Information processing device and information processing method

Abstract

[Problem] To provide an information processing device capable of appropriately classifying an operating mode of an apparatus by using only operation data in which state quantities of the apparatus are recorded in a time series. [Solution] An information processing device 2 comprises: a diagnostic data generation unit 203 which generates diagnostic data by splitting operation data, in which measurement results obtained by using a sensor provided in an apparatus to be diagnosed (vacuum pump 11) to measure a state quantity that varies according to the operation of the apparatus are recorded in a time series, on the basis of a feature of the operation data at the time of a transition in the operating state of the apparatus; and a classification processing unit 204 which, for each of a plurality of operating modes included in an operating mode group, calculates the degree of similarity between the diagnostic data and a plurality of pieces of classification data serving as references for classifying the operating state according to the operating modes, and which classifies the diagnostic data with respect to the operating mode group on the basis of the degree of similarity for each operating mode.

Description

Information Processing Apparatus and Information Processing Method

[0001] The present invention relates to an information processing apparatus and an information processing method.

[0002] In various devices installed in various locations and operated, based on operation data in which state quantities are recorded in time series by sensors provided in the device, monitoring of the occurrence status of abnormalities is performed. At that time, since the device is operated while switching between a plurality of operation modes, in Patent Document 1, a time-series sensor signal is divided for each operation mode based on an event signal, and based on the sensor signal divided for each operation mode, a facility status monitoring system for monitoring the occurrence status of abnormalities is disclosed.

[0003] International Publication No. 2011 / 043108

[0004] In the facility status monitoring system disclosed in Patent Document 1, it is necessary to input an event signal representing the operation, failure, and warning of the device separately from the sensor signal. However, depending on the configuration of the system, there may be cases where the event signal cannot be acquired, or cases where the event signal does not include events effective for monitoring the device. In such cases, it is impossible to appropriately divide the sensor signal for each operation mode, and as a result, it has been difficult to appropriately monitor the occurrence status of abnormalities in the device based on the sensor signal.

[0005] In view of the above problems, an object of the present invention is to provide an information processing apparatus and an information processing method that enable appropriate classification of the operation mode of a device only by operation data in which the state quantity of the device is recorded in time series.

[0006] To achieve the above objective, an information processing device according to one aspect of the present invention includes: a diagnostic data generation unit that generates diagnostic data by dividing operating data, which is recorded in time series from measurement results of state quantities that fluctuate according to the operation of the equipment measured by a sensor provided on the equipment to be diagnosed, at transition points when the operating state of the equipment changes, based on the characteristic quantities of the operating data; and a classification processing unit that calculates the similarity between a plurality of classification data that serve as a criterion for classifying the operating state by a plurality of operating modes included in an operating mode group, and the diagnostic data for each operating mode, and classifies the diagnostic data for the operating mode group based on the similarity for each operating mode.

[0007] According to an information processing device in one aspect of the present invention, operational data, in which the state quantities of the equipment are recorded in a time series, is divided at the point of transition of the operational state based on the characteristic quantities of the operational data, thereby generating diagnostic data, and the diagnostic data is classified into multiple operational modes. Therefore, the operational mode of the equipment can be appropriately classified using only operational data in which the state quantities of the equipment are recorded in a time series.

[0008] Other issues, configurations, and effects will be clarified in the embodiments for carrying out the invention described later.

[0009] This is an overall configuration diagram showing an example of the equipment management system 1 and the semiconductor manufacturing system 10. This is a block diagram showing an example of the vacuum pump 11. This is a block diagram showing an example of the information processing device 2. This is a data configuration diagram showing an example of the equipment database 210. This is a hardware configuration diagram showing an example of the computer 900. This is a flowchart showing an example of the operation of the first classification generation process. This is an explanatory diagram showing an example of the data in the first classification generation process. This is a flowchart showing an example of the operation of the first classification and diagnostic process. This is an explanatory diagram showing an example of the data in the first classification and diagnostic process. This is a flowchart showing an example of the operation of the second classification generation process. This is an explanatory diagram showing an example of the data in the second classification generation process. This is a flowchart showing an example of the operation of the second classification and diagnostic process. This is an explanatory diagram showing an example of the data in the second classification and diagnostic process. This is a flowchart showing an example of the operation of the third classification generation process. This is an explanatory diagram showing an example of the data in the third classification generation process. This is a flowchart showing an example of the operation of the third classification and diagnostic process. This is a flowchart showing an example of the operation of the third classification and diagnostic process. This is an explanatory diagram showing an example of the data in the third classification and diagnostic process. This is an explanatory diagram showing an example of the data in the third classification and diagnostic process. This flowchart shows examples of the operation of the fourth classification, diagnostic, and classification addition processes.

[0010] Hereinafter, embodiments for carrying out the present invention will be described with reference to the drawings. In the following, the scope necessary for explaining how to achieve the objectives of the present invention will be schematically shown, and the scope necessary for explaining the relevant parts of the present invention will be mainly explained, with any parts that are omitted from explanation being based on prior art.

[0011] (Configuration of Equipment Management System 1 and Semiconductor Manufacturing System 10) Figure 1 is an overall configuration diagram showing an example of the equipment management system 1 and the semiconductor manufacturing system 10. The equipment management system 1 functions as a system for diagnosing the operating status and occurrence of abnormalities of the equipment to be diagnosed. In this embodiment, the case in which the equipment to be diagnosed is the vacuum pump 11 provided in the semiconductor manufacturing system 10 will be described.

[0012] The equipment management system 1 comprises, as its main components, an information processing device 2 configured to communicate with the vacuum pump 11, and a maintenance terminal device 3 configured to communicate with the information processing device 2. Each of the information processing device 2 and the maintenance terminal device 3 is, for example, composed of a general-purpose or dedicated computer (see Figure 5 described later), and is configured to send and receive various types of information to and from each other via the network 4.

[0013] Furthermore, the semiconductor manufacturing system 10 includes, as its main components, a vacuum pump 11 as an example of equipment to be diagnosed, a semiconductor manufacturing apparatus 12, and piping 13 connecting the vacuum pump 11 and the semiconductor manufacturing apparatus 12.

[0014] The vacuum pump 11 is equipped with sensors 110 that measure the state quantities of each part of the vacuum pump 11, and controls its operating state based on the various state quantities measured by the sensors 110. The vacuum pump 11 also records the measurement results of the state quantities that fluctuate according to the operation of the vacuum pump 11 measured by the sensors 110 as operating data D2 and transmits it to the information processing device 2.

[0015] The semiconductor manufacturing apparatus 12 comprises a film deposition process chamber 120 and an apparatus control unit 121 that controls the film deposition process chamber 120.

[0016] The film deposition process chamber 120 is connected to a vacuum pump 11 via piping 13. By operating the vacuum pump 11, gas is discharged from the film deposition process chamber 120, maintaining a near-vacuum state. Examples of film deposition methods in the film deposition process chamber 120 include CVD, PVD, ALD, vapor deposition, and sputter deposition.

[0017] The device control unit 121 controls various parts (not shown) of the semiconductor manufacturing apparatus 12 by sequentially switching the gas flow rate and gas type supplied to the film deposition process chamber 120 in order to carry out predetermined processes in each of the processes, such as the film deposition process, implantation process, etching process, ashing process, and heating process. As each process is switched sequentially, the state of the gas in the film deposition process chamber 120 (for example, pressure) changes, and the operating state of the vacuum pump 11 also changes sequentially. In this embodiment, it is assumed that information regarding the manufacturing process being executed by the semiconductor manufacturing apparatus 12 is not transmitted from the device control unit 121 to the vacuum pump 11 by communication or the like.

[0018] The information processing device 2 is a device that sends and receives various types of information between the vacuum pump 11 and the maintenance terminal device 3. The information processing device 2 is composed of, for example, a server-type computer or a cloud-type computer. Alternatively, the information processing device 2 may be composed of a stationary computer (such as a desktop PC) or a portable computer (such as a tablet terminal or smartphone) that is capable of recording various types of data in a memory divided into multiple storage areas.

[0019] The information processing device 2 includes an equipment database 210 capable of registering operating data D2 collected from each vacuum pump 11. The information processing device 2 also generates abnormality data D7 indicating the occurrence of abnormalities in each vacuum pump 11 based on the operating data D2 and transmits it to the maintenance terminal device 3.

[0020] Each vacuum pump 11 is assigned a unique device identification information (device ID) to identify it, and the information processing device 2 manages communication with each vacuum pump 11 and various information based on the device ID assigned to each vacuum pump 11. If a device ID is not uniquely assigned to each vacuum pump 11, the identification of the vacuum pump 11 is not limited to the device ID alone; for example, a combination of the manufacturer, type, and installation location of the vacuum pump 11 may be used. In addition, each maintenance worker for the vacuum pump 11 is assigned a maintenance worker identification information (maintenance worker ID) to identify the maintenance worker, and the information processing device 2 manages communication with each maintenance terminal device 3 and various information based on the maintenance worker ID assigned to each maintenance worker.

[0021] The maintenance terminal device 3 is used by maintenance workers of the vacuum pump 11 to send and receive various types of information with the information processing device 2. The maintenance terminal device 3 is composed of, for example, a stationary computer (desktop PC, etc.) or a portable computer (tablet terminal, smartphone, etc.).

[0022] The maintenance terminal device 3 has programs such as applications and browsers installed, and accepts various input operations and displays various information on its screen. For example, the maintenance terminal device 3 receives abnormal data D7 from the information processing device 2 and displays the abnormal data D7. At that time, the maintenance terminal device 3 displays and transmits various information about each vacuum pump 11 based on the equipment ID assigned to each vacuum pump 11.

[0023] Network 4 is configured by wired communication, wireless communication, or a combination of wired and wireless communication according to any communication standard. Specifically, for example, it can utilize a standardized communication network such as the Internet, a communication network managed within a building such as a local network, or a combination of these communication networks. Furthermore, international standards are typically used as the communication standards for wireless communication. As international standard communication means, methods such as IEEE 802.15.4, IEEE 802.15.1, IEEE 802.15.11a, 11b, 11g, 11n, 11ac, 11ad, ISO / IEC 14513-3-10, and IEEE 802.15.4g can be used. Furthermore, technologies such as Bluetooth®, Bluetooth Low Energy, Wi-Fi, ZigBee®, Sub-GHz, EnOcean®, and LTE can also be used.

[0024] (Configuration of Vacuum Pump 11) Figure 2 is a block diagram showing an example of a vacuum pump 11. In addition to sensors 110 provided on various parts of the vacuum pump 11, the vacuum pump 11 includes a device control unit 111 that controls drive components and electrical components to perform various processes, a device storage unit 112 that can store various information and programs, a device communication unit 113 configured to communicate with the information processing device 2, an inverter 114 connected to a power supply (not shown), a motor 115 connected to the inverter 114, and a rotor 116 connected to the rotating shaft of the motor 115. Note that any type or model of vacuum pump 11 can be used.

[0025] The inverter 114 supplies a drive current to the motor 115, obtained by converting the frequency of the alternating current supplied from the power source. The motor 115 rotates its rotating shaft using the drive current supplied from the inverter 114. The rotor 116 rotates in conjunction with the rotation of the motor 115, drawing in and discharging gas from the piping 13. As a result, the gas is continuously transferred from the piping 13, and the gas in the film deposition process chamber 120 connected to the piping 13 is evacuated.

[0026] The sensor 110 measures state quantities that fluctuate in accordance with the operation of the vacuum pump 11 at predetermined measurement cycles and timings. The sensor 110 consists of, for example, a pressure sensor that measures the pressure value inside the casing housing the rotor 116, a vibration sensor that measures the vibration value of the casing, a rotational speed sensor that measures the rotational speed of the motor 115, and a power sensor that measures the power consumption of the motor 115. The sensor 110 is not limited to the above sensors, and may also be a flow sensor, temperature sensor, speed sensor, acceleration sensor, rotational angular velocity sensor, rotational angular acceleration sensor, torque sensor, current sensor, voltage sensor, sound sensor, etc.

[0027] The equipment storage unit 112 stores setting data D1 for when the vacuum pump 11 is operated and operation data D2 for when the vacuum pump 11 is operated.

[0028] Setting data D1 is data that records the setting values ​​for various setting items when the vacuum pump 11 is in operation. Setting data D1 is used to set the control amount of the inverter 114 and to set the measurement conditions of the sensor 110.

[0029] Operating data D2 is data recorded in time series from the measurement results of state variables measured by the sensor 110 in accordance with the operation of the vacuum pump 11. Specifically, operating data D2 is a time-series recording of pressure values ​​from the pressure sensor, vibration values ​​from the vibration sensor, motor rotation speed from the motor rotation speed sensor, motor power consumption from the motor power sensor, etc.

[0030] The device communication unit 113 functions as a communication interface that sends and receives various types of information to and from the information processing device 2 via the network 4. The device communication unit 113 may also communicate with the information processing device 2 via any relay device, or with the maintenance terminal device 3.

[0031] The equipment control unit 111 acquires state quantities (which may also be processing values) from each sensor 110 according to the measurement conditions of the setting data D1, and controls the inverter 114 based on the state quantities from each sensor 110 and the control quantities from the setting data D1. At that time, the equipment control unit 111 stores the state quantities from each sensor 110 as operation data D2 in the equipment storage unit 112, for example. The equipment control unit 111 then transmits the operation data D2 stored in the equipment storage unit 112 to the information processing device 2. The equipment control unit 111 may also transmit the operation data D2 when a predetermined time has elapsed or when a predetermined amount of data has been accumulated, or it may transmit the operation data D2 in response to a transmission request from the information processing device 2.

[0032] (Configuration of Information Processing Device 2) Figure 3 is a block diagram showing an example of the information processing device 2. The information processing device 2 comprises a management control unit 20 that performs various processing, a management storage unit 21 that can store various information and programs, and a management communication unit 22 configured to communicate with a plurality of vacuum pumps 11.

[0033] The management-side storage unit 21 stores the device database 210, the information processing program 211, the classification data D4, and the transition probability matrix data D5. Details of the classification data D4 and the transition probability matrix data D5 will be described later.

[0034] Figure 4 is a data configuration diagram showing an example of the equipment database 210. The equipment database 210 registers equipment management information D0, setting data D1, and operation data D2 for each vacuum pump 11. In this case, the equipment management information D0, setting data D1, and operation data D2 are managed by the equipment ID. Equipment management information D0 includes, for example, the manufacturer, type, model name, installation date, and installation location of the vacuum pump 11. Note that the data registered as setting data D1 and operation data D2 may differ for each vacuum pump 11. Furthermore, the equipment database 210 may be stored in an external storage device.

[0035] The management communication unit 22 functions as a communication interface that sends and receives various types of information to and from the multiple vacuum pumps 11 via the network 4.

[0036] The management control unit 20 operates according to the information processing program 211 and functions as a data management unit 200, a reference data generation unit 201, a classification generation processing unit 202, a diagnostic data generation unit 203, a classification processing unit 204, a diagnostic processing unit 205, and a classification addition processing unit 206, as shown in Figure 3.

[0037] When the data management unit 200 receives setting data D1 and operation data D2 from the vacuum pump 11, it registers them in the equipment database 210. At that time, the setting data D1 and operation data D2 received from the vacuum pump 11 are assigned an equipment ID indicating the source vacuum pump 11, so the data management unit 200 uses this equipment ID to register the setting data D1 and operation data D2 in the equipment database 210. Note that the setting data D1 may be changeable, for example, via a settings screen, and the changed setting data D1 may be sent to the vacuum pump 11. Details of the other units 201 to 206 will be described later.

[0038] (Hardware configuration of each device) Figure 5 is a hardware configuration diagram showing an example of a computer 900. Each of the vacuum pump 11, information processing device 2, and maintenance terminal device 3 is configured with a general-purpose or dedicated computer 900.

[0039] As shown in Figure 5, the computer 900 comprises, as its main components, a bus 910, a processor 912, a memory 914, an input device 916, an output device 917, a display device 918, a storage device 920, a communication interface unit 922, an external device interface unit 924, an I / O device interface unit 926, and a media input / output unit 928. Note that the above components may be omitted as appropriate depending on the intended use of the computer 900.

[0040] The processor 912 is composed of one or more arithmetic processing units (CPU (Central Processing Unit), MPU (Micro-processing unit), DSP (digital signal processor), GPU (Graphics Processing Unit), etc.) and operates as a control unit that oversees the entire computer 900 (for example, the device control unit 111 or the management control unit 20). The memory 914 stores various information and programs 930 and is composed of volatile memory (DRAM, SRAM, etc.) that functions as main memory, and non-volatile memory (ROM), flash memory, etc.

[0041] The input device 916 consists of, for example, a keyboard, mouse, numeric keypad, or electronic pen, and functions as an input unit. The output device 917 consists of, for example, a sound (voice) output device or a vibration device, and functions as an output unit. The display device 918 consists of, for example, a liquid crystal display, an organic EL display, electronic paper, or a projector, and functions as an output unit. The input device 916 and the display device 918 may be configured integrally, such as a touch panel display. The storage device 920 consists of, for example, an HDD or SSD, and functions as a storage unit (for example, a device storage unit 112 or a management-side storage unit 21). The storage device 920 stores various information necessary for the execution of the operating system and the program 930.

[0042] The communication I / F unit 922 is connected by wired or wireless connection to a network 940 such as the Internet or an intranet (which may be the same as network 4 in Figure 1) and functions as a communication unit (for example, the equipment communication unit 113 or the management communication unit 22) that sends and receives data with other computers according to a predetermined communication standard. The external device I / F unit 924 is connected by wired or wireless connection to external devices 950 such as cameras, printers, scanners, and reader / writers and functions as a communication unit that sends and receives data with the external devices 950 according to a predetermined communication standard. The I / O device I / F unit 926 is connected to I / O devices 960 such as various sensors and actuators and functions as a communication unit that sends and receives various signals and data with the I / O devices 960, for example, detection signals from sensors and control signals to actuators. The media input / output unit 928 consists of, for example, a drive device such as a DVD drive or CD drive, a memory card slot, and a USB connector, and reads and writes data to media (non-temporary storage medium) 970 such as DVDs, CDs, memory cards, and USB memory.

[0043] In the computer 900 having the above configuration, the processor 912 calls and executes the program 930 stored in the storage device 920 in the memory 914, and controls various parts of the computer 900 via the bus 910. The program 930 may also be stored in the memory 914 instead of the storage device 920. The program 930 may be recorded on the media 970 in an installable or executable file format and provided to the computer 900 via the media input / output unit 928. The program 930 may also be provided to the computer 900 by downloading it via the network 940 through the communication interface unit 922. Furthermore, the computer 900 may implement the various functions realized by the processor 912 executing the program 930 using hardware such as an FPGA or ASIC.

[0044] The computer 900 is, for example, a stationary computer or a portable computer, and is an electronic device in any form. The computer 900 may be a client-type computer, a server-type computer, a cloud-type computer, or, for example, an embedded computer called a control panel, a controller (including a microcomputer, a programmable logic controller, a sequencer, etc.).

[0045] (Information Processing Method) Various processes (information processing methods) executed by each part 201-206 of the information processing apparatus 2 will be described. In the following, it will be described assuming that a large number of operation data D2 are already registered in the device database 210.

[0046] (First Classification Generation Process) FIG. 6 is a flowchart showing an operation example of the first classification generation process. FIG. 7 is an explanatory diagram showing a data example in the first classification generation process. In the first classification generation process, classification data D-4 is generated based on the operation data D2 for learning registered in the device database 210. Further, the first classification generation process may be re-executed when a predetermined number of operation data D2 for learning are newly registered in the device database 210, or may be re-executed in response to an instruction from a maintenance operator.

[0047] First, in step S100, the reference data generation unit 201 acquires operation data D2 in which measurement results of state quantities that vary according to the operation of the vacuum pump 11 measured by the sensor 110 provided in the vacuum pump 11 as an example of the device to be learned are recorded in time series. Specifically, the reference data generation unit 201 refers to the device database 210 and acquires the operation data D2 recorded when the vacuum pump 11 in a normal state as the vacuum pump 11 to be learned is operating, as the operation data D2 for learning. A plurality of operation data D2 may be acquired as the operation data D2 for learning.

[0048] Next, in step S110, the reference data generation unit 201 generates reference data D3 as shown in FIG. 7 by dividing the training operation data D2 acquired in step S100 based on the feature amount of the training operation data D2 at the transition time when the operation state of the vacuum pump 11 transitions.

[0049] As the feature amount of the training operation data D2, any feature amount can be used as long as it is a parameter capable of specifying the transition time of the operation state. For example, as the feature amount, an average value, a variance value, a maximum value, a minimum value, a median value, etc. in a predetermined evaluation period among the operation data D2 can be used. Further, as the feature amount, a correlation coefficient, a sum of step differences, a Mahalanobis distance, an Euclidean distance, a dynamic time warping method, a Fourier transform value, an average, a variance, a transition probability matrix, etc. generated by a hidden Markov model when comparing predetermined evaluation periods may be used. Furthermore, as the feature amount, a feature amount of an image converted from the operation data D2 by Fourier transform may be used, or a feature amount of an image converted from the operation data D2 by a recurrence plot may be used.

[0050] The training operation data D2 is time-series data in a predetermined operation period. However, for example, by dividing the time when the change amount of the feature amount exceeds a predetermined threshold as the transition time of the operation state, a plurality of reference data D3 are generated as time-series data in a period shorter than the predetermined operation period, as shown in FIG. 7.

[0051] Next, in step S120, the classification generation processing unit 202 classifies the reference data D3 generated in step S110 for each operation mode based on the feature amount of the reference data D3. For example, the classification generation processing unit 202 groups those with a high similarity between the feature amounts by using the feature amount of each reference data D3, similar to the feature amount when the reference data generation unit 201 generates the reference data D3, thereby classifying each reference data D3.

[0052] Any similarity parameter can be used as the similarity between each reference data D3, as long as it is a parameter that can calculate the similarity between them. For example, the similarity may be the difference between the features of each reference data D3, or it may be MAE (Mean Absolute Error), MSE (Mean Squared Error), RMSE (Root Mean Squared Error), COS similarity (Cosine Similarity), DTW (Dynamic Time Warping), Pearson correlation coefficient, Euclidean distance, etc.

[0053] As a result of the classification by the classification generation processing unit 202, each reference data D3 is classified into one of several operating modes, and these multiple operating modes are defined as an operating mode group. Furthermore, it is presumed that each operating mode corresponds to each manufacturing process performed by the semiconductor manufacturing apparatus 12.

[0054] Then, in step S130, the classification generation processing unit 202 generates a plurality of classification data D4, which will serve as the basis for classifying the operating state of the vacuum pump 11 by a plurality of operating modes, based on the reference data D3 classified for each operating mode in step S120, and stores them in the management side storage unit 21.

[0055] Classification data D4 is data that serves as a criterion for classifying the operating state of the vacuum pump 11 into one of several operating modes when diagnostic operating data D2 is acquired through the operation of the vacuum pump 11. Classification data D4 may be generated, for example, from representative reference data D3 among multiple reference data D3, or by applying statistical processing or synthesis processing to multiple reference data D3, or as feature quantities similar to those described above obtained from multiple reference data D3. Furthermore, classification data D4 may be generated with the time axis normalized, or with the state quantities normalized. In addition, classification data D4 may be generated by combining classification data from multiple vacuum pumps 11, or classification data from other vacuum pumps 11 may be reused.

[0056] In the example shown in Figure 7, multiple reference data D3, which are separated from the driving data D2, are classified into four driving modes A to D, and four classification data D4 corresponding to the four driving modes A to D are generated.

[0057] As described above, the first classification generation process shown in Figure 6 is performed. The classification generation processing unit 202 divides the operation data D2 at the point of transition of the operating state based on the features of the operation data D2, thereby generating classification data D4 which serves as the basis for classifying the operating state. Therefore, even if the operating state of the vacuum pump 11 transitions over an arbitrary period of time, classification data D4 which serves as the basis for classifying the operating mode can be appropriately generated using only the operation data D2.

[0058] (First Classification and Diagnostic Processing) Figure 8 is a flowchart showing an example of the operation of the first classification and diagnostic processing. Figure 9 is an explanatory diagram showing an example of data in the first classification and diagnostic processing. In the first classification and diagnostic processing, the abnormality status of the vacuum pump 11 to be diagnosed is diagnosed based on the diagnostic operation data D2 and the classification data D4 stored in the management-side storage unit 21, and abnormality data D7 is generated. Furthermore, the first classification and diagnostic processing is repeatedly executed each time new diagnostic operation data D2 is acquired.

[0059] First, in step S200, the diagnostic data generation unit 203 acquires operation data D2, which is a time-series recording of measurement results of state quantities that fluctuate according to the operation of the vacuum pump 11, measured by a sensor 110 provided on the vacuum pump 11 as an example of equipment to be diagnosed. Specifically, the diagnostic data generation unit 203 acquires the operation data D2 recorded when the vacuum pump 11 to be diagnosed is in operation as the operation data D2 for diagnosis. At this time, the diagnostic data generation unit 203 may receive the operation data D2 for diagnosis from the vacuum pump 11 to be diagnosed, or, if the operation data D2 for diagnosis is registered in the equipment database 210, it may refer to the equipment database 210.

[0060] Next, in step S210, the diagnostic data generation unit 203 generates diagnostic data D6 by dividing the diagnostic operation data D2 acquired in step S200 at the transition points where the operating state of the vacuum pump 11 changes, based on the characteristic quantities of the operation data D2.

[0061] The features of the diagnostic driving data D2 may be the same as those used in the first classification generation process, or different features may be used. The diagnostic driving data D2 is time-series data over a predetermined driving period. For example, by dividing the data at the point when the change in the features exceeds a predetermined threshold as a transition point in the driving state, diagnostic data D6 is generated as time-series data over a period shorter than the predetermined driving period, as shown in Figure 9. Although Figure 9 shows one set of diagnostic data D6, multiple sets of diagnostic data D6 may be generated.

[0062] Next, in step S220, the classification processing unit 204 calculates the similarity between the multiple classification data D4 and the diagnostic data D6 generated in step S210 for each operating mode. The multiple classification data D4 used are those generated by the first classification generation process and stored in the management-side storage unit 21.

[0063] Any similarity parameter can be used to calculate the similarity between classification data D4 and diagnostic data D6. For example, the similarity may be the difference between the features of classification data D4 and the features of diagnostic data D6, or it may be MAE, MSE, RMSE, COS similarity, DTW (dynamic time stretching), Pearson correlation coefficient, Euclidean distance, etc. If classification data D4 is generated in a normalized state, the diagnostic data D6 may also be normalized in the same way when calculating the similarity. In the example in Figure 9, the similarity to driving modes A and B is calculated to be "1", the similarity to driving mode C is calculated to be "5", and the similarity to driving mode D is calculated to be "3".

[0064] Next, in step S230, the classification processing unit 204 classifies the diagnostic data D6 generated in step S210 into groups of driving modes based on the similarity of each driving mode calculated in step S220. Specifically, the classification processing unit 204 calculates a classification probability for each driving mode based on the similarity of each driving mode, such that the greater the similarity, the higher the classification probability when classifying the diagnostic data D6 into a driving mode.

[0065] The classification probability can be calculated, for example, by summing the similarities for each driving mode and using the ratio of the similarity to that sum. In the example in Figure 9, the classification probabilities for driving modes A and B are "0.1" (= 1 / (1 + 1 + 3 + 5)), the classification probability for driving mode C is "0.5" (= 5 / (1 + 1 + 3 + 5)), and the classification probability for driving mode D is "0.3" (= 3 / (1 + 1 + 3 + 5)). Alternatively, the classification probability may be calculated for the diagnostic data D6 by calculating the distance from the centroid of each cluster for each driving mode and normalizing using the reciprocal of the distance to each cluster.

[0066] Next, in step S240, the diagnostic processing unit 205 calculates the degree of abnormality for each operating mode based on the classification probability for each operating mode calculated in step S230, such that the higher the classification probability, the lower the degree of abnormality of the vacuum pump 11.

[0067] The degree of abnormality can be calculated, for example, as the reciprocal of the classification probability. If the classification probability is "0", the degree of abnormality should be set to an extremely high value. In the example in Figure 9, the degree of abnormality for driving modes A and B is calculated to be "10", the degree of abnormality for driving mode C is "2", and the degree of abnormality for driving mode D is "3.3".

[0068] Next, in steps S250 to S252, the diagnostic processing unit 205 diagnoses the occurrence of an abnormality in the vacuum pump 11 based on the abnormality level for each operating mode calculated in step S240, specifically for the operating mode that satisfies predetermined conditions, or the cumulative abnormality level obtained by accumulating those abnormalities. The operating mode that satisfies the predetermined conditions may be, for example, the operating mode with the highest classification probability, an operating mode with a classification probability higher than a predetermined probability, or a specific operating mode. In this embodiment, steps S250 to S252 will be described in which the diagnosis is made based on the abnormality level for the operating mode with the highest classification probability.

[0069] In step S250, the diagnostic processing unit 205 diagnoses the occurrence of an abnormality in the vacuum pump 11 based on whether the abnormality level (which may be the cumulative abnormality level) for the operating mode with the highest classification probability among the abnormality levels for each operating mode calculated in step S240 exceeds a predetermined standard value. If the abnormality level exceeds the predetermined standard value (step S250: Yes), the diagnostic processing unit 205 diagnoses "abnormality occurred" in step S251, and if the abnormality level does not exceed the predetermined standard value (step S250: No), the diagnostic processing unit 205 diagnoses "no abnormality occurred" in step S252. The diagnostic processing unit 205 then generates abnormality data D7 indicating the occurrence of an abnormality in the vacuum pump 11 as a diagnostic result of the vacuum pump 11 and transmits it to the maintenance terminal device 3. The abnormality data D7 may also be registered in the equipment database 210.

[0070] In the example shown in Figure 9, the diagnostic data D6 is classified into four driving modes A to D, and when an abnormality level of "2" is used for driving mode C, which has the highest classification probability as the driving mode that satisfies the predetermined conditions, the abnormality level of "2" for driving mode C does not exceed the standard value (for example, "3.5"), and therefore the diagnosis is "no abnormality occurred."

[0071] As described above, the first classification and diagnostic processing shown in Figure 8 is performed. The classification processing unit 204 divides the operation data D2 at the point of transition of the operating state based on the features of the operation data D2, thereby generating diagnostic data D6, and the diagnostic data D6 is classified into multiple operating modes. Therefore, even if the operating state of the vacuum pump 11 transitions over an arbitrary period of time, the operating mode can be appropriately classified using only the operation data D2.

[0072] Furthermore, the diagnostic processing unit 205 calculates the degree of abnormality or cumulative degree of abnormality for the operating mode from the diagnostic data D6, and diagnoses the occurrence of an abnormality. Therefore, even if the operating state of the vacuum pump 11 changes over an arbitrary period, the occurrence of an abnormality can be appropriately diagnosed using only the operating data D2.

[0073] (Second Classification Generation Process) Figure 10 is a flowchart showing an example of the operation of the second classification generation process. Figure 11 is an explanatory diagram showing an example of data in the second classification generation process. The second classification generation process differs from the first classification generation process in that it generates first classification data D4 and second classification data D4 based on first and second driving data D2 for learning recorded by the first and second sensors 110, respectively. However, the basic operation is the same, so the explanation will focus on the differences.

[0074] First, in step S300, the reference data generation unit 201 acquires first operating data D2, in which a first state variable (e.g., motor power consumption) is recorded in time series by a first sensor 110 (e.g., motor power sensor) provided on the vacuum pump 11 to be learned, and second operating data D2, in which a second state variable (e.g., vibration value) is recorded in time series by a second sensor 110 (e.g., vibration sensor) provided on the vacuum pump 11 to be learned. Of the multiple sensors 110 provided on the vacuum pump 11, the first sensor 110 is selected to be a sensor suitable for classifying operating modes, and the second sensor 110 is selected to be a sensor suitable for diagnosing the occurrence of abnormalities.

[0075] Next, in step S310, the reference data generation unit 201 divides the first and second training operation data D2 and the second training operation data D2 acquired in step S300 based on the features of the first training operation data D2 at the point of transition of the operating state, thereby generating the first reference data D3 and the second reference data D3 as shown in Figure 11. At this time, the second reference data D3 is divided within the same time window as when the first training operation data D2 is divided into the first reference data D3.

[0076] Next, in step S320, the classification generation processing unit 202 classifies the first reference data D3 generated in step S310 for each operating mode based on the features of the first reference data D3, and also classifies the second reference data D3 generated in step S310 for each operating mode based on the classification result of the first reference data D3 associated with the second reference data D3.

[0077] Then, in step S330, the classification generation processing unit 202 generates a plurality of first classification data D4 and a plurality of second classification data D4 based on the first reference data D3 and the second reference data D3 classified for each operating mode in step S320, and stores them in the management-side storage unit 21.

[0078] In the example shown in Figure 11, multiple first reference data D3 and multiple second reference data D3, which are divided from the first operating data D2 and the second operating data D2 respectively, are classified into four operating modes A to D, and four first classification data D4 and four second classification data D4 corresponding to the four operating modes A to D are generated.

[0079] As described above, the second classification generation process shown in Figure 10 is performed. The classification generation processing unit 202 divides the first operating data D2 and the second operating data D2 at the point of transition of the operating state based on the feature quantities of the first operating data D2, thereby generating the first classification data D4 and the second classification data D4, which serve as the basis for classifying the operating state. Therefore, by using the first operating data D2 recorded by the first sensor 110, which is suitable for classification, and the second operating data D2 recorded by the second sensor 110, which is suitable for diagnosis, in combination, the classification accuracy when classifying the operating mode and the diagnostic accuracy when diagnosing the occurrence of abnormalities can be improved.

[0080] (Second Classification and Diagnostic Processing) Figure 12 is a flowchart showing an example of the operation of the second classification and diagnostic processing. Figure 13 is an explanatory diagram showing an example of data in the second classification and diagnostic processing. The second classification and diagnostic processing differs from the first classification and diagnostic processing in that it diagnoses the occurrence of abnormalities in the vacuum pump 11 to be diagnosed and generates abnormality data D7 based on the first and second operational data D2 for diagnosis recorded by the first and second sensors 110, respectively, and the first and second classification data D4 stored in the management-side storage unit 21. However, the basic operation is the same, so the explanation will focus on the differences.

[0081] First, in step S400, the diagnostic data generation unit 203 acquires first operating data D2 recorded by a first sensor 110 (for example, a motor power sensor) provided on the vacuum pump 11 to be diagnosed, and operating data D2 recorded by a second sensor 110 (for example, a vibration sensor) provided on the vacuum pump 11 to be diagnosed.

[0082] Next, in step S410, the diagnostic data generation unit 203 divides the first and second operational data D2 for diagnosis acquired in step S400 at the point of transition of the operational state, based on the feature quantities of the first operational data D2, thereby generating the first diagnostic data D6 and the second diagnostic data D6, as shown in Figure 13. At this time, the second diagnostic data D6 is divided within the same time window as when the first operational data D2 is divided into the first diagnostic data D6.

[0083] Next, in step S420, the classification processing unit 204 calculates a first similarity between the multiple first classification data D4 generated in the second classification generation process and the first diagnostic data D6 generated in step S410 for each operating mode.

[0084] Next, in step S430, the classification processing unit 204 calculates a second similarity between the multiple second classification data D4 generated in the second classification generation process and the second diagnostic data D6 generated in step S410 for each operating mode.

[0085] Next, in step S440, the classification processing unit 204 classifies the first diagnostic data D6 generated in step S410 into groups of driving modes based on the first similarity for each driving mode calculated in step S420. Specifically, the classification processing unit 204 calculates a first classification probability for each driving mode based on the first similarity for each driving mode, such that the higher the first similarity, the higher the first classification probability when classifying the first diagnostic data D6 into a driving mode.

[0086] Next, in step S450, the classification processing unit 204 classifies the second diagnostic data D6 generated in step S410 into groups of driving modes based on the second similarity for each driving mode calculated in step S430. Specifically, the classification processing unit 204 calculates a second classification probability for each driving mode based on the second similarity for each driving mode, such that the higher the second similarity, the higher the second classification probability when classifying the second diagnostic data D6 into a driving mode.

[0087] Next, in step S460, the diagnostic processing unit 205 calculates the degree of abnormality for each operating mode based on the second classification probability for each operating mode calculated in step S450, such that the higher the second classification probability, the lower the degree of abnormality of the vacuum pump 11.

[0088] Next, in steps S470 to S472, the diagnostic processing unit 205 diagnoses the occurrence of abnormalities in the vacuum pump 11 based on the degree of abnormality for each operating mode, which is determined by the second classification probability calculated in step S460, for operating modes in which the first classification probability calculated in step S440 satisfies a predetermined condition, or based on the cumulative degree of abnormality obtained by accumulating the said degree of abnormality. The operating mode that satisfies the predetermined condition may be, for example, the operating mode in which the first classification probability is the highest, or an operating mode in which the first classification probability is higher than a predetermined probability, or a specific operating mode. In this embodiment, steps S470 to S472 will be described in which the diagnosis is made based on the degree of abnormality for the operating mode in which the first classification probability is the highest.

[0089] In step S470, the diagnostic processing unit 205 diagnoses the occurrence of an abnormality in the vacuum pump 11 based on whether the degree of abnormality (which may be the cumulative degree of abnormality) based on the second classification probability for the operating mode with the highest first classification probability calculated in step S440 exceeds a predetermined standard value, among the degree of abnormality for each operating mode based on the second classification probability calculated in step S460. If the degree of abnormality based on the second classification probability exceeds the predetermined standard value (step S470: Yes), the diagnostic processing unit 205 diagnoses "abnormality occurred" in step S471, and if the degree of abnormality does not exceed the predetermined standard value (step S470: No), the diagnostic processing unit 205 diagnoses "no abnormality occurred" in step S472. The diagnostic processing unit 205 then generates abnormality data D7 indicating the occurrence of an abnormality in the vacuum pump 11 as a diagnostic result of the vacuum pump 11 and transmits it to the maintenance terminal device 3.

[0090] In the example shown in Figure 13, the first diagnostic data D6 and the second diagnostic data D6 are classified into four driving modes A to D, respectively. When an abnormality score of "10" based on the second classification probability for driving mode C, which has the highest classification probability in the first classification, is used as the driving mode that satisfies the predetermined conditions, the abnormality score of "10" based on the second classification probability for driving mode C exceeds the standard value (for example, "3.5"), and therefore, an "abnormality has occurred" diagnosis is made.

[0091] As described above, the second classification and diagnostic processing shown in Figure 12 is performed. The classification processing unit 204 classifies the first diagnostic data D6, generated by dividing the first operating data D2, into multiple operating modes. The diagnostic processing unit 205 calculates the degree of abnormality or cumulative degree of abnormality for each operating mode from the second diagnostic data D6, generated by dividing the second operating data D2, and diagnoses the occurrence of abnormalities. Therefore, by using the first operating data D2 recorded by the first sensor 110, which is suitable for classification, and the second operating data D2 recorded by the second sensor 110, which is suitable for diagnosis, the classification accuracy when classifying operating modes and the diagnostic accuracy when diagnosing the occurrence of abnormalities can be improved.

[0092] (Third Classification Generation Process) Figure 14 is a flowchart showing an example of the operation of the third classification generation process. Figure 15 is an explanatory diagram showing an example of data in the third classification generation process. The third classification generation process differs from the first classification generation process in that it generates classification data D4 and transition probability matrix data D5 based on the training operation data D2 registered in the equipment database 210, but the basic operation is the same, so we will explain the differences in detail.

[0093] First, in step S500, the reference data generation unit 201 acquires the operating data D2 recorded by the sensor 110 provided on the vacuum pump 11 to be learned.

[0094] Next, in step S510, the reference data generation unit 201 divides the training operation data D2 acquired in step S500 at each point in the operation state transition based on the features of the operation data D2, thereby generating reference data D3 before the transition and reference data D3 after the transition, as shown in Figure 15.

[0095] Next, in step S520, the classification generation processing unit 202 classifies the pre-transition reference data D3 and post-transition reference data D3 generated in step S510 for each operating mode based on the feature quantities possessed by the pre-transition reference data D3 and post-transition reference data D3.

[0096] Next, in step S530, the classification generation processing unit 202 generates a plurality of classification data D4 and transition probability matrix data D5 based on the reference data D3 before each transition and the reference data D3 after each transition, which were classified for each operating mode in step S520, and stores them in the management-side storage unit 21.

[0097] The transition probability matrix data D5 is data that records the transition probability for each combination of operating modes when transitioning from one operating mode to another for any combination of operating modes included in the operating mode group. The transition probability in the transition probability matrix data D5 is calculated in step S520 when the reference data D3 before each transition and the reference data D3 after each transition are classified for each operating mode, by aggregating the number of data for the operating modes into which the reference data D3 before the transition was classified and the number of data for the operating modes into which the reference data D3 after the transition was classified.

[0098] The classification generation processing unit 202 may generate anomaly matrix data based on the transition probability matrix data D5. The anomaly matrix data is data that records the degree of anomaly when transitioning from one operating mode to another for any combination of operating modes included in the operating mode group, for each combination of operating modes. The degree of anomaly in the anomaly matrix data is calculated for each combination of operating modes based on the transition probability in the transition probability matrix data D5, such that the higher the transition probability, the lower the transition anomaly of the vacuum pump 11. The transition anomaly in the anomaly matrix data can be calculated, for example, as the reciprocal of the transition probability in the transition probability matrix data D5.

[0099] In the example shown in Figure 15, the reference data D3 before the transition and the reference data D3 after the transition, which are separated from the driving data D2, are classified into four driving modes A to D, and four classification data D4 corresponding to the four driving modes A to D are generated. At the same time, a transition probability matrix data D5 is generated in which the transition probability for any combination of the four driving modes A to D is recorded for each combination of driving modes. In the transition probability matrix data D5 shown in Figure 15, the transition probability from driving mode C to driving mode A is recorded as "0.1", the transition probability from driving mode C to driving mode B is recorded as "0.1", and the transition probability from driving mode C to driving mode D is recorded as "0.8".

[0100] Furthermore, the example in Figure 15 illustrates the case where anomaly matrix data is generated based on the transition probability matrix data D5, recording the degree of transition anomaly for any combination of the four operating modes A to D for each combination of operating modes. In the anomaly matrix data shown in Figure 15, the degree of transition anomaly from operating mode C to operating mode A is recorded as "10", the degree of transition anomaly from operating mode C to operating mode B is recorded as "10", and the degree of transition anomaly from operating mode C to operating mode D is recorded as "1.3".

[0101] In the transition probability matrix data D5 shown in Figure 15, the transition time for the operating mode is assumed to be one, and the data is generated as a two-dimensional matrix. However, it may also be generated as a multi-dimensional matrix of three or more dimensions, assuming multiple transition times for the operating mode. For example, when the transition probability matrix data D5 is generated as a three-dimensional matrix, the classification probabilities for transitions such as the first operating mode, the second operating mode, and the third operating mode are recorded for each combination of the three operating modes.

[0102] As described above, the third classification generation process shown in Figure 14 is performed. The classification generation processing unit 202 divides the operation data D2 at the point of transition of the operating state and generates transition probability matrix data D5, which serves as the basis for classifying the transition status of the operating mode. Therefore, even if the operating state of the vacuum pump 11 transitions over an arbitrary period, the transition probability matrix data D5, which serves as the basis for classifying the transition status of the operating mode, can be appropriately generated using only the operation data D2.

[0103] (Third Classification and Diagnostic Processing) Figures 16A and 16B are flowcharts illustrating examples of the operation of the third classification and diagnostic processing. Figures 17 and 18 are explanatory diagrams showing examples of data in the third classification and diagnostic processing. The third classification and diagnostic processing differs from the first classification and diagnostic processing in that it diagnoses the occurrence of abnormalities in the vacuum pump 11 to be diagnosed and generates abnormality data D7 based on the diagnostic operation data D2 and the classification data D4 and transition probability matrix data D5 stored in the management-side storage unit 21. However, the basic operation is the same, so the differences will be explained in detail.

[0104] First, in step S600, the diagnostic data generation unit 203 acquires the operating data D2 recorded by the sensor 110 installed on the vacuum pump 11 to be diagnosed.

[0105] Next, in step S610, the diagnostic data generation unit 203 divides the diagnostic driving data D2 acquired in step S600 at the point of transition of the driving state based on the feature quantities of the driving data D2, thereby generating diagnostic data D6 before the transition and diagnostic data D6 after the transition, as shown in Figure 17.

[0106] Next, in step S620, the classification processing unit 204 calculates the pre-transition similarity between the multiple classification data D4 generated in the third classification generation process and the pre-transition diagnostic data D6 generated in step S610 for each operating mode.

[0107] Next, in step S621, the classification processing unit 204 calculates the similarity after transition between the multiple classification data D4 generated in the third classification generation process and the transitioned diagnostic data D6 generated in step S610 for each operating mode.

[0108] Next, in step S630, the classification processing unit 204 classifies the pre-transition diagnostic data D6 generated in step S610 into groups of driving modes based on the pre-transition similarity for each driving mode calculated in step S620. Specifically, the classification processing unit 204 calculates a pre-transition classification probability for each driving mode based on the pre-transition similarity for each driving mode, such that the greater the pre-transition similarity, the higher the pre-transition classification probability when classifying the pre-transition diagnostic data D6 into a driving mode.

[0109] Next, in step S631, the classification processing unit 204 classifies the post-transition diagnostic data D6 generated in step S610 into groups of driving modes based on the post-transition similarity for each driving mode calculated in step S621. Specifically, the classification processing unit 204 calculates a second post-transition classification probability for each driving mode based on the post-transition similarity for each driving mode, such that the higher the post-transition similarity, the higher the second post-transition classification probability when the post-transition diagnostic data D6 is classified into a driving mode.

[0110] Next, in step S640, the diagnostic processing unit 205 calculates the degree of abnormality before transition for each operating mode, based on the classification probability before transition for each operating mode calculated in step S630, such that the higher the classification probability before transition, the lower the degree of abnormality of the vacuum pump 11.

[0111] Next, in step S641, the diagnostic processing unit 205 calculates the degree of abnormality after transition for each operating mode, based on the classification probability after transition for each operating mode calculated in step S631, such that the higher the classification probability after transition, the lower the degree of abnormality of the vacuum pump 11.

[0112] Next, in steps S650 to S651, the diagnostic processing unit 205 diagnoses the occurrence of an abnormality in the vacuum pump 11 based on the abnormality level for each operating mode that satisfies predetermined conditions, or the cumulative abnormality level obtained by accumulating the abnormality levels, which are calculated in step S640.

[0113] Next, in steps S660 to S661, the diagnostic processing unit 205 diagnoses the occurrence of an abnormality in the vacuum pump 11 based on the abnormality level after the transition for each operating mode calculated in step S641, specifically for operating modes that meet predetermined conditions, or the cumulative abnormality level obtained by accumulating said abnormality levels.

[0114] Next, if the degree of abnormality before the transition does not exceed a predetermined standard value (step S650: No), and if the degree of abnormality after the transition does not exceed a predetermined standard value (step S660: No), the process proceeds to step S670.

[0115] Next, in step S670, the classification processing unit 204 identifies the transition probability when transitioning from an operating mode whose classification probability before the transition satisfies a predetermined condition to an operating mode whose classification probability after the transition satisfies a predetermined condition, based on the transition probability matrix data D5 generated in the third classification generation process. The operating mode that satisfies the predetermined condition may be, for example, the operating mode with the highest classification probability, an operating mode with a classification probability higher than the predetermined probability, or a specific operating mode. In this embodiment, the case in which the transition probability is identified when transitioning from an operating mode with the highest classification probability before the transition to an operating mode with the highest classification probability after the transition will be described.

[0116] In the examples in Figures 17 and 18, when the driving mode with the highest classification probability is used as the driving mode that satisfies the predetermined conditions, the highest classification probability of "0.5" for the diagnostic data D6 before the transition is driving mode C, and the highest classification probability of "0.5" for the diagnostic data D6 after the transition is driving mode D. Therefore, the transition probability when transitioning from driving mode C to driving mode D is specified as "0.8" in the transition probability matrix data D5.

[0117] Furthermore, if the transition probability matrix data D5 is generated by a multidimensional matrix of three or more dimensions, the diagnostic processing unit 205 can determine the transition probability by classifying the transition status of the driving mode according to the transition probability matrix data D5.

[0118] Next, in step S671, the diagnostic processing unit 205 calculates the degree of transition abnormality based on the transition probability identified in step S670, such that the higher the transition probability, the lower the degree of transition abnormality of the vacuum pump 11. The degree of transition abnormality can be calculated, for example, as the reciprocal of the transition probability.

[0119] Next, in steps S680 to S682, the diagnostic processing unit 205 diagnoses the occurrence of an abnormality in the vacuum pump 11 based on the transition abnormality degree calculated in step S671, or the cumulative transition abnormality degree obtained by accumulating the said transition abnormality degrees. In this embodiment, steps S680 to S682 when diagnosing based on the transition abnormality degree will be described.

[0120] In step S680, the diagnostic processing unit 205 diagnoses the status of an abnormality in the vacuum pump 11 depending on whether the transition abnormality score (which may also be the cumulative transition abnormality score) calculated in step S671 exceeds a predetermined standard value. If the transition abnormality score exceeds the predetermined standard value (step S680: Yes), the diagnostic processing unit 205 diagnoses "abnormality occurred" in step S681, and if the abnormality score does not exceed the predetermined standard value (step S680: No), the diagnostic processing unit 205 diagnoses "no abnormality occurred" in step S682. The diagnostic processing unit 205 then generates abnormality data D7 indicating the status of the abnormality in the vacuum pump 11 as a result of the diagnosis of the vacuum pump 11 and transmits it to the maintenance terminal device 3.

[0121] In the example shown in Figure 18, the abnormality level of "1.3" relative to the transition probability of "0.8" when transitioning from driving mode C to driving mode D does not exceed the standard value (for example, "5"), and therefore the diagnosis is "no abnormality occurred."

[0122] As described above, the third classification and diagnostic processing shown in Figures 16A and 16B is performed. The classification processing unit 204 classifies the pre-transition diagnostic data D6 and post-transition diagnostic data D6, which are generated by the division of the operation data D2, into multiple operation modes, thereby classifying the transition status of the operation modes. Furthermore, the diagnostic processing unit 205 calculates the transition abnormality degree or cumulative transition abnormality degree according to the transition status of the operation modes from the pre-transition diagnostic data D6 and post-transition diagnostic data D6, and diagnoses the occurrence of an abnormality. Therefore, even when the operating state of the vacuum pump 11 transitions over an arbitrary period, the occurrence of an abnormality can be appropriately diagnosed according to the transition status of the operating state.

[0123] (Fourth Classification, Diagnosis, and Classification Addition Process) Figure 19 is a flowchart showing an example of the operation of the fourth classification, diagnosis, and classification addition process. The fourth classification, diagnosis, and classification addition process differs from the first classification and diagnosis process in that, based on the diagnostic operation data D2 and classification data D4, it diagnoses the occurrence of abnormalities in the vacuum pump 11 to be diagnosed without calculating the degree of abnormality, generates abnormality data D7, and adds new classification data D4. However, the basic operation is the same, so we will explain mainly the differences. In particular, steps S700 to S730 shown in Figure 19 are the same as steps S200 to S230 shown in Figure 8, so we will explain from step S740 onwards.

[0124] In step S740, the diagnostic processing unit 205 diagnoses the occurrence of an abnormality in the vacuum pump 11 based on whether the classification probability for the operating mode with the highest classification probability among the classification probabilities for each operating mode calculated in step S730 falls below a predetermined standard value. If the classification probability does not fall below the predetermined standard value (step S740: No), the unit diagnoses "No abnormality occurred" in step S750. If the classification probability falls below the predetermined standard value (step S740: Yes), the unit determines that classification is not possible and proceeds to step S741.

[0125] Next, in step S741, the diagnostic processing unit 205 stores the diagnostic data D6 generated in step S710 in the management storage unit 21 as unclassified diagnostic data D6.

[0126] Next, in step S742, the diagnostic processing unit 205 diagnoses the occurrence of an abnormality in the vacuum pump 11 depending on whether the number of times it could not be classified, which was temporarily stored as unclassified diagnostic data D6 in step S741, exceeds a predetermined standard value. Note that the fourth classification process, diagnostic process, and classification addition process are repeatedly executed each time new diagnostic operation data D2 is acquired, so the number of times it could not be classified increases (increments) each time step S741 is performed on new diagnostic operation data D2. As a result, if the number of times it could not be classified does not exceed the predetermined standard value (step S742: No), it diagnoses "No abnormality occurred" in step S750, and if the number of times it could not be classified exceeds the predetermined standard value (step S742: Yes), it proceeds to step S760.

[0127] Next, in step S760, the diagnostic processing unit 205 diagnoses the occurrence of an abnormality in the vacuum pump 11 depending on whether the unclassified diagnostic data D6, which was temporarily stored in step S741 for the number of times it could not be classified, can be classified into a common operating mode. If the result is that it cannot be classified into a common operating mode (step S760: No), it diagnoses "abnormality occurred" in step S770, and if it can be classified into a common operating mode (step S770: Yes), it proceeds to step S780.

[0128] Next, in step S780, the classification addition processing unit 206 generates classification data D4, which will serve as the basis for a new operating mode, based on the unclassified diagnostic data D6 for the number of times classification was not possible, which was temporarily stored in step S741. This data is stored in the management-side storage unit 21, and the new operating mode is added to the group of operating modes. The unclassified diagnostic data D6 for the number of times classification was not possible consists of multiple diagnostic data D6 for each of the operating modes where the classification probability for the operating mode with the highest classification probability falls below a predetermined standard value, and where the classification probability has been calculated for each mode.

[0129] Furthermore, the classification addition processing unit 206 may, without using the number of times classification is not possible, generate classification data D4 that will serve as the basis for a new operating mode based on the diagnostic data D6 at the time the classification probability was calculated, when the classification probability for the operating mode with the highest classification probability falls below a predetermined standard value, and add the new operating mode to the group of operating modes.

[0130] Furthermore, when the classification addition processing unit 206 adds a new operating mode in step S780, it may confirm with the maintenance worker via the maintenance terminal device 3 whether or not to add the new operating mode. In this case, if the classification addition processing unit 206 receives confirmation from the maintenance terminal device 3 that the new operating mode will not be added, it will not add the new operating mode, and the diagnostic processing unit 205 may diagnose "an abnormality has occurred."

[0131] As described above, the fourth classification process, diagnostic process, and classification addition process shown in Figure 19 are performed. The classification processing unit 204 calculates a classification probability for each operating mode based on the diagnostic data D6 generated by the division of the operating data D2, and the diagnostic processing unit 205 diagnoses the occurrence of an abnormality based on the classification probability for each operating mode. Therefore, even if the operating state of the vacuum pump 11 changes over an arbitrary period, the occurrence of an abnormality can be appropriately diagnosed using only the operating data D2.

[0132] Furthermore, the classification addition processing unit 206 generates classification data D4 as classification data D4 based on the unclassified diagnostic data D6, and the new operating mode is added to the operating mode group. Therefore, even if an unclassified operating state occurs in the vacuum pump 11, the occurrence of the abnormality can be appropriately diagnosed using only the operating data D2.

[0133] (Other Embodiments) The present invention is not limited to the embodiments described above, and can be implemented with various modifications without departing from the spirit of the invention. All of these modifications are included in the technical concept of the present invention.

[0134] In the above embodiment, the device to be diagnosed by the information processing device 2 was described as a vacuum pump 11 of the semiconductor manufacturing system 10, but any device may be used as the device to be diagnosed. For example, the device to be diagnosed may be water supply and drainage equipment, air conditioning equipment, refrigeration and freezing equipment installed in buildings, etc., infrastructure equipment such as electricity, water supply, sewage, gas, and communications equipment including power generation and power distribution equipment, data center equipment, and plant equipment installed in plants such as petroleum refining, power generation, manufacturing, and chemical processes. Furthermore, the device to be diagnosed may not be a pump, but may be, but is not limited to, a compressor, a blower, a turbine, a motor, a cylinder, etc.

[0135] In the above embodiment, the case in which the information processing device 2 is composed of a single device has been described, but it may be composed of multiple devices. Furthermore, each of the parts 200 to 206 of the information processing device 2 may be provided by any of the equipment control unit 111 of the vacuum pump 11, the equipment control unit 121 of the semiconductor manufacturing apparatus 12, and the maintenance terminal device 3, and the various processing (information processing methods) by the information processing device 2 may be performed by any of the vacuum pump 11, semiconductor manufacturing apparatus 12, and maintenance terminal device 3. Moreover, in the above embodiment, the case in which the information processing device 2 is connected to multiple vacuum pumps 11 via a network 4 has been described, but the information processing device 2 may be connected to a single vacuum pump 11, or the information processing device 2 may be mounted on or built into the vacuum pump 11.

[0136] In the above embodiment, the diagnostic processing unit 205 calculated the degree of abnormality and the degree of transition abnormality to diagnose the occurrence of abnormalities in the vacuum pump 11. However, the diagnostic processing unit 205 may also diagnose the occurrence of abnormalities in the vacuum pump 11 based on classification probability and transition probability without calculating the degree of abnormality or the degree of transition abnormality.

[0137] For example, the diagnostic processing unit 205 may diagnose the occurrence of an abnormality in the vacuum pump 11 based on whether the classification probability for an operating mode that satisfies a predetermined condition, calculated by the classification processing unit 204 in step S230, falls below a predetermined standard value, or whether the number of times the classification probability falls below the standard value exceeds a predetermined standard value. The operating mode that satisfies the predetermined condition may, for example, be the operating mode with the highest classification probability, as in this embodiment, or an operating mode with a classification probability higher than the predetermined probability, or a specific operating mode.

[0138] Furthermore, the diagnostic processing unit 205 may diagnose the occurrence of an abnormality in the vacuum pump 11 depending on whether the transition probability identified by the diagnostic processing unit 205 in step S670 falls below a predetermined reference value, or whether the number of times the transition probability falls below the reference value exceeds a predetermined reference value.

[0139] In the above embodiment, the abnormal data D7 generated by the diagnostic processing unit 205 is provided to the maintenance terminal device 3 and displayed on the display screen of the maintenance terminal device 3. However, the abnormal data D7 can be used for any purpose, and may, for example, be used when formulating a maintenance work plan. Alternatively, the abnormal data D7 may be provided to the vacuum pump 11 or the semiconductor manufacturing apparatus 12 and displayed by the vacuum pump 11 or the semiconductor manufacturing apparatus 12.

[0140] In the above embodiment, the abnormal data D7 generated by the diagnostic processing unit 205 was described as being displayed on the maintenance terminal device 3 or registered in the equipment database 210. However, other data handled by the information processing device 2 may also be displayed on the maintenance terminal device 3 or registered in the equipment database 210. Examples of other data include the classification probability and transition probability for each operating mode calculated by the classification processing unit 204, and the degree of abnormality, the degree of transition abnormality, and the number of times classification is not possible for each operating mode calculated by the diagnostic processing unit 205. The above various data may be displayed as graphs showing changes over time, or as statistical values ​​obtained by statistical processing, by repeatedly executing various processes (information processing methods) by the information processing device 2 at different points in time. In this case, for example, the cumulative classification probability obtained by accumulating the classification probabilities, the cumulative transition probability obtained by accumulating the transition probabilities, the cumulative degree of abnormality obtained by accumulating the degree of abnormality, and the cumulative degree of transition abnormality obtained by accumulating the degree of transition abnormality may be displayed.

[0141] In the above embodiment, the case in which the information processing device 2 operates according to the flowcharts shown in Figures 6, 8, 10, 12, 14, 16A, 16B, and 19 has been described. However, the execution order of each step may be changed as appropriate, or some steps may be omitted. In the flowcharts of each figure, steps S100, S110, S300, S310, S500, and S510 correspond to the reference data generation process, steps S120, S130, S320, S330, S520, and S530 correspond to the classification generation process, steps S200, S210, S400, S410, S600, S610, S700, and S710 correspond to the diagnostic data generation process, steps S220, S230, S420 to S450, S620 to S631, S670, S720, and S730 correspond to the classification process, steps S240 to S251, S460 to S471, S640 to S661, S671 to S682, and S740 to S770 correspond to the diagnostic process, and step S780 corresponds to the classification addition process.

[0142] 1... Equipment management system, 2... Information processing device, 3... Maintenance terminal device, 4... Network, 10... Semiconductor manufacturing system, 11... Vacuum pump (equipment), 12... Semiconductor manufacturing equipment, 20... Management side control unit, 21... Management side storage unit, 22... Management side communication unit, 110... Sensor, 111... Equipment control unit, 112... Equipment storage unit, 113... Equipment communication unit, 114... Inverter, 115... Motor, 116... Rotor, 200... Data management unit, 201... Reference data generation unit, 202... Classification generation processing unit, 203... Diagnostic data generation unit, 204... Classification processing unit, 205... Diagnostic processing unit, 206... Classification addition processing unit, 210... Equipment database, 211... Information processing program

Claims

1. An information processing device comprising: a diagnostic data generation unit that generates diagnostic data by dividing operating data, which is recorded in time series from measurement results of state quantities that fluctuate according to the operation of the equipment measured by sensors provided on the equipment to be diagnosed, at transition points when the operating state of the equipment changes, based on the characteristic quantities of the operating data; and a classification processing unit that calculates the similarity between the diagnostic data and a plurality of classification data that serve as a criterion for classifying the operating state by a plurality of operating modes included in an operating mode group, for each operating mode, and classifies the diagnostic data for the operating mode group based on the similarity for each operating mode.

2. The information processing apparatus according to claim 1, wherein the classification processing unit calculates the classification probability for each driving mode based on the similarity for each driving mode such that the greater the similarity, the higher the classification probability when classifying the diagnostic data into the driving mode.

3. The information processing apparatus according to claim 2, comprising a diagnostic processing unit that calculates the degree of abnormality of the equipment based on the classification probability for the operating mode that satisfies predetermined conditions, such that the higher the classification probability, the lower the degree of abnormality of the equipment, and diagnoses the occurrence of abnormalities of the equipment based on the calculated degree of abnormality or the cumulative degree of abnormality obtained by accumulating the degrees of abnormality.

4. The information processing apparatus according to claim 2, further comprising a diagnostic processing unit that diagnoses the occurrence of an abnormality in the equipment, depending on whether the classification probability for the operating mode that satisfies predetermined conditions falls below a predetermined standard value, or whether the number of times the classification probability falls below the standard value exceeds a predetermined standard value.

5. The diagnostic data generation unit generates first diagnostic data and second diagnostic data by dividing first operating data, in which measurement results of a first state quantity measured by a first sensor are recorded in time series, and second operating data, in which measurement results of a second state quantity measured by a second sensor are recorded in time series, based on the feature quantities of the first operating data, at the transition point when the operating state changes. The classification processing unit calculates a first similarity between a plurality of first classification data that serve as a criterion for classifying the operating state by a plurality of operating modes and the first diagnostic data, and a second similarity between a plurality of second classification data that serve as a criterion for classifying the operating state by a plurality of operating modes and the second diagnostic data, for each operating mode. The information processing apparatus according to claim 1, comprising: a diagnostic processing unit that calculates the first classification probability and the second classification probability for each operating mode based on the first similarity and the second similarity for each operating mode such that the greater the first similarity and the second similarity, the higher the first classification probability and the second classification probability when classifying the first diagnostic data and the second diagnostic data to the operating mode, respectively; a diagnostic processing unit that diagnoses the occurrence of an abnormality in the equipment based on the calculated abnormality or the cumulative abnormality obtained by accumulating the abnormalities, based on the second classification probability for the operating mode in which the first classification probability satisfies predetermined conditions, such that the higher the second classification probability, the lower the degree of abnormality of the equipment; and a diagnostic processing unit that diagnoses the occurrence of an abnormality in the equipment based on the calculated abnormality or the cumulative abnormality obtained by accumulating the abnormalities.

6. The diagnostic data generation unit generates diagnostic data before the transition and diagnostic data after the transition by dividing the driving data at the transition point. The classification processing unit calculates the pre-transition similarity between the plurality of classification data and the pre-transition diagnostic data, and the post-transition similarity between the plurality of classification data and the post-transition diagnostic data for each driving mode. Based on the pre-transition similarity and post-transition similarity for each driving mode, the classification probability before the transition and the classification probability after the transition are calculated for each driving mode such that the greater the pre-transition similarity and the post-transition similarity, the higher the classification probability before the transition and the classification probability after the transition when classifying the pre-transition diagnostic data and the post-transition diagnostic data into the driving mode, respectively. The transition probability when transitioning from a driving mode whose pre-transition classification probability satisfies a predetermined condition to a driving mode whose post-transition classification probability satisfies a predetermined condition is identified based on transition probability matrix data recorded for each combination of driving modes included in the driving mode group, where the transition probability when transitioning from one driving mode to the other driving mode is recorded for each combination. The information processing apparatus according to claim 1.

7. The information processing apparatus according to claim 6, comprising a diagnostic processing unit that calculates the degree of abnormality of the equipment based on the transition probability such that the higher the transition probability, the lower the degree of abnormality of the equipment, and diagnoses the occurrence of an abnormality of the equipment based on the calculated degree of abnormality or the cumulative degree of abnormality obtained by accumulating the degrees of abnormality of the equipment.

8. The information processing apparatus according to claim 6, further comprising a diagnostic processing unit that diagnoses the occurrence of an abnormality in the equipment depending on whether the transition probability falls below a predetermined reference value, or whether the number of times the transition probability falls below the reference value exceeds a predetermined reference value.

9. An information processing apparatus according to claim 1, comprising: a reference data generation unit that generates reference data by dividing operating data, which is recorded in time series from measurement results of state quantities that fluctuate according to the operation of the equipment measured by sensors provided on the equipment to be learned, at transition points when the operating state of the equipment changes, based on the feature quantities of the operating data; and a classification generation processing unit that classifies the reference data for each operating mode based on the feature quantities of the reference data, and generates a plurality of classification data based on the reference data for each operating mode.

10. A reference data generation unit that generates first reference data and second reference data by dividing first operating data and second operating data, which are recorded in time series from measurement results of first and second state quantities that fluctuate in accordance with the operation of the device, measured by first and second sensors provided on the device to be learned, at transition points when the operating state of the device changes, based on the feature quantities of the first operating data, thereby generating first reference data and second reference data in association; and a classification generation processing unit that classifies the first reference data for each operating mode based on the feature quantities of the first reference data, and classifies the second reference data for each operating mode based on the classification results of the first reference data associated with the second reference data, and generates a plurality of first classification data and a plurality of second classification data based on the first reference data and the second reference data for each operating mode, respectively.

11. An information processing apparatus according to claim 6, comprising: a reference data generation unit that generates reference data before a transition and reference data after a transition by dividing the operation data, which is recorded in time series from measurement results of state quantities that fluctuate according to the operation of the equipment measured by a sensor provided on the equipment to be learned, at the transition point when the operating state of the equipment changes, based on the feature quantities of the operation data; and a classification generation processing unit that classifies the reference data before a transition and the reference data after a transition for each of the operating modes based on the feature quantities of the reference data before a transition and the reference data after a transition for each of the operating modes, and generates a plurality of classification data and transition probability matrix data based on the reference data before a transition and the reference data after a transition for each of the operating modes.

12. The information processing apparatus according to claim 2, further comprising a classification addition processing unit that, when the classification probability for the operating mode with the highest classification probability falls below a predetermined standard value, generates classification data that serves as a standard for a new operating mode based on the diagnostic data at the time the classification probability was calculated, and adds the new operating mode to the group of operating modes.

13. The information processing apparatus according to claim 12, wherein the classification addition processing unit generates classification data that serves as a basis for a new operating mode based on the multiple diagnostic data at the time the classification probability was calculated, when the number of times the classification probability for the operating mode with the highest classification probability falls below a predetermined threshold exceeds a predetermined threshold, and when the number of diagnostic data at the time the classification probability was calculated falls below a predetermined threshold, the number of times the classification probability for the operating mode with the highest classification probability falls below a predetermined threshold, and the number of times the classification probability for the multiple diagnostic data at the time the classification probability was calculated falls below a predetermined threshold, and adds the new operating mode to the group of operating modes.

14. A computer-based information processing method comprising: a diagnostic data generation step of generating diagnostic data by dividing operating data, which is recorded in time series from measurement results of state quantities that fluctuate according to the operation of the equipment measured by sensors provided on the equipment to be diagnosed, at transition points where the operating state of the equipment changes, based on the characteristic quantities of the operating data; and a classification processing step of calculating the similarity between the diagnostic data and a plurality of classification data that serve as a criterion for classifying the operating state by a plurality of operating modes included in an operating mode group, for each operating mode, and classifying the diagnostic data to the operating mode group based on the similarity for each operating mode.

Images