TWI621951B - Machine sorting device - Google Patents

Abstract

The data acquisition unit (101) acquires machine classification index data from the machine classification index database (201). The Classification Index Quantification Department (102) converts the qualitative data contained in the machine classification index data into quantitative data indicating the similarity between the qualitative data. The machine classification unit (103) uses quantitative data to classify devices according to machine units.

Description

Machine sorting device

The present invention relates to a machine sorting apparatus for classifying equipment according to the machine unit constituting the apparatus.

It is useful to classify devices having the same characteristics into a plurality of devices in a multi-element environment in which elevators and air conditioners are equivalent. For example, in the conventional system described in Patent Document 1, the elevators are classified into the same characteristics in the control of lighting and air conditioning for energy saving of building equipment. In the system described in Patent Document 1, the elevator operation information is used, and the flow rate of the person in the shared unit, the number of people in the room, and the room rate are modeled according to the day of the week and each time zone, and the schedule is controlled. Here, in a building where the elevator operation information cannot be obtained, the classification of the building is performed in order to follow the analysis results of similar buildings having the same characteristics.

[Prior patent documents] [Patent Literature]

Patent Document 1: Japanese Laid-Open Patent Publication No. 2005-104635

In the case of analyzing a fault or an abnormality in a machine constituting a device such as an elevator or an air conditioner, a plurality of sets are performed by machines of the same type and the same characteristics. Classification and analysis of equipment, machines, and detection accuracy of expected failures or anomalies compared to analysis of only one type of equipment.

However, in the conventional method, since it is classified according to the equipment unit of the elevator, even if the machine having the different characteristics is the same type and the same feature according to the equipment unit, there is a problem that the machine cannot be classified. For example, in the case where the door device constituting the elevator A and the door device constituting the elevator B are of the same type but different in characteristics, when the elevator A and the elevator B are used as the equipment of the elevator, the same feature is determined. The machine of one side is classified into the same feature.

In addition, in the conventional method, the index used for the classification is the elevator operation information, the use of the elevator, and the scale. However, in the case of analyzing the fault or the abnormality, the fault history, the equipment, the installation environment of the equipment, and the machine are considered. Classification is carried out with a lot of information such as information replacement, and the classification accuracy is expected to increase. These information are not necessarily quantitative information consisting of numerical values, but also qualitative information containing textual information. However, in the past, in the case of classification based on qualitative data, the evaluation of the similarity of different qualitative data is not considered. As a result, the cause of the abnormality cannot be sufficiently analyzed, and there is a problem that the accuracy of the abnormality detection is lowered.

The present invention has been made to solve the above problems, and an object thereof is to provide a machine sorting device capable of performing analysis of a failure, an abnormality, or the like of a machine with high precision.

The machine classification device of the present invention includes: a data acquisition unit that acquires machine classification index data that is information unique to each device, and the inherent resource The information system is obtained from the monitoring data of each machine of a plurality of devices consisting of one or a plurality of machines; the classification index quantification unit converts the qualitative data contained in the machine classification index data into qualitative data. Quantitative data of the similarity between the two; and the machine classification department uses the quantitative data to classify the equipment according to the machine unit.

The machine sorting device of the present invention is configured to convert qualitative data contained in the machine classification index data into quantitative data indicating the similarity between the qualitative data, and then use the quantitative data to classify the equipment according to the machine unit. Thereby, the analysis of the malfunction or abnormality of the machine can be performed with high precision.

100, 100a‧‧‧ machine sorting device

101‧‧‧Information Acquisition Department

102‧‧‧Classification Indicators Division

103, 103a‧‧‧Machine Classification Department

104‧‧‧Characteristics Transformation Department

200‧‧‧Data collection management device

201‧‧‧ Machine Classification Indicators Database

300‧‧‧Network

400‧‧‧Monitoring objects

Fig. 1 is a configuration diagram of a machine sorting device according to a first embodiment of the present invention.

Fig. 2 is an explanatory view showing an example of maintenance performance data of the machine classification device according to the first embodiment of the present invention.

Fig. 3 is a view showing a hardware configuration of a machine sorting device according to a first embodiment of the present invention.

Fig. 4 is a flowchart showing the machine sorting process of the machine sorting device according to the first embodiment of the present invention.

Fig. 5 is an explanatory view showing an example of quantitative information using the machine sorting device according to the first embodiment of the present invention.

Fig. 6 is a flow chart showing the process of machine classification processing in the case where the device classification device of the first embodiment of the present invention has prior information of qualitative data similarity. Figure.

7A, 7B, and 7C are explanatory views showing an example of prior information using the device classification device according to the first embodiment of the present invention.

Fig. 8 is an explanatory view showing an example of classification of quantitative information of the machine classification device according to the first embodiment of the present invention.

Fig. 9 is an explanatory view showing the feature amounts of the respective machines of the machine sorting device according to the first embodiment of the present invention.

Figure 10 is a configuration diagram of a machine sorting device according to a second embodiment of the present invention.

Hereinafter, in order to explain the present invention in more detail, embodiments of the present invention will be described with reference to the accompanying drawings.

First embodiment

Fig. 1 is a configuration diagram of a monitoring system including the machine sorting device 100 of the present embodiment.

In the monitoring system shown, the device sorting device 100 is connected to the data collection management device 200, and the data collection management device 200 is connected to the monitoring target 400 via the network 300.

The machine classification device 100 includes a material acquisition unit 101, a classification index quantification unit 102, and a device classification unit 103. The data acquisition unit 101 is a processing unit that acquires the device classification index data from the device classification index database 201 managed by the data collection management device 200. The classification index quantification unit 102 is a processing unit that converts the qualitative data contained in the machine classification index data into quantitative data. The machine classifying unit 103 uses the quantitative data generated by the classification index quantifying unit 102, and presses A processing unit that performs device classification according to the machine unit.

The data collection management device 200 collects and monitors the monitoring data from the monitoring target 400 and stores it as the device classification index database 201. The monitoring data stored in the machine classification index database 201 means data directly or indirectly obtained from the monitoring object 400 based on data (for example, maintenance performance data) created by the inspection of the maintenance person of the monitoring object 400 or equipment information. As an example of the machine classification index data stored in the machine classification index database 201, FIG. 2 shows an example of the maintenance performance data by the elevator.

In the second figure, an example of the maintenance performance data obtained from the inspection of the maintainer of one machine of one machine or the equipment information is shown. In the maintenance performance data example, as an example of the data item, the device ID, model ID, machine ID, installation area, operator name, maintenance work contents, and presence or absence of an abnormality are described. The value of these data items is an example. The data items can be changed to store items of maintenance performance data collected from actual equipment and equipment. Moreover, if the equipment and the machine can be distinguished, the data of a plurality of devices and machines can be collected into one table. Furthermore, if the equipment and the machine can be assigned, the data of one of the machines can be divided into a plurality of sheets. In addition, it is also possible to manage the maintenance performance data of the maintenance work in a different manner from the maintenance work in the normal time, the maintenance work in the event of a malfunction or an abnormality. That is, as the machine classification index data stored in the machine classification index database 201, any information is available as long as it is information inherent to the machine.

The monitoring target 400 is, for example, a device composed of one or a plurality of machines of an elevator or an air conditioner. The monitoring target 400 is assumed to have two or more devices consisting of the same device. Alternatively, the monitoring object 400 may be directly connected to the data collection management device 200 without being connected to the network 300. Can also The data collection management device 200 and the device classification device 100 are connected to each other regardless of the connection method between the monitoring target 400 and the data collection management device 200.

Fig. 3 is a block diagram showing the hardware configuration of the machine sorting apparatus of the present embodiment. Fig. 3 shows an example in which the machine sorting device 100 and the data collecting management device 200 of Fig. 1 are formed on a single piece of hardware. The machine sorting device 100 and the data collection management device 200 include a processor 11, a memory 12, a communication I/F (interface) device 13, a memory device 14, and an output device 15. The processor 11 is a processor for implementing the functions of the machine classification device 100 and the data collection management device 200. The program memory of the various programs corresponding to the functions of the memory class 12 device classification device 100 and the data collection management device 200 is a working memory used for data processing by the processor 11 and a memory for expanding the signal data. A memory unit such as a ROM or a RAM used. The communication I/F device 13 is an external communication interface with the network 300 or the like. The memory device 14 is a memory device for storing various materials or programs. The output device 15 is a device for outputting a processing result to the outside.

The processing unit processor 11 executed by the data acquisition unit 101, the classification index quantification unit 102, and the device classification unit 103 in the first drawing reads and executes the program stored in the memory 12. The data stored in the machine classification index database 201 is stored in the memory device 14 from the monitoring target 400 via the network 300 via the communication I/F device 13. The processing result of the machine classifying unit 103 is stored in the memory device 14 as needed, and is output to the outside by the output device 15. Further, the machine sorting device 100 and the material collection management device 200 may be configured on different hardware.

Next, the operation of the machine classification device 100 of the present embodiment will be described.

The data collection management device 200 inputs the machine classification index data obtained from the monitoring target 400 to the machine classification index database 201 continuously or intermittently. The machine classification device 100 acquires and processes the machine classification index data from the machine classification index database 201. Fig. 4 is a flow chart showing the processing of the machine sorting device 100.

First, the material acquisition unit 101 acquires the device classification index data from the device classification index database 201 (step ST1). In addition, in the case where the machine classification index data includes a plurality of data items, the flowchart of FIG. 4 is executed for each data item. For example, as an indicator of the machine classification index data, when the machine ID is input, a list of the classified machine IDs is output. The form of the list is not limited. For example, there is a classification ID assigned to each category, and the category ID corresponding to each machine ID is stored in the output of a list of tables. Further, as an example of another list, there is also a method of creating a file for each category and storing the machine ID belonging to the category in the file.

The classification index quantification unit 102 converts the qualitative data contained in the device classification index data acquired by each device into quantitative data composed of numerical values in a form in which similarity can be determined. In step ST2, it is judged whether or not the quantitative data is based on whether the input machine classification index data is a numerical value, and the subsequent processing is branched. In the case where the step ST2 is quantitative data (in step ST2: YES), the classification index quantifying unit 102 ends the processing. In other words, the machine classification index data input by the classification index quantification unit 102 is directly output to the machine classification unit 103. On the other hand, it is not quantitative in step ST2. In the case of the material (at step ST2: NO), the processing of step ST3 is performed. In step ST3, as the similarity between the qualitative data, the distance between the qualitative data is calculated, and each data is assigned a value corresponding to the distance, thereby serving as quantitative data. The distance between the qualitative data is calculated based on the string analysis method such as the hierarchical cluster analysis of n-gram, and the value corresponding to the distance is used as the quantitative data. Here, the qualitative information is more affected by the distance because it indicates the classification of the more forward text, because the smaller the classification of the text, the smaller the influence on the distance, the smaller the position of the text and the distance. When the relationship of influence is known, it is also possible to perform weighting or the like on a character having a large influence on the distance when calculating the distance. For example, in the case of the character string of the machine ID, if the first half indicates that the major version number is changed, and the second half indicates that the version number is changed minorly, the effect of the preceding character string on the distance is increased.

In Fig. 5, an example of quantitative information on the conversion of qualitative data is shown. In the fifth diagram, as a simple example, the name of the device ID indicates a quantitative example of the case where the main change version number and the minor change version number of the machine are connected by the hyphen "-". In this quantitative data example, the machine IDs of AAA-01, AAA-02, and AAA-03 are similar because the major change version numbers are the same, and only the minor change version numbers are different. Machine IDs BBB-01, BBB-02, AAA-01, AAA-02, and AAA-03 are assigned because they have different version numbers.

In the case where the machine classification index data is quantitative data in step ST2 or after the processing in step ST3, the machine classification unit 103 sorts the values input according to the multivariate analysis method or the mechanical learning method, that is, in multivariate analysis or the like. Each of the devices having a close feature amount (step ST4). About the specific classification example will be Said.

On the other hand, in the case where the prior information is known to be similar to each other in qualitative information, the similarity of the prior information can also be applied. In the case of prior information, it may be distributed only in one part of the qualitative information. For example, among the machine IDs, only the similarity degree of the main change version number or the like is assigned. Moreover, in the prior information, weighting rules for each text position of the qualitative data may also be assigned. For example, in the advance information, the weighted percentage of the major change version number and the minor change version number is assigned. In addition, qualitative information that does not transform into quantitative data can be provided as prior information.

Figure 6 shows the classification index quantification process for the case of prior information. The same processing as in the fourth drawing is attached with the same step number. The classification index quantification unit 102 first determines whether or not the quantitative data is determined in step ST2, and in the case where the data is not quantitative (in step ST2: NO), it is determined whether or not there is a similarity with respect to the input machine classification index data. The prior information of the degree is branched, and the subsequent processing is branched (step ST5). In step ST5, in the case where there is no prior information of the similarity (in step ST5: NO), the processing of step ST3 is carried out in the same manner as in the fourth drawing. In the case of the prior information of the degree of similarity (at step ST5: YES), in the processing of step ST6, the value corresponding to the degree of similarity of the provided prior information is assigned. Here, in the case where the similarity of the prior information is not quantitative information but qualitative data, as in step ST3, as the qualitative similarity between the qualitative data, the distance between the qualitative data is calculated, and the allocation of each data is determined. The value of the distance, by this, as quantitative information. The distance between the qualitative data is calculated based on the method of calculating the distance between words such as the hierarchical analysis of n-gram, and the value corresponding to the distance is used as the quantitative data. Qualitative capital The method of the distance between the materials uses a method different from that of step ST3.

An example of prior information is shown in Figure 7. An example of the similarity of the specified qualitative data is shown in Fig. 7A. Fig. 7A is an example of the similarity of the three characters in the front of the machine ID, indicating that the machine ID is AAA and the BBB machine system is relatively similar, and the machine ID CCC is similar to the machine ID AAA and BBB. The machine is relatively low. Further, an example of the weighting rule for specifying the position of each character of the qualitative data is shown in Fig. 7B. In the seventh embodiment, the weighting rule of each character position of the device ID is set to 10 in order to increase the weight of the first to third characters of the device ID, and the weighting ratio of the fifth to sixth characters is the first to the third. 3 Example where the text is light and set to 1. In addition, since the fourth word is a hyphen, the weighting rule is excluded. Further, an example of designating qualitative data that is not converted into quantitative data is shown in Fig. 7C. Fig. 7C shows a case where the device ID is not quantified. Figures 7A, 7B, and 7C are examples of information specified, and the method of providing information may be changed.

As another example, the free document described as a result of the maintenance work of the machine may be utilized as the machine classification index data. For example, the morpheme analysis extracts the words described in the results of the maintenance work, such as "abnormal", "disposed", "cause is event A", etc., and assigns a similar morpheme to the paper. Values, etc.

In the machine classifying unit 103, each device is classified into each device having similar characteristics, and therefore the quantitative data converted by the classification index quantifying unit 102 of the plurality of machine components is input, and each of the quantitative data is classified into a near value of the quantitative data. machine. Quantitative data can also be entered in only one data item, or multiple data items can be entered together. In step ST4 of Fig. 4 or Fig. 6, the system can also be utilized. A general multivariate analysis technique such as a hierarchical cluster analysis such as a tree diagram or a non-hierarchical cluster analysis such as the k-means method, or a general mechanical learning method such as a support vector machine. An example of classification is shown in Fig. 8.

Fig. 8 is a classification example of quantitative data, and inputs quantitative information of three machines of a plurality of data items, and is used as a feature quantity space in a multivariate analysis method such as principal component analysis, in a two-dimensional scatter map. The feature quantity 1 and the feature quantity 2 are represented in the mode. In Fig. 8, the feature magnitude 801 and the feature magnitude 802 are grouped into one category 804 because the distance on the scatter plot is close. Since the feature amount value 803 is far from the feature amount value 801 and the feature amount value 802 on the scattergram, it indicates that the classification 805 is different from the classification 804. As a method classified in this manner, it is also possible to use the nearest neighbor method which calculates the distance between the feature magnitudes 801, 802, and 803 and sorts according to the threshold of the distance, or k- which determines the number of classifications in advance. General cluster analysis method such as means method.

As a use of the present invention, the failure and abnormality analysis of the machine. For example, in the case of predicting the future failure period in order to establish a maintenance plan for the machine, there is information from the machine, and the probability of future failure (risk of failure) is predicted from the frequency of statistical failure or the tendency of deterioration, and it is estimated that The method of maintenance time.

Here, machines having similar characteristics are highly likely to be similar in frequency or deterioration tendency before failure, so it is also useful to classify similar features for machine classification, which is also useful for predicting the risk of failure. Classification into more similar machines, which leads to an improved prediction accuracy of the risk of failure. The data used to calculate the risk of failure can also be used with the machine classification device 100 of the present embodiment. The information is the same and other materials can be used. When estimating the risk of failure, the risk of failure can be predicted according to the machine unit, but the risk of failure of a plurality of machines such as the correlation of a plurality of machines constituting the equipment can be comprehensively judged, and the risk of failure according to the equipment unit can be predicted.

Next, the effects of the first embodiment will be described. The ninth figure is an example of the characteristic quantity of each machine. The two machines of the machine 1 and the machine 2 collect data from the three devices of the device a, the device b, and the device c, and calculate the feature amount, and then express it in the respective devices. Two-dimensional scatter plots made by two feature quantities. The device feature amount of the machine 1 is indicated by 901, and the device feature amount of the machine 2 is indicated by 902.

In the conventional method, since it is a classification of equipment units, in the case where the equipment a, the equipment b, and the equipment c are classified into devices having the same characteristics, the same classification is made regardless of the characteristics of the machine 1 and the machine 2. On the other hand, in the first embodiment, for example, in the device feature amount 901 of the device 1, when the device a and the device b are set to one type, and the device c is set to another type, the device 2 may be used. The device feature amount 902 sets the device a and the device c into one classification, and sets the device b to another classification or the like according to the machine unit.

By classifying each machine having the same characteristics, it is expected that the accuracy of the failure of the machine, the accuracy of the failure, and the accuracy of the abnormality detection can be improved. In addition, by classifying into similar machines, if a machine finds a fault, an abnormality, etc., the machine having the same characteristics is extracted and maintained, thereby preventing malfunctions and abnormalities in other machines. It is expected that the maintenance of the time schedule according to the maintenance work of each machine will be more efficient. For example, when the torque of the door opening/closing motor of the elevator A is lowered and the door is closed, it is checked whether the door opening and closing motor of the other elevator having the same characteristics has a torque reduction symptom and is protected. By raising, you can expect to reduce failures or accidents. As another example, when it is detected that the torque of the door opening/closing motor of the elevator A is lowered, the door opening and closing motor of the other elevator having the same characteristics may have a torque reduction. However, if it is not necessary to immediately process it, The maintenance work is appropriately discharged into the time course, whereby the work efficiency can be expected.

As described above, the device classification device according to the first embodiment includes a data acquisition unit that acquires device classification index data that is information unique to each device, and the unique information is from each The monitoring data of each machine of the plurality of devices formed by the machine or the plurality of machines; the classification index quantification unit converts the qualitative data contained in the machine classification index data into a quantitative degree indicating the similarity between the qualitative data. The data classification unit and the machine classification department use the quantitative data to classify the equipment according to the machine unit. Therefore, it is possible to analyze the failure or abnormality of the machine with high precision.

Second embodiment

In the first embodiment, the machine classification unit 103 performs classification of each device based on the machine classification index data quantified by the classification index quantification unit 102. In contrast, before inputting the machine classification index data quantified by the classification index quantifying unit 102 to the machine classification unit 103, the model classification unit 103 may be converted into a feature amount in order to emphasize the difference in characteristics of the respective devices. The amount of each machine is classified and described in the second embodiment.

The purpose of the conversion into the feature quantity is to make the difference between the machines clear when the machine classification is performed from a plurality of machine classification index data. In the case of machine classification from only one machine classification indicator data, since similar machine classification index data is assigned a near value when converted into quantitative data, only Classified according to the value of the machine classification indicator data. However, in the case where the machine classification is performed from a plurality of machine classification index data, even if the machine classification index data is a near value machine, there is a case where the other machine classification index data has a far value. In this case, since the value of the machine classification index data is intact, the difference between the machines cannot be clearly known, and the machine classification cannot be performed correctly. Therefore, it is possible to obtain a more accurate classification of the machine by obtaining a feature amount that makes the difference between the machines clear from a plurality of machine classification index data. For example, there are a plurality of machine classification index data such as a model ID and a setting area, and a distance according to the MT method is used as a method of the feature amount. It may be a general method such as a multivariate analysis method such as a feature quantity, for example, a principal component of the principal component analysis, a regression coefficient and an error in the regression analysis, and a similarity degree according to the pattern matching method.

Fig. 10 is a configuration diagram of a monitoring system to which the machine classification device 100a of the second embodiment is applied. The device classification device 100a of the second embodiment includes a material acquisition unit 101, a classification index quantification unit 102, a device classification unit 103a, and a feature amount conversion unit 104. Here, the data acquisition unit 101 and the classification index quantification unit 102 are the same as in the first embodiment. The feature amount conversion unit 104 is a processing unit that converts the machine classification index data quantified by the classification index quantifying unit 102 into a feature amount. The machine classification unit 103a is a processing unit that performs machine classification using the feature amount converted by the feature amount conversion unit 104. Further, in Fig. 10, the data collection management device 200, the network 300, and the monitoring target 400 are the same as those in the first embodiment shown in Fig. 1.

In the machine classification device 100a configured as described above, the feature amount conversion unit 104 converts the quantitative data into the feature amount before inputting the quantitative data generated by the classification index quantifying unit 102 to the device classification unit 103a. machine The device classifying unit 103a acquires the feature amount from the feature amount converting unit 104, and classifies the devices having similar feature values into devices having similar characteristics. In addition, the feature amount conversion unit 104 may not convert all of the quantitative data generated by the classification index quantifying unit 102 into a feature amount, and only convert a part of the data. When only a part of the conversion is performed, the machine classification unit 103a classifies both the converted feature quantity and the quantitative data.

As described above, the device classification device according to the second embodiment includes the data acquisition unit, and acquires device classification index data which is information unique to each device, and the unique information is derived from each. The monitoring data of each machine of the plurality of devices formed by the machine or the plurality of machines; the classification index quantification unit converts the qualitative data contained in the machine classification index data into a quantitative degree indicating the similarity between the qualitative data. The feature quantity conversion unit converts the quantitative data into a feature quantity indicating a difference in characteristics of each machine, and the machine classification part, which is a machine having a similar feature quantity as a machine having a similar feature quantity, and classifying the device according to the machine unit. Therefore, it is possible to perform analysis of malfunction or abnormality of the machine with higher precision.

Further, the present invention can be modified in any combination of the embodiments or any constituent elements of the respective embodiments within the scope of the invention, or any constituent elements are omitted in the respective embodiments.

[Industrial Applicability]

As shown above, the machine sorting apparatus of the present invention classifies each apparatus for a plurality of apparatuses, and is suitable for use in a device in which a plurality of units of different types of elevators or air conditioners are present in different environments. .

Claims (2)

A machine classification device comprising: a data acquisition unit that acquires machine classification index data that is information unique to each device, the inherent information being from a plurality of devices each composed of one or more machines The classification information of the various machines; the classification index quantification department converts the qualitative data contained in the machine classification index data into quantitative data indicating the similarity between the qualitative data; and the machine classification department uses The quantitative data is classified according to the machine unit. A machine classification device comprising: a data acquisition unit that acquires machine classification index data that is information unique to each device, the inherent information being from a plurality of devices each composed of one or more machines The classification information of the various machines; the classification index quantification department converts the qualitative data contained in the machine classification index data into quantitative data indicating the similarity between the qualitative data; the feature quantity conversion unit The quantitative data is converted into a feature quantity indicating a difference in characteristics of the respective machines, and the machine classification unit is a machine having similar feature quantities, and the device is classified according to the machine unit.