus 10 transmits the true observation result of the abnormal value in the observation result to the central control apparatus 20 together with the information representing the approximate curve. For example, when the user performs an operation such as clicking on the position surrounded by the dotted line, the central control device 20 displays the true value of the abnormal value in the observation result received from the measurement device 10. The user can grasp the actual observation result of the section including the abnormal value.

“Probe”: A concept related to a calculation method used when calculating a feature amount of a point. The point sequence is scanned, and the feature amount is finally determined. The calculation of the point feature value and the probe will be described in detail later.
The above is an explanation of terms.

Next, the flow of the approximate curve compression calculation process of the APF unit 603 will be described with reference to FIGS.
FIG. 20 is a first flowchart of approximate curve calculation processing according to an embodiment of the present invention. FIG. 21 is a second flowchart of approximate curve calculation processing according to an embodiment of the present invention.
The APF unit 603 is activated, for example, at a predetermined time interval based on the time information acquired from the Tm 151 (step S20). First, the APF unit 603 reads the observation result information from the BUFFER unit 601 via the Di 612, reads the target frame that is the section to be processed (step S21), and reads the peripheral factor (step S22). Since it is necessary for the approximate curve calculation process, the APF unit 603 reads information before and after (peripheral factor) in addition to the information (target frame) to be targeted this time. The peripheral factor is used to calculate an approximate straight line between points at both ends of the target frame. The APF unit 603 reads the observation result information of a predetermined time interval from the observation result information as a target frame, and reads the observation result information of the predetermined time interval centered on the left and right ends of the target frame, respectively. Factor. Next, the APF unit 603 confirms information loss (step S23). The information deficiency confirmation is a process for confirming whether the target frame and the peripheral factor include information necessary for the approximate curve calculation process. The “information deficiency confirmation” process will be described in detail later with reference to FIG.

Next, the APF unit 603 performs spike removal for removing spikes from the target frame and the peripheral factor after the data loss confirmation process (step S24). The “spike removal” process will be described in detail later with reference to FIG. In addition, the APF unit 603 selects extreme points for the target frame after confirming the data loss (step S25). An extreme point is, for example, a point or an inflection point where the value is maximum or minimum. The APF unit 603 selects extreme points based on thresholds determined for these points.
Next, the APF unit 603 performs edge discovery for the target frame and the peripheral factor after spike removal (step S26), and extracts an edge point group. The “edge discovery” process will be described in detail later with reference to FIG. Next, the APF unit 603 determines whether or not the number of elements included in the extracted edge point group is within the number that can be executed in the “section approximation process” described later (step S27). For example, the APF unit 603 compares the number of elements with a threshold (third threshold), and if the number of elements is equal to or greater than the threshold, the number of elements does not fall within the number that can be used to execute the “section approximation process”. judge. If the APF unit 603 determines that the number of elements is within the number that can be used to execute the “section approximation process”, the APF unit 603 performs table division on the target frame and peripheral factor after spike removal (step S28). The “table division” process will be described in detail later with reference to FIG. Next, the APF unit 603 performs section approximation processing on the target frame and peripheral factor after table division (step S29). The section approximation process is a process for calculating an approximate curve for a point sequence included in a section obtained by dividing the target frame. The “section approximation process” will be described in detail later with reference to FIG.
On the other hand, if it is determined in step S27 that the number of elements does not fall within the number that can be executed for the “section approximation process”, the APF unit 603 performs the processing of the abnormal point sequence for the target frame and peripheral factor after spike removal. This is performed (step S291). The APF unit 603 obtains information obtained by combining two approximate curves for point groups existing on both sides of the abnormal section by processing of the abnormal point sequence, summary information of the abnormal range indicating the existence range of abnormal values, and information of the abnormal point sequence. Generate. The “abnormal point sequence processing” will be described in detail later with reference to FIG.

  Next, the APF unit 603 sets the approximate curve set information generated in step S29 or the approximate curve set generated in step S291 and the summary information of the abnormal range in a predetermined Ap format to generate an Apl file. (Step S292). In addition, the APF unit 603 sets the information on the extreme points selected in step S25 to the generated ApI information. The APF unit 603 outputs the generated ApI information to the APF_POOL unit 604 via the Ap 614. In addition, the APF unit 603 outputs abnormal point sequence information to the DET_LOG unit 605 via As 613. The above is a rough flow of the approximate curve calculation process. Returning to the description of the processing of each component.

  The output result from the APF unit 603 is output from two types of interfaces, Ap 614 (approximation curve or data indicating the existence range of an abnormal value) and As 613 (abnormal value in the observation result of that part when approximation fails). The Information to Ap 614 is always output, and to As 613 is output only when the approximate curve calculation process has failed. The output message of Ap 614 is referred to as Ap1 information in this embodiment. In this embodiment, the description format of Ap1 information is called an Ap format. When information is transmitted to the central control device 20 and stored in the central control device 20, the information is stored in the Ap format.

  The APF_POOL unit 604 accumulates the ApI information output from the APF unit 603 in a queue type. When the CREATE_MESSAGE unit 602 creates a transmission message based on an instruction from the SELECTOR unit 600, the APF_POOL unit 604 is confirmed, and if necessary, the Apl information stored in the APF_POOL unit 604 is added to the message of the immediate observation result. Send. The CREATE_MESSAGE unit 602 determines at what point the ApI information is transmitted. Accumulation of observation results in the APF_POOL unit 604 continues as long as measurement is performed.

  The DET_LOG unit 605 stores the true value of the abnormal value in the observation result determined as the abnormal point sequence. When these values are recorded in the DET_LOG unit 605, the time when the abnormal value in the observation result occurs and the time when it ends are added as index values for search. The DETAIL_SENDER unit 606 compares the specified time with the added index value when a request for the true value of the abnormal value in the observation result is received from the central controller 20, and the target observation result You can get the true value of the outlier. The DET_LOG unit 605 has a function of automatically deleting the true value of the abnormal value in the observation result judged to be relatively unnecessary when the free storage capacity of the DET_LOG unit 605 is smaller than a predetermined threshold value. As a condition for the determination, a method of giving priority to discarding a statistical index of an occurrence value of an observation result or a statistical index of an abnormal value in the observation result, for example, one having a low degree of dispersion may be considered.

  The CREATE_MESSAGE unit 602 generates a transmission message for transmitting the immediate observation result, approximate curve information, and summary information of abnormal values in the observation result to the central control device 20 via the Do 161. The CREATE_MESSAGE unit 602 operates at the time of the information transmission request of Dd611 or As617. The CREATE_MESSAGE unit 602 transmits three types of different information: an immediate observation result, approximate curve information included in ApI information, and a true observation result of an abnormal value in the observation result. Therefore, the CREATE_MESSAGE unit 602 converts these pieces of information into the same message format, forms the message, and adds the type of measurement terminal and the identifier of the measurement terminal in association with the observation result obtained from each measurement terminal 11. Output from Do161. The CREATE_MESSAGE unit 602 refers to the APF_POOL unit 604 at each time point to read and transmit the maximum amount of information that can be transmitted. This is a control method performed when there is very little information to be transmitted and it does not cause a communication load. In particular, when sending the immediate observation result, ApI information is added as much as possible and transmitted. As a result, it is possible to transmit the ApI information stored in as many APF_POOL units 604 as possible to the central control device 20 while minimizing the amount of information communication with the central control device 20. Also, the CREATE_MESSAGE unit 602 has a function of generating a message specifically for ApI information and transmitting it to the central controller 20 when it is determined that the amount of information stored in the APF_POOL unit 604 increases and the storage capacity is insufficient. doing.

  The DETAIL_SENDER unit 606 receives a request from the central control device 20 via the Cm 191. The DETAIL_SENDER unit 606 does not directly reply to this request but outputs a transmission instruction via the interface As 617 and causes the CREATE_MESSAGE unit 602 to reply. Further, the DETAIL_SENDER unit 606 acquires the true value of the abnormal value based on information for designating the section of the abnormal value in the observation result that the user wants to browse and the unique identifier of the request via the As 615. As an example of information that can be used for the unique identifier, the occurrence time of an abnormal value section in the observation result can be considered. At the same time, the type of the measurement terminal 11 can also be a unique identifier. In this case, since the time of the abnormal value in the observation result is different for each measurement terminal 11, only the information of the specific measurement terminal 11 may be transmitted in order to minimize the amount of information returned to the central controller 20. it can. In some cases, the true value of the abnormal value in the corresponding observation result may be already discarded by the DET_LOG unit 605. In this case, the DETAIL_SENDER unit 606 outputs information indicating that the information cannot be collected to the CREATE_MESSAGE unit 602. The DETAIL_SENDER unit 606 also performs processing for converting the true value of the abnormal value in the observation result into a format that can be transmitted by the CREATE_MESSAGE unit 602. In addition, the central controller 20 transmits information specifying the resolution of the true value of the abnormal value in the observation result. The resolution is the ratio of how many of the abnormal values of a certain observation result are transmitted when there is a set of true values. If the resolution is high, the central controller 20 can receive more accurate information. This is an example of a method for reducing the traffic. The DETAIL_SENDER unit 606 outputs information obtained by thinning out information according to the resolution. The method of thinning out information may be thinned out simply by leaving a time interval, or statistical characteristics of observation results may be calculated and only information with a large amount of features may be transmitted. For the feature amount calculation, the same processing as that used when calculating the approximate curve of the APF unit 603 (the processing in FIGS. 44 to 45 described later) can be used.

  Through the above processing, the CU unit 16 transmits the ApI information including the immediate observation result and the approximate curve information and the true value of the abnormal value in the observation result to the central control device 20. Next, the operation of each component of the central controller 20 that has received these pieces of information will be described.

  The MESSAGE_RECOGNIZer unit 21 receives information from each measuring device 10 transmitted via the SwDT 32b, analyzes it, divides it into immediate transmission information or an approximate curve or a true observation result of an abnormal value in the observation result, The divided information is accumulated in a predetermined storage unit such as the APF_POOL unit 22 or the DET_LOG unit 23. The MESSAGE_RECOGNIZer unit 21 stores the immediate transmission information and the approximate curve information in the APF_POOL unit 22 via Ap211 and the true observation result of the abnormal value in the observation result in the DET_LOG unit 23 via As212. In order for the GG unit 24 to visualize and display information from a plurality of measurement devices 10 side by side, each information is given an identifier of the measurement device 10. Due to the fact that the central controller 20 is mounted on a general-purpose computer, the physical termination processing of the information transmission path, which is the basic function of the MESSAGE_RECOGNIZER unit 21, consensus formation for information transmission / reception, code error correction, malice A countermeasure against an unauthorized attack on the present device group by a third party is assumed to use the functions of a lower-level general-purpose computer and the OS thereon. Such functions can use general ones. In addition, a function necessary for the MESSAGE_RECOGNIZer unit 21 includes authentication when receiving information from each measurement device 10. For example, the normality of the measuring device 10 may be checked in preparation for the case where the measuring device 10 installed outdoors is destroyed by a malicious third party and illegally modified. In this case, a mechanism for performing normality confirmation processing with the RU unit 18 of the measurement apparatus 10 is required. A general technique can also be used for the mechanism for checking the normality.

The APF_POOL unit 22 stores information in the Ap format similar to the measurement apparatus 10. However, the APF_POOL unit 22 is provided with a storage area for every observation device connected to the central control device 20. A user who wants to browse information browses information stored in the APF_POOL section 604 of the CU section 16 of the predetermined measuring apparatus by designating the measuring apparatus 10 and its measuring terminal 11 via the GG section 24. be able to.
The information storage format of the DET_LOG unit 23 is the same as that of the measuring apparatus 10. Also in the DET_LOG unit 23, a storage area is provided for each measuring apparatus 10.

The GG unit 24 visualizes and provides observation results to a user who wants to browse information. An example of mounting the GG unit 24 is shown in FIG.
FIG. 22 is a diagram showing an example of a screen image according to an embodiment of the present invention.
FIG. 22 shows an example in which information is visualized and displayed on a Web browser in chronological order for a user who wants to browse information as an example of this implementation. In FIG. 21, an image 121 is an image displayed on the screen of a display device connected to the central control device 20. A browser 122 is displayed in the image 121, and among the plurality of measurement devices 10 connected to the central control device 20 so as to be communicable with each other, the “sensor # 01” and the plurality of measurement terminals 11 included in the sensor # 01 are displayed on the browser 122. Among these, the Ui graph about the observation result of the temperature measured by the “temperature sensor” is displayed.
The observation results illustrated in FIG. 22 are stored in the GG unit 24 and the APF_POOL unit 22 when the user performs an operation to specify “sensor # 01” for the measurement apparatus 10 and “temperature sensor” for the type of measurement terminal, for example. Applicable information is read from the Ap1 information, and a Ui graph is generated and displayed.
Since the GG unit 24 generates an image that is directly operated by the user, the GG unit 24 needs a function of authenticating whether or not the user is valid. Such a function is provided by, for example, a lower OS. Moreover, although it is necessary to generate a display display signal depending on the type of the display device, this is also omitted here because it uses existing technology. In addition, a browser is an example of software for visualizing the Ui graph, and may be displayed by a method using another general technique.

  The COMMAND_GENEATOR section 25 operates by acquiring transmission request information of the true observation result of the abnormal value by the user who has browsed the information generated by the GG section 24 via the Mc215. The COMMAND_GENEATOR section 25 holds information such as whether the measuring device 10 having the specified range can respond immediately, when it is not possible to respond, and under what conditions it can respond, and based on this information, an abnormality The time point at which the transmission request information of the true observation result of the value is actually transmitted to the measuring apparatus 10 is determined. That is, the transmission request information of the true observation result of the abnormal value is immediately sent to the measuring apparatus 10 after the output of the GG unit 24, and the response from the measuring apparatus 10 does not return immediately thereafter. The COMMAND_GENEATOR unit 25 transmits the transmission request information of the true observation result of the abnormal value to the measuring apparatus 10 via the SwCM 32a at the timing determined by itself. In the measuring apparatus 10, the DETAIL_SENDER unit 606 receives this request via the Cm 191, and instructs the CREATE_MESSAGE unit 602 to transmit the true value of the abnormal value in the observation result by the above-described processing. The CREATE_MESSAGE unit 602 outputs the true value of the abnormal value to the MESSAGE_AGGREGATOR unit 17 via the Do 161 based on this instruction. The MESSAGE_AGGREGATOR unit 17 generates a message including the true value of the abnormal value, and the RU unit 18 transmits this message to the central control device 20 via the SrDT 31b. In the central controller 20, the MESSAGE_RECOGNIZer unit 21 receives this message via the SwDT 32b, and outputs the true value of the abnormal value requested by the user to the DET_LOG unit 23 via the As 212. The GG unit 24 reads the true value of the abnormal value requested by the user from the DET_LOG unit 23 via the As 214, and displays the true value of the abnormal value on the display device. Thereby, the user can browse the true value of the abnormal value.

  Hereinafter, the description will be focused on the processing in which detailed description is omitted in FIGS. Of the means for realizing the present embodiment, there are five types of particularly characteristic means. The first is “confirmation of information loss” for determining whether or not the observation result can be processed by the processing of the present embodiment. The second is “edge discovery” for determining whether approximate curve calculation processing is possible for the observation result. The third is “section approximation processing” for converting the observation result into an approximate curve. Fourth, when the observation results are outliers in the observation results that cannot be approximated curve calculation, to treat the observation results as outliers and convert the summary to the outlier summary information of the observation results Is “Abnormal Value Point Sequence Processing”. The fifth is “interactive information display” for easily providing information including approximate curves and abnormal values to a user who wants to browse the information.

First, the “information deficiency confirmation” process will be described. An information loss is a lack of observation result information during a case where the observation results may not be collected over a long period of time due to, for example, the start-up of the measurement device or a failure of the measurement device. In this case, the APF unit 603 handles the information of this section as an abnormal value with no information.
FIG. 23 is a diagram for explaining information loss confirmation processing according to an embodiment of the present invention.
FIG. 23 is an example of an observation result including a section in which information is missing. A section adjacent to the section in which the information shown in FIG. When calculating the approximate curve of this embodiment, information on two sections before and after the point group included in the section to be calculated is necessary. However, the section on the left side of the adjacent section 124 is the observation result missing section 123, and this calculation cannot be performed. Therefore, in this embodiment, the adjacent section 124 is also handled as an abnormal value section. The calculation of the approximate curve resumes from the next quasi-adjacent section 125. In the “confirmation of information loss” process, a process for distinguishing a section treated as an abnormal value section from a normal section is performed.

FIG. 24 is a flowchart of information loss confirmation processing according to an embodiment of the present invention.
First, the APF unit 603 determines whether or not the number of elements of the target frame for the target frame read in step S21 in the flowchart of FIG. 20 includes a sufficient amount for processing in the APF unit 603 such as interval approximation processing. For example, the determination is made by comparing with a predetermined threshold value (first threshold value) serving as a reference for the required amount (step S30). For example, for the observation result deficient section 123 in FIG. 23, the APF unit 603 determines that the number of elements in the section is not a sufficient amount. For the adjacent section 124 and the semi-adjacent section 125, the APF unit 603 determines that the number of elements in the section is a sufficient amount. When it is determined that the number of elements exceeds the required amount, the APF unit 603 determines whether there is a peripheral factor necessary for processing in the APF unit 603 such as interval approximation processing for the peripheral factor read in step S22. (Step S31). For example, when the target frame is the adjacent section 124, the APF unit 603 does not have a necessary peripheral factor in the adjacent section 124 based on the fact that the left section of the adjacent section 124 is the observation result missing section 123. Is determined. Further, for example, for the semi-adjacent section 125, the APF unit 603 determines that the necessary peripherals are included in the semi-adjacent section 125 based on the fact that there are a sufficient number of elements in the adjacent sections on both sides of the semi-adjacent section 125. It is determined that the factor exists. If it is determined that the necessary peripheral factor exists, the APF unit 603 considers the target frame as a normal section, and performs the processing from step S23 onward for the currently read target frame and peripheral factor. On the other hand, if it is determined in step S30 that the number of elements is not a sufficient amount and if it is determined in step S31 that the necessary peripheral factor does not exist, the APF unit 603 treats the currently read target frame as an abnormal section. Determine (step S32). Specifically, the APF unit 603 outputs each element (abnormal point sequence) of the abnormal section to the DET_LOG unit 605 via As 613. In addition, the APF unit 603 sets the summary information of the abnormal section range in the Ap format to generate ApI information (step S33), and outputs it to the APF_POOL unit 604 via Ap614. Thereby, the APF unit 603 can divide the observation result into an abnormal value section and a normal section.

  Next, the “spike removal” process will be described. This process corresponds to step S24 in FIG. For example, information on the agricultural field is information indicating changes in the natural world. Speaking of natural information such as temperature and humidity related to agriculture, true values that are not measured values are continuous values. There is no abrupt change in seconds for each true value. There is also a basic trend with respect to the transition of these true values. For example, in most cases, the temperature is higher in the daytime than in the night, and if the temperature is graphed over time in a day, a mountainous curve is drawn. In the present embodiment, an abnormal value / incorrect value is defined based on such empirical knowledge, and a method of mechanically removing those values from the observation result is used.

FIG. 25 is a flowchart of spike removal processing according to an embodiment of the present invention.
First, the APF unit 603 calculates the angle change rate (step S40). The angle change rate has been described with reference to FIG. Next, the APF unit 603 repeats the determination of whether or not each element of the sequence of angle change sequences obtained by calculating the angle change rate is a spike (step S41). Specifically, the APF unit 603 determines whether or not the absolute value of the difference in angle change rate from the previous point is greater than a predetermined threshold for each element of the angle change sequence number sequence (step S42). ). When larger than the threshold value, the APF unit 603 adds the element (point) to the spike point group (step S43). When the value is equal to or smaller than the threshold value, the APF unit 603 does nothing for the element. The APF unit 603 repeats this determination process for all elements to generate a spike point group. Next, the APF unit 603 performs spike removal for removing elements included in the spike point group from the target frame (step S45), and generates a target frame after the spike removal.

  Since the angle change sequence number sequence indicates the change in angle at that point, the higher the change in angle, the higher the value. In other words, a value having a large absolute value is shown when protruding or depressed from the periphery such as a spike. The notable difference between a spike and an edge is that since only one point of the spike protrudes, the rate of change in angle is high at the front of that point, and it is low at the back of that point. Only one of them will occur. Therefore, detection of edges and spikes is easy. In addition, spikes can be easily distinguished from tables because spikes change the angle change rate very quickly.

FIG. 26 is a diagram illustrating an example of a difference between a spike and a table according to an embodiment of the present invention.
FIG. 26A is a graph showing an example of spikes. FIG.26 (b) has shown the angle change rate of the graph of Fig.26 (a). It can be seen that the angle change rate corresponding to the spike part changes rapidly up and down in a short time. FIG. 26C is a graph showing an example of a table. FIG.26 (d) has shown the angle change rate of the graph of FIG.26 (c). In the angle change rate corresponding to the table portion, it can be seen that the vertical fluctuation occurs with a certain interval.
Spikes do not always occur at only one point, and exist as a minute table. Therefore, in the present invention, a heuristic threshold is given to distinguish between the table and the spike. The points detected as spikes are removed from the original point group. However, when spikes are continuously generated, the processing shifts to the abnormal point sequence processing described later as a special format that does not match the approximate curve pattern. Further, after the information on the spike portion is removed, alternative information may be interpolated. This is a measure against the deviation of the spike information if the measurement time interval is constant.

Next, the “edge discovery” process at the previous stage used when finding the table will be described. This process corresponds to step S26 in FIG. An edge is like one side of a spike that changes up and down (or down and up), and this refers to the point at which the rate of change in angle changes rapidly. The method for obtaining the angle change rate has been described above. An edge point group, which is a set of edge points, is a set of points whose angle change rate is extremely higher than the surroundings. It is only necessary to compare the angle change rates in order. For example, if the difference between the change rates of adjacent points is larger than a threshold (second threshold), it is determined that the edge.
Next, “table processing” will be described. The table is information of a point group having a shape as shown in FIG. A possible cause of such a point cloud is a local change in the measurement target. For example, the concentration is temporarily increased by, for example, carbon dioxide spraying in a house, and after a certain period of time, the skylight is opened to release the air inside the house. In this case, before and after the carbon dioxide is enclosed in the house, the transition follows the external atmosphere. However, while carbon dioxide is being sprayed, the house is cut off from the outside air, creating a unique system inside. In such a case, if local and partial changes are not clearly distinguished from before and after, it becomes difficult to see the trend as a whole excluding the section. Further, when generating an approximate curve, such a table is easily misunderstood as a quadratic function, which causes a decrease in the accuracy of measurement information. Therefore, in this embodiment, the table is detected as part of the edge discovery. Specifically, a method is adopted in which the edges constituting the table are detected and the point cloud is divided before and after the edge points to form separate sections.

FIG. 27 is a diagram illustrating division before and after an edge according to an embodiment of the present invention.
FIG. 27 shows an example in which the original target frame is divided into three based on a table surrounded by edge points in different directions. In the present embodiment, sections are divided before and after the edge, and section approximation processing is performed using each section as a new target frame.

FIG. 28 is a flowchart of table division processing according to an embodiment of the present invention.
As a premise, it is assumed that the APF unit 603 performs edge discovery for the target frame to be processed and generates a set of edge point groups as a result (step S26 in FIG. 21). This process is the process of step S28 in FIG.
First, the APF unit 603 initializes variables in the division process (step S50). Specifically, the AP 603 initializes a variable p_now indicating the current processing target point, a variable E_now indicating the edge of the current point, and a variable E_last indicating the latest edge. Next, the APF unit 603 finds the maximum point and the minimum point when the time elements are compared for the target frame and the peripheral factor after spike removal (step S51). The APF unit 603 stores the coordinate information of the maximum point (x axis: time, y axis: observation result) in p_max and the coordinate information of the minimum point in p_min. Next, the APF unit 603 sets p_min to p_now (step S52). Next, the APF unit 603 determines whether p_now is an edge point (step S53). When p_now is an edge point, the APF unit 603 sets p_now to E_now (step S54). Next, the APF unit 603 compares the direction of the angle change rate of E_now with the angle change rate of E_last, and whether the direction of the angle change rate of E_now is different from the angle change rate of E_last, that is, whether the sign is different. Determination is made (step S55).

When the direction of the angle change rate of E_now is different from E_last, the APF unit 603 adds E_start and p_now to a list variable Table_List that stores table information. Further, the APF unit 603 sets p_now to E_now (step S56).
When the direction of the angle change rate of E_now is equal to E_last, the APF unit 603 adds p_now to the list variable Edge_List that stores edge (non-table) information. Further, the APF unit 603 sets p_now to E_now (step S57).

Next, the APF unit 603 determines whether p_now and p_max are equal (step S58). When p_now is equal to p_max, the APF unit 603 divides the target frame according to Table_List and Edge_List (step S59), and generates a divided target frame.
On the other hand, if p_now is not equal to p_max and if it is determined in step S53 that p_now is not an edge point, the next point is set in p_now (step S591), and the processing from step S53 is repeated.
Thereby, the table can be found from the target frame using the edge point group, and the table portion and other portions can be divided.

Note that clustering is another method for the table discovery process. A method of using clustering will be described with reference to FIGS.
FIG. 29 is a first diagram illustrating a table finding process according to an embodiment of the present invention.
FIG. 30 is a second diagram illustrating the table finding process according to an embodiment of the present invention.
General clustering refers to an act of dividing a group having a plurality of elements into several subgroups. On a clusterable group, a distance can be defined between the elements. For example, the distance can be defined by mapping a point in a two-dimensional space of time and feature amount. In this case, the distance is the Euclidean norm between the two points. The unit vector in the time t direction is normalized according to the difference between the maximum value and the minimum value of the difference diff of the observed values val. This is because diff is a maximum and can take a large value from minus infinity to plus infinity, whereas t is a time difference and remains in a finite range. In the following description, the positions and distances of points are treated as values after normalization. An example of normalization processing is shown in FIG. An example in which clustering is applied to the point group after normalization processing is shown in FIG.

Next, the “section approximation process” will be described with reference to FIG. This process corresponds to step S29 in FIG.
FIG. 31 is a flowchart of section approximation processing according to an embodiment of the present invention.
First, the APF unit 603 calculates a point feature amount for each point in the target frame (step S60). “Point feature amount calculation” will be described later. After obtaining a set of feature values of each point in the target frame, the APF unit 603 applies a sub-block “pattern matching” with respect to those points (step S61). “Pattern matching” will be described later. The APF unit 603 obtains a line shape having a shape closest to the target frame, which is an optimum shape by pattern matching.
In parallel with the “pattern matching”, the APF unit 603 selects candidate points for calculating approximate parameters (step S62). For example, the APF unit 603 selects a point corresponding to a maximum value, a minimum value, a value having a particularly large other feature value, a point corresponding to both ends, or the like from a set of feature values of each point in the target frame. These points are called feature points.
Next, the APF unit 603 calculates approximate parameters for the optimum shape obtained by pattern matching (step S63). In the approximate parameter calculation, based on the optimum shape of the approximate curve, the specific feature points selected in step S62 are substituted, and the values of parameters such as coefficients and constants of the functional expression representing the specific approximate curve are obtained. Ask. For example, the APF unit 603 obtains parameters of the approximate curve by substituting a point having a particularly high feature amount among the feature points into the approximate curve formula. When the complete shape of the approximate curve is obtained, the APF unit 603 converts the expression form into an equation representing the approximate curve using them and some characteristic points (step S64), and outputs the equation.

  Next, “pattern matching” will be described. In the present embodiment, the observation result is converted into an approximate curve, and the parameters of the mathematical formula and the values of characteristic points such as the maximum value and the maximum value are transmitted to the central controller 20. Natural information about agriculture can be approximated by a continuous curve. In a certain section of time, the information does not change so rapidly in the agricultural field, so the curve can have a low dimension. Therefore, when creating an approximate curve of a point sequence of sensing results, which is information, it is divided into sections based on time and approximated to a curve of approximately four dimensions or less.

FIG. 32 is a flowchart of pattern matching according to an embodiment of the present invention.
FIG. 33 is a diagram illustrating compression processing in pattern matching according to an embodiment of the present invention.
FIG. 34 is a diagram for explaining approximate curve finding processing according to an embodiment of the present invention.
First, FP will be described. The FP is a format in which a feature point group can be matched with a pattern. A process for deriving the FP will be described. Since there is no point having the same value at time t, there is no point at the same position, and a direction can be set between two adjacent points. For example the original feature point group _{cp = [p 1, p 2} , ···, p n] when a, the vector representing the direction between the points and _{cv i = p i + 1 -p} i, the column [ _{cv 1,} cv 2, ···, and _{cv n-1].} Next, the maximum (p _max ) / minimum (p _min ) point of the feature point group is found and inserted into the original position of this column. Finally, an information string such as [cv ₁ , cv ₂ , cv _r , p _max , cv _{r + 1} ,..., Cv _s , p _min , cv _{s + 1} ,..., Cv _n−1 ] is obtained. This is called FP.

A flow of pattern matching processing will be described. First, the APF 603 reads the pattern file and finds the maximum number of elements (step S70) and the minimum number of elements (step S71) among all patterns. When the maximum number of elements is specified, it is set to the variable n (step S72). This n represents the length of the pattern to be compared with the matching target.
If the length of the FP is greater than or equal to n, the FP is adjusted to the length n in order to perform a comparison process with the pattern file (compression process) (step S73). Compression is performed by repeating the process of combining adjacent FP elements. As the contents of processing, first, all cv _i are confirmed, and two nearest cv _i are selected. Let these two be (cv _j , cv _k ). Next, an intermediate value is generated from these values. The intermediate value is (cv _{i ′} = (cv _j + cv _k ) / 2). Finally, (cv _j , cv _k ) is removed from the FP, and cv _{i ′} is inserted therein. This is repeated until the length of FP becomes n. An image of these processes is shown in FIG. Points 1 to 10 in FIG. 33 are FPs.

Next, the APF unit 603 reads and compares (matches) the FP compressed as necessary and the pattern file of the pattern length n (step S74). A pattern file configuration method and a pattern to be prepared will be described later. At the time of matching, first, the pattern file and the shape match are confirmed. When the shape of the approximate curve is provisionally determined, a point having a particularly high feature amount among the feature points is substituted into the approximate curve formula to obtain approximate curve parameters. When the parameters of the approximate curve are provisionally determined, the total distance between the approximate curve and the remaining feature points is obtained next time. The reciprocal of this is used as the degree of coincidence (in order to reduce the range of coincidence to a range of 0 to 1). It can be said that the closer the degree of coincidence, the better the approximation. Since some patterns have similar shapes, a plurality of patterns are simultaneously matched for a certain n. Accordingly, the degree of coincidence is recorded, and a curve having the highest degree of coincidence is searched from all patterns. An example of the approximation curve finding process is shown in FIG. In Figure 34, the distance _{d 1} of the point 128 and the curve 126, the distance between the point 128 and the curve 127 _d'1, the distance between the point 129 and the curve 126 is _{d 2,} the distance between the point 129 and the curve 127 with _d'2 is there. APF 603 this time, the matching degree between the feature point and the curve _{126 (1 / (d 1 +} d 2)), compare match degree between the feature point and the curve 127 _{(1 / (d'1 + d'} 2)) Thus, the curve 126 is determined as a more matched curve.
Next, the APF unit 603 compares the existing degree of coincidence calculated so far with the degree of coincidence calculated for the pattern file of the current pattern length = n, and whether the degree of coincidence calculated this time exceeds the existing degree of coincidence. It is determined whether or not (step S75). When the existing matching degree is not exceeded, the APF unit 603 determines a pattern corresponding to the existing matching degree as an optimum pattern. When the existing matching degree is exceeded, the APF unit 603 sets n−1 to n (step S76). Next, the APF unit 603 determines whether or not the newly set n matches the minimum value of the number of elements obtained in step S71 (step S77). If they match, the pattern corresponding to the highest matching degree calculated this time is determined as the optimum pattern. If they do not match, the APF unit 603 repeats the processing from step S73 for the newly set n.

FIG. 35 is a first diagram illustrating a pattern file according to an embodiment of the present invention.
FIG. 36 is a second diagram illustrating a pattern file according to an embodiment of the present invention.
This is a pattern file configuration method, and a file describing the arrangement of known points is called a pattern file. A pattern file is prepared for each curve shape. Examples of specific curve shapes will be described later. The pattern file contains the following information. The first is the allowable range of the entry angle / injection angle. Next, the change direction of the line segment. Finally, the maximum / minimum value, extreme value, inflection point, and change order. FIG. 35 shows information described in the pattern file by taking a cubic curve pattern as an example. FIG. 36 shows an example of the file layout of the pattern file. In this example, a total of 9 elements are described from the 5th line to the 13th line.

FIG. 37 is a first diagram illustrating the types of pattern files according to an embodiment of the present invention.
FIG. 38 is a second diagram for explaining the types of pattern files according to an embodiment of the present invention.
FIG. 39 is a third diagram for explaining the types of pattern files according to an embodiment of the present invention.
FIG. 40 is a fourth diagram illustrating the types of pattern files according to an embodiment of the present invention.
FIG. 37A and FIG. 37B show function patterns represented by linear functions. FIG. 38C to FIG. 38H show functions represented by quadratic functions. FIG. 39 (i) to FIG. 39 (n) show function patterns represented by cubic functions. 40 (o), FIG. 40 (p), FIG. 40 (q), FIG. 40 (r), FIG. 40 (t), and FIG. 40 (u) show functions represented by a quartic function. These patterns are prepared in advance, and the APF unit 603 compares the approximate curve indicated by these patterns with the FP. That is, in this embodiment, the approximate curve is classified into any one of these. In each function pattern shown in FIGS. 37 to 40, the black point is an extreme value, the white point is an inflection point, and the arrows are an entry angle and an exit angle. In the present embodiment, these pattern files are stored in a storage unit (not shown).

Next, “calculation of point feature value” will be described. First, the concept of the point feature amount in this embodiment will be described.
First, the relationship between the approximate curve finally obtained and the original point group will be described. The approximate curve finally obtained in this implementation is drawn so as to pass through the vicinity of the position of each point constituting the section. In other words, each point slightly affects the determination of the approximate curve when drawing the approximate curve. Here, for example, it is assumed that there are a point a and a point b in the point group P giving the approximate curve C. At this time, the approximate curve C ′ calculated from the point group P ′ obtained by removing a from the point group P and the approximate curve C ″ obtained from the point group P ′ obtained by removing b similarly are considered. Here, if C ″ is closer to the original shape of C than C ′, it can be said that C is better reproduced when C ″ is obtained. That is, the point group P ″ retains the characteristics of the original point group P better than the point group P ′. In other words, it can be considered that a lot of information has been lost because the point a has a larger amount of information than the point b which is the removed information. Hereinafter, this relative information amount is referred to as a point feature amount. Here, consider the factors that give the difference in the feature quantity of points. Here, the approximate curve is a line drawn in a two-dimensional space consisting of a time axis and a measurement value axis. Now, if all the observation results are arranged in a straight line in this space, it is sufficient to have information on two points in order to reproduce this straight line. Even if one point is removed from this observation result, the straight line finally completed is the same. It can be said that this removed point has a small feature amount. On the other hand, when all the observation results are arranged on the spike, if the point at the tip of the rapid spike is lost, there will be a large difference in the shape of the approximate curve that is finally created. This point has very many features. In this way, it is the change of the value in the vicinity of the point that has a great influence on the feature amount of the point. This is expressed by the slope of the tangent of the point and its rate of change.

FIG. 41 is a first diagram illustrating calculation of a point feature amount according to an embodiment of the present invention.
FIG. 42 is a second diagram illustrating calculation of a point feature amount according to an embodiment of the present invention.
FIG. 43 is a third diagram illustrating calculation of a point feature amount according to an embodiment of the present invention.
On the other hand, there are cases where it is not possible to distinguish between the slope and the rate of change. For example, in the example of FIG. 41, the points 40 and 41 have a simple enlargement / reduction relationship, and have exactly the same inclination and rate of change. It is the point 41 that determines the shape of the large frame of the section when viewed in the entire section including the points 40 and 41. In this case, the factor causing the difference in influence of the two points in the overall shape is the size of the range in which the point can be said to be characteristic. That is, the length of the distance to other equally characteristic points that would exist to the left and right of that point. This distance is called the surrounding size. If the surrounding size is large, it can be said that the point has a great influence on the shape of the approximate curve near the point. This is because there are no other characteristic points around the point.
From the above, the factors that give the feature amount of the point are the slope and change rate at the point and the size of the surrounding area. Therefore, in this implementation, the minimum value min (Score _target ) is selected from the range that can be taken by the Score _{target of} Expression (1).
Score _target = a ₁ × (norm ₁ + norm ₋₁ ) +
a ₂ × (norm ₂ + norm ₋₂ ) +.
a ₃ × (norm ₃ + norm ₋₃ ) (1)
Here, FIG. 42 shows a ₁ , norm _{1 and the} like of the formula (1). Also, set the parameters _a 1 _~a n, p to following holds. Here, f represents a straight line.
2 × a ₁ + 2 × a ₂ + 2 × a ₃ +... + 2 × a _n = 1
_{a 1 ≧ a 2 ≧ a 3} ≧ ··· ≧ a n = 0
norm _m = ((value of point m) −f (t _m )) ²
m = {p | p = ± 1, ± 2,..., ± n, a _{n = 0} }

  In the present embodiment, a feature amount at a certain point in the target frame is derived as follows. First, in the straight line calculation range centered on a certain point, the inclination of the straight line when the straight line passing through the point minimizes the equation of FIG. 42 is calculated. This is called the slope of the tilt point. After calculating the slopes of all points in the target frame, align them with respect to the time axis. In this point sequence, if the slopes of three adjacent points are almost unchanged, the slope of the center point can be inferred from the slopes of both ends even if there is no center point. Such a point is called a point with a small amount of information. An example of a small amount of information that can be easily estimated from the surroundings is shown in FIG. In FIG. 43, the value of the central point 131 can be estimated from the points 130 and 132. Thus, the point 131 with less information is removed from the point group. Point 131 is a slope with little information.

A method for calculating the minimum value min (Score _target ) of Expression (1) will be described. A certain time is denoted as t _i . The measurement value is referred to as _{V i.} The information (observation results) and calculation processing handled this time are for a two-dimensional space consisting of these two parameters. The approximate straight line can be expressed as N _target (t _i ) = at _i + b. This straight line segment a is the Score _target itself.
A time interval including observation results necessary for obtaining an approximate straight line is referred to as a “score calculation interval”. There is a target somewhere in the score calculation section. Therefore, the score calculation interval is determined by selecting two times so as to include target. Here, the score calculation section is represented as [t _start , t _end ]. t _start ≤ t _target ≤ t _end .

The observation result is a set of a time and a measured value that is uniquely determined by the time (p _i = (t _i , V _i )). Therefore, the elements of the observation result set have an order relationship, and the relationship can be expressed by adding a subscript. As a way of attaching a subscript, here, the target is set to zero and addition is performed for those adjacent in the direction of increasing time, and subtraction is performed for the direction of decreasing. For example, in certain score calculation interval, two at a time prior to target, if there is a time three observations to _post, p _{-2, p -1, p target,} as p _1, p 2, _{p 3} Can be expressed.
Since the observation results are discrete, a finite number of observation results are included in the score calculation section. Here, they are represented as a score calculation point group TC _target . All observation results within the range of the score calculation section excluding _target are entered in TC _target .
TC _target = {p _i | t _start ≦ t _target ≦ t _end }
If the elements are written side by side, it will be as follows. r is the number of observations on the negative side, and s is the number of observations on the positive side. It should be noted that the time interval between these observations is not constant.
TC _target = [p− _r , p− _{r + 1} , p− _{r + 2} ,..., P− _{r + (r−2)} , p− _{r + (r−1)} , p ₁ , p ₂ , p ₃ ,. , P _s−1 , p _s ]
Next, the degree of peeling between the value of the approximate straight line N _target (t _i ) and the actual observation result V _{i at} a certain time t _i is defined as “norm _i” as follows. :
normi _i = (V _i -N _target (t _i )) ²
In order to prevent a point far from the point target from affecting the feature quantity of the target, the norm _{i is} weighted. The weight for a certain point is α _i , and the calculation formula is W (t _i ). Since this weighting needs to be finely adjusted depending on the environment, the following formula W (t _i ) is defined here as an example.

Based on these, the Score _target can be expressed as follows.

It can be understood that this is a quadratic expression having unknown numbers (a, b) when arranged as follows.

Since α _j , t _j , and V _j are given in advance in the above equation (4), they are constants. These constant terms are

Then,
Score _target = A + Ba ² + Cb ² −2bD−2aE + 2abF
Thus, Score _target is a quadratic expression of a and b. Since α _j ≧ 0 and the time is always positive according to the setting of the function, since t _j ≧ 0 and B and C ≧ 0, Score _target has only one minimum point for a and b.
If the Score _target is partially differentiated for a and b, simultaneous equations for the unknowns a and b can be obtained. Since the minimum point is zero,

Solve this to get the desired a.

In this embodiment, the slope of a straight line (min (Score _target ) = a through which the feature amount passes through the point indicated by each observation result and the distance from the other point existing in a predetermined range centered on the point is minimized. ) To calculate. Next, the feature amount calculation will be described.

FIG. 44 is a first flowchart of point feature value calculation according to an embodiment of the present invention.
FIG. 45 is a second flowchart of point feature value calculation according to an embodiment of the present invention.
FIG. 46 is a first diagram illustrating a point probe according to an embodiment of the present invention.
FIG. 47 is a second diagram illustrating a point probe according to an embodiment of the present invention.
First, the APF unit 603 acquires the peripheral factor of the target frame via the Di 612 (Step S80). Next, the APF unit 603 repeats the following process for the target frame from which the spike has been removed (step S81). First, the straight line calculation range for the processing target point in the target frame is acquired from the peripheral factor (step S82). Next, the APF unit 603 calculates the inclination of the approximate straight line of the processing target point using the straight line calculation range (step S83). The APF unit 603 calculates the inclination of the approximate straight line having the highest degree of approximation of a certain processing target point. The APF unit 603 repeats this processing, obtains the slope of the approximate line for all points of the target frame, and generates a set of slopes of the approximate line. Next, the APF unit 603 generates a starting point and a probe for the set of inclinations of the approximate straight lines, and starts the operation of all the probes (step S84). When the APF unit 603 obtains an approximate straight line for each point, the APF unit 603 prepares the probe, operates the probe until all the probes stop, and calculates the feature amount for all the points.

  Here, the probe will be described. As described above, the feature point in the present embodiment refers to a point that greatly affects the shape in determining the approximate curve. Therefore, the feature amount for a certain point needs to be large so that the point satisfies the condition. Also, the number of points used for pattern matching of approximate curves should be as small as possible. Therefore, a definition that gives a large feature amount to a very small part of the point is necessary. Therefore, in this embodiment, the feature amount is calculated using the following idea.

  First, a starting point serving as a reference for the feature amount is selected. This is presumed to be an indispensable point in determining the shape. The feature amount of each point in a certain point group is determined based on this point. For example, points at both ends, maximum / minimum values, extreme values, inflection points, etc. From the relation of the calculation method described later, if there are a large number of points, the time required to calculate the whole will be faster. However, since it is a point selected by estimation, the accuracy of the resulting approximation is affected. Therefore, it is necessary to appropriately determine the number of points in consideration of information to be measured.

Next, a reference value tdiff _k = (setdiff _k , tempdiff _k , degree _k ) of the feature amount is given as a parameter to each starting point. setdiff _k is a feature amount given to the starting point, tempdiff _k is a feature amount of the probe that changes during scanning, and degree _k is an intercept (slope) that the probe refers to when calculating the feature amount. Increasing tempdiff _k increases the possibility that the starting point will be selected as a feature amount later. However, if this is raised too much, the true feature will be buried in the value and it will be difficult to find it, so it is necessary to set an appropriate initial value.

Next, the probe pr _k is produced from the starting point. Pointer loc _k for scanning a row of pr _k point groups [..., p _{i + 1} , p _i , p _{i + 1} , p _{i + 2} ,..., p _j−1 , p _j , p _{j + 1} ,. And a reference value tdiff _k of the feature quantity of the traveling direction vec _k and pr _k . These are hereinafter referred to as pr _k parameters. pr _k is, for example, the points p _i is the chosen starting point, made on both sides. A configuration example of the probe is shown in FIG.
The probe pr _k advances a certain amount for a certain time in the traveling direction vec _k for each processing loop. Let this be ds_min. A method for determining ds_min will be described later. The unit of ds_min is time. Since the sequence of points is not equally spaced with respect to the time axis, even if the probe pr _k moves by ds_min in one loop, it does not necessarily reach the adjacent point. A probe that moves a certain amount over the next point from the current position is called a collision here. The collision between the point and the probe can be easily calculated from the mutual positional relationship and the traveling direction of the probe. When a collision occurs, the point feature quantity diff _i is calculated based on the probe parameters and the point min (Score _target ).

  Only one-way probes are created for points at both ends of the section. Therefore, there is another probe coming in the direction of travel of all the probes. This is called two adjacent probes. Two adjacent probes always collide at some stage as the process proceeds (two probes get over the same point). This is called probe collision. When this state is reached, the two probes have scanned all the points between the starting points. Therefore, these probes are stopped so that they do not advance any further. When all the probes are stopped, the feature amount assignment process is terminated.

Next, how to obtain the optimum ds_min will be described. Since ds_min is the distance that the probe travels once, the time taken to scan the entire section is (t _end −t _start ) × number of probes / ds_min. Therefore, the maximum ds _min that does not get over two points at the same time is set as the maximum ds_min among the lengths of the intervals of the point sequence. The point p _i = (t _i , val _i ) and the interval of the point sequence can be expressed by the following formula as [t _start , t _end ]
ds_min = min ({| p _{i + 1} −p _i || t _start ≦ t _ei <t _end })

Next, a method of calculating the point feature quantity diff _i based on the probe parameters tdiff _k = (setdiff _k , tempdiff _k , degree _k ) and the point intercept degree _i = min (Score _target ) will be described. Degree _k is the current stage section held by the probe. This value is calculated based on the intercept of the points that have been processed with this probe so far. Compare this with degree _i , which is the intercept of the point. If these values are similar, the intercept of this point will not differ much from the point that the probe has processed so far. tempdiff _k is a reference value of the feature value currently held by the probe. tempdiff _k has the maximum value of the feature value of the point that the probe has recently processed. Therefore, if the currently processed point has an intercept similar to the previous one, the previously processed point (this is the point that gave the maximum feature value) represents the feature of the currently processed point. It is considered possible. Therefore, the feature amount of the point currently being processed is lowered. On the other hand, if they differ greatly, the feature amount initially given to the probe is given again, and at the same time, the feature amount of the probe is initialized. Similarly, the process of changing the parameter of pr _k is also performed. If the currently processed point has a significantly different intercept from the previously processed point, information on the intercept of that point is captured. FIG. 47 shows a process for simultaneously calculating the point feature value and changing the probe parameters. Details will be described.

  First, to determine how similar the intercept of the probe and the intercept of the current point is, whether or not the absolute value of the difference is smaller than a certain threshold value α is used. This threshold value α is a constant and is given in advance before the operation of the apparatus. Since the possibility that a certain point is determined to be characteristic increases as α decreases, the number of feature points increases, increasing the load of post-processing pattern matching processing. It is desirable to determine the value of α from the average value of changes in previous observations. β represents the attenuation of the feature value. This indicates that, for example, two points that are sufficiently distant from each other show similar values, but they are irrelevant. The attenuation of the relationship corresponds to the subtraction of β.

As a specific process, when the absolute value of the difference between min (Score _target ) and degree _k is less than or equal to α, the APF unit 603 substitutes (min (Score _target ) −degree _k ) / 2 for degree _k . Also, the APF unit 603 substitutes (degree _k × tempdiff _k ) / 2 for diff _i . Also, the APF unit 603 substitutes tempdiff _k −β for tempdiff _k .
If the absolute value of the difference between min (Score _target ) and degree _k is greater than α, the APF unit 603 substitutes min (Score _target ) for degree _k . Also, the APF unit 603 substitutes setdiff _k for diff _i . Also, the APF unit 603 substitutes setdiff _k for tempdiff _k .

Return to the description of the flowchart. The APF unit 603 repeats the following processing from step S86 to step S894 for each probe included in the set of probes (step S85). First, it is determined whether for a probe _{p r} pr is operating (step S86). If it is determined that it is not operating, the process loop is exited (step S87). When operating, pr is moved by ds_min (step S88). Next, the APF unit 603 determines whether pr collides with an adjacent probe (step S89). If there is a collision, pr is stopped (step S891). Then, the APF unit 603 determines whether there is a point that has collided with pr (step S892). If there is no collision, the process of step S892 is performed. If it is determined in step S892 that there is no collision point, the process loop is exited (step S893). If it is determined in step S892 that there is a point colliding with pr, the APF unit 603 obtains a feature amount of the point p from the parameter of the point p colliding with ppr (step S894). When the above processing is performed for all the probes, feature quantities at the current stage are generated at each point as a result of operating all the probes. When all the probes are stopped (step S895), the APF unit 603 calculates the final feature value diff _i of each point.

Next, the processing of the abnormal point sequence will be described. This process corresponds to step S291 in FIG. When an abnormal value is detected in the measured value, it is desirable to be able to verify whether the abnormal value is really just a measurement error or significant information based on the observation result. The most effective way to do this is to accumulate observation results as they are. Since the observation results have been accumulated to some extent as a characteristic of the verification work, verification is performed on the observation results, so that the immediacy of the target observation results is not required.
Devices that can accumulate observation results in the configuration of this embodiment are the measurement device 10 and the central control device 20. However, in view of the reduction in the amount of transmission information, it is not ideal to store in the central controller 20. This is because in order to accumulate observation results in the central controller 20, the information must be immediately transmitted to the central controller 20 as it is. Therefore, the communication amount is not reduced. Another approach is to accumulate in the measuring device 10. Of course, a method for extracting information from the measuring apparatus 10 is necessary, but in this case, the accumulated observation results need not be transmitted immediately, and may be transmitted at any convenient time.
In the present embodiment, when the measurement apparatus 10 estimates that the observation result is an abnormal value, only the periphery is cut out and stored in the measurement apparatus 10. This allows later verification work. FIG. 48 shows a flowchart of abnormal point sequence processing.

FIG. 48 is a flowchart of abnormal point sequence processing according to an embodiment of the present invention.
First, the APF unit 603 performs the feature amount calculation of the above-described points (step S90). The APF unit 603 generates a set of feature quantities for each point in the target frame. Next, the APF unit 603 searches for a point having a maximum / minimum value among the points (step S91). The APF unit 603 generates a maximum / minimum point and a feature amount for each time. Next, the APF unit 603 divides the target frame (step S92). Specifically, it is divided into a section including the abnormal value (center) and other sections (left and right sections).
Next, the APF unit 603 performs processing on each of the divided target frames. When the divided target frame to be processed is the central target frame after the division, the APF unit 603 finds a point belonging to the upper and lower convex hulls for the central target frame (step S94). The points belonging to the upper and lower convex hulls will be described later. Next, the APF unit 603 performs interval approximation processing on points belonging to the upper and lower convex hulls (step S95). The interval approximation process is as described above. Through the interval approximation process, the APF unit 603 obtains approximate curves of the upper and lower convex hulls. The approximate curves of the upper and lower convex hulls are information indicating the range where the abnormal point sequence exists. Next, the APF unit 603 generates information (abnormal range summary information) indicating the existence range of the abnormal point sequence (step S96). The summary information of the abnormal range is an approximate curve function of the upper and lower convex hulls, the start and end times of the center target frame, and the maximum and minimum values of the observation result in the center target frame.
On the other hand, when the divided target frame to be processed is other than the divided target frame, that is, the target frames at both ends, the APF unit 603 performs section approximation processing (step S97).

FIG. 49 is a first diagram for explaining the processing of the abnormal point sequence according to the embodiment of the present invention.
An example of an abnormality that occurs in sensing in the agricultural field is a sudden change in weather such as a typhoon. For example, it has been found that the amount of sunshine varies greatly during the typhoon. FIG. 49 shows an example of the pattern of the amount of sunlight when the typhoon passes. In this case, although the outline is a mountain, there is a valley in the fine part, and it cannot be said that the information is simply a mountain. If such information is easily approximated, essential information may be lost when analyzing past information based on knowledge discovered at a later date. Even if an approximate curve is generated, the accuracy of the curve is very low.
Therefore, the measuring device of this embodiment does not approximate such a complex shape point sequence, but instead transmits the outline of the shape as a temporary report to the central control device. As for the information of the temporary report part, the observation result is accumulated on the measuring device side. On the other hand, the observation result of the frame made into the approximate curve is discarded. The accumulated observation results are taken out from the measuring device via the central control device when a request is received from a person who wants to browse the information. The outline of the transmitted shape is the maximum and minimum values obtained by approximating the upper and lower convex hulls of the target frame to a curve. An image of the approximate curve of the convex hull is shown in FIG.

FIG. 50 is a second diagram for explaining the processing of the abnormal point sequence according to the embodiment of the present invention.
In FIG. 50, a curve 140 is an approximate curve (upper side) of a convex hull. A curve 141 is a convex hull (upper convex hull) from above. A curved line 142 is a convex hull from below (lower convex hull). A curve 143 is an approximate curve (lower side) of the convex hull.
A convex hull refers to a set of straight lines that wraps a figure. There are various algorithms for obtaining the convex hull. (For example, Luc Devroye, Godfried Toussaint., “A note on linear expected time algorithms for finding convex hulls”, 1981, P361-P366, Computing, vol26). Here, the concept of convex hull is used in parallel with the time axis. It can be seen that all the points included in the abnormal point group are included in at least the region surrounded by the approximate curve of the convex hull. The worst value of the convex hull is a rectangle.
The white dots in FIG. 50 are point groups that constitute the upper convex hull, and the black dots are point groups that constitute the lower convex hull. Since this is a convex hull when the original point cloud is viewed from the direction of the value axis, the entire original point cloud always enters between the straight line groups connecting the upper and lower convex hulls. Each straight line connecting the points constituting the convex hull as a characteristic of the convex hull always increases monotonously upward or downward. Therefore, the approximate curve of the convex hull can be approximated to a quadratic curve or a straight line connecting both ends. However, the maximum and minimum values of the approximate curve may exceed the maximum and minimum values of the point group. An example is shown in FIG.

FIG. 51 is a third diagram for explaining the processing of the abnormal point sequence according to the embodiment of the present invention.
FIG. 51A shows an example in which all points are constituent elements of the lower convex hull. FIG. 51B shows an example where the upper convex hull and the lower convex hull form a rectangle. In this case, the number of points constituting the convex hull is the minimum number. In particular, when the convex hull has a substantially rectangular shape as shown in FIG. 51 (b), the separation between the maximum and minimum values increases. Therefore, in this embodiment, both the approximate curve of the convex hull and the maximum / minimum value of the original point group are calculated as the abnormal point sequence and finally displayed on the GG unit 24. As a result, a user who wants to browse information can grasp the outline of the existence range of the abnormal point sequence more accurately.

FIG. 52 is a fourth diagram illustrating the processing of the abnormal point sequence according to the embodiment of the present invention.
In the processing of the abnormal point sequence of the present embodiment, it is assumed that there is only one section including many abnormal values in the target frame. This indicates the following state. In this implementation, when there are many spike edges in the target frame, it is determined that the section includes an abnormal value. That is, in the present embodiment, when it is determined that the target frame is an abnormal value point group, it is considered that the structure of the target frame is usually as shown in the example of FIG. Therefore, by dividing this target frame into three parts, ie, smooth portions (section 144, section 146) at both ends and a section 145 having a continuous truly abnormal value at the center, the abnormal section can be suppressed to a smaller range. Of these three sections, with respect to the sections 144 and 146, the normal section approximation processing is applied, and the abnormal point sequence processing using the convex hull approximation curve is performed only in the section where the abnormal values corresponding to the section 146 are continuous. . As shown in FIG. 52 (b), the target frame is a smooth section between two or more consecutive abnormal values in the abnormal section 148 in addition to the smooth portions (section 147, section 149) at both ends. It is also conceivable that 148a is sandwiched. In this case, if the number of subdivisions is too much, the number of approximate curves will increase. Therefore, in the description of this embodiment, the subdivision of three or more divisions is not performed. However, this division method is an example and is not limited to this.

Finally, “interactive information display” will be described. In the invention, it is assumed that the information is used as a time series of numbers arranged in order of observation time. Therefore, a two-dimensional graph is used to visualize them. This is realized by a program having a GUI for displaying a line graph or the like, for example.
FIG. 53 is a diagram showing a display example of observation results according to an embodiment of the present invention.
FIG. 53A is an example of a normally displayed Ui graph. The convex hull approximation curve 150 (upper side and lower side) indicates that an abnormal value exists in the observation result in this time zone. When the user clicks a portion surrounded by the convex hull approximate curve 150, the display is switched to the display of FIG. The true value of the abnormal value is displayed in the Ui graph of FIG. 53 (b).
As illustrated in FIGS. 53A and 53B, in this embodiment, when the observation result shows an extreme abnormal value, the portion in the Ui graph is estimated based on a characteristic value. To display a closed area using a set of approximate curves, and to display the information if the user who wants to view the information requests a detailed value of that part. Then, after the user issues a request, the observation result is transmitted from the measuring apparatus 10 and the information is interpolated in the Ui graph. Since the user does not always need information on all the abnormal ranges, the amount of communication can be reduced by transmitting only necessary portions. In addition, transmission of abnormal values is not performed immediately when a request is received from the user, but in the form of a transmission reservation, the status of the power and communication bandwidth of each measuring device 10 is monitored, and there is a margin for those resources You may make it transmit at a certain time.

  In this embodiment, natural information obtained by a measurement device at a remote location, particularly observation results for agricultural applications, is used to compress information when transmitting information from a remote location, using approximate curve calculation processing, Information can be displayed to the person who wants to browse, and the accurate observation result of the part including the abnormal value in the observation result can be referred to by an interactive method if necessary.

  According to the present embodiment, the measuring apparatus 10 converts the information into an approximate curve representing observation results measured in time series and transmits the information to the central control apparatus 20. Therefore, the measuring apparatus 10 does not need to frequently transmit observation results, and can reduce the amount of communication. For example, even if the number of parts of the measuring apparatus 10 is reduced or the quality of the parts is lowered in order to suppress the price, it is not necessary to transmit the observation results to the central control apparatus more frequently. In addition, if it can be converted into an approximate curve, the observation result in that section is deleted, so the storage capacity may be small. In addition, since the observation values can be compressed and transmitted into information that represents an approximate curve with a smaller amount of information, the amount of information to be transmitted increases even if the observation results are measured with a shorter time interval and the observation results are high-definition. There is no. That is, the observation result can be handled with high definition.

There is a failure of the measuring apparatus 10 as a problem in sensing for agriculture. As a type of failure of the measuring apparatus 10, in addition to the case where the function is completely stopped, due to deterioration of the measuring terminal 11 or the like, the true value of natural information is the same, but different observation results are measured. There is a case. As a countermeasure, there is correction of the observation result in the central controller 20. In this case, the observation result is automatically corrected on the central controller 20 side in consideration of such circumstances. However, when these methods are selected, there is an extra expense due to unnecessary replacement of the measuring device that can be originally repaired, and when correction is performed, the true value of natural information and the characteristics of deterioration of the measuring terminal are affected. Deep knowledge is required.
In the present embodiment, since the section where the abnormal value has occurred can be displayed as a convex hull approximation curve, deep knowledge for correction to the observation result is not necessary. Further, the true value of the abnormal value can be displayed as required by the user's requested operation.

  The above-described measuring apparatus 10 has a computer inside. Each process of the measurement apparatus 10 described above is stored in a computer-readable recording medium in the form of a program, and the above process is performed by the computer reading and executing the program. Here, the computer-readable recording medium means a magnetic disk, a magneto-optical disk, a CD-ROM, a DVD-ROM, a semiconductor memory, or the like. Alternatively, the computer program may be distributed to the computer via a communication line, and the computer that has received the distribution may execute the program.

  The program may be for realizing a part of the functions described above. Furthermore, what can implement | achieve the function mentioned above in combination with the program already recorded on the computer system, what is called a difference file (difference program) may be sufficient.

  In addition, it is possible to appropriately replace the components in the above-described embodiments with known components without departing from the spirit of the present invention. The technical scope of the present invention is not limited to the above-described embodiment, and various modifications can be made without departing from the spirit of the present invention. For example, the present invention can be applied to an observation system in a field where there is a demand for observing natural information by remote sensing in commercial / industrial areas closely related to the natural environment such as agriculture, livestock, fisheries, and fishery. Further, in the service industry where the natural environment affects sales, such as convenience stores and restaurants, it is conceivable to use the remote natural information measurement system of this embodiment in a field where there is a demand for grasping the trend of natural information.

The measurement terminal 11 corresponds to the sensor 101, the AD_CONV unit 13 corresponds to the sensor information acquisition unit 102, and the CU unit 16 (SELECTOR unit 600) corresponds to the immediate transmission determination unit 103. The CU unit 16 (APF unit 603) corresponds to the approximate curve generation unit 104, and the CU unit 16 (DETAIL_SENDER unit 606) corresponds to the abnormal value transmission control unit 106. The CP unit 19 corresponds to the instruction information analysis unit 105, the RU unit 18 corresponds to the communication unit 107, the TIMING unit 15 corresponds to the timer 108, the BUFFER unit 601, the APF_POOL unit 604, and the DET_LOG unit 605 are storage units. 109.
The MESSAGE_RECOGNIZer unit 21 corresponds to the acquired information identification unit 202, the GG unit 24 corresponds to the image generation unit 203, the COMMAND_GENETOROR unit 25 corresponds to the instruction information generation unit 204, and the APF_POOL unit 22 and the DET_LOG unit 23 are storage units. This corresponds to 205.

DESCRIPTION OF SYMBOLS 1 ... Natural information measurement system, 3 ... Network, 10 ... Measuring apparatus, 101 ... Sensor, 102 ... Sensor information acquisition part, 103 ... Immediate transmission determination part, 104 ... Approximate curve generation unit, 105 ... Instruction information analysis unit, 106 ... Abnormal value transmission control unit, 107 ... Communication unit, 108 ... Timer, 109 ... Storage unit, 11 ... Measurement terminal , 12 ... Interface Pr, 13 ... AD_CONV part, 14 ... Interface Di, 15 ... TIMING part, 151 ... Interface Tm151, 16 ... CU part, 161 ... Interface Do, 17 ... MESSAGE_AGGREGATOR part, 171a ... Interface Am, 18 ... RU part, 181 ... Interface Mc, 1 ... CP part, 191 ... interface Cm, 20 ... central control unit, 201 ... input receiving part, 202 ... acquisition information identifying part, 203 ... image generating part, 204 ... Instruction information generation unit, 205 ... storage unit, 206 ... communication unit, 21 ... MESSAGE_RECOGNIZer unit, 211 ... interface Ap, 212 ... interface As, 213 ... interface Ap, 214 ... Interface As, 215 ... Interface Mc, 216 ... Interface Ui, 22 ... APF_POOL section, 23 ... DET_LOG section, 24 ... GG section, 25 ... COMMAND_GENETOROR section, 600 ... SELECTOR part, 601... BUFFER part, 602. CREATE_MESSAGE section, 603... APF section, 604... APF_POOL section, 605... DET_LOG section, 606... DETAIL_SENDER section, 611... Interface Dd, 612. As, 614 ... interface Ap, 615 ... interface As, 616 ... interface Ap, 617 ... interface As

Claims (10)

An approximate curve generation unit that acquires observation results measured by the sensor and generates information representing an approximate curve that is a curve that approximates a set of points indicated by the observation results in time series;
A communication unit that transmits information representing the approximate curve generated by the approximate curve generation unit;
A measuring apparatus comprising: The approximate curve generation unit exists in a predetermined range centered on the point through the point indicated by the observation result of the feature amount corresponding to the observation result at each time included in the time-series observation result. Calculate using the slope of the straight line that minimizes the distance to other points, and generate information representing an approximate curve based on the set of features.
The measuring apparatus according to claim 1. The approximate curve generation unit calculates a change rate of each point in a set of points indicated by the observation result based on an observation result corresponding to the point and observation results corresponding to points before and after the point, and the change rate When there is a set of points surrounded by two points with different directions of change of the information, the point set is divided by the two points, and information representing an approximated curve for each of the divided point sets Generate
The measuring apparatus according to claim 2. The approximate curve generation unit is configured when the number of observation results included in a target section for generating information representing an approximate curve in the time-series observation results is equal to or less than a predetermined first threshold, or indicated by the observation results An observation result in which the rate of change calculated based on the observation result corresponding to each point in the set of points and the observation result corresponding to the points before and after the point is equal to or greater than a predetermined second threshold is set in the target section When three or more thresholds are included, it is determined that an abnormal value is included in the section.
The measuring device according to claim 2 or claim 3. When the approximate curve generation unit determines that the section includes an abnormal value, the approximate curve generation unit generates information representing a curve that includes the abnormal value in the section,
The communication unit transmits information of a curve that convexly envelops the abnormal value instead of information representing an approximate curve for the section.
The measuring apparatus according to claim 4. Based on a request for transmission of an abnormal value included in the observation result, an abnormal value transmission control unit that performs control to transmit the observation result included in the section determined to include the abnormal value,
The measuring apparatus according to claim 4 or 5, further comprising: An immediate transmission determination unit that determines whether it is necessary to immediately transmit the observation result by comparing with a predetermined threshold;
The communication unit immediately transmits an observation result determined by the immediate transmission determination unit to be immediately transmitted together with information indicating the approximate curve that has been generated by the approximate curve generation unit,
The measuring apparatus according to any one of claims 1 to 6. One or more measuring devices according to any one of claims 1 to 7;
A central control device comprising: an image generation unit that generates a graph of an approximate curve based on information representing the approximate curve received from the measurement device;
Measuring system. Obtaining observation results measured by the sensor, generating information representing an approximate curve that is a curve that approximates a set of points indicated by the observation results in time series,
A measurement result collecting method for transmitting information representing the generated approximate curve. Measuring device computer,
Means for acquiring observation results measured by a sensor and generating information representing an approximate curve that is a curve that approximates a set of points indicated by the observation results in time series;
Means for transmitting information representing the generated approximate curve;
Program to function as.

Images