Apparatus, system, method and computer readable medium

Abstract

The present application provides a measurement data recording device, a generating device, a system, a device, a method, and a computer readable medium, the measurement data recording device including: an acquisition unit that acquires time-series measurement data that measures a state of an object; a recording unit that records the acquired measurement data; a measurement data transmission unit that transmits the measurement data recorded in the recording unit to a generating device that generates a judgment model that judges whether or not the state of the object is good; a model reception unit that receives the judgment model generated by the generating device; and a judgment unit that judges whether or not the state of the object is good based on newly acquired measurement data using the received judgment model.

Description

TECHNICAL FIELD

[0001] The present application relates to a measurement data recording device, a generating device, a system, a device, a method, and a computer-readable medium. BACKGROUND

[0002] Patent Document 1 describes “calculating a predicted value including at least one or more measurable values that can be measured using the generated learning model”.

[0003] PRIOR ART DOCUMENTS

[0004] Patent Document 1: Japanese Patent Publication No. 2019-124596

[0005] Patent Document 2: Japanese Patent Publication No. 2011-8735

[0006] Patent Document 3: Japanese Patent Publication No. Hei 7-280603 SUMMARY

[0007] In a first aspect of the present application, a measurement data recording device is provided. The measurement data recording device can include an acquisition unit that acquires measurement data of a time series of a state of an object. The measurement data recording device can include a recording unit that records the acquired measurement data. The measurement data recording device can include a measurement data transmission unit that transmits the measurement data recorded in the recording unit to a generating device that generates a judgment model that judges whether or not the state of the object is good. The measurement data recording device can include a model reception unit that receives the judgment model generated by the generating device. The measurement data recording device can include a judgment unit that judges whether or not the state of the object is good based on newly acquired measurement data using the received judgment model.

[0008] The measurement data recording device can include a decryption unit that decrypts the judgment model encrypted and received by the model reception unit.

[0009] Information indicating the legitimacy of the judgment model can be received together with the judgment model by the model reception unit, and the judgment model can be used by the judgment unit.

[0010] The judgment unit can judge whether or not the state of the object when the time series of measurement data is measured is good using the judgment model based on the newly acquired time series of measurement data.

[0011] The judgment unit can judge using a difference between a judgment threshold value in the judgment model and an index value that is acquired from the judgment model and indicates the state of the object based on a value corresponding to a measurement value of the measurement data being input, and output the judgment result and at least the most recent difference.

[0012] The determination unit can use the determination model to determine whether the state of the subject at the time of measurement of the measurement data is good or not each time new measurement data is acquired.

[0013] In a second aspect of the present application, a generation device is provided. The generation device can include an acquisition unit that acquires measurement data from a measurement data recording device that records measurement data regarding a time series of a state of a subject. The generation device can include a generation unit that generates a determination model for determining whether the state of the subject is good or not based on the measurement data using the acquired measurement data of the time series. The generation device can include a transmission unit that transmits the determination model to the measurement data recording device.

[0014] The generation device can include an encryption unit that encrypts the determination model generated by the generation unit. The transmission unit can transmit the encrypted determination model to the measurement data recording device.

[0015] The transmission unit can transmit information indicating the legitimacy of the determination model to the measurement data recording device together with the determination model.

[0016] The generation device can include a conversion unit that converts the determination model generated by the generation unit into code of a process executed in the measurement data recording device. The transmission unit can transmit the determination model converted by the conversion unit to the measurement data recording device.

[0017] The generation unit can have a learning processing unit that generates the determination model by learning processing using learning data containing the measurement data of the time series.

[0018] The measurement data of the time series can be measurement data in a process performed repeatedly in the subject each time. The generation unit can have a DTW processing unit that, for each kind of measurement data, performs processing of making the time width of measurement data of another time series consistent with one time series of measurement data acquired by the acquisition unit, determining a DTW path in which the distance between each point of the one time series of measurement data and each point of the measurement data of the other time series is the shortest, and calculating at least one of a parameter indicating the DTW distance between the one time series of measurement data and the measurement data of the other time series when the above processing is performed, a parameter indicating the number of measurement data in each point of the measurement data of the other time series that indicates measurement data shifted in the time axis direction in the DTW path, and a parameter indicating the number of measurement data that is not shifted in the time axis direction in the DTW path. The learning processing unit can generate the determination model using the learning data containing the at least one parameter.

[0019] The generation device can include an analysis section that performs cluster analysis on a plurality of time-series measurement data measured when the state of the subject is not good. The generation device can include a determination section that determines that one time-series measurement data should be classified in a cluster in accordance with the one time-series measurement data being judged as the state of the subject not being good by the judgment model.

[0020] The time-series measurement data can be measurement data in a process performed repeatedly in the subject each time. The generation section can have a statistical processing section that generates, for each kind of measurement data, an allowable range of measurement data over time in the process as the judgment model by statistical processing on a plurality of time-series measurement data acquired by the acquisition section.

[0021] The generation section can have a DTW processing section that, for each kind of measurement data, minimizes waveform differences of other time-series measurement data with respect to one time-series measurement data by dynamic time warping with respect to the one time-series measurement data to make time widths of the other time-series measurement data consistent. The statistical processing section can perform statistical processing on a plurality of time-series measurement data processed by the DTW processing section.

[0022] In a third aspect of the present application, a system is provided. The system can include the measurement data recording device of the first aspect. The system can include the generation device of the second aspect.

[0023] In a fourth aspect of the present application, a device is provided. The device can include an acquisition section that acquires measurement data from a measurement data recording device that records time-series measurement data on a state of a subject. The device can include a learning processing section that generates a judgment model that judges whether the state of the subject is good or not from the measurement data by learning processing using learning data containing the time-series measurement data. The device can include an analysis section that performs cluster analysis on a plurality of time-series measurement data measured when the state of the subject is not good. The device can include a determination section that determines that one time-series measurement data should be classified in a cluster in accordance with the one time-series measurement data being judged as the state of the subject not being good by the judgment model.

[0024] In a fifth aspect of the present invention, a method is provided. The method can include a measurement data acquisition stage that acquires measurement data that measures a time series of a state of an object. The method can include a measurement data recording stage that records the acquired measurement data. The method can include a measurement data transmission stage that transmits the measurement data recorded in the measurement data recording stage to a generation device that generates a judgment model that judges whether the state of the object is good or not. The method can include a model reception stage that receives the judgment model generated by the generation device. The method can include a judgment stage that judges whether the state of the object is good or not based on newly acquired measurement data using the received judgment model.

[0025] In a sixth aspect of the present invention, a method is provided. The method can include a measurement data acquisition stage that acquires measurement data from a measurement data recording device that records measurement data that measures a time series of a state of an object. The method can include a generation stage that generates a judgment model that judges whether the state of the object is good or not based on the acquired time series of measurement data. The method can include a transmission stage that transmits the judgment model to the measurement data recording device.

[0026] In a seventh aspect of the present invention, a method is provided. The method can include a measurement data acquisition stage that acquires measurement data from a measurement data recording device that records measurement data that measures a time series of a state of an object. The method can include a learning process stage that generates a judgment model that judges whether the state of the object is good or not based on a learning process using learning data that includes the time series of measurement data. The method can include an analysis stage that performs cluster analysis on a plurality of time series of measurement data measured when the state of the object is not good. The method can include a determination stage that determines a cluster to which one time series of measurement data, for which the state of the object is judged to be not good by the judgment model, should be classified.

[0027] In an eighth aspect of the present invention, a computer-readable medium having a program recorded thereon is provided. The computer can function as an acquisition unit that acquires measurement data that measures a time series of a state of an object by executing the program. The computer can function as a recording unit that records the acquired measurement data by executing the program. The computer can function as a measurement data transmission unit that transmits the measurement data recorded in the recording unit to a generation device that generates a judgment model that judges whether the state of the object is good or not by executing the program. The computer can function as a model reception unit that receives the judgment model generated by the generation device by executing the program. The computer can function as a judgment unit that judges whether the state of the object is good or not based on newly acquired measurement data using the received judgment model by executing the program.

[0028] In a ninth aspect of the present invention, a computer readable medium having a program recorded thereon is provided. A computer can function as a acquisition section that acquires measurement data from a measurement data recording device that records measurement data on a time series of a state of an object, by executing the program. The computer can function as a generation section that generates a judgment model that judges whether the state of the object is good or not from the measurement data, using the acquired measurement data on the time series, by executing the program. The computer can function as a transmission section that transmits the judgment model to the measurement data recording device, by executing the program.

[0029] In a tenth aspect of the present invention, a computer readable medium having a program recorded thereon is provided. A computer can function as a acquisition section that acquires measurement data from a measurement data recording device that records measurement data on a time series of a state of an object, by executing the program. The computer can function as a learning processing section that generates a judgment model that judges whether the state of the object is good or not from the measurement data, by using a learning processing of learning data that contains the measurement data on the time series, by executing the program. The computer can function as an analysis section that performs cluster analysis on a plurality of time series of measurement data measured when the state of the object is not good, by executing the program. The computer can function as a determination section that determines a cluster to which one time series of measurement data, which is judged that the state of the object is not good by inputting the one time series of measurement data to the judgment model, should be classified, by executing the program.

[0030] In addition, the above summary of the invention does not list all the necessary features of the invention. Furthermore, sub-combinations of these feature groups can also be inventions. BRIEF DESCRIPTION OF DRAWINGS

[0031] Figure 1 A system 1 of the first embodiment is shown.

[0032] Figure 2 An action of the system 1 is shown.

[0033] Figure 3 A transition of an operation index is shown.

[0034] Figure 4 A system 1A of the second embodiment is shown.

[0035] Figure 5 A distribution of the clustered batch file 430 is shown.

[0036] Figure 6 An example of a screen output by the generation device 5A is shown.

[0037] Figure 7Another example of a screen output by the generation device 5A.

[0038] Figure 8 A system 1B of the third embodiment is shown.

[0039] Figure 9 An alarm curve is shown.

[0040] Figure 10 An example of a computer 2200 that can implement the present application in whole or in part is shown.

[0041] Reference Signs

[0042] 1 system, 2 facility, 4 measurement data recording device, 5 generation device, 20 sensor, 21 equipment, 40 acquisition section, 41 measurement data recording section, 43 storage section, 45 communication section, 46 DTW processing section, 48 determination section, 51 communication section, 52 storage section, 53 generation section, 58A classification section, 430 batch processing file, 431 determination model, 451 measurement data transmission section, 452 model reception section, 453 decryption section, 510 acquisition section, 511 conversion section, 512 encryption section, 513 transmission section, 531 DTW processing section, 532 learning processing section, 533B statistical processing section, 581 analysis section, 582 setting section, 583 determination section, 2200 computer, 2201 DVD-ROM, 2210 main controller, 2212 CPU, 2214 RAM, 2216 graphics controller, 2218 display device, 2220 input / output controller, 2222 communication interface, 2224 hard disk drive, 2226 DVD-ROM drive, 2230 ROM, 2240 input / output chip, 2242 keyboard. DETAILED DESCRIPTION

[0043] The present application will be described below through embodiments of the application, but the following embodiments do not limit the application involved in the claims. Furthermore, not all combinations of features described in the embodiments are necessary for the solution of the application.

[0044] [1. First Embodiment]

[0045] [1.1. Configuration of System 1]

[0046] Figure 1 A system 1 of the first embodiment is shown. The system 1 supports status monitoring of an object, including a facility 2, a measurement data recording device 4, and a generation device 5 as examples of the object.

[0047] [1.1-1. Facility 2]

[0048] The facility 2 is provided with one or more sensors 20. A plurality of devices 21 can also be provided in the facility 2.

[0049] The facility 2 can be, for example, a factory or a complex device that complexes a plurality of devices 21. As a factory, in addition to a chemical or biological industrial factory, for example, a factory that manages and controls a wellhead and its periphery of a gas field or an oil field, a factory that manages and controls power generation of hydropower, thermal power, atomic power, and the like, a factory that manages and controls environmental power generation of solar power or wind power, and a factory that manages and controls waterworks or a dam, and the like can be listed.

[0050] [1.1-1-1. Devices 21]

[0051] Each device 21 is an instrument, a machine, or a device, and can be, for example, an actuator such as a valve, a pump, a heater, a fan, a motor, a switch, or the like that controls at least one physical quantity such as pressure, temperature, pH, speed, flow rate, and the like in a process of the facility 2. Each device 21 can be of a different kind from each other, or at least a part of two or more devices 21 can be of the same kind. In the present embodiment, as an example, the devices 21 can be controlled from the outside by a wired or wireless method, but the devices 21 can also be controlled manually.

[0052] [1.1-1-2. Sensors 20]

[0053] Each sensor 20 measures a state of the facility 2. The sensor 20 can measure at least one physical quantity such as pressure, temperature, pH, speed, flow rate, and the like. In addition, the sensor 20 can also measure a production amount of the facility 2, a proportion of a mixed impurity, an operation state of each device 21, and the like. Each sensor 20 can be of a different kind from each other, or at least a part of two or more sensors 20 can be of the same kind. Each sensor 20 can supply measurement data to the measurement data recording device 4.

[0054] In addition, communication between the sensor 20 and the measurement data recording device 4 can be performed, for example, by a wireless communication protocol of the ISA (International Society of Automation), and can be performed, as an example, by ISA100, HART (Highway Addressable Remote Transducer) (registered trademark), BRAIN (registered trademark), FOUNDATION Fieldbus, PROFIBUS, and the like.

[0055] [1.1-2. Measurement data recording device 4]

[0056] The measurement data recording device 4 records the measurement data of each sensor 20. The measurement data recording device 4 has an acquisition section 40, a measurement data recording section 41, a storage section 43, a communication section 45, a DTW processing section 46, and a determination section 48.

[0057] [1.1-2-1. Acquisition section 40]

[0058] The acquisition section 40 acquires the measurement data of the time series of the state of the facility 2 that is measured. The acquisition section 40 can sequentially acquire the measurement data from each sensor 20. The acquisition section 40 can acquire measurement data of a plurality of kinds (also referred to as channels). The channels of the measurement data can differ for each sensor 20 in this embodiment as an example, but can also differ depending on the physical quantity that is the object. The acquisition section 40 can supply the acquired measurement data to the measurement data recording section 41.

[0059] [1.1-2-2. Measurement data recording section 41]

[0060] The measurement data recording section 41 records the measurement data acquired by the acquisition section 40. The measurement data recording section 41 can record the measurement data in the storage section 43.

[0061] The measurement data recording section 41 can record the measurement data in the order of the time series. For example, the measurement data recording section 41 can collectively record the measurement data of the time series in a single file (also referred to as a batch file 430) each time a process that is repeatedly executed in the facility 2 (also referred to as a batch process) is performed. A plurality of measurement data of the time series (also referred to as batch data) of each channel in the batch process is stored in each batch file 430. The measurement data recording section 41 can file the measurement data supplied from the acquisition section 40 and record it in accordance with the input of a trigger signal.

[0062] Here, the batch process can be, for example, a manufacturing process of a product such as a tire, or a heating process of a cooked food or a furnace. Furthermore, the trigger signal can be supplied to the measurement data recording device 4 in accordance with the end of the batch process in the facility 2, can be supplied from the equipment 21 or the sensor 20 of the facility 2, or can be supplied from the outside (for example, from an operator).

[0063] The measurement data recording section 41 can attach the measurement time, the identification information of the sensor 20 that measured the measurement data, and record it in the storage section 43. The measurement time of the measurement data can be the time when the measurement data is measured, and can indicate the measurement time within the batch process. In this embodiment, the measurement time can indicate the elapsed time from the start time of the batch process as an example. The measurement time of the measurement data can also be the acquisition time when the measurement data is acquired by the acquisition section 40. In addition, the measurement time and the identification information of the sensor 20 can also be attached in advance to the measurement data supplied from the sensor 20.

[0064] Further, the measurement data recording section 41 can further attach various information in the measurement data, the batch data, the batch file 430 according to an operation from the outside by an operator or the like. As an example, the attached information can be a kind of a physical quantity indicated by the measurement data (as an example, a Celsius temperature), a unit (as an example, °C), various setting contents and control conditions in the facility 2, and presence or absence of generation of an alarm and a generation time, and the like.

[0065] [1.1-2-3. Storage section 43]

[0066] The storage section 43 stores various information. For example, the storage section 43 can store the measurement data (in the present embodiment, as an example, the batch file 430 in which time-series measurement data measured by each sensor 20 in the execution of the batch is summarized) supplied from the measurement data recording section 41, and the judgment model 431 supplied from the generation device 5 described later. In addition, the judgment model 431 can be stored in the storage section 43 according to supply from the generation device 5 to the measurement data recording device 4 via the communication section 45.

[0067] [1.1-2-3-1. Judgment model 431]

[0068] The judgment model 431 judges whether or not the state of the facility 2 is good according to the measurement data. The judgment model 431 can output a judgment result of whether or not the state is good according to input of the measurement data. In the present embodiment, as an example, the judgment model 431 can perform the good or not judgment according to the time-series measurement data (that is, the batch data) included in the batch file 430.

[0069] The judgment model 431 can be a functional expression in which a value corresponding to a measurement value of the measurement data is a variable. As an example, the functional expression of the judgment model 431 can be the following expression (1) or expression (2).

[0070] f(x) = A x X1 + B x X2 + C x X3 + D x X4... (1)

[0071] f(x) = (A x X1 + B x X2) / C x X3... (2)

[0072] Here, "A", "B",... in the expression are coefficients, and can indicate a specific numerical value such as "4.3", "3.5", and the like. Further, "X1", "X2",... in the expression are values corresponding to the measurement value of the measurement data, and can be the measurement value itself of the measurement data of each sensor or each measurement time, can be an average or a variance of the measurement value included in the batch data, or can be a parameter calculated by the DTW processing section 46 described later.

[0073] The value obtained from the judgment model 431 (as an example of a function in the present embodiment) according to the measured value of the measurement data can be an example of an index value indicating the state of the facility 2 (also referred to as a state index value). The judgment model 431 can output a judgment result corresponding to the difference (also referred to as an operation index) between the state index value and a judgment threshold value. In the present embodiment, as an example, the judgment model 431 can use zero as the judgment threshold value, and output a judgment result that the state is good according to the operation index being a positive value.

[0074] In addition, the judgment model 431 stored in the storage section 43 can be expressed in a code of a process executable in the measurement data recording device 4. In the present embodiment, as an example, the judgment model 431 is expressed in C language, but can be expressed in another language such as Basic.

[0075] [1.1-2-4. Communication section 45]

[0076] The communication section 45 performs communication with the generation device 5. The communication section 45 can perform wireless communication, or can perform wired communication. The communication section 45 can perform communication via a relay device (as an example, a mobile terminal such as a smartphone, a fixed gateway device). The communication section 45 has a measurement data transmission section 451, a model reception section 452, and a decryption section 453.

[0077] [1.1-2-4-1. Measurement data transmission section 451]

[0078] The measurement data transmission section 451 transmits the measurement data recorded in the storage section 43 to the generation device 5. The measurement data transmission section 451 can also transmit the measurement data as a batch processing file 430.

[0079] [1.1-2-4-2. Model reception section 452]

[0080] The model reception section 452 receives the judgment model 431 generated by the generation device 5. The model reception section 452 can be able to use the judgment model 431 from the judgment section 48 according to information indicating the legality of the judgment model 431 received together with the judgment model 431. For example, the model reception section 452 can supply the judgment model 431 to the storage section 43 according to the information indicating the legality received together with the judgment model 431. However, the model reception section 452 can also store the judgment model 431 in the storage section 43 regardless of whether the information indicating the legality is received, and set the judgment model 431 to be accessible from the judgment section 48 according to the information indicating the legality being received. The information indicating the legality can be an electronic signature, or can be header information attached to the beginning of a communication packet.

[0081] In addition, in the present embodiment, as an example, the model receiving section 452 receives the encrypted judgment model 431. Therefore, the model receiving section 452 can supply the received judgment model 431 to the storage section 43 via the decryption section 453.

[0082] [1.1-2-4-3. Decryption section 453]

[0083] The decryption section 453 decrypts the judgment model 431 that is encrypted and received by the model receiving section 452. The decryption section 453 can perform decryption using a decryption key corresponding to the key with which encryption is performed in the generation device 5. The decryption section 453 can supply the decrypted judgment model 431 to the storage section 43.

[0084] [1.1-2-5. DTW processing section 46]

[0085] The DTW processing section 46 performs DTW processing on the time-series measurement data (i.e., batch data). The DTW processing can be processing in which, for each channel of the measurement data, the time width of the other time-series measurement data (also referred to as object batch data) is made uniform with respect to one time-series measurement data (also referred to as reference batch data), and the distance of each point of one time-series measurement data and each point of the other time-series measurement data is determined to be the minimum DTW path. On the basis of this or instead of this, the DTW processing can be processing in which the time width of the object batch data is made uniform with respect to the reference batch data, and the waveform difference between the reference batch data and the object batch data is minimized.

[0086] Here, the object batch data can be batch data included in the latest batch file 430. The reference batch data can be selected from batch data in which the facility 2 is in good condition, and as an example, can be batch data selected by an operator, can be batch data in which the time width from the beginning to the end among a plurality of batch data is a central value, or can be batch data in which the sum of the DTW distances from other batch data is the minimum among a plurality of batch data. In the case where the channels of the measurement data are a plurality, in other words, in the case where a plurality of batch data are included in each batch file 430, the reference batch data of each channel can be batch data included in the same batch file 430.

[0087] The DTW processing section 46 can read out the reference batch processing data and the object batch processing data from the storage section 43 and perform DTW processing in accordance with the input of the trigger signal. The DTW processing section 46 can supply the object batch processing data for which the waveform difference with respect to the reference batch processing data is minimized (also referred to as the object batch processing data for which DTW processing has been performed) to the determination section 48. In addition, the DTW processing section 46 can also supply at least one parameter calculated through DTW processing to the determination section 48.

[0088] The parameter calculated by the DTW processing section 46 can be the DTW distance between the reference batch processing data and the object batch processing data. On this basis or instead, the parameter calculated by the DTW processing section 46 can be the number of measurement data included in the measurement data of each point of the object batch processing data that is shifted in the time axis direction in the DTW path. On this basis or instead, the parameter calculated by the DTW processing section 46 can be the number of measurement data included in the measurement data of each point of the object batch processing data that is not shifted in the time axis direction in the DTW path. The shift of the measurement data in the time axis direction refers to, as an example in the present embodiment, a shift of the measurement time indicating the measurement timing backward, but on this basis or instead, can also be a shift of the measurement time indicating the measurement timing forward.

[0089] [1.1-2-6. Determination section 48]

[0090] The determination section 48 determines whether the state of the facility 2 is good or not based on the newly acquired measurement data using the determination model 431 in the storage section 43. The determination section 48 can determine whether the state of the facility 2 at the time of measurement of the object batch processing data is good or not based on the newly acquired time series measurement data, that is, the object batch processing data. For example, the determination section 48 can determine based on the batch processing data for which DTW processing has been performed by the DTW processing section 46. In addition, the determination section 48 can determine based on each batch processing data in the latest batch processing file 430, or based on the values (the measurement values themselves, the average or variance of the measurement values included in the batch processing data, the parameters calculated by the DTW processing section 46) corresponding to each measurement data in the batch processing data. For example, the determination section 48 can determine using an operation index that is the difference between the determination threshold in the determination model 431 and the state index value acquired from the determination model 431 based on the measurement value of each measurement data, and in the present embodiment, as an example, can determine that the state of the facility 2 is good in the case where the operation index is positive, and determine that the state is not good in the case where the operation index is negative. In addition, in the case where the average or variance of the measurement values included in the batch processing data is used for determination, the determination section 48 can calculate the same.

[0091] The determination unit 48 can output the determination result. Further, the determination unit 48 can output at least one of the latest operation indexes. As an example, the determination unit 48 can output the operation indexes for the latest batch processing, or the operation indexes for the latest two batch processings, respectively, and can further output the operation indexes for the batch processing performed four times before that. The latest two batch processings can be the last batch processing and the batch processing performed one time before the last batch processing.

[0092] [1.1-3. Generation device 5]

[0093] The generation device 5 generates the determination model 431. The generation device 5 has a communication unit 51, a storage unit 52, and a generation unit 53. In addition, the generation device 5 can be one or a plurality of computers, can be constituted by a PC or the like, and can be realized by cloud computing.

[0094] [1.1-3-1. Communication unit 51]

[0095] The communication unit 51 performs communication with the measurement data recording device 4. The communication unit 51 can perform wireless communication, or can perform wired communication. The communication unit 51 has a retrieval unit 510, a conversion unit 511, an encryption unit 512, and a transmission unit 513.

[0096] [1.1-3-1-1. Retrieval unit 510]

[0097] The retrieval unit 510 retrieves the measurement data from the measurement data recording device 4. The retrieval unit 510 can supply the retrieved measurement data to the storage unit 52.

[0098] [1.1-3-1-2. Conversion unit 511]

[0099] The conversion unit 511 converts the determination model 431 generated by the generation unit 53 described later into a code (as an example, a code in C language in the present embodiment) of a process executed in the measurement data recording device 4. The conversion unit 511 can supply the converted determination model 431 to the encryption unit 512.

[0100] [1.1-3-1-3. Encryption unit 512]

[0101] The encryption unit 512 encrypts the determination model 431. In the present embodiment, as an example, the encryption unit 512 can encrypt the determination model 431 encoded by the conversion unit 511. The encryption unit 512 can supply the encrypted determination model 431 to the transmission unit 513.

[0102] [1.1-3-1-4. Transmission unit 513]

[0103] The transmission section 513 transmits the judgment model 431 to the measurement data recording apparatus 4. The transmission section 513 can transmit the judgment model 431 converted by the conversion section 511. The transmission section 513 can transmit the judgment model 431 encrypted by the encryption section 512.

[0104] Further, the transmission section 513 can transmit information indicating the legality of the judgment model 431 together with the judgment model 431. The transmission section 513 can transmit an electronic signature indicating the legality, header information of a communication packet, together with the judgment model 431.

[0105] [1.1-3-2. Storage section 52]

[0106] The storage section 52 stores various information. For example, the storage section 52 can store the measurement data supplied from the acquisition section 510 and the judgment model 431 supplied from the generation section 53.

[0107] The storage section 52 can store the measurement data as a batch file 430. In the measurement data stored in the storage section 52, a label indicating whether the state of the facility 2 at the time of measuring the measurement data is good or not can be attached. The label can be set by an operator via an input section not shown, or can be attached in advance by the measurement data recording section 41 of the measurement data recording apparatus 4 as information indicating whether an alarm has occurred or not.

[0108] The judgment model 431 stored in the storage section 52 can be represented not in a code of a process executable in the measurement data recording apparatus 4, but in a code of a process executable in the generation apparatus 5.

[0109] [1.1-3-3. Generation section 53]

[0110] The generation section 53 generates the judgment model 431 using the time-series measurement data acquired by the acquisition section 510. The generation section 53 has a DTW processing section 531 and a learning processing section 532.

[0111] [1.1-3-3-1. DTW processing section 531]

[0112] The DTW processing section 531 can perform DTW processing on the time-series measurement data (i.e., batch data). The DTW processing section 531 can perform, for each channel of the measurement data, processing of aligning the time width of one time-series measurement data as the target batch data with respect to the other time-series measurement data as the reference batch data read out from the storage section 52, determining a DTW path in which the distance between each point of the reference batch data and each point of the target batch data becomes the minimum, and calculating at least one parameter. In addition, the DTW processing section 531 can perform processing of aligning the time width of the target batch data with respect to the reference batch data, and minimizing the waveform difference between the reference batch data and the target batch data.

[0113] The DTW processing section 531 can supply, for each batch file 430, the plurality of batch data on which the DTW processing has been performed and the calculated parameter to the learning processing section 532. The parameter calculated by the DTW processing section 531 can be the same kind of parameter as the parameter calculated by the DTW processing section 46 of the measurement data recording apparatus 4 (as an example in the present embodiment, the DTW distance, the number of measurement data shifted in the time axis direction in the DTW path, and the number of measurement data not shifted in the time axis direction in the DTW path).

[0114] Here, the reference batch data can be selected from the batch data in which the facility 2 is in good condition, and as an example, can be batch data selected by an operator, batch data in which the time width from the beginning to the end among a plurality of batch data is a central value, or batch data in which the sum of the DTW distance with respect to other batch data among a plurality of batch data is the minimum. The reference batch data used by the DTW processing section 46 of the measurement data recording apparatus 4 and the reference batch data used by the DTW processing section 531 of the generation apparatus 5 can be the same or different. In the case where the channels of the measurement data are a plurality of channels, in other words, in the case where a plurality of batch data is included in each batch file 430, the reference batch data for each channel can be batch data included in the same batch file. The target batch data can be time-series measurement data in which the facility 2 is in good condition, or time-series measurement data in which the facility 2 is in poor condition.

[0115] [1.1-3-3-2. Learning processing section 532]

[0116] The learning processing section 532 generates the determination model 431 by using a learning process of learning data containing time-series measurement data (i.e., batch data). In the measurement data contained in the learning data, a label indicating whether the state of the facility 2 is good or not can be attached. Further, in the learning data, for each channel of the measurement data, an average or a variance of the time-series measurement data, a parameter calculated by the DTW processing section 531 (as an example in the present embodiment, a DTW distance, a number of measurement data shifted in the time axis direction in a DTW path, a number of measurement data not shifted in the time axis direction in the DTW path) can be contained. In a case where the average or the variance of the measurement data is contained in the learning data, the learning processing section 532 can calculate it and contain it in the learning data. The learning processing section 532 can supply the generated determination model 431 to the storage section 52 to be stored, and to the conversion section 511 of the communication section 51.

[0117] According to the measurement data recording apparatus 4 in the system 1 described above, the measurement data is transmitted to the generation apparatus 5 that generates the determination model 431 determining whether the state of the facility 2 is good or not, and the determination model 431 received from the generation apparatus 5 is used to determine whether the state of the facility 2 is good or not based on newly acquired measurement data. Therefore, it is possible to determine whether the state of the facility 2 is good or not using the generated determination model 431 without generating the determination model 431 in the measurement data recording apparatus 4.

[0118] Further, since the encrypted determination model 431 is decrypted, it is possible to prevent the content of the determination model 431 received by the measurement data recording apparatus 4 from being changed. Therefore, it is possible to prevent the determination model 431 changed by a third party having malice from being used in the determination section 48.

[0119] Further, according to the information indicating the legitimacy of the determination model 431 received together with the determination model 431, it is possible to use the determination model 431 from the determination section 48, and therefore, it is possible to prevent the determination model 431 generated by a third party having malice from being used in the determination section 48.

[0120] Further, according to the newly acquired time-series measurement data, whether the state of the facility 2 at the time of measuring the time-series measurement data is good or not is determined, and therefore, compared to a case where determination is made based on measurement data at one time, it is possible to accurately determine the state of the facility 2.

[0121] Further, determination is made using the difference between the determination threshold in the determination model 431 and the state index value of the facility 2, i.e., the operation index, and the determination result and at least the latest operation index are output, and therefore, it is possible to accurately predict the future state of the facility 2.

[0122] Further, according to the generation device 5 in the system 1, the measurement data is acquired from the measurement data recording device 4, the judgment model 431 is generated, and is transmitted to the measurement data recording device 4. Therefore, it is possible to generate the judgment model 431 in the measurement data recording device 4, and to judge whether the state of the facility 2 is good or not using the generated judgment model 431.

[0123] Further, since the encrypted judgment model 431 is transmitted, it is possible to prevent the content of the judgment model 431 from being changed before being received by the measurement data recording device 4. Therefore, it is possible to prevent the judgment model 431 changed by a third party having malice from being used in the judgment section 48.

[0124] Further, since the information indicating the legitimacy of the judgment model 431 is transmitted together with the judgment model 431, it is possible to prevent the judgment model 431 generated by a third party having malice from being used in the judgment section 48.

[0125] Further, the generated judgment model 431 is converted into a code of a process executable in the measurement data recording device 4 and is transmitted, and therefore it is possible to omit the encoding of the judgment model 431 on the measurement data recording device 4 side.

[0126] Further, the judgment model 431 is generated by using a learning process of learning data containing the measurement data in time series, and therefore, compared to a case where the measurement data of one time is used for the learning data, it is possible to generate the judgment model 431 capable of accurately judging the state of the facility 2.

[0127] Further, at least one of the parameter indicating the DTW distance, the parameter indicating the number of measurement data shifted in the time axis direction in the DTW path, and the parameter indicating the number of measurement data not shifted in the time axis direction in the DTW path is used to generate the judgment model 431. Therefore, unlike a case where the judgment model 431 is generated using only the measurement data of each time contained in the measurement data in time series, the average, or the variance thereof, it is possible to generate the judgment model 431 taking into account the deviation of the measurement time of each measurement data, and therefore it is possible to improve the judgment accuracy of the judgment model 431.

[0128] [1.2. Action of the system 1]

[0129] Figure 2 The action of the system 1 is indicated. The measurement data recording device 4 and the generation device 5 of the system 1 generate the judgment model 431 by performing the processes of steps S11 to S31, and judge whether the state of the facility 2 is good or not. In addition, the action can be started in conjunction with the start of the facility 2.

[0130] In step Sll, the acquisition section 40 of the measurement data recording apparatus 4 acquires the time-series measurement data that measures the state of the facility 2 from each sensor 20.

[0131] In step S13, the measurement data recording section 41 records the measurement data acquired by the acquisition section 40 in the storage section 43. The measurement data recording section 41 can store the batch file 430 of the measurement data in the storage section 43 each time a batch process is performed by the facility 2.

[0132] In step S15, the measurement data transmission section 451 transmits the measurement data recorded in the storage section 43 to the generation apparatus 5. In the present embodiment, as an example, the measurement data transmission section 451 transmits the batch file 430 of the measurement data respectively according to an operation from the outside by an operator or the like, but can also transmit each time the batch file 430 is created, and can also periodically transmit the created batch file 430. Thereby, in step S17, the acquisition section 510 of the generation apparatus 5 acquires the measurement data from the measurement data recording apparatus 4. The acquisition section 510 can supply the acquired measurement data to the storage section 52.

[0133] In step S19, the generation section 53 generates the judgment model 431 using the acquired time-series measurement data. In the present embodiment, as an example, the generation section 53 generates the judgment model 431 according to an operation from the outside by an operator or the like, but can also generate the judgment model 431 according to the batch file 430 stored in the storage section 52 reaching a reference number.

[0134] In the generation of the judgment model 431, the DTW processing section 531 of the generation section 53 can calculate the DTW distance, the number of measurement data shifted in the time axis direction in the DTW path, and the number of measurement data not shifted in the time axis direction in the DTW path by performing, for each channel of the measurement data, a process of minimizing the waveform difference of the other time-series measurement data as the object batch data with respect to the measurement data as the reference batch data read out from the storage section 52 by the dynamic time warping method.

[0135] Further, the learning processing section 532 can generate the judgment model 431 by a learning process using learning data containing the time-series measurement data (i.e., the batch data). In the measurement data contained in the learning data, a label indicating whether the state of the facility 2 is good or not can be attached. In the learning data, the average or variance of the time-series measurement data, and the parameters calculated by the DTW processing section can be contained for each channel of the measurement data.

[0136] The learning processing section 532 can store at least one functional expression (as an example, the above-described expressions (1), (2)) that becomes the determination model 431 in advance, and can generate the determination model 431 by adjusting the coefficients of the functional expression through learning processing. For example, the learning processing section 532 can adjust the coefficients of the relational expression so as to improve the correlation between the difference between the value of the formula when the value corresponding to the measurement values of the measurement data in each batch processing file 430 is input to the functional expression of the determination model 431 and the determination threshold and the good or bad state of the facility 2 indicated by the label attached to the batch processing file 430.

[0137] In addition, the value corresponding to the measurement values of the measurement data can be the measurement values themselves of the measurement data, can be the average or variance of the measurement values of each measurement data included in the batch processing data, or can be a parameter calculated by the DTW processing section 531. In the case where the learning processing section 532 stores a plurality of functional expressions, the functional expression to be adjusted can be automatically selected, or can be selected according to an operation from the outside by an operator or the like.

[0138] After the determination model 431 is generated, in step S21, the transmission section 513 transmits the determination model 431 to the measurement data recording apparatus 4. In the present embodiment, as an example, the transmission section 513 can transmit the determination model 431 converted into a code executable in the measurement data recording apparatus 4 by the conversion section 511 and encrypted by the encryption section 512 together with information indicating the legality (as an example, an electronic signature, header information attached to the beginning of a communication packet).

[0139] In step S23, the model receiving section 452 of the measurement data recording apparatus 4 receives the determination model 431 generated by the generation apparatus 5. The model receiving section 452, upon receiving the determination model 431 together with information indicating the legality of the determination model 431, stores the determination model 431 in the storage section 43 after decrypting the determination model 431 in the decryption section 453, and can be utilized from the determination section 48.

[0140] In step S25, the acquisition section 40 re-acquires the time-series measurement data, and in step S27, the measurement data recording section 41 records the acquired measurement data in the storage section 43. The measurement data recording section 41 can store the batch processing file 430 of the measurement data in the storage section 43 each time a batch processing is performed by the facility 2.

[0141] In step S29, the determination section 48 determines the state of the facility 2 to be good or bad using the received determination model 431 based on the batch processing file 430 of the newly acquired measurement data. The determination section 48 can determine based on each batch processing data in the batch processing file 430 in which DTW processing is performed by the DTW processing section 46. In the present embodiment, as an example, the determination section 48 can determine using the performance index.

[0142] Also, in step S31, the judging section 48 outputs the judgment result and at least the latest operation index. The judging section 48 can output the progress of the latest multiple operation indexes. After the process of step S31 ends, the measurement data recording device 4 can cause the process to shift to the above-described step S25. The latest multiple operation indexes can be multiple operation indexes in order close to the current time.

[0143] [1.3. Progress of operation index]

[0144] Figure 3 The graph indicates the progress of the operation index. The vertical axis in the graph indicates the operation index, and the horizontal axis indicates the elapsed time when the batch process is repeatedly executed or the number of times of the batch process. Further, the solid line indicates the operation index corresponding to the measurement data, and the broken line indicates the operation index predicted by the operator at the time of the next batch process. Thus, by the progress of the operation index, it is possible to predict whether the facility 2 is good or not.

[0145] [1.4. Modification]

[0146] In addition, in the above-described first embodiment, it is described that the model receiving section 452 of the measurement data recording device 4 receives the judgment model 431 from the generation device 5, but it is also possible to receive parameters indicating the contents of the judgment model 431. For example, in the storage section 43 of the measurement data recording device 4, multiple function expressions into which coefficients can be substituted can be stored, and the model receiving section 452 can receive the expression number of any one of these function expressions and the values of the respective coefficients in the function expression of the expression number from the generation device 5. In this case, the measurement data recording device 4 can generate the judgment model 431 in the storage section 43 according to these parameters.

[0147] Further, it is described that the model receiving section 452 receives the judgment model 431 expressed in the code of the process executable in the measurement data recording device 4, but it is also possible to receive the judgment model 431 not expressed in the code of the executable process. In this case, the measurement data recording device 4 can convert the judgment model 431 into executable using a compiler not shown and store it in the storage section 43.

[0148] Further, it is described that the measurement data recording device 4 has the judgment model 431 in the storage section 43, but it is also possible not to have the judgment model 431. In this case, the generation device 5 can store the generated judgment model 431 in an external storage device connected externally to the measurement data recording device 4, and the judging section 48 of the measurement data recording device 4 can perform the judgment using the judgment model 431 in the external storage device.

[0149] [2. Second embodiment]

[0150] Figure 4 This describes system 1A according to a second embodiment. System 1A includes a measurement data recording device 4A and a data generation device 5A. Furthermore, in system 1A of this embodiment, [the following is a description of the second embodiment, which is not directly related to the system description and can be omitted]. Figure 1 The components of System 1 shown are generally the same, with the same reference numerals and descriptions omitted.

[0151] [2.1. Measurement data recording device 4A]

[0152] The measurement data recording device 4A sends the batch file 430 of measurement data from the communication unit 45 to the generation device 5.

[0153] [2.2. Generation Apparatus 5A]

[0154] The generating device 5A has a generating unit 53A, a judging unit 48A, and a sorting unit 58A.

[0155] [2.2-1. Generation Section 53A]

[0156] The generation unit 53A includes a DTW processing unit 531A. Before generating the determination model 431, the DTW processing unit 531A performs the same processing as the DTW processing unit 531 in the first embodiment.

[0157] Furthermore, after generating the judgment model 431, the DTW processing unit 531A also performs DTW processing on the newly acquired time series measurement data (i.e., batch data). For example, the DTW processing unit 531 can read the reference batch data and the target batch data from the storage unit 52 and perform DTW processing on each channel of the measurement data, based on operations performed externally by an operator or other personnel. The reference batch data before the generation of the judgment model 431 and the reference batch data after the generation of the judgment model 431 can be the same or different. The target batch data can be the batch data contained in the latest batch file 430.

[0158] After generating the judgment model 431, the DTW processing unit 531 can supply the batch data of the objects that have undergone DTW processing to the judgment unit 48A. In addition, the DTW processing unit 531 can also supply at least one parameter calculated by DTW processing to the judgment unit 48A.

[0159] [2.2-2. Decision Unit 48A]

[0160] The judgment unit 48A performs the same processing as the judgment unit 48 of the measurement data recording device 4 in the embodiment. Based on this, the judgment unit 48A can supply the judgment result and batch processing file 430 to the classification unit 58A if it determines that the facility 2 is in a bad condition.

[0161] [2.2-3. Classification section 58A]

[0162] The classification section 58A classifies the abnormal state of the facility 2. The classification section 58A has an analysis section 581, a setting section 582, and a determination section 583.

[0163] [2.2-3-1. Analysis section 581]

[0164] The analysis section 581 performs cluster analysis on a plurality of time-series measurement data measured when the state of the facility 2 is not good.

[0165] The analysis section 581 can perform cluster analysis on the batch files 430 measured when the state of the facility 2 is not good, which are stored in the storage section 52. The batch files 430 measured when the state of the facility 2 is not good can be batch files 430 including measurement data to which a label indicating that the state of the facility 2 is not good is added, or batch files 430 of measurement data measured when the state of the facility 2 is determined to be not good by the operator or the determination section 48A.

[0166] The analysis section 581 can perform cluster analysis on a plurality of batch files 430, and generate a plurality of clusters in which the batch files 430 are classified. The analysis section 581 can supply, to the setting section 582, the identification information (also referred to as a batch file ID) of the batch files 430 belonging to each cluster, for each cluster generated.

[0167] [2.2-3-2. Setting section 582]

[0168] The setting section 582 sets identification information (also referred to as a cluster ID) for the clusters of measurement data classified by the analysis section 581, based on an operation performed from the outside by the operator or the like. The setting section 582 can store the batch file IDs of the batch files 430 of each cluster supplied from the analysis section 581 in the storage section 52 in correspondence with the cluster IDs of the clusters.

[0169] The cluster ID can indicate the type of the abnormal state of the facility 2, and in the present embodiment, for example, can be set by the operator to indicate a factor that deteriorates the state of the facility 2, or the name of the abnormal state.

[0170] In addition, the cluster ID can be set by the operator after the cluster is generated by the analysis section 581, or can be set by the operator before the cluster is generated. In the case where the cluster ID is set before the cluster is generated, the cluster ID set by the operator for at least one of the batch files 430 before classification can be set for the cluster including the batch files 430.

[0171] [2.2-3-3. Determination section 583]

[0172] The determination section 583 determines the cluster to which the one time series of measurement data should be classified, based on the one time series of measurement data input to the judgment model 431, as a result of which the state of the facility 2 is judged to be poor. In the present embodiment, as an example, the distribution range of the measurement data within the batch file 430 can be stored in advance for each cluster, and the determination section 583 can determine the cluster to which the batch file 430 should be included, based on which cluster the measurement data of the batch file 430 is included in the distribution range of the measurement data of the cluster, in the case where the batch data of the new batch file 430 is input to the judgment model 431 and the state of the facility 2 is judged to be poor. Instead, the determination section 583 can also cause the analysis section 581 to perform cluster analysis again, and determine the cluster to which the batch file 430 is classified.

[0173] The determination section 583 can output the content or the distribution chart of each batch file 430 included in the determined cluster, or can output the cluster ID set for the determined cluster. In the case where the distribution chart of the batch file 430 included in the determined cluster is output and displayed, the determination section 583 can display the cluster in a recognizable manner, and can also display the distribution of the batch file 430 included in the other cluster together. Further, the determination section 583 can also display the cluster ID corresponding to each cluster. Thus, the cluster ID and the distribution of the batch file 430 are displayed for each cluster. In the case where the cluster ID is not set, the display area of the cluster ID can be displayed as a blank, and the cluster ID can be set by the setting section 582, based on the input of the cluster ID in the blank by the operator.

[0174] According to the generation device 5A of the system 1A described above, the cluster to which the one time series of measurement data should be classified is determined, based on the one time series of measurement data input to the judgment model 431, as a result of which the state of the facility 2 is judged to be poor. Thus, in the case where the state is judged to be poor based on the time series of measurement data, the trend or the cause of the poor, or the name of the poor state can be determined.

[0175] Further, since the identification information set for the determined cluster (the cluster ID indicating the cause of the poor or the name of the poor state, as an example in the present embodiment) is output, the name of the poor state or the cause thereof can be easily determined.

[0176] [2.3. Distribution chart of batch file]

[0177] Figure 5is a distribution diagram of the batch files 430 after clustering. In this diagram, each batch file 430 is represented as a plot, and the clusters are set with cluster IDs such as "temperature abnormality", "pressure abnormality", and "gas leakage". In a case where any one of the clusters is determined by the determination section 583, the display color of the cluster can be changed from other clusters, and the like, whereby the clusters are displayed with recognition.

[0178] [2.4. Example of output screen]

[0179] Figure 6 is an example of a screen output by the generation device 5A. As a result of the determination of the batch files 430, the generation device 5A can output whether the state of the facility 2 is good or not "NG", an operation index "-0.246" calculated from the batch files 430, a time of determination by the determination section 48A, a cluster ID of a cluster to which the batch files 430 should be classified (an abnormality factor in this diagram) "gas leakage", and the like.

[0180] Figure 7 is another example of a screen output by the generation device 5A. The generation device 5A can output the determination result, the operation index, the cluster ID of a cluster to which the batch files 430 should be classified (an abnormality factor in this diagram), and the like for each batch file 430. In addition, in this diagram, the information of each batch file 430 is displayed in the order of determination by the determination section 48A.

[0181] [2.5. Modified example]

[0182] In addition, in the above-described second embodiment, the generation device 5A determines whether the state is good or not, but the measurement data recording device 4A can determine whether the state is good or not using the determination model 431 as in the above-described first embodiment. In this case, the cluster ID of the state of failure can be determined by the generation device 5A in accordance with the determination of the state of failure in the measurement data recording device 4A.

[0183] In addition, the generation device 5A determines the cluster ID of the state of failure, but the measurement data recording device 4A can determine the cluster ID of the state of failure. In this case, the measurement data recording device 4A can have the determination section 583 and can have the classification section 58A.

[0184] In addition, the generation device 5A has the setting section 582, but can not have the setting section 582. In this case, the determination section 583 can output the content or the distribution diagram of each batch file 430 included in the determined cluster. In this case, the name of the state of failure, the cause of the failure, and the trend can be determined from the content of the output batch file 430.

[0185] [3. Third embodiment]

[0186] Figure 8 A system 1B according to a third embodiment is shown. The system 1B includes a measurement data recording device 4B and a generation device 5B. In addition, in the system 1B according to the present embodiment, the same components as those of the system 1 shown in FIG. 1 are denoted by the same reference numerals and the explanation thereof is omitted. Figure 1

[0187] [3.1. Measurement data recording device 4B]

[0188] The measurement data recording device 4B has a storage section 43B and a determination section 48B.

[0189] [3.1-1. Storage section 43B]

[0190] The storage section 43B stores a determination model 431B.

[0191] The determination model 431B determines whether the state of the facility 2 is good or not based on the measurement data. The determination model 431B can output a determination result of good or not based on the input of the measurement data. In the present embodiment, as an example, the determination model 431B can determine good or not based on the measurement data measured at an arbitrary measurement time (as an example, the measurement data measured most recently).

[0192] The determination model 431B can indicate an allowable range (also referred to as an alarm curve) of the measurement data according to the elapsed time within a process (as an example, a batch process in the present embodiment) performed in the facility 2. The determination model 431B can indicate the allowable range of the measurement value for each of the elapsed times of the process for each channel of the measurement data. The allowable range of the measurement data can have at least one of an upper limit value and a lower limit value.

[0193] [3.1-2. Determination section 48B]

[0194] The determination section 48B uses the determination model 431B to determine whether the state of the facility 2 at the time of measuring the measurement data is good or not each time the measurement data is acquired. The determination section 48B can sequentially acquire the measurement data from the acquisition section 40 after the start of the process performed in the facility 2, and calculate the elapsed time from the start of the process performed in the facility 2 to the measurement of the measurement data.

[0195] The determination section 48B can perform determination by comparing the upper and lower limit values of the allowable range according to the elapsed time, that is, the determination threshold, with the measurement value of the measurement data. If the measurement data is within the allowable range as indicated by the comparison result, the determination section 48B can determine the state as good, and if outside the allowable range, can determine the state as not good. The determination section 48B can output the determination result.

[0196] ​[3.2. Generation device 5B]

[0197] The generation device 5B has a storage section 52B and a generation section 53B.

[0198] [3.2-1. Storage section 52B]

[0199] The storage section 52B stores the determination model 431B. The determination model 431B stored in the storage section 52B can not be expressed in a code of a process executable in the measurement data recording device 4B, but can be expressed in a code of a process executable in the generation device 5B.

[0200] [3.2-2. Generation section 53B]

[0201] The generation section 53B generates the determination model 431B using the time-series measurement data acquired by the acquisition section 510. The generation section 53B has a DTW processing section 531B and a statistical processing section 533B.

[0202] [3.2-2-1. DTW processing section 531B]

[0203] The DTW processing section 531B, like the DTW processing section 531 of the embodiment, performs DTW processing on the batch data for each channel of the measurement data. The DTW processing section 531B can supply the batch data in which the measurement time is made consistent with respect to the reference batch data to the statistical processing section 533B. However, the DTW processing section 531B can calculate the DTW distance, the number of measurement data shifted in the time axis direction in the DTW path, and the number of measurement data not shifted in the time axis direction in the DTW path by the DTW processing, and not supply to the statistical processing section 533B.

[0204] [3.2-2-2. Statistical processing section 533B]

[0205] The statistical processing section 533B generates, as the determination model 431B, the allowable range of the measurement data by elapsed time within the batch, that is, the alarm curve, by statistical processing on a plurality of batch data acquired by the acquisition section 510 for each channel of the measurement data. The statistical processing section 533B can perform statistical processing on a plurality of batch data processed by the DTW processing section 531B. The statistical processing section 533B can determine the upper limit value and the lower limit value of the allowable range by statistical processing by elapsed time for each channel of the measurement data.

[0206] For example, the statistical processing section 533B can extract the measurement data of each measurement time from the measurement data measured when the state of the facility 2 is good, and determine the upper limit value and the lower limit value of each elapsed time in such a manner that all measurement values are included in the allowable range.

[0207] Further, the statistical processing section 533B can extract the measurement data of each measurement time from the measurement data measured when the state of the facility 2 is good, and determine the upper limit value and the lower limit value of each elapsed time in a manner that the distribution of the measurement values is regarded as a normal distribution, and the measurement values in the 1σ interval, the 2σ interval, the 3σ interval converging to the distribution are included in the allowable range. Note that σ is the standard deviation of the measurement values.

[0208] Further, the statistical processing section 533B can extract the measurement data of each measurement time from the measurement data measured when the state of the facility 2 is good, and calculate the average and the variance of the measurement values, and determine the upper limit value and the lower limit value based on the calculation result. In this case, the statistical processing section 533B can determine the upper limit value and the lower limit value in a manner that the average value is the intermediate value of the upper and lower limit values, and the greater the variance, the greater the width of the upper and lower limit values, and the smaller the variance, the smaller the width of the upper and lower limit values.

[0209] The statistical processing section 533B can supply the generated judgment model 431B to the storage section 52B and store it, and supply it to the measurement data recording device 4B via the communication section 51.

[0210] According to the measurement data recording device 4B in the above system 1B, using the judgment model 431B, whether or not the state of the facility 2 at the time of measuring the measurement data is good is judged each time the measurement data is acquired. Therefore, it is possible to obtain the judgment result successively without waiting until the measurement data is complete.

[0211] Further, according to the generation device 5B in the system 1B, by performing statistical processing on the plurality of time series of measurement data classified by kind, the allowable range of the measurement data according to the elapsed time within the processing performed in the facility 2 is generated as the judgment model 431, so it is possible to generate the judgment model 431 that can judge whether or not the state of the facility 2 at the time of measuring the measurement data is good each time the measurement data is acquired.

[0212] Further, statistical processing is performed on the plurality of time series of measurement data processed by the DTW processing section 531, so it is possible to make the measurement times coincide to improve the efficiency of the statistical processing.

[0213] [3.3. Example of alarm curve]

[0214] Figure 9 The alarm curve is shown. The horizontal axis in the graph indicates the elapsed time in the batch, and the vertical axis indicates the measurement value. The alarm curve indicated by the judgment model 431B can have the upper limit value and the lower limit value of the measurement data according to the elapsed time in the batch. Note that in the graph, it is judged that the state of the facility 2 is not good because the measurement value exceeds the upper limit value.

[0215] [4. Modified example]

[0216] In addition, in the first, third and fourth embodiments described above, the generation device 5, 5B generates either one of the determination model 431 and the determination model 431B, but can generate both. In this case, the measurement data recording device 4, 4B can use both the determination models 431, 431B to determine the state of the facility 2.

[0217] Further, in the first to third embodiments described above, the object is described as the facility 2, but can be another subject. For example, the object can be a product manufactured in the facility 2, can be a natural object such as a living body, can be a natural environment such as weather, terrain, or can be a natural phenomenon such as a chemical reaction, a biochemical reaction.

[0218] Further, the various embodiments of the present application can be described with reference to flowcharts and block diagrams in which the modules can represent (1) stages of a process that performs an operation or (2) portions of an apparatus that have a role of performing an operation. The specific stages and portions can be implemented by dedicated circuits, programmable circuits supplied with computer-readable instructions stored on a computer-readable medium, and / or processors supplied with computer-readable instructions stored on a computer-readable medium. The dedicated circuits can include digital and / or analog hardware circuits, and can include integrated circuits (ICs) and / or discrete circuits. The programmable circuits can include reconfigurable hardware circuits including logical AND, logical OR, logical XOR, logical NAND, logical NOR, and other logical operations, flip-flops, registers, field programmable gate arrays (FPGAs), programmable logic arrays (PLAs), and the like memory elements.

[0219] The computer-readable medium can include any tangible device that is capable of storing instructions for execution by an appropriate device, and has a computer-readable medium with instructions stored therein, including a product including instructions capable of being executed to make means for performing operations specified by a flowchart or a block diagram. Examples of the computer-readable medium can include electronic storage media, magnetic storage media, optical storage media, electromagnetic storage media, semiconductor storage media, and the like. More specific examples of the computer-readable medium can include a soft (registered trademark) disk, a magnetic disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or a flash memory), an electrically erasable programmable read-only memory (EEPROM), a static random access memory (SRAM), a compact disc read-only memory (CD-ROM), a digital versatile disk (DVD), a Blu-ray (RTM) disc, a memory stick, an integrated circuit card, and the like.

[0220] Computer readable instructions include any of source code, object code, and target code described by any combination of one or more programming languages, including an assembly language, an instruction set architecture (ISA) instruction, machine instructions, a machine dependent instruction, microcode, firmware instructions, state-setting data, or any combination of a Smalltalk (registered trademark), JAVA (registered trademark), C++, and an object oriented programming language, and a "C" programming language or a similar programming language that is a procedural programming language.

[0221] Computer readable instructions can be provided to a processor of a general purpose computer, special purpose computer, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the operations specified by the flow diagram or block diagram. Examples of processors include a computer processor, a processing unit, a microprocessor, a digital signal processor, a controller, a microcontroller, a microcomputer, and so on.

[0222] Figure 10 The computer 2200 is an example of a machine that can implement the various ways of the present application in whole or in part. The computer 2200 is capable of functioning as the operations associated with or performing one or more parts of the apparatus of the embodiments of the present application, or executing the operations or the one or more parts, by programs installed in the computer 2200, and / or the computer 2200 is capable of executing the processes or the stages of the embodiments of the present application. In order for the computer 2200 to perform the specific operations associated with some or all of the modules of the flowcharts and block diagrams described in the present specification, such programs can be executed by the CPU 2212.

[0223] The computer 2200 of the present embodiment includes the CPU 2212, the RAM 2214, the graphics controller 2216, and the display device 2218, which are connected to each other through the host controller 2210. The computer 2200 further includes the communication interface 2222, the hard disk drive 2224, the DVD-ROM drive 2226, and the input / output units such as an IC card drive, which are connected to the host controller 2210 via the input / output controller 2220. The computer further includes the ROM 2230 and the conventional input / output units such as a keyboard 2242, which are connected to the input / output controller 2220 via the input / output chip 2240.

[0224] The CPU 2212 operates in accordance with the programs stored in the ROM 2230 and the RAM 2214, thereby controlling the units. The graphics controller 2216 acquires the image data generated by the CPU 2212 in a frame buffer or the like provided in the RAM 2214 or in itself, and displays the image data on the display device 2218.

[0225] The communication interface 2222 is capable of communicating with other electronic devices via a network. The hard disk drive 2224 stores programs and data used by the CPU 2212 within the computer 2200. The DVD-ROM drive 2226 reads programs or data from the DVD-ROM 2201, and provides the programs or data to the hard disk drive 2224 via the RAM 2214. The IC card drive reads and / or writes programs and data from and to an IC card.

[0226] The ROM 2230 stores therein a boot program or the like to be executed by the computer 2200 at the time of activation and / or a program depending on the hardware of the computer 2200. The input / output chip 2240 can also connect various input / output units to the input / output controller 2220 via a parallel port, a serial port, a keyboard port, a mouse port, or the like.

[0227] A program is provided by a computer-readable medium such as the DVD-ROM 2201 or an IC card. The program is read from the computer-readable medium, and installed in the hard disk drive 2224, the RAM 2214, or the ROM 2230, which are also examples of computer-readable media, and executed by the CPU 2212. Information processing described within these programs is read to the computer 2200, thereby bringing about cooperation between the programs and the various types of hardware resources described above. An apparatus or a method can be constituted by accompanying the use of the computer 2200 to realize the operation or processing of information.

[0228] For example, in the case where communication is performed between the computer 2200 and an external device, the CPU 2212 can execute a communication program loaded in the RAM 2214, and instruct the communication interface 2222 to the communication processing based on the processing described in the communication program. The communication interface 2222 reads transmission data stored in a transmission buffer processing area provided in the RAM 2214, the hard disk drive 2224, the DVD-ROM 2201, or a recording medium such as an IC card, transmits the read transmission data to a network, or writes reception data received from a network to a reception buffer processing area provided on a recording medium, or the like, under the control of the CPU 2212.

[0229] Further, the CPU 2212 can read all or a necessary part of a file or a database stored in the hard disk drive 2224, the DVD-ROM drive 2226 (the DVD-ROM 2201), an external recording medium such as an IC card, to the RAM 2214, and execute various types of processing on the data on the RAM 2214. Then, the CPU 2212 writes the processed data back to the external recording medium.

[0230] Various types of information such as various types of programs, data, tables, and databases can be stored in the recording medium and subjected to information processing. The CPU 2212 performs various types of processing described throughout the present disclosure on data read from the RAM 2214, including various types of operations, information processing, conditional judgments, conditional branching, unconditional branching, search / replacement of information, and the like, specified by an instruction sequence of a program, and writes the results back to the RAM 2214. Further, the CPU 2212 can search for information in a file, a database, or the like within the recording medium. For example, in a case where a plurality of entries each having an attribute value of a first attribute associated with an attribute value of a second attribute are stored within the recording medium, the CPU 2212 can search for an entry that coincides with a condition that specifies an attribute value of the first attribute from among the plurality of entries, and read an attribute value of the second attribute stored within the entry, thereby acquiring an attribute value of the second attribute associated with the first attribute that satisfies a predetermined condition.

[0231] The program or software module described above can be stored in a computer-readable medium on or near the computer 2200. Further, a recording medium such as a hard disk or a RAM provided within a server system connected to a dedicated communication network or the Internet can be used as the computer-readable medium, whereby the program is provided to the computer 2200 via the network.

[0232] The above describes the present application by way of embodiments, but the technical scope of the present application is not limited to the range described in the above embodiments. It is apparent that various changes or improvements can be made to the above embodiments by those skilled in the art. Such changes or improvements made to the embodiments are also included in the technical scope of the present application according to the recitations of the claims.

[0233] The order of execution of each processing such as actions, processes, steps, and stages of the apparatuses, systems, programs, and methods shown in the claims, the specification, and the drawings is not particularly indicated as "earlier," "before," or the like, and it should be noted that the order can be implemented in any order as long as the output of the previous processing is not used in the subsequent processing. With respect to the flow of actions in the claims, the specification, and the drawings, even if the description is made using "first," "next," or the like for the convenience of explanation, it does not mean that the order must be implemented in that order.

Claims

1. A measurement data recording device, characterized in that... include: The acquisition unit acquires time-series measurement data of the object's state. The recording department records the acquired measurement data; The measurement data transmission unit sends the measurement data recorded in the recording unit to the generation device that generates a judgment model for judging whether the object is in good or bad condition; The model receiving unit receives the judgment model generated by the generating device; as well as The judgment unit uses the received judgment model to determine whether the object is in good or bad condition based on the newly acquired measurement data.

2. The measurement data recording device according to claim 1, characterized in that, This includes a decryption unit that decrypts the encrypted judgment model received by the model receiving unit.

3. The measurement data recording device according to claim 1 or 2, characterized in that, The model receiving unit receives information indicating the legality of the judgment model along with the judgment model, and can then utilize the judgment model from the judgment unit.

4. The measurement data recording device according to claim 1 or 2, characterized in that, The judgment unit uses the judgment model to determine whether the object is in good or bad condition when the measurement data of the newly acquired time series is measured, based on the newly acquired time series measurement data.

5. The measurement data recording device according to claim 1 or 2, characterized in that, The judgment unit uses the difference between the judgment threshold in the judgment model and the index value to make a judgment, and outputs the judgment result and at least the most recent difference. The index value is obtained from the judgment model based on the value corresponding to the measurement value of the measurement data and represents the state of the object.

6. The measurement data recording device according to claim 1 or 2, characterized in that, The judgment unit uses the judgment model to determine whether the object is in good or bad condition each time new measurement data is acquired.

7. A generating apparatus, characterized in that... include: The acquisition unit acquires the measurement data from a measurement data recording device that records a time series of measurement data about the state of the object; The generation unit uses the acquired time-series measurement data to generate a judgment model that determines whether the object is in a good or bad state based on the measurement data; and The sending unit sends the judgment model to the measurement data recording device.

8. The generating apparatus according to claim 7, characterized in that, This includes an encryption unit that encrypts the judgment model generated by the generation unit. The sending unit sends the encrypted judgment model to the measurement data recording device.

9. The generating apparatus according to claim 7 or 8, characterized in that, The sending unit sends information indicating the legality of the judgment model, along with the judgment model, to the measurement data recording device.

10. The generating apparatus according to claim 7 or 8, characterized in that, The device includes a conversion unit that converts the judgment model generated by the generation unit into processing code that is executed in the measurement data recording device. The transmitting unit sends the judgment model converted by the conversion unit to the measurement data recording device.

11. The generating apparatus according to claim 7 or 8, characterized in that, The generation unit has a learning processing unit, which generates the judgment model by processing learning data, the learning data including time series measurement data.

12. The generating apparatus according to claim 11, characterized in that, The time series measurement data is the measurement data from each repeated process performed on the object. The generation unit has a DTW processing unit, which processes each type of measurement data as follows: Using a dynamic time scaling method, relative to a time series of measurement data acquired by the acquisition unit, the time width of other time series of measurement data is made consistent; the distance between each point of the measurement data in the one time series and each point of the measurement data in the other time series is determined to be the shortest DTW path; and the DTW processing unit calculates at least one of the following parameters when performing the above processing: a parameter representing the DTW distance between the measurement data in the one time series and the measurement data in the other time series; a parameter representing the number of measurement data points in the other time series that are offset in the time axis direction along the DTW path; and a parameter representing the number of measurement data points in the DTW path that are not offset in the time axis direction. The learning processing unit uses learning data containing at least one parameter to generate the judgment model.

13. The generating apparatus according to claim 11, characterized in that... include: The analysis unit performs cluster analysis on multiple time-series measurement data measured when the object is in a poor state; as well as The determination unit, based on a time series measurement data input into the judgment model, determines that the state of the object is not good, and determines the cluster to which the time series measurement data should be classified.

14. The generating apparatus according to claim 7 or 8, characterized in that, The time series measurement data is the measurement data from each repeated process performed on the object. The generation unit has a statistical processing unit, which generates a judgment model for each type of measurement data by performing statistical processing on multiple time series measurement data obtained by the acquisition unit. The statistical processing unit performs statistical processing on the measurement data of multiple time series obtained by the acquisition unit.

15. The generating apparatus according to claim 14, characterized in that, The generation unit includes a DTW processing unit. For each type of measurement data, the DTW processing unit uses a dynamic time scaling method to make the time width of other time series measurement data consistent relative to one time series measurement data acquired by the acquisition unit, thereby minimizing the waveform differences between the other time series measurement data relative to the one time series measurement data. The statistical processing unit performs statistical processing on the measurement data of multiple time series processed by the DTW processing unit.

16. A system, characterized in that... include: The measurement data recording device as described in any one of claims 1 to 6; as well as The generating apparatus as described in any one of claims 7 to 15.

17. An apparatus, characterized in that... include: The acquisition unit acquires the measurement data from a measurement data recording device that records a time series of measurement data about the state of the object; The learning processing unit generates a judgment model based on the measurement data to determine whether the object is in a good or bad state. The analysis unit performs cluster analysis on multiple time-series measurement data measured when the object is in a poor state; as well as The determination unit, based on a time series measurement data input into the judgment model, determines that the state of the object is not good, and determines the cluster to which the time series measurement data should be classified.

18. A method, characterized in that include: In the acquisition phase, time-series measurement data of the object's state are obtained; During the recording phase, the acquired measurement data is recorded; During the measurement data transmission phase, the measurement data recorded in the recording phase is sent to the generation device that generates a judgment model to determine whether the object is in good or bad condition. In the model receiving stage, the judgment model generated by the generating device is received; as well as In the judgment phase, the received judgment model is used to determine whether the object is in good or bad condition based on the newly acquired measurement data.

19. A method, characterized in that include: In the acquisition phase, the measurement data is acquired from a measurement data recording device that records a time series of measurement data about the state of the object; In the generation phase, the acquired time-series measurement data is used to generate a judgment model that determines whether the object is in good or bad condition based on the measurement data. as well as During the sending phase, the judgment model is sent to the measurement data recording device.

20. A method, characterized in that include: In the acquisition phase, the measurement data is acquired from a measurement data recording device that records a time series of measurement data about the state of the object; In the learning and processing phase, a judgment model is generated by using learning data containing time-series measurement data to determine whether the object is in a good or bad state based on the measurement data. During the analysis phase, cluster analysis is performed on multiple time-series measurement data obtained when the object is in a poor state. as well as In the determination phase, based on the time series measurement data input into the judgment model, the object is judged to be in a poor state, and the cluster to which the time series measurement data should be classified is determined.

21. A computer-readable medium containing a program, characterized in that, The computer performs the functions of an acquisition unit, a recording unit, a measurement data transmission unit, a model receiving unit, and a judgment unit by executing the program. The acquisition unit acquires time-series measurement data of the object's state. The recording unit records the acquired measurement data. The measurement data sending unit sends the measurement data recorded in the recording unit to the generation device that generates a judgment model to determine whether the object is in good or bad condition. The model receiving unit receives the judgment model generated by the generating device. The judgment unit uses the received judgment model to determine whether the object is in good or bad condition based on the newly acquired measurement data.

22. A computer-readable medium containing a program, characterized in that, The computer performs the functions of an acquisition unit, a generation unit, and a transmission unit by executing the program. The acquisition unit acquires the measurement data from a measurement data recording device that records a time series of measurement data about the state of the object. The generation unit uses the acquired time-series measurement data to generate a judgment model that determines whether the object is in a good or bad state based on the measurement data. The sending unit sends the judgment model to the measurement data recording device.

23. A computer-readable medium containing a program, characterized in that, The computer performs the functions of an acquisition unit, a learning and processing unit, an analysis unit, and a determination unit by executing the program. The acquisition unit acquires the measurement data from a measurement data recording device that records a time series of measurement data about the state of the object; The learning processing unit generates a judgment model based on the measurement data to determine whether the object is in a good or bad state. This model is achieved by processing learning data that includes time-series measurement data. The analysis unit performs cluster analysis on multiple time-series measurement data obtained when the object is in a poor state. The determining unit inputs a time series measurement data into the judgment model and determines that the object is in a bad state, and determines that the time series measurement data should be classified into a cluster.

Images