Data analysis processing system, terminal device, data analysis processing method, and program

Abstract

To implement a data analysis processing system capable of performing data analysis processing accurately while reducing a data communication amount in a system comprising a terminal device for acquiring data and a server device for collecting, via a communication network, the data acquired by the terminal device.SOLUTION: In a data analysis processing system, a terminal determines whether processing target data (data whose normality / abnormality can be judged) input to the terminal is data in a normal state or data in an abnormal state and generates communication data depending on a result of the determination before transmitting the communication data from the terminal device to a server device, which enables effective suppression of increase in a communication amount between the terminal device and the server device. In the data analysis processing system, the data in the abnormal state is transmitted to the server device after data compression processing is applied, which enables the server device to acquire securely even the data in the abnormal state.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to a technology for performing highly accurate data analysis processing while reducing data communication volume in a system including a terminal device that acquires data and a server device that collects the data acquired by the terminal device via a communication network. [Background technology]

[0002] In recent years, a technology has been developed that uses edge computing and other technologies to transmit data acquired by a terminal device (e.g., an edge terminal) via a network to a server device, and the server device performs processing (e.g., data analysis processing) using a machine learning model on the data collected from the terminal device (see, for example, Patent Document 1). [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Publication No. 2022-127144 Summary of the Invention [Problem to be solved by the invention]

[0004] In the technology disclosed in Patent Document 1, data acquired by a terminal device (e.g., an edge terminal) is transmitted to a server device via a network, and the server device performs a predetermined process (inference process) on the data collected from the terminal device using a machine learning model. Therefore, the data acquired by the terminal device (e.g., an edge terminal) must be valid. In the technology disclosed in Patent Document 1, the server device calculates the likelihood of the inference process using the machine learning model. If the calculated likelihood is equal to or less than a threshold, the server device requests the terminal device to adjust the data compression rate (increase the data compression rate) and retransmit the acquired data to the server device. In response to the request from the server, the terminal device acquires the data with the increased data compression rate, retransmits the data to the server device, and again executes a process to calculate the likelihood of the acquired data. Therefore, the technology disclosed in Patent Document 1 has a problem in that the number of retransmissions increases, and further, the number of processes (inference process, data likelihood calculation process) on the server side increases, which may result in a large amount of data communication between the terminal device and the server device.

[0005] In view of the above problems, the present invention aims to provide a data analysis processing system, terminal device, server device, data analysis processing method, and program that can perform data analysis processing with high accuracy while reducing the amount of data communication in a system that includes a terminal device that acquires data and a server device that collects the data acquired by the terminal device via a communication network. [Means for solving the problem]

[0006] In order to solve the above problems, a representative example (one aspect) of the invention disclosed in this application is a data analysis processing system including a terminal device and a server device. The terminal device includes a status determination processing unit, a data classification processing unit, and a first communication interface unit. The server device includes a second communication interface unit and a status analysis processing unit.

[0007] The state determination processing unit determines the state of the data to be processed, and obtains the result data of the determination as state determination data.

[0008] If the state determination data indicates a normal state, the data classification processing unit executes a data classification process on the data to be processed, and obtains an index indicating the state of the data to be processed.

[0009] The first communication interface is a communication interface capable of transmitting, to the server device, communication data including the index and / or the data to be processed when the status determination data does not indicate a normal status.

[0010] The second communication interface unit receives communication data transmitted from the terminal device.

[0011] If the communication data received by the second communication interface unit includes an index and / or processing target data, the status analysis processing unit inputs the index and / or processing target data and performs status analysis processing. [Effects of the Invention]

[0012] According to the present invention, in a system including a terminal device that acquires data and a server device that collects the data acquired by the terminal device via a communication network, it is possible to realize a data analysis processing system, terminal device, server device, data analysis processing method, and program that can perform data analysis processing with high accuracy while reducing the amount of data communication. [Brief explanation of the drawings]

[0013] [Figure 1] FIG. 1 is a schematic configuration diagram of a data analysis processing system 1000 according to a first embodiment. [Figure 2] 10 is a flowchart of a learning process for a trainable model (a model for a two-class classification task) installed in the state determination processing unit 11. [Figure 3]FIG. 1 is a diagram for explaining the learning process of a trainable model (a model for a two-class classification task) mounted in the state determination processing unit 11, and the inference process (processing using a trained model). [Figure 4] 1 is a flowchart of a learning process for a trainable model (a model for a multi-class classification (N-class classification) task) installed in the data classification processing unit 12. [Figure 5] FIG. 2 is a diagram for explaining the learning process of a trainable model (a model for a multi-class classification (N-class classification) task) installed in the data classification processing unit 12. [Figure 6] FIG. 1 is a diagram for explaining the inference process of a trained model (model for a multi-class classification (N-class classification) task) installed in the data classification processing unit 12. [Figure 7] FIG. 10 is a schematic configuration diagram of a data analysis processing system 1000A according to a first modified example of the first embodiment. [Figure 8] FIG. 10 is a schematic configuration diagram of a data analysis processing system 1000B according to a second modified example of the first embodiment. [Figure 9] A diagram showing the CPU bus configuration. DETAILED DESCRIPTION OF THE INVENTION

[0014] [First embodiment] The first embodiment will be described below with reference to the drawings.

[0015] <1.1: Data analysis processing system configuration> FIG. 1 is a schematic configuration diagram of a data analysis processing system 1000 according to the first embodiment.

[0016] 1, the data analysis processing system 1000 includes a terminal device 100, a server device 200, and a data storage unit DB1. The terminal device 100 and the server device 200 are connected to each other via a communication network so as to be able to communicate with each other.

[0017] As shown in FIG. 1, the terminal device 100 includes a state determination processing unit 11, a selector SEL1, a data classification processing unit 12, a data compression processing unit 13, and a first communication interface .

[0018] The state determination processing unit 11 is a functional unit including a trained model (a trained model that performs a two-class classification task) acquired by machine learning, and receives data Din (for example, data that can identify a normal state / abnormal state). The state determination processing unit 11 receives the data Din and inputs it to the trained model (a trained model that performs a two-class classification task), and acquires data (state determination processing data) output from the trained model. The state determination processing unit 11 then generates a switching signal sel1 for switching the selector SEL1 based on the data (state determination processing data) output from the trained model, and outputs the generated switching signal sel1 to the selector SEL1.

[0019] The selector SEL1 is a one-input, two-output selector that receives data Din. The selector SEL1 also receives a switching signal sel1 output from the state determination processing unit 11, and outputs the data Din to the data classification processing unit 12 as data D10 or outputs the data Din to the data compression processing unit 13 as data D11 in accordance with the switching signal sel1.

[0020] The data classification processing unit 12 is a functional unit including a trained model (a trained model that performs a multi-class classification task) acquired by machine learning, and receives data D10 output from the selector SEL1. The data classification processing unit 12 inputs the data D10 to the trained model (a trained model that performs a multi-class classification task), identifies an index based on the data output from the trained model, and outputs data including the identified index as data D20 to the first communication interface 14.

[0021] The data compression processing unit 13 receives the data D11 output from the selector SEL1, performs data compression processing on the data D11, and outputs the data after the data compression processing to the first communication interface 14 as data D21.

[0022] The first communication interface 14 is a communication interface for transmitting and receiving data to and from an external device (for example, the server device 200) via a wired and / or wireless communication network. As shown in Fig. 1, the first communication interface 14 is connected to the data classification processing unit 12 and the data compression processing unit 13. The first communication interface 14 receives data D20 output from the data classification processing unit 12, generates communication data Dtx(D20) including the data D20 (communication data (communication signal) in a format that can be transmitted to an external device via a wired and / or wireless communication network), and transmits the generated communication data Dtx(D20) to the server device 200. In addition, the first communication interface 14 inputs data D21 output from the data compression processing unit 13, generates communication data Dtx(D21) including the data D21 (communication data (communication signal) in a format that can be transmitted to an external device via a wired and / or wireless communication network), and transmits the generated communication data Dtx(D21) to the server device 200.

[0023] As shown in FIG. 1, the server device 200 includes a second communication interface 21, a data decompression processing unit 22, and a status analysis processing unit .

[0024] The second communication interface 21 is a communication interface for transmitting and receiving data to and from an external device (for example, the terminal device 100) via a wired and / or wireless communication network. As shown in FIG. 1, the second communication interface 21 is connected to a data decompression processing unit 22 and a status analysis processing unit 23. The second communication interface 21 receives communication data transmitted from an external device (for example, the terminal device 100). For example, the second communication interface 21 receives communication data Dtx (D20) transmitted from the terminal device 100, acquires data D20 included in the communication data Dtx (D20), and outputs the acquired data D20 to the status analysis processing unit 23 as data D30.

[0025] In addition, the second communication interface 21 receives communication data Dtx (D21) transmitted from the terminal device 100, acquires data D21 contained in the communication data Dtx (D21), and outputs the acquired data D21 to the data decompression processing unit 22 as data D31.

[0026] The data decompression processing unit 22 receives the data D31 output from the second communication interface 21, performs data decompression processing on the data D31, and acquires the data after the data decompression processing. The data decompression processing unit 22 then outputs the data after the data decompression processing to the state analysis processing unit 23 as data D41.

[0027] The state analysis processing unit 23 receives the data D30 output from the second communication interface 21, and calculates the index included in the data D30 (the index acquired at time step k is defined as “Idx (k) "), and outputs data including the acquired index and information on the time step t at which the index was acquired as data D4A to the data storage unit DB1, and stores the data in the data storage unit DB1.

[0028] Furthermore, the state analysis processing unit 23 receives the data D41 output from the data decompression processing unit 22. When the state analysis processing unit 23 receives the data D41, the state analysis processing unit 23 calculates an index (Idx) from n time steps (n: number of prior steps) before the time step t at which the data D41 was acquired to one time step before the time step t. (t-1) ~Idx (t-n) ) corresponding data D (t-1) ~D (t-n) The status analysis processing unit 23 reads out the index Idx from the data storage unit DB1. k When reading data corresponding to (k: natural number, k represents a time step) from the data storage unit DB1, the index Idx (k) A request command Req(Idx) for reading data corresponding to the (k) ) to the data storage unit DB1, and the request command Req(Idx (k) ) data obtained (read) according to Res(Idx (k) )(=D (k) ).

[0029] The data storage unit DB1 is a functional unit that can store data, and reads predetermined data from a predetermined storage area in response to an external read command, and / or writes predetermined data to a predetermined storage area in response to an external write command. The data storage unit DB1 stores data D4A (index Idx) output from the status analysis processing unit 23. t In addition, the data read command Req (including Idx) from the state analysis processing unit 23 is stored in the memory in accordance with the instruction (data write command) from the state analysis processing unit 23. (k) ) according to Idx (k) The data corresponding to the index specified by is read, and the data containing that data is (k) )(=D (k) ) to the state analysis processing unit 23.

[0030] <1.2: Operation of the data analysis processing system> The operation of the data analysis processing system 1000 configured as above will now be described.

[0031] FIG. 2 is a flowchart of the learning process of a trainable model (a model for a two-class classification task) installed in the state determination processing unit 11.

[0032] FIG. 3 is a diagram for explaining the learning process of a trainable model (a model for a two-class classification task) installed in the state determination processing unit 11, and the inference process (processing using a trained model).

[0033] FIG. 4 is a flowchart of the learning process of a trainable model (a model for a multi-class classification (N-class classification) task) installed in the data classification processing unit 12.

[0034] FIG. 5 is a diagram for explaining the learning process of a trainable model (a model for a multi-class classification (N-class classification) task) installed in the data classification processing unit 12.

[0035] FIG. 6 is a diagram for explaining the inference process of a trained model (model for a multi-class classification (N-class classification) task) installed in the data classification processing unit 12.

[0036] For ease of explanation, the operation of the data analysis processing system 1000 will be described below for the case where the data to be processed (data that can be considered as normal / abnormal data) in the data analysis processing system 1000 is image data of a manufactured product (as an example).

[0037] (1.2.1: Learning process of trainable model (model for two-class classification task) installed in state determination processing unit 11) First, the learning process of the trainable model (model for two-class classification task) installed in the state determination processing unit 11 will be described with reference to the flowchart of FIG.

[0038] (Step S11): In step S11, a process of acquiring data in a normal state is executed. For example, an imaging device (not shown) is used to acquire data in a normal state (image data of a manufactured product in a normal state), and the acquired data is stored in the data storage unit DB1.

[0039] (Step S12): In step S12, a learning process for a two-class classification learning model (learnable model) is executed. Specifically, as shown in FIG. 3, normal state data (image data of a manufactured product in a normal state) stored in a data storage unit DB1 is input to the two-class classification learning model, and the output of the two-class classification learning model, i.e., classification result data, is obtained. The two-class classification learning model has two output nodes in the output layer, with the first output node being a node that outputs the probability that the input data is normal and the second output node being a node that outputs the probability that the input data is abnormal. Then, an error (e.g., cross-entropy error) between the data of the output node (two-dimensional vector) and the correct label (a two-dimensional vector, a vector (one-hot vector) in which the first-dimensional data (probability of being normal) is 1.0 and the second-dimensional data (probability of being abnormal) is 0.0) is obtained. Then, the parameters of the two-class classification learning model are updated using the obtained error, for example, by backpropagation. This process is repeated until the error (e.g., cross-entropy error) between the output node data (two-dimensional vector) and the correct label becomes equal to or less than a predetermined value, or until the variation of the error falls within a predetermined range, and the learning is converged. When it is determined that the learning has converged, the parameters set in the two-class classification trainable model are acquired as the optimal parameters.

[0040] (Step S13): In step S13, a two-class classification trained model (a two-class classification model in which the optimal parameters acquired in step S12 are set) is installed in the state determination processing unit 11. Specifically, the parameters of the two-class classification trained model in the state determination processing unit 11 are set to the optimal parameters acquired in step S12. As a result, as shown in the lower diagram of FIG. 3, when image data (input data) of a manufactured product is input to the above-mentioned two-class classification trained model, normal / abnormal determination result data of the input image data of the manufactured product (indicating the probability that the value of the first node of the output node is normal) is output from the two-class classification trained model.

[0041] (1.2.2: Learning process of trainable model (model for multi-class classification (N-class classification) task) installed in data classification processing unit 12) Next, the learning process of the trainable model (model for multi-class classification (N-class classification) task) installed in the data classification processing unit 12 will be described with reference to the flowchart of FIG.

[0042] (Step S21): In step S21, a process of acquiring data in a normal state is executed. For example, an imaging device (not shown) is used to acquire data in the normal state (image data of a manufactured product in a normal state), and the acquired data is stored in the data storage unit DB1. Note that step S21 may be shared (or common) with step S11.

[0043] (Step S22): In step S22, a classification process for the normal state data (a process of assigning an index to each class (annotation process)) is executed. For example, the normal state data (image data of a manufactured product in a normal state) is classified into N states (k-th normal state (k: natural number, 1≦k≦N)), and a process is executed to assign the k-th index Idxk to data determined to be in the k-th state (k-th normal state). After the above process, data (indexed data) linking the normal state data (image data of a manufactured product in a normal state) with the index Idxk of that data is stored in the data storage unit DB1.

[0044] (Step S23): In step S23, a learning process for a multi-class (N-class) classification learning model (learnable model) is executed. Specifically, as shown in FIG. 5, normal state data (image data of a manufactured product in a normal state) with index Idxk stored in the data storage unit DB1 is input to the multi-class (N-class) classification learnable model, and the output of the multi-class (N-class) classification learnable model, i.e., classification result data, is obtained. Note that the multi-class (N-class) classification learnable model has N output nodes in the output layer, and the kth output node is a node that outputs the probability that the input data is data in the kth normal state (k: natural number, 1≦k≦N). Then, an error (e.g., cross-entropy error) between the data of the output node (N-dimensional vector) and the correct label (an N-dimensional vector, a vector (one-hot vector) indicating the probability that the kth-dimensional data is data in the kth normal state) is obtained. Then, the parameters of the multi-class (N-class) classification learnable model are updated using the obtained error, for example, by error backpropagation. This process is repeated until the error (e.g., cross-entropy error) between the output node data (N-dimensional vector) and the correct label becomes equal to or less than a predetermined value, or until the variation of the error falls within a predetermined range, and the learning is converged. When it is determined that the learning has converged, the parameters set in the multi-class (N-class) classification learning model are acquired as the optimal parameters.

[0045] (Step S24): In step S24, a multi-class (N-class) classification trained model (an N-class classification model in which the optimal parameters acquired in step S23 are set) is installed in the data classification processing unit 12. Specifically, the parameters of the multi-class (N-class) classification trained model of the data classification processing unit 12 are set to the optimal parameters acquired in step S23. As a result, as shown in FIG. 6, when image data (input data) of a manufactured product is input to the above-mentioned multi-class (N-class) classification trained model, determination result data indicating the probability that the input image data of the manufactured product is in the k-th normal state is output from the multi-class (N-class) classification trained model.

[0046] (1.2.3: Specific Operation of Data Analysis Processing System 1000) Next, the specific operation of the data analysis processing system 1000 will be described.

[0047] At time t (time step t), for example, image data of a manufactured product captured by an imaging device (not shown) is input to the terminal device 100 as data Din.

[0048] The state determination processing unit 11 inputs the data Din to a trained model (a trained model that performs a two-class classification task) and acquires data (state determination processing data) output from the trained model. Then, the state determination processing unit 11 generates a switching signal sel1 for switching the selector SEL1 based on the data (state determination processing data) output from the trained model.

[0049] Specifically, when the data output from the trained model (state determination processing data) indicates that the input data Din is normal state data (when the value of the first output node (node that outputs the probability of a normal state) of the trained model is greater than the value of the second output node (node that outputs the probability of an abnormal state) of the trained model), the state determination processing unit 11 generates a switching signal sel1 that selects terminal 0 of the selector SEL1 and outputs the generated switching signal sel1 to the selector SEL1.

[0050] On the other hand, when the data output from the trained model (state determination processing data) indicates that the input data Din is data for an abnormal state (when the value of the first output node (node that outputs the probability of a normal state) of the trained model is less than or equal to the value of the second output node (node that outputs the probability of an abnormal state) of the trained model), the state determination processing unit 11 generates a switching signal sel1 that selects terminal 1 of the selector SEL1 and outputs the generated switching signal sel1 to the selector SEL1.

[0051] When the switching signal sel1 output from the state determination processing unit 11 is a signal (a selection signal that selects terminal 0) when it is determined that the data Din is data in a normal state, the selector SEL1 selects terminal 0 and outputs the data Din (data in a normal state) to the data classification processing unit 12 as data D10.

[0052] On the other hand, when the switching signal sel1 output from the state determination processing unit 11 is a signal (a selection signal that selects terminal 1) when it is determined that the data Din is data in an abnormal state, terminal 1 is selected and the data Din (data in an abnormal state) is output to the data compression processing unit 13 as data D11.

[0053] When data D10 (data in a normal state) is output from the selector SEL1 (when the state determination processing unit 11 determines that the data Din is data in a normal state), the data classification processing unit 12 inputs the data D10 (data in a normal state) to a trained model (a trained model that performs an N-class classification task) and acquires data output from the trained model. Then, the data classification processing unit 12 identifies a node that shows the maximum value among the output node values of the trained model (a trained model that performs an N-class classification task) and identifies an index corresponding to the node. In other words, when the node that shows the maximum value among the output node values of the trained model (a trained model that performs an N-class classification task) is the k-th node, the data classification processing unit 12 determines that the input data D10 is likely to be k-th (k: natural number, 1≦k≦N) data in a normal state, and sets the index of the data D10 to Idxk.

[0054] Then, the data classification processing unit 12 classifies the index Idx (the index Idx of the time step t) specified above into (t) ) is output to the first communication interface 14 as data D20.

[0055] When data D11 (abnormal state data) is output from the selector SEL1 (when the state determination processing unit 11 determines that the data Din is abnormal state data), the data compression processing unit 13 performs data compression processing on the data D11 (abnormal state data) and outputs the data after the data compression processing (data including data cmpData obtained by data compression of the abnormal state data Din) to the first communication interface 14 as data D21.

[0056] When the first communication interface 14 receives data D20 from the data classification processing unit 12 (when the state determination processing unit 11 determines that the data Din is normal state data), it generates communication data Dtx(D20) including the data D20 (communication data (communication signal) in a format that can be transmitted to an external device via a wired and / or wireless communication network), and transmits the generated communication data Dtx(D20) to the server device 200.

[0057] In addition, when the first communication interface 14 receives data D21 from the data compression processing unit 13 (when the state determination processing unit 11 determines that the data Din is data in an abnormal state), it generates communication data Dtx(D21) including the data D21 (communication data (communication signal) in a format that can be transmitted to an external device via a wired and / or wireless communication network), and transmits the generated communication data Dtx(D21) to the server device 200.

[0058] When the second communication interface 21 of the server device 200 receives communication data Dtx(D20) from the terminal device 100 (when communication data including data of an index of data in a normal state is received), it acquires the data D20 contained in the communication data Dtx(D20) and outputs the acquired data D20 to the status analysis processing unit 23 as data D30.

[0059] In addition, when the second communication interface 21 receives communication data Dtx (D21) from the terminal device 100 (when it receives communication data including data after data compression processing of data in an abnormal state), it acquires the data D21 included in the communication data Dtx (D21) and outputs the acquired data D21 to the data decompression processing unit 22 as data D31.

[0060] When the data D30 output from the second communication interface 21 is input (when the terminal device 100 acquires data of a normal state), the state analysis processing unit 23 receives the index included in the data D30 (the index Idx of the time step t) and (t)Then, the state analysis processing unit 23 acquires the acquired index (index Idx (t) ) and information on the time step t at which the index was acquired, is output as data D4A to the data storage unit DB1 and stored in the data storage unit DB1.

[0061] When the data decompression processing unit 22 receives data D31 output from the second communication interface 21, it performs data decompression processing on the data D31 and acquires the data after the data decompression processing. As a result, the data decompression processing unit 22 acquires the original data (data before the data compression processing) of the data D21 after the data compression processing of the data in the abnormal state. Note that the data compression processing performed by the data compression processing unit 13 and the data decompression processing performed by the data decompression processing unit 22 are preferably lossless conversion processes.

[0062] Then, the data decompression processing unit 22 outputs the data after the data decompression processing to the state analysis processing unit 23 as data D41.

[0063] When the data D41 is input (when the terminal device 100 acquires data on an abnormal state), the state analysis processing unit 23 calculates an index (Idx) from n time steps (n: number of prior steps) before the time step t at which the data D41 is acquired to one time step before the time step t. (t-1) ~Idx (t-n) ) corresponding data D (t-1) ~D (t-n) The state analysis processing unit 23 reads out data corresponding to the input data Din (image data in this embodiment) from the data storage unit DB1. (k) A request command Req(Idx) for reading data corresponding to the (k) ) to the data storage unit DB1, and the request command Req(Idx (k) ) data obtained (read) according to Res(Idx (k) )(=D (k)) is acquired. As a result, the state analysis processing unit 23 acquires data D (t-1) ~D (t-n) Get.

[0064] The state analysis processing unit 23 then extracts data D41 (data at time step t) output from the data decompression processing unit 22, which is data acquired at time step t (image data in this embodiment), and data D42 (data at time step t) acquired as described above from time step tn to time step t-1. (t-1) ~D (t-n) Using these, a state analysis process (for example, a time series analysis until an abnormality occurs) is performed (for example, an analysis process is performed on the time series data from n time steps before the time step at which the abnormal state occurs to the time step t at which the data on the abnormal state is acquired).

[0065] Then, the state analysis processing unit 23 acquires the processing result data of the state analysis processing (for example, the time series analysis until the occurrence of the abnormality) as data Dout.

[0066] As described above, the data analysis processing system 1000 can perform analysis processing (including time series analysis up until the occurrence of an abnormality) on data acquired by the terminal device 100 at time t (time step t).

[0067] Data analysis processing can also be performed for time steps t+1 (the time (time step) when data Din is next acquired by terminal device 100) and thereafter (time steps t+2, t+3, ...) by performing the same processing as above in the data analysis processing system 1000.

[0068] <Summary> As described above, in the data analysis processing system 1000, the terminal device 100 determines whether data input to the terminal device 100 (data that can be considered normal / abnormal) is normal state data or abnormal state data, generates communication data according to the result of the determination, and transmits the data from the terminal device 100 to the server device, thereby effectively suppressing an increase in the amount of communication between the terminal device 100 and the server device 200. That is, in the data analysis processing system 1000, (1) when the input data is determined to be normal state data, an index Idx that specifies what normal state the data is in (if the data is determined to be the kth normal state out of N normal states, an index Idxk (k: natural number, 1≦k≦N)) is obtained, and the obtained index Idx (for time step t, the index Idx (t) ) is transmitted from the terminal device 100 to the server device 200, and (2) if the input data is determined to be data in an abnormal state, a data compression process is performed on the data, the data after the data compression process is obtained, and the communication data including the obtained data after the data compression process is transmitted from the terminal device 100 to the server device 200.

[0069] Therefore, in the data analysis processing system 1000, (1) when data in a normal state is acquired, the index Idx (for time step t, the index Idx (t)), is transmitted from the terminal device 100 to the server device 200, the amount of communication between the terminal device 100 and the server device 200 is small, and (2) only when abnormal state data is acquired, communication data (large-capacity communication data) including data after data compression processing of the abnormal state data is transmitted from the terminal device 100 to the server device 200. In other words, in the data analysis processing system 1000, for abnormal state data that requires detailed analysis, communication data (large-capacity communication data) including data after data compression processing is transmitted to ensure that the server device 200 can perform detailed analysis processing, and for normal state data, an index Idx (in the case of time step t, the index Idx) for identifying one of N states in the normal state data is transmitted. (t) ) (for example, communication data not including the processing target data itself or data after the data compression processing) is transmitted from the terminal device 100 to the server device 200, thereby reducing the amount of data communication (effectively suppressing an increase in the amount of data communication). Furthermore, in the data analysis processing system 1000, data of an abnormal state that requires detailed analysis is reliably transmitted from the terminal device 100 to the server device 200, so that the server device 200 can perform data analysis processing with high accuracy. Furthermore, in the data analysis processing system 1000, for data of a normal state, the server device 200 acquires data corresponding to the index Idx received from the terminal device 100 from the data storage unit DB1 at any timing, so that the server device 200 can always acquire data of a normal state and data of an abnormal state in chronological order. Therefore, for example, in the data analysis processing system 1000, if data of a normal state continues and then an abnormal state occurs, data of the normal state that continued before the abnormal state occurred (data of the normal state that continued before the abnormal state occurred is the index Idx stored in the data storage unit DB1) can be transmitted. (k)By performing time-series data analysis processing using the time series data (which can be obtained based on the above) and data on the timing at which an abnormal state occurred, it is possible to perform highly accurate analysis processing (for example, understanding the state transitions leading up to the occurrence of an abnormal state, investigating the cause of the abnormal state, etc.).

[0070] In this way, the data analysis processing system 1000 can perform data analysis processing with high accuracy while reducing the amount of data communication between the terminal device that acquires the data and the server device that collects the data acquired by the terminal device via a communication network.

[0071] <First Modification> Next, a first modified example of the first embodiment will be described. Note that the same parts as those in the above embodiment are given the same reference numerals, and detailed description thereof will be omitted.

[0072] FIG. 7 is a schematic configuration diagram of a data analysis processing system 1000A according to a first modified example of the first embodiment.

[0073] As shown in Figure 7, the data analysis processing system 1000A of this modified example has a configuration in which, in the data analysis processing system 1000 of the first embodiment, the terminal device 100 is replaced with a terminal device 100A, the server device 200 is replaced with a server device 200A, and the data storage unit DB1 is replaced with a data storage unit DB1A.

[0074] The terminal device 100A of this modification has a configuration in which the data classification processing unit 12 in the terminal device 100 of the first embodiment is replaced with a data classification processing unit 12A, and the data compression processing unit 13 is replaced with a data compression processing unit 13A.

[0075] The data classification processing unit 12A has the same functions as the data classification processing unit 12, and further, if it determines in the data classification processing that it is difficult to identify the index Idxk from the input data D10, it outputs to the data compression processing unit 13A (1) the data D10 and (2) transmission instruction information Req_send that instructs the data compression processing unit 13A to perform data compression processing on the data D10 and output the data after the data compression processing to the first communication interface 14.

[0076] The data compression processing unit 13A has the same functions as the data compression processing unit 13, and further has the following functions: The data compression processing unit 13A receives data D10 and transmission instruction information Req_send output from the data classification processing unit 12A, and when the data D10 and transmission instruction information Req_send are received, performs data compression processing on the data D10 and outputs data including the compressed data and the transmission instruction information Req_send to the first communication interface 14 as data D21.

[0077] The server device 200A of this modification has a configuration in which the second communication interface 21 in the server device 200 of the first embodiment is replaced with a second communication interface 21A, and the data decompression processing unit 22 is replaced with a data decompression processing unit 22A.

[0078] The second communication interface 21A has the same functions as the second communication interface 21, and further has the following functions: That is, when the second communication interface 21A receives communication data Dtx(D21) transmitted from the terminal device 100A when the data classification processing unit 12A determines that it is difficult to identify the index Idxk, the second communication interface 21A extracts the data D21 included in the communication data Dtx(D21), further extracts the transmission instruction information Req_send included in the data D21, and outputs data including the extracted data D21 and the transmission instruction information Req_send as data D31 to the data decompression processing unit 22A.

[0079] The data decompression processing unit 22A has the same functions as the data decompression processing unit 22, and further has the following functions. That is, when the server device 200A receives communication data Dtx (D21) transmitted from the terminal device 100A when the data classification processing unit 12A determines that it is difficult to identify the index Idxk, the data decompression processing unit 22A inputs data D31 including the data D21 and transmission instruction information Req_send from the second communication interface 21A. Then, when the transmission instruction information Req_send is input, the data decompression processing unit 22A performs data decompression processing on the data D21 (data after compression processing of the data D10) included in the data D31, and acquires data equivalent to the data D10. Then, the data decompression processing unit 22A outputs data including the acquired data and an index indicating that the data is in a new state (for example, index IdxN+1 indicating the N+1th normal state) (or data to which an index indicating that the data is in a new state (for example, index IdxN+1 indicating the N+1th normal state) has been assigned (data equivalent to data D10)) as data D42 to the data storage unit DB1A, and stores the data D42 in the data storage unit DB1.

[0080] The data storage unit DB1A has the same function as the data storage unit DB1, and further has the function of inputting the data D42 output from the data decompression processing unit 22A and storing the data D42.

[0081] The operation of the data analysis processing system 1000A of this modified example configured as above will be described below. Note that a description of the same parts as those in the first embodiment will be omitted.

[0082] A case will be described in which data Din is input to the state determination processing unit 11, the state determination processing unit 11 determines that the data is normal, and the data Din is input to the data classification processing unit 12A as data D10. In this case, if the data classification processing unit 12A determines that none of the output node values of the trained model (N-class classification trained model) are particularly high and that it is difficult to determine which of the N classes the data should be classified into (if it is difficult to determine which of the N normal states the data is in), the data classification processing unit 12A outputs to the data compression processing unit 13A (1) the data D10 and (2) transmission instruction information Req_send that instructs the data compression processing unit 13A to perform data compression processing on the data D10 and output the data after data compression processing to the first communication interface 14.

[0083] In the above case, if there are no particularly high values in any of the output node values of the trained model (N-class classification trained model) of the data classification processing unit 12A, it is highly likely that the data D10 is new data in a normal state. For example, if there are no particularly high values in any of the output node values of the trained model (N-class classification trained model) of the data classification processing unit 12A and the values of two or more of the top output nodes indicate similar values (values within a predetermined range), it is highly likely that data in an unexpected state has been input to the trained model.

[0084] Therefore, if the data classification processing unit 12A determines that it is difficult to determine which of the N classes the input data D10 should be classified into, it determines that there is a high possibility that the data D10 is new normal state data, and transmits the data (compressed data) to the server device 200A, stores it in the data storage unit DB1A, and uses it, for example, as data for re-learning.

[0085] When data D10 and transmission instruction information Req_send are input from the data classification processing unit 12A, the data compression processing unit 13A performs data compression processing on the data D10 and outputs data including the data after the data compression processing and the transmission instruction information Req_send as data D21 to the first communication interface 14.

[0086] The data D21 generated as described above is then transmitted as communication data Dtx(D21) to the server device 200A via the first communication interface 14.

[0087] The second communication interface 21A of the server device 200A receives the communication data Dtx(D21). Specifically, when the second communication interface 21A receives the communication data Dtx(D21) from the terminal device 100A, which was transmitted when the data classification processing unit 12A determined that it was difficult to identify the index Idxk, the second communication interface 21A extracts the data D21 included in the communication data Dtx(D21), further extracts the transmission instruction information Req_send included in the data D21, and outputs data including the extracted data D21 and the transmission instruction information Req_send as data D31 to the data decompression processing unit 22A.

[0088] The data decompression processing unit 22A receives data D31 including data D21 and transmission instruction information Req_send from the second communication interface 21A. When the data decompression processing unit 22A receives the transmission instruction information Req_send, the data decompression processing unit 22A performs data decompression processing on the data D21 included in the data D31 (data obtained after the data compression processing of data D10) to acquire data equivalent to data D10. The data decompression processing unit 22A then outputs data including the acquired data and an index indicating that the data is in a new state (for example, the index IdxN+1 indicating the (N+1)th normal state) (or data (data equivalent to data D10) to which an index indicating a new state (for example, the index IdxN+1 indicating the (N+1)th normal state) has been assigned) as data D42 to the data storage unit DB1A, and stores the data D42 in the data storage unit DB1.

[0089] Then, at a predetermined timing, data to which an index indicating a new state (for example, index IdxN+1 indicating the (N+1)th normal state) stored in the data storage unit DB1A is assigned and data previously stored in the data storage unit DB1A (data to which indices indicating the first to Nth states are assigned) are used to perform a training process for a trainable model (model for a multi-class classification task) to be mounted in the data classification processing unit 12A, as described in (1.2.2: Training process of a trainable model (model for a multi-class classification (N-class classification) task) to be mounted in the data classification processing unit 12A). In this case, the training process is performed so that the classification tasks for the added classes can be performed. For example, if the number of added classes is i, the training process for the trainable model to be mounted in the data classification processing unit 12A can be performed so that the task of N+i-class classification can be performed. Then, the trained model (trained model for N+i-class classification), to which the optimal parameters acquired after the training process are set, is mounted in the data classification processing unit 12A. This enables N+i class classification, that is, processing using a trained model for multi-class classification that can determine (identify) N+i normal states.

[0090] As described above, in the data analysis processing system 1000A of this modified example, when data (unknown data) that is not in a known normal state is input, the data is stored in the data storage unit DB1A, and re-learning is performed at a predetermined timing to obtain a trained model (a model to be installed in the data classification processing unit 12A) that corresponds to the unknown data, making it possible to handle the unknown data.

[0091] <<Second Modification>> Next, a second modified example of the first embodiment will be described. Note that the same parts as those in the above embodiment are given the same reference numerals, and detailed description thereof will be omitted.

[0092] FIG. 8 is a schematic configuration diagram of a data analysis processing system 1000B according to a second modified example of the first embodiment.

[0093] As shown in FIG. 8, a data analysis processing system 1000B of this modified example has a configuration in which the terminal device 100 in the data analysis processing system 1000 of the first embodiment is replaced with a terminal device 100B.

[0094] The terminal device 100B of this modification has a configuration in which the data classification processing unit 12 in the terminal device 100 of the first embodiment is replaced with a data classification processing unit 12B, and further a terminal data storage unit 15 is added.

[0095] The data classification processing unit 12B has the same functions as the data classification processing unit 12, and further, if it determines in the data classification process that it is difficult to identify the index Idxk from the input data D10, it outputs the data D10 to the terminal data storage unit 15 as data D20A.

[0096] The operation of the data analysis processing system 1000B of this modified example configured as above will be described below. Note that a description of the same parts as those in the first embodiment will be omitted.

[0097] A case will be described in which data Din is input to the state determination processing unit 11, the state determination processing unit 11 determines that the data is normal, and the data Din is input to the data classification processing unit 12B as data D10. In this case, if the data classification processing unit 12B determines that none of the output node values of the trained model (N-class classification trained model) are particularly high and that it is difficult to determine which of the N classes the data should be classified into (if it is difficult to determine which of the N normal states the data is in), the data classification processing unit 12B outputs the data D10 to the terminal data storage unit 15 as data D20A. Furthermore, the data classification processing unit 12B outputs data D20 including information indicating that the data could not be classified by the trained model of the data classification processing unit 12B, or D20 including data including an index indicating a class other than the N class (for example, a dummy index or an index of a specific value (a value other than 1 to N (for example, -999))) to the first communication interface 14. Then, communication data Dtx(D20) including the data D20 is transmitted from the terminal device 100B to the server device 200. Then, the server device 200 determines from the data D20 included in the communication data Dtx(D20) that there was data that could not be classified by the data classification processing unit 12B of the terminal device 100B.

[0098] In the above case, if there are no particularly high values in any of the output node values of the trained model (N-class classification trained model) of the data classification processing unit 12A, it is highly likely that the data D10 is new data in a normal state. For example, if there are no particularly high values in any of the output node values of the trained model (N-class classification trained model) of the data classification processing unit 12A and the values of two or more of the top output nodes indicate similar values (values within a predetermined range), it is highly likely that data in an unexpected state has been input to the trained model.

[0099] Therefore, when the data classification processing unit 12B determines that it is difficult to determine which of the N classes the input data D10 should be classified into, it determines that there is a high possibility that the data D10 is data in a new normal state, outputs the data (compressed data) to the terminal data storage unit 15, and stores the data in the terminal data storage unit 15. The terminal data storage unit 15 assigns an index to the data D10 indicating that it is a new state (for example, an index IdxN+1 indicating the (N+1)th normal state), stores the data D10, and uses it as data for re-learning to be executed at a predetermined timing.

[0100] Then, at a predetermined timing, data to which an index indicating a new state (for example, index IdxN+1 indicating the (N+1)th normal state) stored in the terminal data storage unit 15 is assigned and data previously stored in the data storage unit DB1A (data to which indices indicating the first to Nth states are assigned) are used to perform a training process for a trainable model (model for a multi-class classification task) to be mounted in the data classification processing unit 12, as described in (1.2.2: Training process of a trainable model (model for a multi-class classification (N-class classification) task) to be mounted in the data classification processing unit 12A). In this case, the training process is performed so that the classification tasks for the added classes can be performed. For example, if the number of added classes is i, a training process is performed for a trainable model to be mounted in the data classification processing unit 12B so that the task of N+i-class classification can be performed. Then, a trained model (trained model for N+i-class classification) to which the optimal parameters acquired after the training process are set is mounted in the data classification processing unit 12B. This allows for N+i class classification, i.e., processing using a trained model for multi-class classification that can determine (identify) N+i normal states. The data acquired by the terminal data storage unit 15 is stored in the data storage unit DB1.

[0101] As described above, in the data analysis processing system 1000B of this modified example, when data (unknown data) that is not in a known normal state is input, the data is stored in the terminal data storage unit 15, and re-learning is performed at a predetermined timing to obtain a trained model (a model to be installed in the data classification processing unit 12B) that corresponds to the unknown data, thereby making it possible to handle the unknown data.

[0102] [Other embodiments] In the above embodiment (including the modified example), the data compression processing unit 13, 13A of the terminal device performs data compression processing on the input data D11, but this is not necessarily limited to this. For example, if the data Din is not large data, the data compression processing unit 13, 13A may not perform data compression processing on the input data D11 (the input data D11 may be output as data D21 as is). In this case, the data compression processing unit 13, 13A may be omitted from the terminal device 100, 100A.

[0103] When the data compression processing unit 13, 13A does not perform data compression processing on the input data D11, the data decompression processing unit 22, 23A of the server device does not perform data decompression processing on the data D31, and outputs the input data D31 as is as data D41 to the status analysis processing unit 23, 23A. Also, when the data compression processing unit 13, 13A does not perform data compression processing on the input data D11, the data decompression processing unit 22, 23A may be omitted from the server device 200, 200A.

[0104] In the above embodiments (including variations), in the data analysis processing systems 1000, 1000A, and 1000B, the terminal device 100 and the server device 200, the terminal device 100A and the server device 200A, and the terminal device 100B and the server device 200 are connected via a wired and / or wireless communication network (in the case of a 1:1 connection), but this is not limited to this, and in the data analysis processing systems 1000, 1000A, and 1000B, n (n: natural number) terminal devices and m (m: natural number) server devices may be connected (n:m connection) via a wired and / or wireless communication network, or a mixed wired and wireless communication network.

[0105] In addition, in the above embodiments (including variant examples), we have described the use of a learnable model in the data analysis processing systems 1000, 1000A, and 1000B, but the learnable model may also be a machine learning model, a deep learning model, a neural network model, or other model.

[0106] Furthermore, each block (each functional unit) of the data analysis processing systems 1000, 1000A, 1000B, the terminal devices 100, 100A, 100B, and the server devices 200, 200A described in the above embodiments (including modifications) may be individually implemented as a single chip using a semiconductor device such as an LSI, or may be integrated into a single chip to include some or all of the blocks. Furthermore, each block (each functional unit) of the data analysis processing systems 1000, 1000A, 1000B, the terminal devices 100, 100A, 100B, and the server devices 200, 200A described in the above embodiments (including modifications) may be realized by multiple semiconductor devices such as LSIs.

[0107] Although we have referred to it as an LSI here, it may also be called an IC, system LSI, super LSI, or ultra LSI depending on the level of integration.

[0108] Furthermore, the method of integration is not limited to LSI, but may be realized by dedicated circuits or general-purpose processors. It is also possible to use FPGAs (Field Programmable Gate Arrays), which can be programmed after the LSI is manufactured, or reconfigurable processors, which allow the connections and settings of circuit cells inside the LSI to be reconfigured.

[0109] Furthermore, part or all of the processing of each functional block in each of the above embodiments (including modified examples) may be realized by a program. And part or all of the processing of each functional block in each of the above embodiments is performed by a central processing unit (CPU) in a computer. Furthermore, the programs for performing each processing are stored in a storage device such as a hard disk or ROM, and are executed in the ROM or read out to the RAM.

[0110] Furthermore, each process in the above-described embodiment (including modifications) may be realized by hardware, or by software (including cases where it is realized together with an OS (operating system), middleware, or a predetermined library). Furthermore, it may be realized by a combination of software and hardware.

[0111] For example, when each functional unit of the above embodiment (including the modified examples) is realized by software, each functional unit may be realized by software processing using the hardware configuration shown in FIG. 9 (for example, a hardware configuration in which a CPU, GPU, processor, ROM, RAM, memory, input unit, output unit, etc. are connected via a bus).

[0112] Furthermore, when each functional unit of the above embodiment (including modified examples) is realized by software, the software may be realized using a single computer having the hardware configuration shown in Figure 9, or may be realized by distributed processing using multiple computers.

[0113] Furthermore, the execution order of the processing method in the above embodiment (including the modified examples) is not necessarily limited to the description of the above embodiment, and the execution order can be changed within the scope of the gist of the invention. Furthermore, in the processing method in the above embodiment, some steps may be executed in parallel with other steps within the scope of the gist of the invention. Furthermore, in the processing method in the above embodiment, processes executed in parallel may be executed serially (sequentially).

[0114] The scope of the present invention includes a computer program for causing a computer to execute the above-described method and a computer-readable recording medium having the program recorded thereon, including, for example, a flexible disk, a hard disk, a CD-ROM, an MO, a DVD, a DVD-ROM, a DVD-RAM, a large-capacity DVD, a next-generation DVD, and a semiconductor memory.

[0115] The computer program is not limited to one recorded on the recording medium, but may be one transmitted via a telecommunications line, a wireless or wired communication line, a network such as the Internet, or the like.

[0116] Furthermore, in the description of this specification and the claims, "optimal" refers to the best state or the state that is brought to the best, and the "optimal parameters" of a system (model) refer to parameters that make the value of the objective function of the system (model) the optimal value. The "optimal value" is the maximum value when the system is in a better state as the value of the system's objective function increases, and is the minimum value when the system is in a better state as the value of the system's objective function decreases. The "optimal value" may also be an extreme value. The "optimal value" may also allow for a predetermined error (measurement error, quantization error, etc.), and may be a value within a predetermined range (a range that can be considered to have sufficiently converged).

[0117] The term "part" may also include the concept of "circuitry." A circuitry may be realized in whole or in part by hardware, software, or a combination of hardware and software.

[0118] The functions of the elements disclosed herein may be implemented using circuitry or processing circuitry, including general-purpose processors, special-purpose processors, integrated circuits, ASICs ("application-specific integrated circuits"), conventional circuitry, and / or combinations thereof, configured to perform the disclosed elements or programmed to perform the disclosed functions. A processor is considered to be processing circuitry or circuitry when it includes transistors and other circuitry therein. In this disclosure, a circuitry, unit, or means is hardware that performs the recited function or hardware programmed to perform the function. The hardware may be any hardware disclosed herein or other known hardware that is programmed to perform or configured to perform the recited function. When the hardware is a processor, which may be considered as a type of circuitry, the circuitry, means, or unit is a combination of hardware and software, software used to configure the hardware, and / or processor.

[0119] The specific configuration of the present invention is not limited to the above-described embodiment, and various changes and modifications are possible without departing from the gist of the invention.

[0120] [Note] The present invention can also be realized as follows. A first invention is a data analysis processing system including a terminal device and a server device. The terminal device includes a status determination processing unit, a data classification processing unit, a data compression processing unit, and a first communication interface unit. The server device includes a second communication interface unit and a status analysis processing unit.

[0121] The state determination processing unit determines the state of the data to be processed, and obtains the result data of the determination as state determination data.

[0122] If the state determination data indicates a normal state, the data classification processing unit executes a data classification process on the data to be processed, and obtains an index indicating the state of the data to be processed.

[0123] The first communication interface is a communication interface capable of transmitting, to the server device, communication data including the index and / or the data to be processed when the status determination data does not indicate a normal status.

[0124] The second communication interface unit receives communication data transmitted from the terminal device.

[0125] If the communication data received by the second communication interface unit includes an index and / or processing target data, the status analysis processing unit inputs the index and / or processing target data and performs status analysis processing.

[0126] In this data analysis processing system, the terminal device determines whether the data to be processed (data that can be considered normal / abnormal) input to the terminal device is normal or abnormal, generates communication data according to the result of the determination, and transmits the generated communication data from the terminal device to the server device, thereby effectively suppressing an increase in the amount of communication between the terminal device and the server device. That is, in this data analysis processing system, (1) if the data to be processed is determined to be normal, communication data including an index that specifies the normal state of the data is transmitted from the terminal device to the server device, and (2) if the data to be processed is determined to be abnormal, communication data including the data to be processed is transmitted from the terminal device to the server device.

[0127] Therefore, in this data analysis processing system, an increase in the amount of communication between the terminal device and the server device can be effectively suppressed.

[0128] The second invention is the first invention, wherein the terminal device further includes a data compression processing unit that, when the status determination data does not indicate a normal status, performs data compression processing on the data to be processed and obtains data after the data compression processing of the data to be processed.

[0129] The first communication interface unit can transmit the communication data including the index and / or the data after data compression processing to the server device.

[0130] The server device further includes a data decompression processing unit that, when the communication data received by the second communication interface unit is communication data including data after data compression processing, acquires the data after data compression processing from the communication data and performs data decompression processing on the acquired data after data compression processing to acquire the data after data decompression processing.

[0131] When the data decompression processing unit acquires data after the data decompression processing, the status analysis processing unit inputs the index contained in the communication data received by the second communication interface unit and / or the data after the data decompression processing, and performs status analysis processing.

[0132] In this data analysis processing system, the terminal device determines whether data to be processed (data that can be considered normal / abnormal) input to the terminal device is normal or abnormal, generates communication data according to the result of the determination, and transmits the generated communication data from the terminal device to the server device, thereby effectively suppressing an increase in the amount of communication between the terminal device and the server device. That is, in this data analysis processing system, (1) if the data to be processed is determined to be normal, communication data including an index that specifies the normal state of the data is transmitted from the terminal device to the server device, and (2) if the data to be processed is determined to be abnormal, data compression processing is performed on the data, the compressed data is obtained, and communication data including the obtained compressed data is transmitted from the terminal device to the server device.

[0133] Therefore, in this data analysis processing system, an increase in the amount of communication between the terminal device and the server device can be effectively suppressed.

[0134] Furthermore, in this data analysis processing system, abnormal state data that requires detailed analysis is reliably transmitted from the terminal device to the server device, allowing the server device to perform data analysis processing with high accuracy, and for normal state data, the server device obtains data corresponding to the index received from the terminal device from the data storage unit, allowing the server device to always obtain normal state data and abnormal state data in chronological order. Therefore, for example, in this data analysis processing system, if normal state data continues and then an abnormal state occurs, time-series data analysis processing can be performed using the normal state data that continued before the abnormal state occurred and the data at the time the abnormal state occurred, allowing for highly accurate analysis processing (e.g., understanding the state transitions leading up to the occurrence of the abnormal state, investigating the cause of the abnormal state, etc.).

[0135] In this way, the data analysis processing system can perform data analysis processing with high accuracy while reducing the amount of data communication between the terminal device that acquires data and the server device that collects the data acquired by the terminal device via a communication network. The data to be processed by the data analysis processing system may be any data that can be determined as normal or abnormal, such as image data, video data, audio data, noise data, temperature data, pressure data, velocity data, acceleration data, and data of a predetermined physical quantity acquired by a predetermined sensor, etc., which can be determined as normal or abnormal.

[0136] A third invention is the first invention, in which the server device is used together with a data storage device. (1) When the communication data received by the second communication interface unit is communication data including an index, the status analysis processing unit stores the index included in the communication data in the data storage device; (2) When the communication data received by the second communication interface unit is communication data including the data to be processed, the status analysis processing unit acquires, from the data storage device, normal status data corresponding to the index stored in the data storage device. The status analysis processing unit performs status analysis processing based on the normal status data and the processing target data.

[0137] As a result, in this data analysis processing system, when data in a normal state is acquired, only the index is stored in the data storage unit, and when an abnormal state occurs, data corresponding to the index of the past normal state can be acquired from the data storage unit. Then, for example, time-series data analysis processing can be performed using the data in the past normal state and the data when the abnormal state occurred.

[0138] A fourth aspect of the present invention is the second aspect of the present invention, in which the server device is used together with a data storage device. (1) When the communication data received by the second communication interface unit is communication data including an index, the status analysis processing unit stores the index included in the communication data in the data storage device; (2) When the communication data received by the second communication interface unit is communication data including data after data compression processing, the data decompression processing unit acquires the data after data compression processing from the communication data, and performs data decompression processing on the acquired data after data compression processing to acquire data after data decompression processing; The status analysis processing unit acquires normal status data corresponding to the index stored in the data storage device from the data storage device. The state analysis processing unit performs state analysis processing based on the normal state data and the data after data decompression processing.

[0139] As a result, in this data analysis processing system, when data in a normal state is acquired, only the index is stored in the data storage unit, and when an abnormal state occurs, data corresponding to the index of the past normal state can be acquired from the data storage unit. Then, for example, time-series data analysis processing can be performed using the data in the past normal state and the data when the abnormal state occurred (data after data expansion processing).

[0140] The fifth invention is the second or fourth invention, wherein when the data classification processing unit determines that there are multiple index candidates indicating the state of the data to be processed, the data classification processing unit outputs the data to be processed and data transmission instruction information to the data compression processing unit, instructing the data to be compressed and transmitted to the server device.

[0141] The data compression processing unit obtains data after the data compression processing by performing data compression processing on the data to be processed output from the data classification processing unit based on the data transmission instruction information, and outputs the obtained data after the data compression processing and data including the transmission instruction information to the first communication interface unit as data after the data compression processing with transmission instruction information.

[0142] The first communication interface unit transmits the communication data including the data compressed with transmission instruction information input from the data compression processing unit to the server device.

[0143] The second communication interface receives the communication data transmitted from the first communication interface, and outputs the data compressed with transmission instruction information included in the communication data to the data decompression processor.

[0144] The data decompression processing unit acquires transmission instruction information from the data output from the second communication interface unit, and when the transmission instruction information is acquired, acquires data after data decompression processing by performing a data decompression process on the data after data compression processing included in the data after data compression processing with transmission instruction information, and acquires the acquired data after data decompression processing as new normal state data.

[0145] As a result, in this data analysis processing system, when data (unknown data) that is not in a known normal state is input, the data is stored in the data storage unit, and, for example, re-learning is performed at a predetermined timing to obtain a trained model that corresponds to the unknown data, and the trained model is installed in the data classification processing unit, thereby making it possible to handle unknown data.

[0146] A sixth aspect of the present invention is any one of the first to fourth aspects of the present invention, further comprising a terminal data storage unit.

[0147] If the data classification processing unit determines that there are multiple index candidates that indicate the state of the data to be processed, The data classification processing unit outputs the processing target data to the terminal data storage unit, The terminal data storage unit stores the data output from the data classification processing unit as new normal state data.

[0148] As a result, in this data analysis processing system, when data (unknown data) that is not in a known normal state is input, the data is stored in the terminal data memory unit, and, for example, re-learning is performed at a predetermined timing to obtain a trained model that corresponds to the unknown data, and the obtained trained model can be installed, for example, in a data classification processing unit, thereby making it possible to handle unknown data.

[0149] A seventh aspect of the present invention is a terminal device used in the data analysis processing system according to any one of the first to fourth aspects of the present invention.

[0150] An eighth aspect of the present invention is a server device used in the data analysis processing system according to any one of the first to fourth aspects of the present invention.

[0151] A ninth invention is a data analysis processing method executed using a data analysis processing system having a terminal device and a server device, and includes a status determination processing step, a data classification processing step, a first communication step, a second communication step, and a status analysis processing step.

[0152] In the state determination processing step, the terminal device determines the state of the data to be processed, and acquires the result data of the determination as state determination data.

[0153] In the data classification processing step, if the state determination data indicates a normal state, the terminal device executes a data classification process on the data to be processed and obtains an index indicating the state of the data to be processed.

[0154] In the first communication step, the terminal device transmits, to the server device, communication data including the index and / or the data to be processed when the status determination data does not indicate a normal status.

[0155] In the second communication step, the server device receives the communication data transmitted from the terminal device.

[0156] In the status analysis processing step, if the communication data received in the second communication step includes an index and / or processing target data, the server device inputs the index and / or processing target data and performs status analysis processing.

[0157] This makes it possible to realize a data analysis processing method that has the same effects as the first aspect of the invention.

[0158] A tenth invention is a data analysis processing method executed using a data analysis processing system that includes a terminal device and a server device and is used in conjunction with a data storage device that can store normal state data, and that includes a state determination processing step, a data classification processing step, a first communication step, a second communication step, a data acquisition processing step, a data decompression processing step, and a state analysis processing step.

[0159] In the state determination processing step, the terminal device determines the state of the data to be processed, and acquires the result data of the determination as state determination data.

[0160] In the data classification processing step, if the state determination data indicates a normal state, the terminal device executes a data classification process on the data to be processed and obtains an index indicating the state of the data to be processed.

[0161] In the data compression processing step, if the status determination data does not indicate a normal status, the terminal device executes data compression processing on the processing target data and acquires data after data compression processing of the processing target data.

[0162] In the first communication step, the terminal device transmits communication data including the index and / or the data after data compression processing to the server device.

[0163] In the second communication step, the server device receives the communication data transmitted from the terminal device.

[0164] In the data acquisition processing step, if the communication data received in the second communication step is communication data including an index, the server device acquires the index from the communication data and acquires normal state data corresponding to the index from the data storage device.

[0165] In the data decompression processing step, if the communication data received in the second communication step is communication data including data after data compression processing, the server device obtains the data after data compression processing from the communication data and performs data decompression processing on the obtained data after data compression processing, thereby obtaining data after data decompression processing.

[0166] In the status analysis processing step, when data after data decompression processing is obtained by the data decompression processing step, the server device performs status analysis processing based on the index contained in the communication data received by the second communication step and / or the data after data decompression processing.

[0167] This makes it possible to realize a data analysis processing method that has the same effects as the second aspect of the invention.

[0168] An eleventh aspect of the present invention is a program for executing the data analysis processing method of the ninth or tenth aspect of the present invention on a computer.

[0169] This makes it possible to realize a program for executing on a computer a data analysis processing method that has the same effects as the ninth or tenth invention. [Explanation of symbols]

[0170] 1000, 1000A, 1000B Data Analysis and Processing System 100, 100A, 100B terminal equipment 200, 200A server equipment 11. State determination processing unit 12, 12A, 12B Data classification processing section 13 Data compression processing section 14 First communication interface 15. Terminal data storage unit 21 Second communication interface 22 Data decompression processing unit 23 Status analysis processing unit DB1, DB1A data storage unit

Claims

1. A terminal device; a server device; A data analysis processing system comprising: The terminal device a state determination processing unit that determines a state of the data to be processed and acquires result data of the determination as state determination data; a data classification processing unit that executes a data classification process on the processing target data when the state determination data indicates a normal state, and acquires an index indicating a state of the processing target data; a first communication interface unit that is a communication interface capable of transmitting, to the server device, communication data including the index and / or the processing target data when the status determination data does not indicate a normal status; Equipped with The server device a second communication interface unit for receiving communication data transmitted from the terminal device; a status analysis processing unit that, when the index and / or the processing target data is included in the communication data received by the second communication interface unit, inputs the index and / or the processing target data and performs a status analysis process; Equipped with Data analysis processing system.

2. The terminal device a data compression processing unit that performs a data compression process on the processing target data when the state determination data does not indicate a normal state, and acquires data after the data compression process on the processing target data; the first communication interface unit is capable of transmitting communication data including the index and / or the data after data compression processing to the server device; The server device a data decompression processing unit that, when the communication data received by the second communication interface unit is communication data including the data after data compression processing, acquires the data after data compression processing from the communication data and performs data decompression processing on the acquired data after data compression processing to acquire data after data decompression processing, The state analysis processing unit When the data decompression processing unit acquires the data after the data decompression processing, the data decompression processing unit inputs the index and / or the data after the data decompression processing included in the communication data received by the second communication interface unit, and performs a state analysis process. The data analysis processing system according to claim 1 .

3. the server device is used together with a data storage device; (1) When the communication data received by the second communication interface unit is communication data including the index, the status analysis processing unit stores the index included in the communication data in the data storage device; (2) When the communication data received by the second communication interface unit is communication data including the processing target data, The status analysis processing unit acquires normal status data corresponding to the index stored in the data storage device from the data storage device. the status analysis processing unit performs the status analysis processing based on the normal status data and the processing target data. The data analysis processing system according to claim 1 .

4. the server device is used together with a data storage device; (1) When the communication data received by the second communication interface unit is communication data including the index, the status analysis processing unit stores the index included in the communication data in the data storage device; (2) When the communication data received by the second communication interface unit is communication data including the data after the data compression processing, the data decompression processing unit acquires the data after the data compression processing from the communication data, and performs a data decompression process on the acquired data after the data compression processing to acquire data after the data decompression processing; The status analysis processing unit acquires normal status data corresponding to the index stored in the data storage device from the data storage device. the state analysis processing unit performs the state analysis processing based on the normal state data and the data after the data decompression processing. The data analysis processing system according to claim 2 .

5. If the data classification processing unit determines that there are multiple candidates for an index indicating the state of the processing target data, the data classification processing unit outputs the processing target data and data transmission instruction information instructing the data compression processing unit to transmit data of the processing target data after data compression to the server device; the data compression processing unit acquires data after the data compression processing by performing a data compression process on the processing target data output from the data classification processing unit based on the data transmission instruction information, and outputs data including the acquired data after the data compression processing and the transmission instruction information to the first communication interface unit as data after the data compression processing with transmission instruction information; the first communication interface unit transmits communication data including the data compressed with the transmission instruction information input from the data compression processing unit to the server device; the second communication interface unit receives the communication data transmitted from the first communication interface unit, and outputs the data compressed with the transmission instruction information included in the communication data to the data decompression processing unit; the data decompression processing unit acquires the transmission instruction information from the data output from the second communication interface unit, and when the transmission instruction information is acquired, executes a data decompression process on the data after the data compression process included in the data after the data compression process with transmission instruction information to acquire data after the data decompression process, and acquires the acquired data after the data decompression process as new data in a normal state.

5. The data analysis processing system according to claim 2 or 4.

6. Further comprising a terminal data storage unit, If the data classification processing unit determines that there are multiple candidates for an index indicating the state of the processing target data, the data classification processing unit outputs the processing target data to the terminal data storage unit; The terminal data storage unit stores the data output from the data classification processing unit as new normal state data.

5. The data analysis processing system according to claim 1.

7. 5. The terminal device used in the data analysis processing system according to claim 1.

8. The server device used in the data analysis processing system according to any one of claims 1 to 4.

9. A data analysis processing method executed by using a data analysis processing system including a terminal device and a server device, a status determination processing step in which the terminal device determines the status of the data to be processed and acquires result data of the determination as status determination data; a data classification processing step in which, when the state determination data indicates a normal state, the terminal device executes a data classification process on the processing target data and acquires an index indicating a state of the processing target data; a first communication step in which the terminal device transmits, to the server device, communication data including the index and / or the processing target data when the status determination data does not indicate a normal status; Equipped with a second communication step in which the server device receives communication data transmitted from the terminal device; a status analysis processing step in which, when the index and / or the processing target data is included in the communication data received in the second communication step, the server device inputs the index and / or the processing target data and performs a status analysis process; Equipped with Data analysis processing methods.

10. A data analysis processing method executed by using a data analysis processing system including a terminal device and a server device, a status determination processing step in which the terminal device determines the status of the data to be processed and acquires result data of the determination as status determination data; a data classification processing step in which, when the state determination data indicates a normal state, the terminal device executes a data classification process on the processing target data and acquires an index indicating a state of the processing target data; a data compression processing step in which, when the status determination data does not indicate a normal status, the terminal device performs a data compression process on the processing target data and acquires data after the data compression process on the processing target data; a first communication step in which the terminal device transmits communication data including the index and / or the data after data compression processing to the server device; Equipped with a second communication step in which the server device receives communication data transmitted from the terminal device; a data decompression processing step in which, when the communication data received in the second communication step is communication data including the data after data compression processing, the server device acquires the data after data compression processing from the communication data and performs data decompression processing on the acquired data after data compression processing to acquire data after data decompression processing; a status analysis processing step in which, when the data after the data decompression processing is acquired by the data decompression processing step, the server device performs a status analysis processing based on the index and / or the data after the data decompression processing included in the communication data received by the second communication step; Equipped with Data analysis processing methods.

11. A program for executing the data analysis processing method according to claim 9 or 10 on a computer.

Images