Analysis system and method

Abstract

An analysis system (1) analyzes data related to equipment (60) and comprises a processor (11) and a display (50). The processor (11) acquires data related to the life cycle of first equipment (60). The life cycle includes a plurality of processes. The data related to the life cycle includes process data related to the processes of the life cycle. The processor (11) uses the process data and a first diagnostic result, which was output from a first equipment diagnostic model (31) for diagnosing the first equipment (60), to analyze, for each process, a factor level by which the process data contributes to the first diagnostic result, and displays an analysis result on the display (50).

Description

Analysis System and Method

[0001] The present disclosure relates to an analysis system and method.

[0002] Conventionally, various device diagnosis models for diagnosing devices have been developed. For example, Japanese Patent Application Laid-Open No. 2022-105916 (Patent Document 1) describes a method for diagnosing a device using a machine learning model, a method for replacing an updated model obtained by retraining a machine learning model with the original machine learning model to prevent degradation of the accuracy of the machine learning model, and a method for reducing the update time of the updated model.

[0003] Japanese Patent Application Laid-Open No. 2022-105916

[0004] Incidentally, a device is placed in an actual operating environment after going through a plurality of processes such as a development process and a manufacturing process. After being placed in the actual operating environment, the device is subject to maintenance as necessary. As such, the life cycle of a device includes various processes from birth to disposal.

[0005] However, conventionally, the relationship between the cause of a diagnosis result output from a device diagnosis model and a process has been unclear. For this reason, when some diagnosis result is obtained from a device diagnosis model, a user cannot identify which process the main cause of the diagnosis result exists in. Therefore, there is a problem that the user cannot utilize the diagnosis result for improving the process of the device.

[0006] An object of the present disclosure is to enable a user to grasp the relationship between a diagnosis result output from a device diagnosis model and a plurality of processes in the life cycle of a device.

[0007] One aspect of the present disclosure is an analysis system for analyzing data relating to equipment, comprising a processor and a display, wherein the processor acquires data relating to the lifecycle of a first piece of equipment, the lifecycle includes a plurality of processes, the lifecycle data includes process data relating to the processes of the lifecycle, and the processor uses a first diagnostic result output from a first piece of equipment diagnostic model for diagnosing the first piece of equipment and the process data to analyze the degree to which the process data contributes to the first diagnostic result for each process, and displays the analysis results on the display.

[0008] Another aspect of the present disclosure is a method for analyzing data relating to an apparatus, the method comprising the step of obtaining data relating to the lifecycle of a first apparatus, which is performed by a computer, wherein the lifecycle comprises a plurality of processes, and the data relating to the lifecycle comprises process data relating to the processes of the lifecycle, and the method further comprises the step of using a first diagnostic result output from a first apparatus diagnostic model for diagnosing the first apparatus and the process data, which is performed by a computer, to analyze the degree to which the process data contributes to the first diagnostic result for each process, and displaying the analysis results on a display.

[0009] According to this disclosure, users can understand the relationship between diagnostic results output from the device diagnostic model and multiple processes in the device's lifecycle.

[0010] This figure shows the overall configuration of the analysis system related to this embodiment. This figure conceptually shows the configuration of the analysis device shown in Figure 1. This figure conceptually shows the lifecycle data catalog model. This figure conceptually shows the lifecycle data catalog. This figure shows an example of ID correspondence data stored in the ID correspondence storage unit. This block diagram shows the procedure for calculating analysis results based on lifecycle data and performance estimates. This flowchart shows the procedure for collecting process data by ID. This figure shows an example of an information management screen displayed on the display. This figure shows an example of an information upload screen displayed on the terminal device. This flowchart shows the procedure for determining performance estimates. This figure shows an example of a graph created when normal judgment data and abnormal judgment data are classified based on actual operation performance estimates and test operation performance estimates. This figure conceptually shows the procedure for generating a regression model used to calculate the degree of abnormality. This flowchart shows the procedure for calculating and displaying the degree of abnormality. This figure shows an example of a graph of the degree of abnormality displayed on the display. This figure shows some examples of analysis information that can be identified by using the analysis system.

[0011] This embodiment will be described in detail below with reference to the drawings. Note that the same or corresponding parts in the drawings are denoted by the same reference numerals, and their descriptions will not be repeated.

[0012] Figure 1 shows the overall configuration of the analysis system 1 according to this embodiment. The analysis system 1 includes an analysis device 100 and a display 50. The analysis device 100 may be a server. The analysis system 1 may also include an input device 40. The analysis system 1 may also include one or more devices 60 connected via a network NW such as the Internet, and one or more terminal devices 70 connected via the network NW. Thus, the analysis system 1 may also be a communication system.

[0013] The input device 40 may include a keyboard and a mouse. The display 50 may be, for example, a liquid crystal display or an organic EL display. The user inputs various information to the analysis device 100 using the input device 40. The analysis device 100 receives the information input from the input device 40.

[0014] The equipment 60 may be, for example, electrical equipment such as an air conditioner and a water heater. The analysis device 100 diagnoses the equipment 60 based on the operating data output from the equipment 60. The equipment 60 is deployed to the actual operating environment after going through various processes such as parts procurement, development, and other processes. After being deployed to the actual operating environment, the equipment 60 is subject to maintenance as needed. Thus, the lifecycle of the equipment 60 includes various processes from its creation to its disposal. The terminal device 70 transmits data related to the processes included in the lifecycle of the equipment 60 to the analysis device 100.

[0015] Hereinafter, data relating to such processes will be referred to as "process data," and a series of "process data" will be collectively referred to as "lifecycle data." Process data may include, for example, data relating to the components of equipment 60, data relating to the equipment 60 itself, test data used when developing equipment 60, data on the transportation route used when transporting equipment 60 to the actual operating environment, and data showing the results of maintenance of equipment 60.

[0016] The terminal devices 70 may be held by the person in charge of the process. For example, the person in charge of parts procurement, the development person, and the transportation person may each hold a terminal device 70. In other words, the analysis system 1 may have terminal devices 70 for each process. The analysis system 1 may have terminal devices 70 for each process and for each piece of equipment 60. For example, if the number of pieces of equipment 60 to be analyzed is 3 and the number of processes for each piece of equipment 60 is 10, the number of terminal devices 70 may be 30.

[0017] Figure 1 shows an analysis system 1 configured such that terminal devices 70 are provided for each process and each piece of equipment 60. In this embodiment, the equipment 60 and the process data corresponding to the equipment 60 are referred to as an "analysis target set". Figure 1 shows analysis target sets DS1 and DS2 as examples of multiple analysis target sets.

[0018] Furthermore, the terminal device 70 may transmit process data corresponding to each of the multiple processes to the analysis device 100. For example, if the number of devices 60 to be analyzed is 3 and the number of processes in each device 60 is 10, the number of terminal devices 70 may be the same as the number of devices 60. In this case, each terminal device 70 may transmit all the process data of the corresponding device 60 to the analysis device 100. Thus, in this disclosure, the number of terminal devices 70 does not have to be uniformly determined by the number of devices 60 and the number of processes.

[0019] However, in the following description, an analysis system 1 will be explained as an example, in which terminal devices 70 are provided for each process and each piece of equipment 60.

[0020] The analysis device 100 comprises an arithmetic unit 10 and storage devices 20 and 30. The arithmetic unit 10 comprises a processor 11, a memory 12, and an input / output interface 13. The arithmetic unit 10 may be, for example, a microcomputer. The processor 11 is an arithmetic unit that executes various processes according to various programs. In this embodiment, it is assumed that the processing of the arithmetic unit 10 described below is also understood as the processing of the processor 11.

[0021] The processor 11 may include, for example, at least one of the following: CPU (Central Processing Unit), FPGA (Field Programmable Gate Array), GPU (Graphics Processing Unit), and MPU (Multi Processing Unit).

[0022] Memory 12 includes memory for storing programs executed by the processor 11, and working memory. Memory 12 includes volatile memory such as DRAM (dynamic random access memory) and SRAM (static random access memory), non-volatile memory such as ROM (Read Only Memory) and flash memory. Memory 12 may also be an HDD (Hard Disk Drive) and an SSD (Solid State Drive), etc.

[0023] The storage devices 20 and 30 may be non-volatile memory such as HDDs and SSDs. The storage device 20 stores the database 201. The database 201 includes a catalog model database 21, a process database 22, and an ID correspondence relationship database 23.

[0024] The catalog model database 21 stores a lifecycle data catalog model that includes classification criteria for process data. The process database 22 stores process data. When the computing unit 10 acquires process data from the terminal device 70, it stores the process data in the process database 22 for each device 60 according to the classification criteria of the lifecycle data catalog model.

[0025] The ID correspondence database 23 contains information for identifying process data by set of data to be analyzed. The computing unit 10 uses the information registered in the ID correspondence database 23 to store process data in the process database 22, for example, separately for sets of data to be analyzed, DS1 and DS2.

[0026] The storage 30 stores an instrument diagnostic model 31. The instrument diagnostic model 31 is an example of a diagnostic model for diagnosing the instrument 60. The storage 30 stores multiple instrument diagnostic models 31, each prepared for a different set of analysis targets. That is, the storage 30 stores at least an instrument diagnostic model 31 corresponding to the analysis target set DS1 and an instrument diagnostic model 31 corresponding to the analysis target set DS2. However, the instrument diagnostic model 31 may also be stored in the storage 30 for each model of instrument 60. The instrument diagnostic model 31 may be used in common for all analysis target sets. The instrument diagnostic model 31 may be a rule-based diagnostic model, such as those found in expert systems.

[0027] The device diagnostic model 31 may be, for example, an estimation model trained by machine learning. In the following description, this embodiment will be explained assuming that the device diagnostic model 31 is an estimation model trained by machine learning.

[0028] Machine learning may employ techniques such as deep learning. The equipment diagnostic model 31 may be a model trained by supervised learning. In that case, for example, the dependent variable may be information regarding the presence or absence of abnormalities in the equipment 60, and the independent variable may be time-series operating data output from the equipment 60 when it is in operation. The dependent variable may not be information regarding the presence or absence of abnormalities in the equipment 60, but rather information that helps the user estimate the likelihood of failure.

[0029] For example, if the equipment 60 is an air conditioner including a compressor, the explanatory variable may be time-series data relating to the operating frequency of the compressor, and the dependent variable may be the power consumption or air conditioning capacity (cooling capacity or heating capacity) of the equipment 60. The explanatory variable can be any information relating to the mechanical operation of the equipment 60. For example, instead of the operating frequency, the temperature of the components of the equipment 60, the pressure inside the equipment 60, etc., may be adopted as explanatory variables.

[0030] More specifically, the objective variable may include information regarding compressor wear, fracture, or oil leaks. Alternatively, the objective variable may include information regarding the presence or absence of abnormalities in the refrigerant circuit. More specifically, the objective variable may include information regarding misconnections or refrigerant leaks in the refrigerant circuit.

[0031] The device diagnostic model 31 may be located in a cloud connected to the analysis device 100 via a network NW. In this case, the computing device 10 may access the device diagnostic model 31 via the network NW.

[0032] The computing unit 10 inputs the operating data acquired from the equipment 60 into the equipment diagnostic model 31 corresponding to the equipment 60. The equipment diagnostic model 31 outputs a diagnostic result for the equipment 60 based on the operating data. The diagnostic result output from the equipment diagnostic model 31 is, for example, an estimated performance value of the equipment 60. The estimated performance value of the equipment 60 may be the power consumption of the equipment 60, etc.

[0033] The computing unit 10 analyzes the degree of factors contributing to the performance estimate for each process, based on the estimated performance of the equipment 60 included in the analysis target set DS1 and the process data corresponding to the analysis target set DS1, and displays the analysis results on the display 50.

[0034] When a diagnostic result is obtained from the equipment diagnostic model 31, the user can identify which process is the main cause of the diagnostic result by referring to the analysis results. Therefore, according to this embodiment, the user can understand the relationship between the diagnostic result output from the equipment diagnostic model 31 and multiple processes in the equipment's lifecycle. This allows the user to utilize the diagnostic result to improve the equipment's processes.

[0035] Figure 2 is a conceptual diagram showing the configuration of the analysis device 100 shown in Figure 1. Figure 3 is a conceptual diagram showing the lifecycle data catalog model 210. Figure 4 is a conceptual diagram showing the lifecycle data catalog 221. Figure 5 is a diagram showing an example of ID correspondence relationship data stored in the ID correspondence relationship storage unit 230.

[0036] As already explained, the analysis device 100 has an input / output interface 13 and an instrument diagnostic model 31. The input / output interface 13 communicates with the input device 40, the display 50, the instrument 60, and the terminal device 70.

[0037] As shown in Figure 2, the analysis device 100 conceptually includes an ID assignment unit 101, a trace unit 102, a quality analysis unit 103, a model generation unit 104, a performance estimation unit 105, and an ID correspondence relationship storage unit 230. The ID assignment unit 101, trace unit 102, quality analysis unit 103, model generation unit 104, and performance estimation unit 105 are implemented by the arithmetic unit 10 shown in Figure 1. The ID correspondence relationship storage unit 230 is implemented by the storage 20 shown in Figure 1.

[0038] The analysis device 100 includes a lifecycle data catalog model 210, model development data 106, and performance estimation data 107. These are included, for example, in the database 201 shown in Figure 1. In particular, the lifecycle data catalog model 210 is stored in the catalog model database 21 shown in Figure 1 as a catalog model that includes classification criteria for process data.

[0039] The lifecycle data 220 shown in Figure 2 is stored in the process database 22 shown in Figure 1. The lifecycle data 220 includes a lifecycle data catalog 221, procurement, development, and manufacturing data 222, transportation and construction data 223, test run / actual operation data 224, and maintenance data 225. The process database 22 stores corresponding process data for each of the processes: procurement (parts procurement), development, manufacturing, test run, actual operation, and maintenance. The process database 22 may also include databases configured for each of these processes.

[0040] The model generation unit 104 generates an equipment diagnostic model 31 using the model development data 106. The model development data 106 is, for example, a training dataset for training the equipment diagnostic model 31 using machine learning. The trained equipment diagnostic model 31 takes the test operation / actual operation data 224 as input data and outputs performance estimates such as the power consumption value of the equipment 60 to the performance estimation unit 105. In this way, the equipment diagnostic model 31 is an estimation model that has been trained to output estimated values ​​as diagnostic results based on the input data. The performance estimation unit 105 saves the performance estimates as performance estimation data 107, for example, in the database 201 (see Figure 1).

[0041] The ID assignment unit 101 assigns an ID to the process data according to the set to be analyzed. The ID assignment unit 101 further assigns an ID to the equipment diagnostic model 31, the model development data 106, and the performance estimation data according to the set to be analyzed. The ID assignment unit 101 refers to the classification criteria of the lifecycle data catalog model 210 and assigns IDs to the various types of data according to the classification criteria.

[0042] As a result, the performance estimation data 107 and the lifecycle data (process data) 220 are associated with IDs that conform to the classification criteria. The ID correspondence relationship storage unit 230 stores the correspondence between various types of data, including process data, and IDs for each set of data to be analyzed.

[0043] The trace unit 102 acquires the ID correspondence relationships stored in the ID correspondence relationship storage unit 230, the lifecycle data 220, and the performance estimation data 107. The trace unit 102 outputs the acquired data to the input / output interface 13 and the quality analysis unit 103. The quality analysis unit 103 uses the data acquired by the trace unit 102 to analyze the quality of the lifecycle data 220 by process data. The quality analysis unit 103 outputs the analysis results to the input / output interface 13. Details of the operation of the quality analysis unit 103 will be described later.

[0044] With reference to FIG. 3, a lifecycle data catalog model 210 will be described. As shown in FIG. 3, the lifecycle data catalog model 210 indicates classification criteria for process data. FIG. 3 shows, as an example of a plurality of processes defined by the lifecycle data catalog model 210, a parts procurement process, a development process, a manufacturing process, a transportation process, a construction process, a test operation process, an actual operation process, and a maintenance process.

[0045] Each of the plurality of process data includes one or more information sets. The information set includes "ID", "data item", "data item type", and "data item requirement". For example, FIG. 3 shows one information set corresponding to the parts procurement process, and two information sets corresponding to the test operation process.

[0046] "ID" is identification information for identifying a data item. "Data item" is the name of the data item. "Data item type" indicates the characteristic of data. "Data item requirement" is the required number of data or the collection period. "Data item requirement" is an example of a collection amount criterion. A person in charge related to each process transmits process data from the terminal device 70 to the analysis device 100 so as to satisfy the "data item requirement".

[0047] Thus, the lifecycle data catalog model 210 indicates classification criteria and collection amount criteria for process data. The lifecycle data catalog model 210 is an example of a catalog model. The lifecycle data catalog model 210 is stored in the catalog model database 21 shown in FIG. 1.

[0048] The parts procurement process relates to parts procurement of the device 60. FIG. 3 shows, as an example of parts procurement process data, the part name and component ID of the device 60. Further, it is shown that the characteristic of data related to the part is categorical, and that the required number of data related to the part is one. The lifecycle data catalog model defines process data of the parts procurement process for each component of the device 60.

[0049] The development process involves the development of device 60. Figure 3 shows an example of development process data, including the name and ID of the training data for device 60. It also indicates that the training data is time-series and that the required data collection period is one week. Furthermore, Figure 3 shows an example of development process data, including the model name and model ID of device 60. It also indicates that the data for the model is categorical and that one data item is required for that model.

[0050] The manufacturing process relates to the production of equipment 60. Figure 3 shows an example of manufacturing process data, including the compressor model number and equipment ID of equipment 60. For example, if equipment 60 is an air conditioner, equipment 60 may be identified by its compressor model number. In this case, the equipment ID is identification information for identifying the equipment 60 itself. Figure 3 further shows that the characteristics of the compressor model number are categorical and that only one data item is required for that compressor model number.

[0051] The transportation process involves the transportation of equipment 60. Figure 3 shows an example of transportation process data, including the transportation route and transportation ID of equipment 60. Furthermore, it is shown that the characteristics of the transportation route data are categorical, and that only one data entry is required for the transportation route. The transportation route data may include air and sea routes, etc. The transportation route data may also include the name of the transportation company.

[0052] The installation process relates to the installation of equipment 60 in an actual operating environment. Figure 3 shows an example of installation process data, including the installation conditions and installation ID of equipment 60. Furthermore, it is shown that the characteristics of the installation condition data are categorical, and that only one data point is required for the installation condition data. The installation condition data may include information about the installation environment of equipment 60. The installation condition data may also include the name of the contractor.

[0053] The test operation process relates to a test run conducted before the equipment 60 is put into actual operation in the actual operating environment. The test run may, for example, be an operation conducted over a certain period (e.g., two weeks) before actual operation, after the equipment 60 has been installed in the actual operating environment.

[0054] Figure 3 shows an example of test operation process data, specifically the test operation data period and test operation data ID for equipment 60. Furthermore, it indicates that the data for that test operation data period is time-series, and that the required data collection period for that test operation data period is one day. Additionally, Figure 3 shows an example of test operation process data, specifically the performance estimation data period and performance estimation data ID for equipment 60. Furthermore, it indicates that the data for that performance estimation data period is time-series, and that the required data collection period for that performance estimation data period is one day.

[0055] The actual operation process relates to the actual operation of equipment 60. Figure 3 shows the actual operation data period and actual operation data ID for equipment 60 as an example of experimental operation process data. Furthermore, it is shown that the data characteristics of the actual operation data period are time-series, and that the required data collection period for the actual operation data period is one year. Furthermore, Figure 3 shows the performance estimation data period and performance estimation data ID for equipment 60 as an example of actual operation process data. Furthermore, it is shown that the data characteristics of the performance estimation data period are time-series, and that the required data collection period for the performance estimation data period is one year.

[0056] The maintenance process concerns maintenance performed after the equipment 60 has been deployed in the actual operating environment. Figure 3 shows an example of maintenance process data, including the wiring check image name and maintenance data ID for checking the wiring inside the equipment 60. Furthermore, it is shown that the data characteristics of the image name are that it is an image, and that the required number of data items for that image name is one.

[0057] The lifecycle data catalog 221 will be explained with reference to Figure 4. The lifecycle data catalog 221 is a catalog of lifecycle data generated according to the definition of the lifecycle data catalog model 210. For example, Figure 4 shows component procurement process data where the component ID is "BBB-1111". The data item name of this component procurement process data is "Steel Plate", and the data item type is "Category Cal".

[0058] In the lifecycle data catalog 221, "data item performance" is used instead of "data item requirements" as shown in Figure 3. "Data item performance" indicates the data performance against the requirements defined in "data item requirements". If the "data item performance" does not meet the requirements defined in "data item requirements", the analysis device 100 may prompt the person in charge of the corresponding process to send process data to the terminal device 70.

[0059] As shown above, Figure 4 illustrates lifecycle data generated according to the definition of the lifecycle data catalog model 210 shown in Figure 3.

[0060] For example, Figure 4 shows development process data with training data ID "CCC-4112" and development process data with equipment model ID "DDD-1111". The former is training data, and the latter is equipment model.

[0061] Figure 4 shows manufacturing process data for a manufacturing process, where the equipment ID is "AAA-1111". This manufacturing process data corresponds, for example, to the compressor model number "XYZ-1234". Figure 4 also shows transportation process data for a transportation process, where the transportation ID is "EEE-1242". This transportation process data corresponds, for example, to "Route F".

[0062] Figure 4 shows construction process data with construction ID "FFF-2167" for the construction process. This construction process data corresponds to, for example, "Construction Condition A". Figure 4 also shows test operation process data with test operation data IDs "GGG-1111" and "ZZZ-1111" for the test operation process. For example, the former test operation process data corresponds to the test operation data period "2032 / 6 / 7". Figure 4 also shows maintenance process data with maintenance data ID "III-1111" for the maintenance process. This maintenance process data corresponds to the wiring check image names "Wiring Images a, b, c".

[0063] Figure 4 shows actual operation process data with an actual operation data ID of "HHH-1111" and actual operation process data with a performance estimation data ID of "ZZZ-1113" for the actual operation process. The "data item performance" for these actual operation process data is 90 days. On the other hand, in the lifecycle data catalog model 210 shown in Figure 3, the corresponding "data item requirement" is 1 year. Therefore, the actual operation process data shown in Figure 4 does not meet the "data item requirement". For example, it is expected that these actual operation process data will meet the "data item requirement" when the actual operation period has increased from 90 days to 1 year.

[0064] As illustrated with Figure 2, the lifecycle data 220 includes the lifecycle data catalog 221. The lifecycle data 220 includes, as detailed data according to the lifecycle data catalog 221, procurement, development, and manufacturing data 222, transportation and construction data 223, test run / actual operation data 224, and maintenance data 225.

[0065] Figure 5 shows an example of ID correspondence data stored in the ID correspondence storage unit 230. As shown in Figure 5, the ID correspondence storage unit 230 stores ID correspondences by process (parts procurement, development, manufacturing, etc.).

[0066] For example, Figure 5 shows component IDs such as "BBB-1111" and "BBB-1213" as part of the parts procurement process. The process data for the parts procurement process is information about the parts of the equipment 60. The construction ID such as "FFF-2167" is shown as part of the construction process. The process data for the construction process is construction information when the equipment 60 is installed in the actual operating environment. The construction information may include information such as the name of the construction company and the construction environment.

[0067] "EEE-1242" is shown as a transport ID related to the transport process. The process data for the transport process is transport information when the equipment 60 is transported to the actual operating environment. The transport information may include, for example, the name of the transporter, the transport route, and the transport environment. "AAA-1111" and "AAA-1112" are shown as equipment IDs related to the manufacturing process. The process data for the manufacturing process may be information related to the equipment 60 itself. In this case, "AAA-1111" and "AAA-1112" are IDs corresponding to the equipment 60 itself.

[0068] Figure 5 shows the device diagnostic model ID and training data ID as IDs related to the development process. Here, it is assumed that the process data of the development process is information related to the device diagnostic model 31 and training data used for machine learning of the device diagnostic model 31. Therefore, "DDD-1111" and "DDD-1112," etc., which correspond to the device diagnostic model ID, are IDs corresponding to the device diagnostic model 31, and "CCC-4112" and "CCC-4113," etc., which correspond to the training data ID, are IDs corresponding to the training data.

[0069] Similarly, the ID correspondence storage unit 230 stores IDs for process data related to the test operation process, the actual operation process, and the maintenance process, respectively. Note that in Figure 5, an example of the configuration of IDs corresponding to the maintenance process is omitted. For example, the IDs corresponding to the maintenance process may be IDs corresponding to the name of the maintenance company that performed the maintenance on the equipment 60, the maintenance environment, the timing of the maintenance, and the number of maintenance sessions.

[0070] The ID correspondence storage unit 230 also stores performance estimation data IDs. The performance estimation data IDs are associated with the performance estimation data 107 shown in Figure 2.

[0071] As shown by the arrows in Figure 5, the equipment ID is stored in the ID correspondence relationship storage unit 230 in association with other IDs (component ID, construction ID, transportation ID, equipment diagnostic model ID, etc.). Therefore, the arithmetic unit 10 can identify the correspondence between process data and performance estimation data 107 for each piece of equipment 60 by referring to the ID correspondence relationship storage unit 230.

[0072] Figure 6 is a block diagram showing the procedure for calculating analysis results based on lifecycle data and performance estimates. The processes of the tracing unit 102 and the quality analysis unit 103 will be specifically explained using Figure 6.

[0073] As described above, the equipment 60 is deployed to the actual operating environment after going through various processes such as parts procurement, development, and other processes. In this embodiment, throughout the lifecycle of the equipment 60, process data corresponding to each process is collected and stored in the analysis device 100 as data that constitutes lifecycle data 220.

[0074] Equipment 60, deployed in the actual operating environment, transmits operating data to the analysis device 100. The operating data is input to the equipment diagnostic model 31. The equipment diagnostic model 31 outputs performance estimates as diagnostic results. The performance estimates are stored in the analysis device 100 as performance estimate data 107. The trace unit 102 acquires the performance estimate data 107 and the lifecycle data 220. At this time, the trace unit 102 refers to the ID correspondence relationship storage unit 230 and identifies the correspondence between the performance estimate data 107 and the lifecycle data 220.

[0075] The quality analysis unit 103 obtains the data to be analyzed from the trace unit 102. The data to be analyzed is a set of performance estimation data 107 and lifecycle data 220 identified by the trace unit 102. The quality analysis unit 103 determines whether the performance estimation data is abnormal or not. The quality analysis unit 103 may, for example, use a threshold to determine whether the performance estimation data is abnormal or not. If the performance estimation data 107 is abnormal, the quality analysis unit 103 further analyzes the degree of the abnormality (degree of abnormality) for each process data.

[0076] The quality analysis unit 103 may also analyze the factor degree for each process data even when the performance estimation data 107 is normal. In this case, the factor degree corresponds to the "normal factor degree". The quality analysis unit 103 outputs the analysis results. For example, the quality analysis unit 103 may display the analysis results on the display 50. The quality analysis unit 103 may display the abnormal factor degree on the display 50. The quality analysis unit 103 may display the normal factor degree on the display 50.

[0077] If the user obtains a diagnostic result indicating an abnormality from the equipment diagnostic model 31, they can refer to the abnormality factor score to identify which process is the main cause of the diagnostic result and improve that process. Alternatively, if the user obtains a diagnostic result indicating a normal state from the equipment diagnostic model 31, they can refer to the normal state factor score to identify which process is the main cause of the diagnostic result and, for example, proceed with the development of another piece of equipment 60 taking that process into consideration.

[0078] Figure 7 is a flowchart showing the procedure for collecting process data by ID. Figure 8 is a diagram showing an example of the information management screen displayed on the display 50. Figure 9 is a diagram showing an example of the information upload screen displayed on the terminal device 70. The arithmetic unit 10 executes the process described below according to the flowchart shown in Figure 7 and displays the information management screen shown in Figure 8 on the display 50.

[0079] First, the arithmetic unit 10 sets an ID for each process data corresponding to the device 60 (step S1). Next, the arithmetic unit 10 stores the ID correspondence relationships for each device 60 in the ID correspondence relationship storage unit 230 (step S2). Next, the arithmetic unit 10 receives process data from the terminal devices 70 corresponding to each process (step S3). In other words, the arithmetic unit 10 acquires process data for each process. For example, the arithmetic unit 10 is configured to acquire first process data corresponding to a first process from a first terminal device among the multiple terminal devices 70, and second process data corresponding to a second process from a second terminal device among the multiple terminal devices 70.

[0080] Next, the arithmetic unit 10 stores process data in the process database 22 for each piece of equipment 60 and each process (step S4). Next, the arithmetic unit 10 calculates the sufficiency rate of process data for each piece of equipment 60 and each process (step S5).

[0081] The process data sufficiency rate is calculated based on the data item requirements defined in the lifecycle data catalog model 210 (see Figure 3) and the actual data item data registered in the lifecycle data catalog 221 (see Figure 4). For example, if the data item requirements and actual data item data for a certain process data are 10 and 5 respectively, the sufficiency rate is calculated to be 50%.

[0082] Next, the arithmetic unit 10 displays an information management screen that includes the process data sufficiency rate (step S6). Figure 8 shows an example of the information management screen. Note that the information management screen illustrated in Figure 8 shows the data sufficiency rate for a single process and the data sufficiency rate for two processes combined (for example, a development process and a manufacturing process). However, the arithmetic unit 10 may also display an information management screen on the display 50 that shows the data sufficiency rate for each process.

[0083] As shown in Figure 8, the information management screen includes a button 501 that corresponds to a data request. If a user wants to request the transmission of process data for which the data sufficiency rate is less than 100%, they press the corresponding button 501 with their finger. The arithmetic unit 10 accepts the user operation of pressing the button 501 with a finger as a "process data request" (step S7). In this case, the arithmetic unit 10 requests the transmission of the process data to the corresponding terminal device 70 (step S8), and completes the processing based on this flowchart.

[0084] The information management screen, which includes button 501, is an example of an interface that accepts requests to send process data. Button 501 is arranged on the information management screen for each process. Therefore, the information management screen can accept, for example, a first transmission request for first process data and a second transmission request for second process data. The arithmetic unit 10 requests the terminal device 70 corresponding to the first process data to send the first process data in accordance with the first transmission request, and requests the terminal device 70 corresponding to the second process data to send the second process data in accordance with the second transmission request.

[0085] Upon receiving a request to transmit process data, terminal device 70 displays a message or other information prompting the person holding terminal device 70 to transmit the process data. The person in charge enters the process data into the information upload screen of terminal device 70 and then transmits the process data to the analysis device 100. Figure 9 shows an example of the information upload screen displayed on terminal device 70.

[0086] As shown in Figure 9, the information upload screen includes the model of the equipment 60, equipment ID, process, and person in charge information. The person in charge enters multiple process data corresponding to the process into item 1, item 2, item 3, etc. After finishing entering the process data, the person in charge presses the upload button 701 with their finger. This sends the entered process data to the analysis device 100. If the acquired process data is construction process data, the analysis device 100 saves it as transportation / construction data 223.

[0087] Furthermore, the terminal device 70 displays the information upload screen not only when it receives a request to transmit process data, but also when it detects an operation by the person in charge. The person in charge is required to transmit the process data to the analysis device 100 without delay as soon as the process data is ready. If there is a delay in transmitting the process data, the completion rate will decrease. In this case, a user referring to the information management screen can press the corresponding button 501 (see Figure 8) to prompt the person in charge to transmit the process data. The processes described above using Figures 7 to 9 are performed separately for each piece of equipment 60.

[0088] Figure 10 is a flowchart showing the procedure for determining the performance estimate. The computing unit 10 executes the processes described below according to the flowchart shown in Figure 10 and determines the performance estimate output from the equipment diagnostic model 31.

[0089] First, the computing unit 10 acquires the performance estimate output from the equipment diagnostic model 31 (step S11). Next, the computing unit 10 compares the performance estimate with a threshold and determines whether the performance estimate is normal or abnormal (step S12).

[0090] If the device 60 is an air conditioner, the performance estimate may be the percentage that exceeds the design value of the amount of electricity required to achieve the air conditioning capacity. In this case, the computing device 10 may determine that the performance estimate is abnormal if the percentage that exceeds the design value of the amount of electricity (performance estimate) is higher than a threshold.

[0091] Next, the arithmetic unit 10 stores the determination result in the memory 12 (step S13), and completes the processing based on this flowchart. If the arithmetic unit 10 determines that the performance estimate is abnormal, it executes a process to calculate the degree of abnormality. The process described above using Figure 10 is performed separately for each piece of equipment 60.

[0092] Figure 11 shows an example of a graph created when normal judgment data and abnormal judgment data are classified based on the estimated actual operating performance and the estimated test operation performance. The graph plots the performance estimates output from the equipment diagnostic model 31.

[0093] Normal judgment data is data in which the equipment 60 is determined to be normal based on the performance estimate output from the equipment diagnostic model 31. Abnormal judgment data is data in which the equipment 60 is determined to be abnormal based on the performance estimate output from the equipment diagnostic model 31. In Figure 11, "Cooling LP" is one of several types of cooling modes. As illustrated in Figure 11, each of the normal judgment data and abnormal judgment data includes the equipment ID and process data.

[0094] The quality analysis unit 103 may also determine whether the performance estimate is abnormal by comparing the reference design data with the test operation data. The quality analysis unit 103 may also determine whether the performance estimate is abnormal by comparing the reference design data with the actual operation data. The quality analysis unit 103 may also determine whether the performance estimate is abnormal by comparing the reference test operation data with the actual operation data. For example, the quality analysis unit 103 may determine that it is abnormal if, despite operating the equipment 60 under the same operating standards, the performance of the equipment 60 during actual operation is lower than the performance of the equipment 60 during test operation by more than a threshold.

[0095] Figure 12 is a conceptual diagram showing the procedure for generating the regression model 302 used to calculate the anomaly factor score. Figure 13 is a flowchart showing the procedure for calculating and displaying the anomaly factor score. Figure 14 is a diagram showing an example of the anomaly factor score graph displayed on the display 50. The procedure for calculating and displaying the anomaly factor score will be explained below with reference to Figures 12 to 14.

[0096] As shown in Figure 12, the analysis device 100 includes a training unit 301 for generating a regression model 302. The training unit 301 is comprised of, for example, a computing unit 10. The computing unit 10 generates a regression model 302 that reflects the state of the target equipment 60 and calculates the degree of anomaly by analyzing the parameters of the generated regression model 302.

[0097] As shown in Figure 13, the calculation unit 10 sets the target equipment 60 (step S21). The target equipment 60 refers to the equipment for which the degree of abnormality is to be calculated. Next, the calculation unit 10 determines a group of equipment similar to the target equipment 60 (step S22).

[0098] The device group is used to generate a regression model 302 that reflects the state of the target device 60. The device group is a subset of the total set of all devices 60 stored in the database 201. The device group may be determined by a pre-configured logic or by user selection. Instead of all the lifecycle data for each device 60 included in the device group, lifecycle data for a specific period may be used. The specific period may be determined by a pre-configured logic or by user selection based on the values ​​of the lifecycle data.

[0099] The criteria for narrowing down the lifecycle data used to generate the regression model 302, that is, the "selection criteria" for the equipment to be included in the equipment group, may be equipment similar to the model of the target equipment 60. Alternatively, the "selection criteria" may be equipment similar to the system configuration of the target equipment 60. For example, if the target equipment 60 is an air conditioner, the similarity of the system configuration may be determined based on the number of indoor units connected to the outdoor unit and the length of the piping.

[0100] The "selection criteria" may be that the equipment is installed in an environment similar to that of the target equipment 60 (for example, industry and installation area). The "selection criteria" may also be that the equipment has a maintenance history (number of years installed, history of maintenance of certain parts) similar to that of the target equipment 60. The "selection criteria" may also be that the equipment has operating conditions (operating mode, etc.) similar to that of the target equipment 60. Based on one or more of the "selection criteria" exemplified above, the equipment to be included in the equipment group and the adoption period of the life cycle data may be selected.

[0101] Next, the computing unit 10 generates a regression model 302 using the performance estimate as the target variable and the lifecycle data 220 as the explanatory variables (step S23). As shown in Figure 12, the training unit 301 receives the lifecycle data 220 as the explanatory variable and the performance estimate as the target variable. The lifecycle data 220 includes multiple process databases. The training unit 301 uses these explanatory and target variables as a dataset to train the regression model 302 using supervised learning. The training unit 301 trains the regression model 302 with the dataset corresponding to each of the devices 60 included in the device group.

[0102] Next, the computing unit 10 analyzes the parameters of the regression model 302 and calculates the degree of anomaly factor for each process (step S24). More specifically, the computing unit 10 analyzes the parameters of the trained regression model 302 and analyzes the variables that contribute most to explaining the performance estimate. In this way, the computing unit 10 identifies which of the multiple types of process data included in the lifecycle data 220 contribute most to the performance estimate. The computing unit 10 calculates the degree of anomaly factor using the identified results. In this way, the computing unit 10 analyzes the degree of anomaly factor by analyzing the parameters of the regression model 302, which is trained with process data as explanatory variables and the estimate as the dependent variable.

[0103] Furthermore, even if there are no abnormalities in the performance estimates, the arithmetic unit 10 may analyze the factor degree (normal factor degree) using the performance estimates and process data. In this way, the arithmetic unit 10 may analyze both abnormal factor degrees and normal factor degrees.

[0104] Next, the arithmetic unit 10 stores the calculation results in the memory 12 (step S25). Then, the arithmetic unit 10 displays the degree of abnormality for each process on the display 50 (step S26), and finishes the processing based on this flowchart. As a result, a graph showing the degree of abnormality is displayed on the display 50. Figure 14 shows an example of the graph.

[0105] In this way, the computing unit 10 uses the diagnostic results (performance estimates) output from the equipment diagnostic model 31 for diagnosing the equipment 60 and process data to analyze the degree to which process data contributes to the diagnostic results for each process, and displays the analysis results on the display 50. Note that, as shown in Figure 1, if multiple equipment 60 are arranged in the analysis system 1, the computing unit 10 executes the process shown in Figure 13 for each piece of equipment 60.

[0106] If there is an anomaly in the performance estimate, the computing unit 10 analyzes the degree of anomaly contributing to the performance estimate for each process and displays a graph on the display 50 as exemplified in Figure 14. Figure 14 shows the degree of anomaly contributing to each of the following: construction conditions, transport route, part Y lot, part Z lot, and piping length. Construction conditions, transport route, part Y lot, part Z lot, and piping length are examples of process data.

[0107] For example, by referring to the graph shown in Figure 14, the user can identify that the main cause of abnormality in the target equipment 60 lies in the construction conditions. By referring to the transportation and construction data 223 (see Figure 2), the user can understand the details of the construction conditions identified as the main cause of abnormality. By reviewing the construction conditions, the user can prevent similar abnormalities from occurring in similar equipment 60. In addition to the graph of the degree of abnormality, the calculation unit 10 may also display the diagnostic results of the equipment diagnostic model 31 on the display 50. The process described above using Figures 12 to 14 is performed for each piece of equipment 60.

[0108] Figure 15 shows some examples of analytical information that can be identified using the analysis system 1. As shown in Figure 15, the lifecycle of the equipment 60 includes various processes, from parts procurement to operation and maintenance. For example, the operation (commissioning, actual operation) process or the maintenance process may include data obtained by remotely monitoring the equipment 60. Such data may include various types of data, such as data that ensures the normal operation of the equipment 60 and data that indicates the detection of anomalies in the equipment 60.

[0109] The analysis device 100 manages the process data for each of the multiple processes using an ID assignment unit 101, an ID correspondence relationship storage unit 230, and a trace unit 102, etc. Therefore, by applying this embodiment, the analysis device 100 can, for example, determine that the operating accuracy of the equipment 60 has deteriorated three months after the actual operation of the equipment 60 has started, when the construction conditions are A. Alternatively, the analysis device 100 can, for example, determine that the operating accuracy of the equipment 60 has deteriorated one month after the actual operation of the equipment 60 has started, when the transport route is F. Alternatively, the analysis device 100 can, for example, determine that the operating accuracy of the equipment 60 has deteriorated six months after the actual operation of the equipment 60 has started, when lot Z of part Y was used in the equipment 60.

[0110] The embodiments disclosed herein should be considered in all respects to be illustrative and not restrictive. The scope of this disclosure is indicated by the claims rather than by the description of the embodiments above, and all modifications within the meaning and scope of the claims are intended to be included.

[0111] 1 Analysis system, 10 Calculation unit, 11 Processor, 12 Memory, 13 Input / Output interface, 20, 30 Storage, 21 Catalog model database, 22 Process database, 23 ID correspondence relationship database, 31 Equipment diagnostic model, 40 Input device, 50 Display, 60 Equipment, 70 Terminal device, 100 Analysis device, 101 ID assignment unit, 102 Trace unit, 103 Quality analysis unit, 104 Model generation unit, 105 Performance estimation unit, 106 Model development data, 107 Performance estimation data, 201 Database, 301 Training unit, 302 Regression model, 210 Lifecycle data catalog model, 220 Lifecycle data, 221 Lifecycle data catalog, 222 Procurement / development / manufacturing data, 223 Transportation / construction data, 224 Test run / actual operation data, 225 Maintenance data, 501, 701 Buttons, DS1, DS2 The set to be analyzed is the network (NW).

Claims

1. An analysis system for analyzing data relating to equipment, comprising a processor and a display, wherein the processor acquires data relating to the lifecycle of a first piece of equipment, the lifecycle includes a plurality of processes, the data relating to the lifecycle includes process data relating to the processes of the lifecycle, and the processor uses a first diagnostic result output from a first piece of equipment diagnostic model for diagnosing the first piece of equipment and the process data to analyze the degree to which the process data contributes to the first diagnostic result for each process, and displays the analysis results on the display.

2. The analysis system according to claim 1, wherein the processor analyzes the degree of the cause when it determines that there is an abnormality in the first device based on the first diagnostic result.

3. The analysis system according to claim 1 or 2, wherein the plurality of processes include any one of the following: a process relating to the components of the first equipment, a process relating to the development of the first equipment, a process relating to the manufacture of the first equipment, a process relating to the transportation of the first equipment, a process relating to the installation of the first equipment, a process relating to the test run of the first equipment, a process relating to the actual operation of the first equipment, and a process relating to the maintenance of the first equipment.

4. The analysis system according to any one of claims 1 to 3, further comprising: a catalog model database that stores catalog models indicating classification criteria and collection quantity criteria for the process data for each of the plurality of processes; and a process database that stores the process data, wherein the processor stores the process data in the process database for each process according to the classification criteria, calculates the degree of sufficiency of the process data stored in the process database based on the collection quantity criteria, and displays the degree of sufficiency on the display.

5. The analysis system according to any one of claims 1 to 4, wherein the plurality of processes include a first process and a second process, the processor is configured to acquire first process data corresponding to the first process from a first terminal device and to acquire second process data corresponding to the second process from a second terminal device, and further comprises an interface for receiving a first transmission request for the first process data and a second transmission request for the second process data, the processor requests the first terminal device to transmit the first process data in accordance with the first transmission request and requests the second terminal device to transmit the second process data in accordance with the second transmission request.

6. The analysis system according to any one of claims 1 to 5, wherein the processor acquires process data relating to the second device for each process in the lifecycle relating to the second device, stores the correspondence between the first device diagnostic model and the process data relating to the first device, and the correspondence between the second device diagnostic model for diagnosing the second device and the process data relating to the second device, and uses the second diagnostic result output from the second device diagnostic model and the process data to analyze the degree to which the process data contributes to the second diagnostic result for each process, and displays the analysis results on the display.

7. The analysis system according to any one of claims 1 to 6, wherein the first equipment diagnostic model is an estimation model trained to output an estimated value as the first diagnostic result based on input data, the input data includes operating data of the first equipment, and the processor analyzes the factorization using the estimated value and the process data.

8. The analysis system according to claim 7, wherein the processor analyzes the factorization by analyzing the parameters of a regression model trained with the process data as explanatory variables and the estimated values ​​as the dependent variable.

9. A method for analyzing data relating to a device, the method comprising a plurality of steps performed by a computer, the plurality of steps comprising a step of acquiring data relating to the lifecycle of a first device, the lifecycle comprising a plurality of processes, the data relating to the lifecycle comprising process data relating to the processes of the lifecycle, and the plurality of steps comprising analyzing, for each process, the degree to which the process data contributes to the first diagnostic result, using a first diagnostic result output from a first device diagnostic model for diagnosing the first device and the process data, and displaying the analysis results on a display.

Images