JP6979477B2 - Status determination device, status determination method and computer program - Google Patents

Description

The present invention relates to a state determination device, a state determination method, and a computer program.

Conventionally, in equipment equipped with multiple devices consisting of mechanical parts such as motors, pumps, bearings and gears, vibrations generated during operation of the equipment are measured to monitor the operating status of the equipment, and the equipment is worn or damaged. (See, for example, Non-Patent Document 1). The vibration of a device composed of mechanical parts gradually increases during operation due to wear of parts, deterioration of lubricating oil, etc. with time of use. If the operation is continued in this state, the risk of equipment failure increases.

Keisuke Hashizume, Akitoshi Takeuchi, Yuzuru Tanaka, "Condition Monitoring System (CMS) Initiatives", [online], NTN TECHNICAL REVIEW No.82 (2014), [Search on March 3, 2nd year of Reiwa], Internet <URL: https://www.ntn.co.jp/japan/products/review/pdf/NTN_TR82_P074.pdf ＞

By the way, in the current equipment inspection, for example, the magnitude of vibration (vibration value) can be measured using a commercially available vibration meter, and in the abnormality determination, for example, the absolute standard of the vibration value indicated by ISO10816-1 or the like can be referred to. many. However, the actual magnitude of vibration varies greatly depending on the target equipment. Furthermore, even with the same equipment, the vibration generated differs depending on the load. In addition, the measured vibration value varies greatly depending on the location where the vibration meter is installed.
As described above, even when the abnormality diagnosis of the equipment is performed simply by the magnitude of the vibration value, the result of the abnormality diagnosis can be greatly affected by various factors. Therefore, it is necessary to determine the state of the equipment by a method suitable for the equipment for each diagnosis, which causes a problem that it takes time and effort.

In view of the above circumstances, an object of the present invention is to provide a technique capable of easily determining the state of equipment regardless of the equipment or the place of diagnosis.

One aspect of the present invention is installed in a reference value setting unit for setting reference data as a reference in any of a stopped state in the equipment, an excessive state at the start, or a state in a steady operation, and a facility to be diagnosed. Using a plurality of detection data including information on the vibration detected by the detected sensor, a probability distribution representing the existence probability of the amplitude indicated by the information on the vibration included in the detection data and one of the reference data. A data selection unit that performs an operation to obtain a difference from the probability distribution representing the probability of existence of the amplitude in the state, and selects detection data whose value of the operation result is less than a predetermined threshold value as data used for determining the state of the equipment. It is a state determination device including a state determination unit for determining the degree of abnormality of the equipment using the selected data.

One aspect of the present invention is the above-mentioned state determination device, further including a noise removing unit for extracting noise from the detection data to extract the vibration component inherent in the machine, and the data selection unit is the reference data. The data to be used for the state determination is selected from the detection data from which noise has been removed.

One aspect of the present invention is the above-mentioned state determination device, in which the state determination unit performs regression prediction using the selected data and calculates at least a period until it is estimated that an abnormality occurs.

One aspect of the present invention is the above-mentioned state determination device, in which the data selection unit compares a probability distribution obtained from the detection data with a plurality of probability distributions obtained from data in a plurality of operating states. After determining the motion state of the detected data from the calculation result, the calculation is performed to compare the probability distribution obtained from the detected data with the probability distribution obtained from the reference data in the determined motion state. The detection data whose resulting value is less than a predetermined threshold is selected as the data used for the state determination.

One aspect of the present invention is the above-mentioned state determination device, in which the reference value setting unit further sets a reference for the number of data used for processing by the state determination unit, and the state determination unit is set. The state of the equipment is determined using the numerical data.

One aspect of the present invention includes a reference value setting step for setting reference data as a reference in any of a stopped state in the equipment, an excessive state at the start, or a state in a steady operation, and installation in the equipment to be diagnosed. Using a plurality of detection data including information on the vibration detected by the detected sensor, a probability distribution representing the existence probability of the amplitude indicated by the information on the vibration included in the detection data and one of the reference data. A data selection step of performing an operation to obtain a difference from a probability distribution representing the probability of existence of an amplitude in a state, and selecting detection data whose value of the operation result is less than a predetermined threshold value as data used for determining the state of the equipment. It is a state determination method including a state determination step for determining the degree of abnormality of the equipment using the selected data.

One aspect of the present invention includes a reference value setting step for setting reference data as a reference in any of a stopped state in the equipment, an excessive state at the start, or a state in a steady operation, and installation in the equipment to be diagnosed. Using a plurality of detection data including information on the vibration detected by the detected sensor, a probability distribution representing the existence probability of the amplitude indicated by the information on the vibration included in the detection data and one of the reference data. A data selection step in which an operation is performed to obtain a difference from a probability distribution representing the probability of existence of an amplitude in a state, and detection data in which the value of the operation result is less than a predetermined threshold value is selected as data used for determining the state of the equipment. It is a computer program for causing a computer to execute a state determination step of determining an abnormality degree of the equipment using the selected data.

INDUSTRIAL APPLICABILITY According to the present invention, it is possible to easily determine the state of equipment regardless of the equipment or the place of diagnosis.

It is a block diagram which shows the structure of the state determination system in this invention. It is a schematic block diagram which shows the internal structure of the vibration detection apparatus in this embodiment. It is a schematic block diagram which shows the functional structure of the state determination apparatus in this embodiment. It is a flowchart which shows the flow of the state determination processing of the state determination apparatus in this embodiment. It is a figure which shows an example of the display screen displayed on the display part in this embodiment. It is a figure which shows an example of the display screen displayed on the display part in this embodiment.

Hereinafter, an embodiment of the present invention will be described with reference to the drawings.
FIG. 1 is a configuration diagram showing the configuration of the state determination system 100 in the present invention.
The state determination system 100 includes a vibration detection device 10, a relay device 20, a data server 40, and a state determination device 50. The vibration detection device 10 and the relay device 20 are wirelessly connected to each other so as to be communicable. The relay device 20 and the data server 40 and the data server 40 and the state determination device 50 are connected by wire so as to be communicable.

The vibration detection device 10, the relay device 20, the data server 40, and the state determination device 50 are installed at the site of the customer's plant equipped with the diagnosis target equipment 60. The diagnosis target equipment 60 is equipment that is the target of the state determination. The state determination referred to here is a determination regarding the operating operation of the equipment 60 to be diagnosed. For example, the operating operation of the equipment 60 to be diagnosed is normal, a state requiring a little attention, an abnormal state, or the like. In this embodiment, the plants include industrial plants such as chemicals, plants that manage and control wells such as gas fields and oil fields and their surroundings, plants that manage and control power generation such as hydropower, thermal power, and nuclear power, solar power, and the like. There are plants that manage and control environmental power generation such as wind power, and plants that manage and control water and sewage and dams.

Although FIG. 1 shows only one diagnostic target equipment 60 in the plant, a plurality of diagnostic target equipment 60 may be installed in the plant. In this case, at least one vibration detection device 10 is installed for each diagnosis target equipment 60.

The vibration detection device 10 is installed in a part of the diagnosis target equipment 60, and detects the vibration of the installed diagnosis target equipment 60. The vibration detection device 10 performs signal processing on the detected vibration data, and relays the data after the signal processing (hereinafter referred to as “detection data”) at a predetermined timing (for example, every 10 to 60 minutes). Send to. Here, the vibration data is data having a time length capable of detecting, for example, at least 100 sample points. The detection data transmitted by the vibration detection device 10 includes at least the identification information of the vibration detection device 10, the time information at which the vibration data is detected (hereinafter referred to as “detection time information”), and the time waveform of the acceleration. The time waveform of acceleration is shown by the time on the horizontal axis and the amplitude value on the vertical axis. The detection data transmitted by the vibration detection device 10 may include temperature information around the vibration detection device 10, information on the remaining battery level of the vibration detection device 10, and information on the radio wave intensity.

The relay device 20 relays the detection data transmitted from the vibration detection device 10 to the data server 40. The relay device 20 is, for example, a gateway.

The data server 40 stores the detection data transmitted from the relay device 20. The data server 40 transmits the accumulated detection data to the state determination device 50 at the timing when the request from the state determination device 50 is made or at the predetermined timing. The data server 40 is configured by using an information processing device such as a personal computer.

The state determination device 50 determines the state of the diagnosis target equipment 60 by using the detection data transmitted from the data server 40. The state determination device 50 is configured by using an information processing device such as a personal computer.

FIG. 2 is a schematic block diagram showing an internal configuration of the vibration detection device 10 according to the present embodiment.
The vibration detection device 10 is composed of a head portion 1 and a main body portion 2. The head portion 1 and the main body portion 2 are connected by wire. The head portion 1 is installed in the equipment to be diagnosed 60.
The head portion 1 includes a sensor 11. The sensor 11 detects the vibration data of the equipment 60 to be diagnosed. The sensor 11 is a vibration sensor, for example, an acceleration sensor, a speed sensor, or a displacement sensor. The sensor 11 outputs the identification information, the detection time information, and the detected vibration data of the vibration detection device 10 to the main body 2.

The main body 2 includes a signal processing unit 12, a communication unit 13, and a power supply unit 14.
The signal processing unit 12 performs signal processing on the vibration data detected by the sensor 11. Specifically, in addition to the vibration data, the signal processing unit 12 obtains identification information of the vibration detection device 10, detection time information, ambient temperature information of the sensor 11, and capacitor voltage information (remaining battery level) of the power supply unit 14. It is associated and output to the communication unit 13.

The communication unit 13 transmits the detection data output from the signal processing unit 12 to the relay device 20 at a predetermined timing. The communication unit 13 transmits the detection data to the relay device 20 by, for example, LoRa (registered trademark) WAN (Low Power Wide Area Network), which is one of the communication methods of LPWA (Low Power Wide Area). The communication method used by the communication unit 13 does not have to be limited to LoRaWAN, and any communication method may be used as long as it is a low power communication method.

The power supply unit 14 supplies the sensor 11, the signal processing unit 12, and the communication unit 13 with the electric power required for operation. The power supply unit 14 acquires, for example, the electric power required for operation by generating electric power by a built-in photovoltaic element. The photovoltaic element is composed of a combination of a dye-sensitized solar cell and a super capacitor (electric double layer capacitor). By using a dye-sensitized solar cell and a supercapacitor (electric double layer capacitor), it is possible to generate electricity even in a low illuminance environment (for example, 100 to 5000 lp). Furthermore, by complying with explosion-proof standards that do not use storage batteries (intrinsically safe explosion-proof), it can be installed in chemical plants that require explosion-proof specifications.

FIG. 3 is a schematic block diagram showing a functional configuration of the state determination device 50 in the present embodiment.
The state determination device 50 includes a communication unit 51, a display unit 52, an operation unit 53, a control unit 54, and a storage unit 55.
The communication unit 51 communicates with the data server 40. The communication unit 51 receives the detection data transmitted from the data server 40.

The display unit 52 is an image display device such as a liquid crystal display or an organic EL (Electro Luminescence) display. The display unit 52 displays various information. The display unit 52 displays, for example, information on the state of the equipment 60 to be diagnosed and the result of the state determination. The display unit 52 may be an interface for connecting the image display device to the state determination device 50. In this case, the display unit 52 generates a video signal for displaying information about the state of the equipment 60 to be diagnosed and the result of the state determination, and outputs the video signal to the image display device connected to the display unit 52.

The operation unit 53 is an input device for inputting a user's instruction to the state determination device 50. The operation unit 53 accepts inputs such as an information display instruction and a setting change instruction. The operation unit 53 is configured by using an existing input device such as a keyboard, a touch panel, and a button. Further, the operation unit 53 may be an interface for connecting the input device to the state determination device 50. In this case, the operation unit 53 inputs the input signal generated in response to the user's input in the input device to the state determination device 50.

The control unit 54 is configured by using a processor such as a CPU (Central Processing Unit) and a memory. By executing the program, the control unit 54 realizes the functions of the data acquisition unit 541, the noise removal unit 542, the data selection unit 543, the reference value setting unit 544, the state determination unit 545, and the display control unit 546. Some of these functions (for example, data acquisition unit 541, noise removal unit 542, data selection unit 543, reference value setting unit 544, and state determination unit 545) do not need to be mounted on the state determination device 50 in advance. , It may be realized by installing an additional application program in the state determination device 50.

The data acquisition unit 541 acquires the detection data. The data acquisition unit 541 may actively acquire the detection data or passively acquire the detection data. When the data acquisition unit 541 actively acquires the detection data, the data acquisition unit 541 acquires the detection data by transmitting the detection data acquisition instruction to the data server 40 via the communication unit 51. When the data acquisition unit 541 passively acquires the detection data, the data acquisition unit 541 acquires the detection data received by the communication unit 51 at a predetermined timing.

The noise removing unit 542 extracts the vibration component inherent in the machine by removing noise from the detection data acquired by the data acquisition unit 541. The noise removed by the noise removing unit 542 is a vibration component and a gravitational acceleration measured when acceleration is not applied when the sensor 11 is installed. The noise removal unit 542 stores the detection data after noise removal (hereinafter referred to as “noise removal detection data”) in the storage unit 55 and outputs the detection data to the data selection unit 543. The noise reduction detection data includes data of the time waveform of the acceleration from which noise has been removed.

When LoRaWAN is used as the communication method as in the present embodiment, since it is very short discrete data, the value of the data varies depending on the acquisition timing. Such discrete data is susceptible to noise. Therefore, in order to analyze the continuous data in the same manner, various noise removal processes are required in the noise removal unit 542.

The data selection unit 543 selects the data in the operating state of the diagnosis target equipment 60 used for the state determination by using the noise removal detection data. The equipment 60 to be diagnosed has an operating state such as a stopped state, an excessive state at the time of starting, and a state at the time of steady operation. The data selection unit 543 selects the data during steady operation as the data of the operating state of the equipment 60 to be diagnosed used for the state determination. At this time, the data selection unit 543 selects the data in the operating state of the diagnosis target equipment 60 used for the state determination by using the noise removal detection data. More specifically, first, the data selection unit 543 performs an operation to compare the probability distribution obtained from the noise removal detection data with the probability distribution obtained from the reference data. Then, the data selection unit 543 selects the noise removal detection data in which the value of the calculation result is less than a predetermined threshold value as the data of the operating state of the diagnosis target equipment 60 used for the state determination. In this embodiment, KL (Kullback-Leibler) divergence is used as an operation for comparing the probability distribution obtained from the detection data with the probability distribution obtained from the reference data.

The reference value setting unit 544 registers the reference data for each operating state of the equipment 60 to be diagnosed in the storage unit 55. The reference value setting unit 544 may register the data for each operating state selected by the user's operation in the storage unit 55 as reference data, or specify the data for each operating state based on the past detection data. Then, it may be registered in the storage unit 55. The reference data for each operating state is used to calculate the probability distribution of reference values (hereinafter referred to as "reference distribution"). The reference data for each operating state is the data of the amplitude waveform of the acceleration in each operating state.

The state determination unit 545 determines the state of the equipment to be diagnosed 60 using the noise removal detection data selected by the data selection unit 543. For example, the state determination unit 545 performs regression prediction using the selected noise removal detection data, and calculates at least a period until it is estimated that an abnormality occurs.

The display control unit 546 controls the display of the display unit 52. For example, the display control unit 546 uses the data stored in the storage unit 55 to generate screen data for display on the display unit 52, and causes the display unit 52 to display the generated screen data.

The storage unit 55 stores equipment information 550, detection data 551, reference data 552, and determination result 553.
The equipment information 550 is information about the equipment 60 to be diagnosed. For example, the equipment information 550 is identification information for identifying the equipment to be diagnosed 60, identification information of the sensor 11 of the vibration detection device 10 installed in the equipment 60 to be diagnosed, and information associated with the installation location. The equipment information 550 may be stored in the storage unit 55 in the form of a table.
The detection data 551 is the noise removal detection data accumulated by the noise removal unit 542. As the detection data 551, the sensor identification information and the detection time information may be associated with the noise removal detection data.

The reference data 552 is data of the amplitude waveform of the acceleration in each operating state registered in advance by the reference value setting unit 544 for each operating state of the equipment 60 to be diagnosed.
The determination result 553 is data showing the result of the state determination by the state determination unit 545. For example, as the determination result 553, there is a result by regression prediction, information on a state at the time of measurement of the equipment 60 to be diagnosed, and the like.

FIG. 4 is a flowchart showing the flow of the state determination process of the state determination device 50 in the present embodiment.
The data acquisition unit 541 acquires the detection data received by the communication unit 51 (step S101). The data acquisition unit 541 outputs the acquired detection data to the noise removal unit 542. The noise removing unit 542 removes noise from the detection data output from the data acquisition unit 541 (step S102). Specifically, first, the noise removing unit 542 calculates the average value of the acceleration based on the time waveform of the acceleration included in the detection data. Then, the noise removing unit 542 excludes noise from the detection data by subtracting the calculated average value of the acceleration (the average value based on the time waveform of the acceleration) from the time waveform of the acceleration included in the detection data. This makes it possible to unify the standard of the absolute value of the amount of change in the amplitude of acceleration. The noise removal unit 542 stores the noise removal detection data in the storage unit 55 and outputs the noise removal detection data to the data selection unit 543. The noise removing unit 542 executes the process of step S102 for each detection data.

The data selection unit 543 calculates a probability distribution for comparison (hereinafter referred to as “comparison distribution”) using the noise reduction detection data (step S103). As a method of calculating the probability distribution, first, the data selection unit 543 creates a frequency distribution (histogram) of a data string (here, a time waveform of acceleration) of noise removal detection data obtained from the noise removal unit 542. The horizontal axis of the histogram is the maximum value from the minimum value of the acceleration amplitude of the target data string, and the width of the horizontal axis is the value obtained by subtracting the minimum value from the maximum value of the acceleration amplitude. Set the step size on the horizontal axis to a sufficiently fine value so that the shape of the histogram does not become rough. Further, when the vertical axis of the histogram is 0, the result diverges infinitely during the calculation in KL divergence. Therefore, when the vertical axis becomes 0, the data selection unit 543 adds a preset offset value of a sufficiently small value. The data selection unit 543 converts the frequency distribution into a probability distribution, setting all the components of the created histogram to 1.

The data selection unit 543 calculates the probability distribution of the reference value using the reference data 552 stored in the storage unit 55 (step S104). In the present embodiment, it is assumed that the data selection unit 543 knows in advance the motion state of the equipment 60 to be diagnosed. In this case, the data selection unit 543 uses the data of the amplitude waveform of the acceleration of the equipment 60 to be diagnosed in the operating state. The data selection unit 543 calculates the reference distribution by performing the same processing as the processing at the time of calculating the comparative distribution. The data selection unit 543 calculates the difference between the comparative distribution and the reference distribution based on the following equation (1) (step S105).

In equation (1), P (i) represents a comparative distribution and Q (i) represents a reference distribution. i represents the amplitude value. The data selection unit 543 determines whether or not the difference between the comparative distribution and the reference distribution is equal to or greater than the threshold value (step S106). When the difference between the comparative distribution and the reference distribution is equal to or greater than the threshold value (step S106-YES), the data selection unit 543 deletes the noise reduction detection data of the comparative distribution in which the difference from the reference distribution is equal to or greater than the threshold value (step). S107). At this time, the data selection unit 543 may also delete the noise removal detection data of the comparative distribution in which the difference from the reference distribution is equal to or larger than the threshold value among the noise reduction detection data stored in the storage unit 55.
On the other hand, when the difference between the comparative distribution and the reference distribution is less than the threshold value (step S106-NO), the data selection unit 543 states the noise removal detection data of the comparative distribution in which the difference from the reference distribution is less than the threshold value. Output to the determination unit 545.

The fact that the difference from the reference distribution is equal to or greater than the threshold value in the process of step S106 means that the noise removal detection data may be detection data of an operating state different from the operating state in which the reference data from which the reference distribution was generated was obtained. Highly sex. On the other hand, if the difference from the reference distribution is less than the threshold value, it is possible that the noise reduction detection data is the detection data in the same operating state as the operating state in which the reference data from which the reference distribution was generated was obtained. expensive. The data selection unit 543 makes a determination in step S106 in order to determine whether or not to use the noise removal detection data for regression prediction.

The state determination unit 545 determines whether or not there is data for regression prediction (step S108). Specifically, the state determination unit 545 determines that there is data for regression prediction when there is a number of noise removal detection data required for performing regression prediction, and is required for performing regression prediction. If there is no number of noise removal detection data, it is determined that there is no data for regression prediction. At least two or more denoising detection data are required to perform regression prediction.

When there is no data for regression prediction (step S108-NO), the state determination device 50 executes the processes after step S101.
When there is data for regression prediction (step S108-YES), the state determination unit 545 performs regression prediction using the number of noise removal detection data required for performing regression prediction (step S109). For example, the state determination unit 545 performs regression prediction by the least squares method. In this case, first, the state determination unit 545 acquires the value having the largest acceleration amplitude as the acceleration peak value using one noise removal detection data, and integrates the noise removal detection data (time waveform of the acceleration with noise removed). Then, the calculated effective value is acquired as the speed RMS value. The state determination unit 545 performs this processing on the noise removal detection data used for the regression prediction.

Then, the state determination unit 545 performs regression prediction using each of the acquired acceleration peak value and RMS value. The state determination unit 545 determines the current state of the equipment 60 to be diagnosed from the acceleration peak value and the RMS value obtained from the latest noise removal detection data used for the regression prediction. Further, the state determination unit 545 determines the state of the equipment 60 to be diagnosed in the future from the straight line obtained by the regression prediction.

For example, the state determination unit 545 determines whether the current state of the equipment 60 to be diagnosed is any of the standard, caution, and danger states. Criteria, caution and danger criteria are appropriately set by the user. In the order of standard, caution and danger, it indicates that an abnormality has occurred in the operating state of the equipment 60 to be diagnosed, or that there is a high possibility that an abnormality will occur. The state determination unit 545 records the determination result in the storage unit 55 (step S110). At this time, the state determination unit 545 stores the acquired acceleration peak value and RMS value, the noise removal detection data, and the determination result in the storage unit 55 in association with each other. The state determination device 50 generates screen data using the equipment information 550 recorded in the storage unit 55 and the determination result 553 according to the operation of the user, and displays the generated screen data on the display unit 52.

5 and 6 are diagrams showing an example of a display screen displayed on the display unit 52 in the present embodiment.
On the display screen of the display unit 52, a selection area 61 for selecting information to be displayed and a display area in which the selected information is displayed are displayed. In the selection area 61 shown in FIGS. 5 and 6, a button for displaying a failure prediction list and a button for displaying a graph are displayed. The information of the failure prediction list is displayed in FIG. 5, and the graph is displayed in FIG. 6. The measurement time shown in FIG. 5 represents the latest time measured at the installation location where the vibration detection device 10 is installed. The validity period shown in FIG. 5 represents the period of data used for displaying the graph. The graph shown in FIG. 6 is displayed using only the data within the valid period.

As shown in FIG. 5, the display control unit 546 displays the state of each diagnosis target equipment 60 on the display unit 52 in a different display mode according to the determination result by the state determination unit 545. Further, when the display control unit 546 estimates the dangerous time by the state determination unit 545, the display control unit 546 displays the dangerous time as the occurrence time. Further, when the button for displaying the graph is selected, the display control unit 546 displays the transition of the vibration of the target equipment 60 to be diagnosed as a graph, as shown in FIG. The transition line 62-1 in FIG. 6 represents the transition from the past to the present within the valid period of the acceleration peak value, and the transition line 62-2 represents the transition from the past to the present within the valid period of the RMS value. Further, the transition line 63-1 in FIG. 6 represents the future transition of the acceleration peak value estimated by the regression prediction of the state determination unit 545, and the transition line 63-2 is the RMS estimated by the regression prediction of the state determination unit 545. It shows the future transition of the value.

According to the state determination device 50 configured as described above, it is possible to easily determine the state of the equipment regardless of the equipment or the diagnosis place. Specifically, the state determination device 50 sets reference data as a reference for a specific operating state, and determines the state from the detection data including information on the vibration detected by the sensor 11 installed in the equipment to be diagnosed 60. The data to be used is selected, and the state of the equipment 60 to be diagnosed is determined using the selected data. In this way, reference data that serves as a reference for a specific operating state is set according to the equipment and the diagnosis location, and necessary data is selected based on the set reference data. Therefore, appropriate data is selected as the data for a specific operating condition at that location. This makes it possible to determine a specific operating state of the equipment regardless of the equipment or the diagnosis location. Therefore, it is possible to easily determine the state of the equipment regardless of the equipment or the diagnosis place.

The state determination device 50 distinguishes between the load fluctuation of the equipment and the detection data during the transient operation by using the detection data of the equipment 60 to be diagnosed during the steady operation. Specifically, the state determination device 50 uses the acceleration data of vibration during steady operation as a reference value, and calculates the probability distribution from the amplitude value of the acceleration data of the reference value. Next, the state determination device 50 uses the acceleration data of the vibration other than the reference value as the comparison value, and calculates the probability distribution from the amplitude value of the acceleration data of the comparison value. The state determination device 50 converts the difference between the reference distribution and the probability distribution of the comparison value into an index value by KL divergence, and excludes the index value in a transient state such as a stop or a start / stop. This distinguishes only the data during steady operation. With this configuration, the operating condition can be estimated without installing a new sensor other than the acceleration sensor (for example, a clamp meter for measuring the load current, a temperature sensor for measuring the machine temperature, etc.). An operation stop state can be mentioned as an operation situation that can be estimated and excluded. The data in the stopped state is data that should be excluded when analyzing the tendency of equipment vibration. When no mechanical vibration is generated due to the stoppage of operation, the acceleration amplitude is small and exhibits aperiodic behavior. On the other hand, when mechanical vibration is generated by operation, the acceleration amplitude is large and has a periodic vibration component according to the rotation frequency of the machine. When the above two patterns are converted into a probability distribution and converted into index values by KL divergence, it can be confirmed that the difference between the two is large.

Conventionally, in the field, skilled workers have confirmed the operating noise and diagnosed from the measured values using a vibration meter (commercially available), judged the condition of the equipment, and adjusted or replaced the parts as necessary. However, such a diagnosis is highly dependent on the experience and knowledge of skilled workers, and it is difficult for all people to make a diagnosis at the same level as skilled workers. In particular, the difficulty of diagnosing the equipment using the vibration data is that the vibration value greatly differs depending on the equipment as described above, and also greatly differs depending on the difference in load and the mounting method of the sensor. In addition, it is considered that the tendency of the vibration value to change differs depending on the cause of the increase in vibration (wear, deterioration of lubricating oil, small scratches). Due to this background, the experience of a skilled worker is required for diagnosis. However, there is concern that diagnostic know-how will disappear due to the aging and retirement of skilled workers, which has become a social problem.
On the other hand, in the state determination system 100 in the present embodiment, the reference data is registered in advance by the reference value setting unit 544, and the difference between the data obtained by the state determination unit 545 and the reference data is obtained. Determine the operating status. This makes it possible to determine the state with high accuracy even if the skilled worker lacks experience and knowledge.

By using a photovoltaic element (combination of a dye-sensitized solar cell and a super capacitor) as a power supply source in the power supply unit 14, the vibration detection device 10 enables power generation under dark indoor lighting. It will be possible to operate without using commercial power or batteries. Further, the vibration detection device 10 can operate using the electric power generated internally, and communicates wirelessly. Therefore, it can be easily installed in the equipment to be diagnosed 60, and the vibration detection device 10 itself can be easily moved.

The state determination unit 545 performs linear regression by the least squares method for noise removal detection data (acceleration peak value, velocity effective value) from an arbitrary time point in the past (for example, a steady operation state after periodic repair and maintenance) to the present. And calculate the slope and intercept of the regression line. Then, the state determination unit 545 estimates an arbitrary point in the future (for example, the acceleration peak value after one month and the velocity effective value) based on the calculated slope of the regression line and the intercept. The state determination unit 545 collates the estimated value by linear regression with the threshold value, and if it exceeds the threshold value, determines that it is caution or dangerous. The display control unit 546 displays the standard, caution, and danger on the display unit 52 in different display modes. This makes it possible to implement preventive maintenance and efficiently formulate maintenance plans before the failure becomes large.

As another pattern of vibration abnormality, when a minute scratch is generated on a mechanical part, a sudden change may occur in the vibration during operation, which may cause an equipment failure in a short time. Conventionally, periodic inspections performed on-site cannot quickly detect the occurrence of such sudden abnormal vibrations. On the other hand, as a method of capturing sudden changes in vibration, there is a method of constantly installing a large number of vibration sensors in the equipment and collecting and monitoring vibration data in real time, in addition to the periodic inspection. However, in order to monitor sudden vibrations that occur infrequently, it is necessary to install a monitoring device and constantly monitor the vibrations, but such a method is costly and is limited to important equipment such as large power generation equipment. Has been done.
On the other hand, in the state determination system 100 in the present embodiment, the power supply unit 14 independently generates power, so that the power supply source is not limited. Furthermore, it can be easily installed in a site where there is no commercial power supply or communication network environment, and it is easy to relocate the installation location. Therefore, the cost can be suppressed.

<Modification example>
The data server 40 and the state determination device 50 may be integrated and configured. A part of the functional part of the state determination device 50 may be mounted on another device. For example, the display control unit 546 and the display unit 52 included in the state determination device 50 may be mounted on another device. In this case, another device including the display control unit 546 and the display unit 52 requests the state determination device 50 for data to be displayed on the display unit 52. The data to be displayed on the display unit 52 is, for example, data necessary for displaying FIGS. 5 and 6.

In the present embodiment, the combination of the dye-sensitized solar cell and the supercapacitor has been described as an example of the power supply unit 14, but any power supply unit 14 can supply the power required for the operation of the vibration detection device 10. It may be. For example, in an environment where a commercial power source can be secured, the power supply unit 14 may supply the electric power supplied from the commercial power source to each functional unit. For example, the electric power supply unit 14 may supply the electric power supplied from the battery to each functional unit.

The noise removing unit 542 may perform an averaging process as a specific method for removing noise. This makes it possible to reduce variations in data values due to acquisition timing. In addition, the averaging process enhances the natural vibrations buried in the noise, making it possible to make up for the lack of data time.

The data selection unit 543 calculates the difference between the probability distribution obtained from the reference data and the probability distribution obtained from the detection data using KL divergence, but other methods may be used. For example, the data selection unit 543 may calculate the difference between the probability distribution obtained from the reference data and the probability distribution obtained from the detection data by using methods such as Pearson distance, relative Pearson distance, and L2 distance. good.

In the above embodiment, the state determination device 50 describes the configuration in which the processing from step S102 to step S107 is repeated every time the detection data is obtained, but the state determination device 50 can obtain a predetermined number or more of detection data. At that point, the process of step S102 may be started. The predetermined number may be, for example, a number capable of regression prediction or a preset number.

In the above embodiment, the data selection unit 543 shows the configuration using the noise reduction detection data in the steady state, but the data selection unit 543 first corresponds to the noise reduction detection data obtained from the plurality of motion states. After classifying into the exercise state group, the data at the time of the operating state of the equipment to be diagnosed 60 used for the state determination may be selected by using the reference data of the classified group. Specifically, the data selection unit 543 is obtained from each of the probability distribution (comparative distribution) obtained from the noise removal detection data and the data in different operating states (transient state such as stop and start, steady operation). Compare each of multiple probability distributions. As a result, the data selection unit 543 classifies which operating state data the noise reduction detection data corresponds to. The data selection unit 543 determines the motion state of the noise removal detection data from the calculation result, and then performs an operation to compare the comparative distribution with the probability distribution (reference distribution) obtained from the reference data in the determined motion state. .. Then, the data selection unit 543 selects the noise removal detection data in which the value of the calculation result is less than a predetermined threshold value as the data in the operating state of the diagnosis target equipment 60 used for the state determination.
With this configuration, it is possible to detect abnormal vibration not only in the steady state but also in the transient state and the stopped state. As a method of classifying into groups having a common probability distribution, a K-means method known as a clustering method may be used, or a trained model obtained by machine learning may be used. For the trained model, for example, when classifying the operating state in advance, the worker may input the operating state and learn the operating state by deep learning or the like as the correct answer.

The above-described embodiment is characterized in that it handles detection data with a small amount of data, but when the amount of data is extremely small or the vibration changes irregularly, it is obtained from the detection data each time measurement is performed. It is assumed that the probability distributions will be different, and the vibration monitoring accuracy may decrease. Therefore, the reference value setting unit 544 may set a reference for the number of data used for processing by the state determination unit. When configured in this way, the state determination unit 545 determines the state of the diagnosis target equipment 60 using the set number of data. More specifically, the reference value setting unit 544 calculates the average value and the variance value of the probability distribution obtained from the detection data. Further, the reference value setting unit 544 sets an appropriate threshold value for the time variation of the average value and the variance value of the probability distribution of the detection data. When the time variation of the average value and the variance value of the probability distribution is larger than the threshold value, it is judged that the data length of the detection data is insufficient for stable monitoring, and the probability distribution is calculated by increasing the amount of detection data. When there is an upper limit to the data length that can be transmitted by the vibration detection device 10, the probability distribution is calculated by combining the detection data divided into a plurality of pieces. In the present embodiment using the probability distribution, the divided detection data does not need to be continuous in time, and the probability distribution can be obtained from a plurality of detection data observed discretely. The reference value setting unit 544 increases the number of detected data observed discretely to 1, 2, ..., And the time variation of the average value and the variance value of the probability distribution of the detected data becomes equal to or less than the set threshold value. The number of data can be selected.
With such a configuration, it becomes possible to improve the vibration monitoring accuracy.

Although the embodiments of the present invention have been described in detail with reference to the drawings, the specific configuration is not limited to this embodiment, and includes designs and the like within a range that does not deviate from the gist of the present invention.

10 ... Vibration detection device, 20 ... Relay device, 40 ... Data server, 50 ... Status determination device, 11 ... Sensor, 12 ... Signal processing unit, 13 ... Communication unit, 14 ... Power supply unit, 51 ... Communication unit, 52 ... Display unit, 53 ... Operation unit, 54 ... Control unit, 55 ... Storage unit, 541 ... Data acquisition unit, 542 ... Noise removal unit, 543 ... Data selection unit, 544 ... Reference value setting unit, 545 ... Status determination unit, 546 … Display control unit

Claims (7)

A reference value setting unit that sets reference data that serves as a reference for either the stopped state, the excessive state at the start, or the state during steady operation in the equipment.
Using a plurality of detection data including information on vibration detected by the sensor installed in the equipment to be diagnosed, the probability distribution representing the existence probability of the amplitude indicated by the information on the vibration included in the detection data and the reference data. An operation is performed to obtain a difference from the probability distribution representing the existence probability of the amplitude in any of the states shown in, and the detection data in which the value of the operation result is less than a predetermined threshold is selected as the data to be used for determining the state of the equipment. Data selection section and
A state determination unit that determines the degree of abnormality of the equipment using the selected data, and
A state determination device comprising. It is further equipped with a noise removing unit that extracts the vibration component inherent in the machine by removing noise from the detection data.
The data selection unit selects data to be used for the state determination from the detection data from which noise has been removed based on the reference data.
The state determination device according to claim 1. The state determination unit performs regression prediction using the selected data, and calculates at least a period until it is estimated that an abnormality occurs.
The state determination device according to claim 1 or 2. The data selection unit performs an operation to compare the probability distribution obtained from the detected data with each of the plurality of probability distributions obtained from the data in a plurality of operating states, and determines the motion state of the detected data from the calculation result. After that, an operation is performed to compare the probability distribution obtained from the detection data with the probability distribution obtained from the reference data in the determined exercise state, and the detection data in which the value of the calculation result is less than a predetermined threshold is determined. Select as the data to be used for
The state determination device according to any one of claims 1 to 3. The reference value setting unit further sets a reference for the number of data used for processing by the state determination unit.
The state determination unit determines the state of the equipment using a set number of data.
The state determination device according to any one of claims 1 to 4. A reference value setting step for setting reference data that serves as a reference for either a stopped state, an excessive state at the start, or a state during steady operation in the equipment.
Using a plurality of detection data including information on vibration detected by the sensor installed in the equipment to be diagnosed, the probability distribution representing the existence probability of the amplitude indicated by the information on the vibration included in the detection data and the reference data. An operation is performed to obtain a difference from the probability distribution representing the existence probability of the amplitude in any of the states shown in, and the detection data in which the value of the operation result is less than a predetermined threshold is selected as the data to be used for determining the state of the equipment. Data selection steps to be performed and
A state determination step for determining the degree of abnormality of the equipment using the selected data, and a state determination step.
A state determination method having. A reference value setting step for setting reference data that serves as a reference for either a stopped state, an excessive state at the start, or a state during steady operation in the equipment.
Using a plurality of detection data including information on vibration detected by the sensor installed in the equipment to be diagnosed, the probability distribution representing the existence probability of the amplitude indicated by the information on the vibration included in the detection data and the reference data. An operation is performed to obtain a difference from the probability distribution representing the existence probability of the amplitude in any of the states shown in, and the detection data in which the value of the operation result is less than a predetermined threshold is selected as the data to be used for determining the state of the equipment. Data selection steps to be performed and
A state determination step for determining the degree of abnormality of the equipment using the selected data, and a state determination step.
A computer program that lets your computer run.

Images