Information processing method, information processing system, and computer program

Abstract

This information processing method involves a process, executed by a computer, of acquiring time-series measurement data of a plurality of power storage elements, and generating compressed data by performing compression processing on difference data between the acquired measurement data of each of the power storage elements and reference measurement data.

Description

Information processing method, information processing system, and computer program

[0001] The present invention relates to an information processing method, an information processing system, and a computer program.

[0002] Energy storage devices are widely used in uninterruptible power supplies, DC power supplies, etc. In addition, the use of energy storage devices is expanding in large-scale systems that store electricity generated by renewable energy or existing power generation systems. Large-scale systems use multiple energy storage devices.

[0003] For stable operation of a system using energy storage elements, it is important to understand the state and lifespan of the energy storage elements. Regarding methods for diagnosing the state and predicting the lifespan of energy storage elements, various methods have been proposed, such as those using measurement data such as voltage, current, and temperature observed during charging and discharging of the energy storage elements, with the aim of improving accuracy (see, for example, Patent Document 1).

[0004] JP 2013-003115 A

[0005] A large-scale energy storage system is constructed using a large number of energy storage elements. For example, a large-scale photovoltaic power generation system known as a mega solar power generation system is equipped with an extremely large number of energy storage elements, numbering in the millions, far exceeding the number of energy storage elements used, for example, for powering a vehicle or for auxiliary machinery. In such a large-scale energy storage system, if data is measured for each energy storage element at predetermined intervals, the amount of measurement data obtained will be enormous. It is not easy to collectively and accurately grasp the vast amount of measurement data, and performing various processes on the entire vast amount of measurement data requires a considerable amount of processing time. Therefore, a technology suitable for processing multiple pieces of measurement data is desired.

[0006] An object of the present disclosure is to provide a technique suitable for processing using multiple pieces of measurement data.

[0007] An information processing method according to one aspect of the present disclosure involves a computer acquiring time-series measurement data for multiple storage elements, and performing a compression process on the differential data between the acquired measurement data for each storage element and reference measurement data to generate compressed data.

[0008] According to the present disclosure, it is possible to provide a technique suitable for processing using a plurality of measurement data.

[0009] FIG. 1 is a diagram illustrating an overview of a remote monitoring system according to an embodiment of the present invention. FIG. 2 is a block diagram illustrating a configuration example of a power generation system. FIG. 3 is a block diagram illustrating a configuration example of an information processing device. FIG. 4 is a block diagram illustrating a configuration example of a terminal device. FIG. 5 is a functional block diagram illustrating a configuration example of an information processing device. FIG. 6 is a diagram illustrating an example of compressed data. FIG. 7 is a schematic diagram illustrating an example of a screen displayed on a display unit of a terminal device. FIG. 8 is a flowchart illustrating an example of a processing procedure executed by an information processing device.

[0010] (1) In an information processing method according to one aspect of the present disclosure, a computer acquires time-series measurement data for a plurality of energy storage elements, and performs a compression process on the differential data between the acquired measurement data for each energy storage element and the reference measurement data to generate compressed data.

[0011] According to the information processing method described in (1) above, measurement data acquired from a plurality of energy storage elements included in a large-scale energy storage system can be converted into compressed data. In this specification, compression processing refers to lossy compression, not lossless compression, which can be compressed and decompressed. By compressing the measurement data to reduce the information in the original data, a huge amount of measurement data can be converted into an easy-to-handle form and provided as data suitable for subsequent processing. By generating compressed data, unnecessary noise can be reduced and the accuracy of subsequent data analysis can be improved. By compressing the difference data between the measurement data and reference measurement data rather than compressing the measurement data itself, the deviation from the reference state can be reflected in the compressed data, allowing the behavior of each energy storage element in the population to be appropriately understood.

[0012] (2) In the information processing method described in (1) above, an abnormality in the storage element may be detected based on the generated compressed data.

[0013] According to the information processing method described in (2) above, the interpretability of anomaly detection can be improved by using compressed data obtained by compressing measurement data. By detecting anomalies based on measurement data as relative values ​​with respect to a reference state, false detections can be reduced and anomalies can be detected with high accuracy. Using compressed data can suppress an increase in the processing load associated with a large-scale energy storage system, leading to efficient anomaly detection.

[0014] (3) In the information processing method described in (1) or (2) above, distribution information that visualizes the distribution of the generated compressed data may be output.

[0015] According to the information processing method described in (3) above, the distribution of compressed data can be visualized and presented, allowing the user to visually and clearly recognize the distribution of compressed data. By presenting the compressed information, the state of each storage element can be easily grasped at a glance.

[0016] (4) In the information processing method described in any one of (1) to (3) above, the compressed data may be classified into normal clusters and abnormal clusters using a clustering technique, and an abnormality in the storage element may be detected based on the classification results.

[0017] According to the information processing method described in (4) above, anomaly detection can be easily and accurately performed using a clustering method. By using a clustering method, which is a type of machine learning method, anomalies can be detected from features based on the obtained measurement data, reducing the dependency on individual users in anomaly detection. By using compressed data, the load of clustering can be reduced and accuracy can be improved compared to when the measurement data itself is used. By using an unsupervised learning algorithm, the preparation of training data is unnecessary, allowing unknown anomalies to be recognized with high accuracy.

[0018] (5) In the information processing method described in (4) above, the normal clusters and abnormal clusters may be superimposed in a visually distinguishable manner on distribution information that visualizes the distribution of the compressed data and output.

[0019] According to the information processing method described in (5) above, the distribution status of compressed data and anomaly detection results can be presented in association with each other, allowing the user to visually clearly recognize the information. By visualizing and presenting the anomaly detection results, the explainability of the anomaly detection is improved.

[0020] (6) In the information processing method described in any one of (1) to (5) above, the differential data may be generated by calculating the difference between the average or median of the measurement data of the multiple storage elements at the same time as the reference measurement data and the measurement data of each storage element.

[0021] According to the information processing method described in (6) above, reference measurement data can be calculated based on measurement data of a plurality of storage elements. By using a relative value with respect to the reference measurement data that takes into account the actual state of a plurality of storage elements, the accuracy of subsequent processing such as anomaly detection can be improved. Anomaly storage elements that exhibit behavior that deviates from the group can be detected with high accuracy.

[0022] (7) In the information processing method according to any one of (1) to (6) above, the compression process may be performed using principal component analysis, UMAP, t-SNE, or an autoencoder.

[0023] According to the information processing method described in (7) above, the dimension of measurement data can be easily and accurately reduced using a machine learning technique. Since data can be compressed while preserving the original feature values, compressed data that appropriately reflects the behavior of each storage element can be generated. By using an unsupervised learning algorithm, the preparation of training data is unnecessary, and the characteristics of diversely changing measurement data can be accurately recognized.

[0024] (8) In the information processing method according to any one of (1) to (7) above, the measurement data may include at least one of a voltage, a temperature, and a state of charge (SOC) of the storage element.

[0025] According to the information processing method described in (8) above, by using data on voltage, temperature, SOC, or a combination thereof, which are sensitive to changes and reflect the internal state of the storage element well, among the measurement data, it is possible to provide data that properly represents the state of the storage element.

[0026] (9) In the information processing method described in any one of (2) to (8) above, an abnormality in the storage element may be further detected using a storage element simulator that estimates measurement data of the storage element based on the internal state of the storage element.

[0027] According to the information processing method described in (9) above, in addition to the anomaly detection by the above-mentioned technique, anomaly detection can be performed using an energy storage element simulator, thereby enabling multifaceted judgment. Because the measurement data of an energy storage element depends on the internal state of the energy storage element, false detections can be reduced by using a simulator that can take the internal state of the energy storage element into account in anomaly detection. However, since it is generally difficult for a simulator to handle internal states that engineers have not anticipated, the reliability of anomaly detection for both anticipated and unknown anomalies can be improved by using anomaly detection processing using a machine learning technique in combination with a method for detecting anomalies by focusing on the feature quantities of the measurement data itself to be detected.

[0028] (10) An information processing system according to one aspect of the present disclosure includes an acquisition unit that acquires time-series measurement data for a plurality of storage elements, and a generation unit that generates compressed data by performing a compression process on differential data between the acquired measurement data for each storage element and reference measurement data.

[0029] (11) A computer program according to one aspect of the present disclosure causes a computer to acquire time-series measurement data for a plurality of energy storage elements, and to perform a compression process on differential data between the acquired measurement data for each energy storage element and reference measurement data to generate compressed data.

[0030] Hereinafter, the present disclosure will be specifically described with reference to the drawings showing embodiments thereof.

[0031] (First embodiment) Fig. 1 shows an overview of a remote monitoring system 100 according to this embodiment. Information relating to the energy storage elements included in a power generation system 200 in the remote monitoring system 100 can be remotely accessed. The remote monitoring system 100 is an example of an information processing system. The remote monitoring system 100 includes an information processing device 50 as a main device. The information processing device 50 collects information from the power generation system 200 that is the target of remote monitoring. The information processing device 50 is connected to a network N1 such as the Internet. A terminal device 60 and the power generation system 200 are also connected to the network N1.

[0032] The information processing device 50 and the terminal device 60 are not limited to being separate devices, and may be, for example, a single processing device that serves as both the information processing device 50 and the terminal device 60. The information processing device 50 and / or the terminal device 60 may be integrated into any one of the power generation systems 200. The number of power generation systems 200 may be one or three or more.

[0033] FIG. 2 is a block diagram showing an example configuration of a power generation system 200. Power generation devices such as a solar power generation system and a wind power generation system are not shown. The power generation system 200 includes a communication device 10, a domain management device 30, and a power storage unit (domain) 40. The server device 20 is connected to the communication device 10 via a network N2. The power storage unit 40 may include multiple banks 41. The power storage unit 40 is housed in a battery panel and used in, for example, a thermal power generation system, a mega solar power generation system, a wind power generation system, an uninterruptible power supply (UPS), a stabilized power supply system for railways, and the like. A configuration including the communication device 10, the domain management device 30, and the power storage unit 40 is called a power storage system. The power storage system may also include a power conditioner (not shown). The power storage unit 40 is not limited to industrial use, and may also be for home use.

[0034] The business operator designs, installs, operates, and maintains a power storage system including a communication device 10, a domain management device 30, and a power storage unit 40, and can remotely monitor the power storage system using a remote monitoring system 100.

[0035] The communication device 10 includes a control unit 11, a storage unit 12, a first communication unit 13, and a second communication unit 14. The control unit 11 is composed of a CPU (Central Processing Unit) and the like, and controls the entire communication device 10 using built-in memories such as a ROM (Read Only Memory) and a RAM (Random Access Memory).

[0036] The storage unit 12 includes a non-volatile storage device such as a flash memory, etc. The storage unit 12 can store required information, for example, information obtained by processing by the control unit 11.

[0037] The first communication unit 13 includes a communication interface that realizes communication with the domain management device 30 or the battery management device 44. The control unit 11 can communicate with the domain management device 30 through the first communication unit 13.

[0038] The second communication unit 14 includes a communication interface that realizes communication via the network N2. The control unit 11 can communicate with the server device 20 through the second communication unit 14.

[0039] The domain management device 30 uses a predetermined communication interface to send and receive information to and from each bank 41. The storage unit 12 can store measurement data acquired via the domain management device 30.

[0040] The server device 20 can collect measurement data of the power storage system from the communication device 10. The measurement data includes measurement values ​​such as current, voltage, and temperature of each power storage element in the power storage system. The server device 20 may store the collected measurement data by dividing it into groups for each power storage element. The server device 20 can transmit the measurement data to the information processing device 50 via the networks N2 and N1. Note that the networks N1 and N2 may be a single communication network.

[0041] The bank 41 is made up of multiple storage modules connected in series, and includes a battery management unit (BMU) 44, multiple storage modules 42, and a measurement board (CMU: Cell Management Unit) 43 provided in each storage module 42.

[0042] The storage module 42 has a plurality of storage cells connected in series. The power generation system 200 is a large-scale storage system including a large number of storage cells, for example, more than one million cells. The storage cells are an example of a storage element. The storage elements are preferably rechargeable, such as secondary batteries such as lead-acid batteries and lithium-ion batteries, or capacitors. Some of the storage elements may be non-rechargeable primary batteries. The storage element may refer to the storage module 42, the bank 41, or a domain in which banks 41 are connected in parallel.

[0043] The measurement board 43 acquires measurement data relating to each power storage cell of the power storage module 42. The measurement data can be repeatedly acquired at an appropriate cycle, such as every 0.1 seconds, 0.5 seconds, or 1 second.

[0044] The battery management device 44 can communicate with the measurement board 43 equipped with a communication function via serial communication and can acquire measurement data detected by the measurement board 43. The battery management device 44 can send and receive information to and from the domain management device 30. The domain management device 30 aggregates measurement data from the battery management devices 44 of the banks belonging to the domain. The domain management device 30 outputs the aggregated measurement data to the communication device 10. In this way, the communication device 10 can acquire and store measurement data of the power storage unit 40 via the domain management device 30.

[0045] The communication device 10 transmits, at a predetermined timing (for example, at a regular interval, or when the amount of data satisfies a predetermined condition, etc.), measurement data stored since the previous timing to the server device 20. The communication device 10 associates the measurement data with identification information (for example, a cell ID) of the power storage cell and transmits the measurement data. The communication device 10 may transmit all measurement data obtained via the domain management device 30, or may transmit measurement data thinned out at a predetermined rate.

[0046] The information processing device 50 acquires measurement data of the power storage cells provided in the power generation system 200 via the server device 20, and monitors the state of the power generation system 200 based on the acquired measurement data. The information processing device 50 converts the acquired measurement data into an easy-to-handle format using a machine learning technique, and performs abnormality detection of the power storage cells using the converted data. The information processing device 50 presents the results of various analysis processes to the user via the terminal device 60. In this specification, machine learning refers to machine learning in a broad sense, including multivariate analysis such as clustering and principal component analysis.

[0047] 3 is a block diagram showing an example configuration of an information processing device 50. The information processing device 50 is a computer capable of various information processing and transmitting and receiving information, such as a server computer, a personal computer, or a quantum computer. The information processing device 50 may be a multi-computer consisting of multiple computers, or may be a virtual machine virtually constructed by software. The information processing device 50 includes a control unit 51, a storage unit 52, and a communication unit 53.

[0048] The control unit 51 is an arithmetic circuit including a CPU, a GPU (Graphics Processing Unit), a ROM, a RAM, etc. The CPU or GPU included in the control unit 51 executes various computer programs stored in the ROM or the storage unit 52, and controls the operation of each of the above-mentioned hardware components. The control unit 51 may also include functions such as a timer that measures the elapsed time from when an instruction to start measurement is given until when an instruction to end measurement is given, a counter that counts numbers, and a clock that outputs date and time information.

[0049] The storage unit 52 includes a non-volatile storage device such as a flash memory, a hard disk drive, etc. The storage unit 52 stores various computer programs and data referenced by the control unit 51. The storage unit 52 may be an external storage device connected to the information processing device 50.

[0050] The storage unit 52 of this embodiment stores a program 521 for causing a computer to execute processing using measurement data, and a measurement DB (Data Base) 522 as data necessary for executing this program 521 .

[0051] The measurement DB 522 is a database that stores measurement data received from the power generation system 200. As described above, the measurement data includes measurement values ​​such as the current, voltage, and temperature of the storage cells in the power generation system 200. The measurement data includes data when the storage cells are charged or discharged. The measurement data may include not only measurements such as the current, voltage, and temperature measured by the measurement board 43, but also calculated values ​​(e.g., SOC) calculated using these measurement values. The measurement DB 522 stores records that link information such as the identification information of the storage cells, the measurement date and time, and the measurement values, using, for example, an ID for identifying the measurement data as a key. The measurement DB 522 may also store results of anomaly detection, etc. Each time the control unit 51 receives measurement data transmitted from the server device 20, it stores the received measurement data in the measurement DB 522 in chronological order.

[0052] A computer program (program product) including the program 521 may be provided by a non-transitory recording medium 5A on which the computer program is readably recorded. The recording medium 5A is a portable memory such as a CD-ROM, a USB memory, or an SD (Secure Digital) card. The control unit 51 reads the desired computer program from the recording medium 5A using a reading device (not shown) and stores the read computer program in the memory unit 52. Alternatively, the computer program may be provided via communication. The program 521 may be a single computer program or may be composed of multiple computer programs, and may be executed on a single computer or on multiple computers interconnected by a communication network.

[0053] The communication unit 53 includes a communication interface that realizes communication via the network N1. The control unit 51 receives measurement data transmitted from the power generation system 200 through the communication unit 53. The control unit 51 transmits various processing results to an external device such as a terminal device 60 through the communication unit 53.

[0054] The information processing device 50 may include a display unit that displays various information, an operation unit that accepts user operations, and the like.

[0055] 4 is a block diagram showing an example configuration of the terminal device 60. The terminal device 60 is a computer capable of various information processing and transmitting and receiving information, such as a personal computer, a smartphone, or a tablet terminal. The terminal device 60 is used by a user such as a manager, maintenance person, or operator of the battery system of the power generation system 200. The terminal device 60 includes a control unit 61, a storage unit 62, a communication unit 63, a display unit 64, an operation unit 65, etc.

[0056] The control unit 61 is an arithmetic circuit including a CPU, a ROM, a RAM, etc. The CPU or GPU included in the control unit 61 executes various computer programs stored in the ROM or the storage unit 62, and controls the operation of each of the above-mentioned hardware components.

[0057] The storage unit 62 includes a non-volatile storage device such as a flash memory, a hard disk drive, etc. The storage unit 62 stores various computer programs and data referenced by the control unit 61. The control unit 61 displays various processing results provided by the information processing device 50 on the display unit 64 based on the computer programs stored in the storage unit 62.

[0058] The communication unit 63 includes a communication interface that realizes communication via the network N1. The control unit 61 transmits and receives information to and from the information processing device 50 through the communication unit 63.

[0059] The display unit 64 includes a display device such as a liquid crystal display or an organic electroluminescent display (OELD). The display unit 64 displays information to be notified to the user in accordance with instructions from the control unit 61. The display unit 64 may be interpreted as a notification unit and may notify the user by other means such as audio.

[0060] The operation unit 65 is an interface that accepts user operations. The operation unit 65 includes, for example, a keyboard, a touch panel device with a built-in display, a speaker, a microphone, etc. The operation unit 65 accepts operation input from the user and sends a control signal corresponding to the operation content to the control unit 61.

[0061] 5 is a functional block diagram showing an example configuration of the information processing device 50. The control unit 51 of the information processing device 50 reads and executes a program 521 stored in the storage unit 52, thereby realizing the functions of an acquisition unit 511, a first generation unit 512, a second generation unit 513, an abnormality detection unit 514, and an output unit 515. Each functional unit of the information processing device 50 may be realized by software, by hardware, or by a combination thereof.

[0062] The acquisition unit 511 acquires time-series measurement data for a predetermined period of time set in advance for a plurality of power storage cells in the power generation system 200. The measurement data is associated with identification information of the power storage cell. The acquisition unit 511 may acquire the measurement data transmitted from the server device 20 via the communication unit 53, or may acquire the measurement data by reading out the measurement data stored in the measurement DB 522. The predetermined period of time may be, for example, one hour, one day, one month, etc.

[0063] The measurement data includes one or more data selected from the voltage, temperature, SOC, current, and power of the storage cell. From the viewpoint of improving the accuracy of anomaly detection, the measurement data preferably includes at least one of voltage, temperature, and SOC, and more preferably includes voltage. In this embodiment, voltage data of the storage cell is acquired as the measurement data.

[0064] The acquiring unit 511 may acquire measurement data from all the power storage cells in the power generation system 200, but by selecting a predetermined number of power storage cells from among all the power storage cells as the power storage cells from which measurement data is to be acquired, the processing load can be reduced. The power storage cells from which measurement data is to be acquired may be selected in advance according to a predetermined rule or manually. The power storage cells from which measurement data is to be acquired are preferably, for example, power storage cells that represent the load or temperature in each unit when the power generation system 200 is divided into specific units (for example, banks). The number of power storage cells from which measurement data is to be acquired can be determined taking into account the total number of power storage cells in the power generation system 200.

[0065] The first generation unit 512 generates difference data representing the difference between the voltage value of each storage cell and a reference voltage value based on the voltage data of each storage cell acquired by the acquisition unit 511. The first generation unit 512 calculates the difference between the voltage value of the storage cell at the same point in time (the same measurement time) and the reference voltage value for each measurement time, thereby obtaining time-series difference data. The reference voltage value is, for example, the average or median of the voltage values ​​of all storage cells at the same measurement time. The difference data is high-dimensional data having a number of data points corresponding to the number of times the voltage values ​​are measured.

[0066] The second generation unit 513 uses a predetermined compression method to compress the differential data generated by the first generation unit 512 and generate compressed data. Examples of compression methods include dimensionality reduction methods, such as principal component analysis (PCA), uniform manifold approximation and projection (UMAP), t-distributed stochastic neighbor embedding (t-SNE), and autoencoders. Principal component analysis is a preferred dimensionality reduction method. The second generation unit 513 generates low-dimensional data by reducing the dimensionality of the differential data through compression processing. When presenting the compressed data to a user using distribution information, which will be described later, the compressed data is preferably reduced to two or three dimensions.

[0067] Fig. 6 is a diagram showing an example of compressed data. Fig. 6 shows an example in which measurement data for each storage cell over a predetermined period is compressed into two dimensions using principal component analysis. In Fig. 6, the compressed data for each storage cell is mapped onto a two-dimensional coordinate system with the horizontal axis representing the first principal component and the vertical axis representing the second principal component. As conceptually shown in Fig. 6, voltage data for a predetermined period corresponding to a certain storage cell is converted into compressed data represented by a single plot.

[0068] When performing dimensionality reduction using principal component analysis, the second generating unit 513 may determine principal components to use based on the cumulative contribution rate of each component. The second generating unit 513 may extract a predetermined number of principal components in descending order of cumulative contribution rate, thereby compressing the differential data to a predetermined number of dimensions. The second generating unit 513 may extract principal components whose cumulative contribution rate is equal to or greater than a predetermined threshold.

[0069] It is preferable that one compressed data set is generated based on the difference data for a predetermined period that is the unit of measurement data acquisition, but if the predetermined period is relatively long, the predetermined period may be divided into preset time units and compressed data may be generated for each time unit. The unit of generation of compressed data may be set appropriately depending on the purpose of the analysis process using the measurement data.

[0070] The second generating unit 513 may perform compression processing by combining a plurality of compression methods. For example, the second generating unit 513 may perform principal component analysis and then further compress the dimensions by UMAP.

[0071] The anomaly detection unit 514 detects an anomaly in the power storage cell based on the compressed data of each power storage cell generated by the second generation unit 513. The anomaly detection unit 514 of this embodiment performs anomaly detection using k-means, which is one of the clustering methods.

[0072] K-means is a type of unsupervised machine learning that automatically classifies each compressed data into a predetermined number of clusters. In this embodiment, as shown in Fig. 6, by specifying three cluster numbers, each compressed data is classified into one normal cluster and the remaining (two) abnormal clusters. In Fig. 6, the clusters surrounded by solid lines are normal clusters, and the clusters surrounded by dashed lines are abnormal clusters.

[0073] In k-means, the distance (e.g., Euclidean distance, Mahalanobis distance, etc.) between the compressed data and the center of each cluster is calculated, and the compressed data is classified into the cluster with the smallest calculated distance. Of the multiple clusters generated, the anomaly detection unit 514 can regard the cluster with the largest number of pieces of compressed data as a normal cluster, and the remaining clusters as abnormal clusters. The anomaly detection unit 514 identifies abnormal storage cells based on the clustering results of the compressed data of each storage cell.

[0074] In the above, anomaly detection is performed using k-means, but the anomaly detection method is not limited to this as long as it is possible to identify the presence or absence of anomalies in the compressed data of each storage cell. The anomaly detection unit 514 may also use methods such as hierarchical clustering, k-nearest neighbors, an autoregressive model, a neural network, or a support vector machine. The anomaly detection unit 514 may also use a combination of multiple anomaly detection methods.

[0075] The output unit 515 outputs to the terminal device 60 screen information indicating the compressed data generated by the second generation unit 513 and the abnormality detection result by the abnormality detection unit 514. The control unit 61 of the terminal device 60 displays a screen indicating the compressed data and the abnormality detection result on the display unit 64 based on the screen information transmitted through the output unit 515.

[0076] 7 is a schematic diagram showing an example of a screen 640 displayed on the display unit 64 of the terminal device 60. The screen 640 includes a result display unit 641 that displays information about a storage cell that has been detected as abnormal, and a data display unit 642 that displays compressed data of each storage cell.

[0077] The output unit 515 receives the abnormality detection result from the abnormality detection unit 514, and causes the power storage cell detected as abnormal to be displayed in a visually distinguishable form on the result display unit 641. In Fig. 7 , the result display unit 641 displays the cell ID of the power storage cell detected as abnormal, and an illustration showing its location within the power generation system 200.

[0078] The output unit 515 receives the compressed data from the second generation unit 513 and generates distribution information that visualizes the distribution of the received compressed data. The output unit 515 displays the generated distribution information on the data display unit 642. The distribution information is, for example, a distribution diagram in which the compressed data of each storage cell is mapped on an n-dimensional coordinate system with the first axis (horizontal axis) representing the first principal component, the second axis (vertical axis) representing the second principal component, the n-th axis representing the n-th principal component, and so on. FIG. 7 shows a two-dimensional distribution diagram.

[0079] Information visually indicating the clustering results or anomaly detection results by the anomaly detection unit 514 may be superimposed on the distribution map. In Fig. 7, the clustering results are visualized by displaying the compressed data on the distribution map using markers with different colors or shading (different shading in Fig. 7) for each cluster. In addition, the anomaly detection results are visually indicated by displaying a solid line on the distribution map that separates normal cluster regions from abnormal cluster regions.

[0080] Markers indicating each compressed data on the distribution map may be configured to be selectable. For example, when the output unit 515 receives a selection of a marker on the distribution map via the terminal device 60, the output unit 515 reads out the identification information of the energy storage cell corresponding to the received marker, the measurement data before compression, etc., based on the information stored in the storage unit 52. As shown in FIG. 7 , the output unit 515 displays the read identification information and measurement data in a pop-up format or in a separate window.

[0081] The information processing device 50 may be configured to perform processes up to generating compressed data without including the anomaly detection unit 514. In this case, only the compressed data is output as the processing result. Anomaly detection may be performed by the user based on the compressed data provided via the terminal device 60.

[0082] 8 is a flowchart showing an example of a processing procedure executed by the information processing device 50. The processing in each of the following flowcharts is executed by the control unit 51 in accordance with a program 521 stored in the storage unit 52 of the information processing device 50. The control unit 51 periodically executes the processing procedure shown below.

[0083] The control unit 51 of the information processing device 50 acquires time-series measurement data for a predetermined period of time from each of the plurality of power storage cells included in the power generation system 200 (step S11). The measurement data is, for example, voltage data.

[0084] The control unit 51 generates differential data for each storage cell by calculating the difference between the voltage value of each storage cell and a reference voltage value based on the acquired voltage data for each storage cell (step S12). The differential data is time-series data of voltage differences. The control unit 51 obtains the reference voltage value for each measurement time, for example, by calculating the average or median of the voltage values ​​of all storage cells at the same measurement time. The control unit 51 generates differential data for the target storage cell for which differential data is to be generated by calculating the difference between the reference voltage value at each measurement time and the voltage value of the target storage cell for which differential data is to be generated. Differential data is generated for each of the multiple storage cells.

[0085] In the above, a case has been described in which the average value or median value of the measurement data of all the energy storage cells is used as an example of the reference value (the reference voltage value in this embodiment). However, this is not limited to this. A reference value may be set in advance (using a separate experimental value), an updated reference value obtained by appropriately learning and updating the reference value may be used, a value obtained by correcting the reference value using a definition formula, the average value or median value of measurement data from multiple energy storage cells randomly selected by a computer, or the average value or median value of multiple representative energy storage cells selected in advance. Alternatively, an ideal reference value calculated by simulation may be used as the reference value. These reference values ​​may be selected depending on the situation and the scene.

[0086] The control unit 51 performs compression processing on the generated differential data using principal component analysis to generate compressed data for each power storage cell (step S13). The control unit 51 generates compressed data that has been dimensionally compressed to two dimensions using principal component analysis, for example.

[0087] The control unit 51 performs anomaly detection using a clustering method based on the compressed data of each power storage cell generated by the second generation unit 513 (step S14). The control unit 51 classifies the compressed data of each power storage cell into normal clusters and abnormal clusters. Based on the classification result, the control unit 51 detects a power storage cell corresponding to compressed data belonging to an abnormal cluster as an abnormal power storage cell.

[0088] The control unit 51 generates a screen including a distribution map showing the distribution of the compressed data of each storage cell and the abnormality detection result (step S15). The control unit 51 generates a screen including a distribution map in which the compressed data of each storage cell is mapped on a two-dimensional coordinate system and information indicating the storage elements detected as abnormal. The generation of the distribution map may be performed simultaneously with the generation of the compressed data or the clustering process. The control unit 51 visually displays the clustering result on the distribution map and displays normal clusters and abnormal clusters in a distinguishable manner.

[0089] The control unit 51 outputs the generated distribution map and the screen including the abnormality detection result to the terminal device 60 (step S16), and causes the screen to be displayed on the display unit 64 of the terminal device 60. The control unit 51 then ends the series of processes.

[0090] Second Embodiment An information processing device 50 according to a second embodiment performs anomaly detection processing using a power storage cell (power storage element) simulator in addition to anomaly detection using the machine learning technique described in the first embodiment. The following mainly describes the above differences.

[0091] The energy storage cell simulator refers to a simulator constructed to simulate the behavior of an energy storage cell. The energy storage cell simulator can output measurement data of the energy storage cell based on the internal state quantities of the energy storage cell. The internal state quantities provided to the energy storage cell simulator may include, for example, the SOC, internal temperature, positive electrode capacity, negative electrode capacity, and capacity imbalance of the energy storage cell. The capacity imbalance refers to the difference in capacity between the positive electrode and the negative electrode of the energy storage cell, at which charged ions can reversibly move in and out of the electrodes.

[0092] The energy storage cell simulator may further include a usage history of the energy storage cell as an input element. The usage history refers to information indicating a usage pattern (how the energy storage cell is used). The usage history may include, for example, information indicating a change in the power or current (load) of the energy storage cell over a predetermined period of time, information indicating a change in the environmental temperature over a predetermined period of time, etc.

[0093] For ease of explanation, the energy storage cell that is the target of the anomaly detection process will be referred to as a target cell, and the cell that serves as a comparison standard when performing the anomaly detection will be referred to as a representative cell. The representative cell is selected from the energy storage cells in the power generation system 200, taking into consideration, for example, the arrangement in the power generation system 200 and measurement data such as voltage, current, and temperature.

[0094] The information processing device 50 of the second embodiment performs anomaly detection using a power storage cell simulator in addition to anomaly detection using a machine learning technique in step S14 of the first embodiment.

[0095] Anomaly detection methods using a power storage cell simulator are not limited to the above, but an example is shown below. The information processing device 50 obtains the usage history of each of the representative cell and the target cell based on measurement data including the voltage, current, and temperature of the representative cell and the target cell. The information processing device 50 obtains internal state quantities corresponding to the acquired measurement data and usage history for each of the representative cell and the target cell using the power storage cell simulator. The information processing device 50 sets provisional values ​​for the internal state quantities, and searches for optimal values ​​for the internal state quantities based on the provisional values ​​of the set internal state quantities and the identified usage history so that the measurement data output from the power storage cell simulator approximates the actual measurement data. The information processing device 50 searches for optimal values ​​for the internal state quantities using a known optimization method, such as a genetic algorithm, a Nelder-Mead method, or a gradient method.

[0096] The information processing device 50 detects an abnormality in the target cell by determining whether the difference between the obtained internal state quantity of the representative cell and the internal state quantity of the target cell is less than a preset threshold. If the difference in the internal state quantity is less than the threshold, the information processing device 50 determines that the target cell is normal, and if the difference in the internal state quantity is equal to or greater than the threshold, the information processing device 50 determines that the target cell is abnormal.

[0097] In the above-described embodiment, an example has been described in which each process in the above flowchart is executed by the information processing device 50. Alternatively, some or all of the above-described processes may be executed by another processing entity, such as the domain management device 30, the server device 20, or the terminal device 60.

[0098] In the above-described embodiment, the remote monitoring system shown in Fig. 1 has been described, but the information processing method, the information processing system, and the computer program may be applied to a mobile object (such as an automobile, a train, an airplane, or a ship). Also, for example, in addition to the remote system shown in Fig. 1, all systems may be mounted (for example, all systems may be mounted on a mobile object) to complete the system.

[0099] The embodiments disclosed herein are illustrative in all respects and should not be considered limiting. The technical features described in each embodiment can be combined with each other, and the scope of the present invention is intended to include all modifications within the scope of the claims and equivalents thereto. The sequences shown in each embodiment are not limited, and within the scope of no contradiction, each processing step may be executed in a different order, or multiple processes may be executed in parallel. The entity that performs each process is not limited, and within the scope of no contradiction, the process of each device may be executed by another device.

[0100] The matters described in each embodiment can be combined with each other. Furthermore, the independent claims and dependent claims described in the claims can be combined with each other in any and all combinations, regardless of the reference format. Furthermore, the claims use a format in which a claim references two or more other claims (multiple claim format), but this is not limited to this. A multiple claim (multi-multi claim) that references at least one other multiple claim may also be used.

[0101] 100 Remote monitoring system (information processing system) 200 Power generation system 10 Communication device 11 Control unit 12 Memory unit 13 First communication unit 14 Second communication unit 20 Server device 30 Domain management device 40 Power storage unit 41 Bank 42 Power storage module 43 Measurement board 44 Battery management device 50 Information processing device 51 Control unit 52 Memory unit 53 Communication unit 521 Program 522 Measurement DB 5A Recording medium 60 Terminal device 61 Control unit 62 Memory unit 63 Communication unit 64 Display unit 65 Operation unit

Claims

1. An information processing method in which a computer executes a process of acquiring time-series measurement data for a plurality of storage elements, and generating compressed data by performing a compression process on the differential data between the acquired measurement data for each storage element and a reference measurement data.

2. The information processing method according to claim 1, further comprising detecting an abnormality in the storage element based on the generated compressed data.

3. The information processing method according to claim 1 or 2, further comprising outputting distribution information that visualizes the distribution of the generated compressed data.

4. The information processing method according to claim 1 or 2, further comprising classifying the compressed data into normal clusters and abnormal clusters using a clustering technique, and detecting an abnormality in the storage element based on the classification results.

5. The information processing method according to claim 4, further comprising the step of outputting the normal clusters and the abnormal clusters in a visually distinguishable form superimposed on distribution information that visualizes the distribution of the compressed data.

6. An information processing method as described in claim 1 or claim 2, wherein the differential data is generated by calculating the difference between the average or median of the measurement data of the multiple storage elements at the same time point as the reference measurement data and the measurement data of each storage element.

7. The information processing method according to claim 1 or 2, wherein the compression process is performed using principal component analysis, UMAP, t-SNE or an autoencoder.

8. The information processing method according to claim 1 or 2, wherein the measurement data includes at least one of a voltage, a temperature, and an SOC of a storage element.

9. The information processing method according to claim 2, further comprising the step of detecting an abnormality in the storage element by using a storage element simulator that estimates measurement data of the storage element based on an internal state of the storage element.

10. An information processing system comprising: an acquisition unit that acquires time-series measurement data for a plurality of storage elements; and a generation unit that generates compressed data by performing compression processing on differential data between the acquired measurement data for each storage element and reference measurement data.

11. A computer program that causes a computer to execute a process of acquiring time-series measurement data for a plurality of storage elements, and generating compressed data by performing a compression process on the differential data between the acquired measurement data for each storage element and the reference measurement data.

Images