Information processing device, information processing method, program and data structure

Abstract

To provide a technology capable of predicting abnormality of a storage battery with high accuracy by using data of the battery measured in periodically inspecting a storage battery system.SOLUTION: A program causes a computer to function as each means of an information processing device for detecting an abnormality sign relating to a plurality of storage batteries constituting an uninterruptible power supply. The information processing device includes acquisition means for acquiring a dataset, for a specific storage battery, to be recorded in each periodical inspection performed at a prescribed time interval and including information on a measurement result to the specific storage battery and information on a use state of the specific storage battery, and determination means for determining the existence / nonexistence of an abnormality sign of the specific storage battery by inputting the dataset for the specific storage battery to a first learning model which has learned in advance. The information on the measurement result to the specific storage battery includes a change degree of an internal resistance value of the specific storage battery in each prescribed time interval, and the information on the use state of the specific storage battery includes the number of use days from a charging date.SELECTED DRAWING: Figure 3

Description

Technical Field

[0001] The present invention relates to an information processing apparatus, an information processing method, a program, and a data structure.

Background Art

[0002] Conventionally, in a system using a power storage module in which a plurality of power storage cells are connected in series, a technique for determining the deterioration of the power storage cells is known (Patent Document 1). In the technique proposed in Patent Document 1, based on the power storage element information, the measured value time series data, and the predicted value time series data of the power storage cells, a learning model outputs a determination of the deterioration of the power storage cells and the presence or absence of an environmental abnormality.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] By the way, for example, in a data center, a battery system as an uninterruptible power supply device may be used in order to enable the computers installed in the data center to be always operational. A battery system as an uninterruptible power supply device charges a plurality of batteries and supplies the power of the batteries to the equipment in the data center when a power outage occurs.

[0005] In a battery system, for example, there is a configuration in which 100 or more batteries are connected in series. In such a battery system, if an abnormality occurs in one battery and a disconnection occurs, there is a risk of losing the power supply function. Therefore, it is necessary to detect an early sign of abnormality in the battery and perform appropriate maintenance such as replacement.

[0006] On the other hand, in countries where the frequency of power outages is low, it is rare for a battery system to operate due to an actual power outage. Nevertheless, if the battery system is configured to provide sensors for each battery to constantly monitor the status of the battery, the configuration of the battery system becomes complex and expensive. Even if the configuration of the battery system is not constantly monitored by sensors, it may be possible to appropriately maintain the battery system by manually performing regular inspections of the battery at predetermined time intervals (e.g., once a month to several months).

[0007] Against this background, in a battery system, a technology that can accurately predict abnormalities in a battery while utilizing battery data measured during regular inspections is desired.

[0008] The present invention has been made in view of the above problems, and an object thereof is to realize a technology capable of accurately predicting abnormalities in a battery using battery data measured during regular inspections of the battery system.

Means for Solving the Problems

[0009] To solve this problem, for example, the program of the present invention has the following configuration. That is, A program that causes a computer to function as each means of an information processing device for detecting an abnormality prediction related to a plurality of batteries constituting an uninterruptible power supply device, wherein the information processing device includes: An acquisition means for acquiring a data set for a specific battery including information regarding a measurement result for the specific battery and information regarding the usage status of the specific battery, which are recorded for each regular inspection performed at a predetermined time interval; A determination means for determining the presence or absence of an abnormality prediction for the specific battery by inputting the data set for the specific battery into a first pre-learned learning model, The information regarding the measurement results for the specific storage battery includes the degree of change in the internal resistance value of the specific storage battery at each of the predetermined time intervals, and the information regarding the usage status of the specific storage battery includes the number of days of use since the charging date, which is characterized by this.

Advantages of the Invention

[0010] According to the present invention, it becomes possible to accurately predict an abnormality of a storage battery by using data of the storage battery measured at the time of regular inspection of the storage battery system.

Brief Description of the Drawings

[0011]

Figure 1

Figure 2A

Figure 2B

Figure 2C

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7A

Figure 7B

Figure 8

Figure 9A

Figure 9B

Figure 10

Figure 11

Mode for Carrying Out the Invention

[0012] Hereinafter, embodiments will be described in detail with reference to the accompanying drawings. Note that the following embodiments do not limit the invention according to the claims, and not all combinations of the features described in the embodiments are essential for the invention. Two or more of the plurality of features described in the embodiments may be arbitrarily combined. Also, the same or similar configurations are given the same reference numerals, and duplicate descriptions are omitted.

[0013] (Overview of the Abnormal Sign Detection System for Storage Batteries) With reference to FIG. 1, the overview of the abnormal sign detection system for storage batteries will be described. The abnormal sign detection system 10 for storage batteries includes an information processing apparatus 100 which is an information processing server and a communication apparatus 210. The communication apparatus 210 is, for example, a personal computer or a mobile terminal arranged in the data center 200. The communication apparatus 210 receives data input regarding measurement results and usage status of the storage batteries by the maintenance staff of the storage battery group 230, and transmits information regarding the input measurement results and usage status to the information processing apparatus 100 via a network. Further, the communication apparatus 210 may receive data input by the maintenance staff of the storage battery group 230 or the administrator of the abnormal sign detection system 10 for storage batteries, and transmit correct data (described later) in the learning data to the information processing apparatus 100.

[0014] The information processing apparatus 100 receives and records information regarding measurement results and information regarding usage status from the respective communication apparatuses 210 of a plurality of data centers, and detects an abnormal sign of the storage battery based on a machine learning model.

[0015] In data center 200, a server group 220 that executes arbitrary processing (for example, processing of a tenant using server resources of a server group) is operating. The server group 220 normally operates with power supplied from a power source (not shown) in the data center 200. The battery group 230 includes a plurality of storage batteries and functions as an uninterruptible power supply device to supply power to the server group 220 in an emergency such as a power outage.

[0016] The battery group 230 in the data center 200 is composed of, for example, a plurality of battery systems (herein referred to as a UPS battery system) as shown in FIG. 2A. Each of the plurality of UPS battery systems is distinguished, for example, as "UPS No. 1", "UPS No. 2", "UPS No. 3", and "UPS reserve", and can independently supply power to the server group 220. The "UPS reserve" system operates, for example, when at least one of "UPS No. 1", "UPS No. 2", and "UPS No. 3" fails, requires maintenance, or cannot supply the necessary power due to insufficient power supply.

[0017] Each of the UPS battery systems includes a plurality of battery groups 240 such as a battery A group and a battery B group. These battery groups 240 may be connected in series. The battery group 240 is configured to generate a required voltage by connecting a plurality of storage batteries 250 in series, for example, as shown in FIG. 2B. When a disconnection or the like due to an abnormality occurs in one storage battery 250, the function as a UPS power supply device may be lost in one UPS battery system. Therefore, accurately detecting an abnormal sign for each storage battery is an important issue for functioning as a UPS power supply device.

[0018] (Detection of Abnormal Signs of Storage Batteries) Referring to FIG. 2C, the detection of an abnormal sign of the storage battery will be described. FIG. 2C shows an example of the transition of the internal resistance values of several individual storage batteries. The horizontal axis of FIG. 2C represents time, and the vertical axis represents the internal resistance value (mΩ). Graph 261 shows the transition of the internal resistance values measured at times t-3, ···, t, and t+1 for a specific storage battery. Times t-3, ···, t, and t+1 correspond to the timings at the time of measurement in the regularly performed periodic inspections. That is, graph 261 is a graph of the measurement results for the storage battery recorded in the periodic inspection. Graph 262 and graph 263 also represent the measurement results for individual storage batteries.

[0019] In the conventional abnormal detection method, when the internal resistance value of the storage battery exceeds a predetermined threshold value 264, it is determined that an abnormal sign is indicated, and the storage battery is replaced or inspected. However, as shown in graph 262, when the inspection timing is t, it does not exceed the threshold value, but if the internal resistance value of the storage battery rapidly increases between the inspection timings t and t+1, the storage battery may fail before the inspection timing t+1.

[0020] In response to such a problem, in the present embodiment, in order to detect a storage battery whose internal resistance value will exceed the threshold value in the future, in addition to the internal resistance value, the transition of the change in the internal resistance value is taken into account to detect an abnormal sign of the storage battery. Actually, for the internal resistance value of a normal storage battery, even if it increases with time, the degree of increase decreases. On the other hand, the internal resistance value of a storage battery in which an abnormality occurs increases acceleratively (the internal resistance value increases and the degree of increase also increases). The present embodiment focuses on such a transition tendency of the change in the internal resistance value.

[0021] In addition, in the detection of abnormal signs according to the present embodiment, a machine learning model is trained using, as learning data, data that can take into account the transition of changes in the internal resistance value, and the presence or absence of abnormal signs is determined by the machine learning model. For example, in the example of FIG. 2C, the data at times t, t-1, and t-2 are input into a (trained) machine learning model, and it is determined whether an abnormal sign (exceeding the threshold value 264) appears by time t+1. It is determined that there is no abnormal sign for the storage battery corresponding to graphs 261 and 263, and it is determined that there is an abnormal sign for the storage battery corresponding to graph 262.

[0022] FIG. 3 shows an example of the learning data (also referred to as teacher data) used in the present embodiment. This learning data is stored in a learning data DB, which will be described later, of the information processing apparatus 100. The learning data includes, for example, a storage battery identification ID, information regarding each measurement result corresponding to times t-2 to t, information regarding the usage status, and a determination.

[0023] "Determination" represents correct answer data (also referred to as a label). Based on past periodic inspections and the fact of abnormal occurrences, "NG" (indicating the occurrence of an abnormality) is assigned to the storage battery that exceeded the threshold value 264 between times t and t+1. On the other hand, "OK" (indicating that no abnormality occurred) is assigned to the storage battery that did not exceed the threshold value 264 between times t and t+1.

[0024] The battery identification ID is an ID for identifying each battery. It can not only simply identify the battery, but also, for example, make it possible to specify in which data center the battery is located, in which UPS battery system it is located, and which battery in the UPS battery system it is. In this way, when using the learning data, it is possible to execute the learning of the model using the learning data of a specific data center, or execute the learning of the model using the learning data related to the batteries of a specific UPS battery system. When different batteries (that is, batteries manufactured by different manufacturers or different types of batteries with different characteristics) are arranged for each data center or for each UPS battery, it is possible to execute the learning of the model for each battery with different properties, and a learning model suitable for each battery with different properties can be realized.

[0025] The measurement data corresponding to time t - 2 includes the internal resistance value (mΩ) of the target battery at time t - 2 and the slope of the internal resistance value between time t - 1 and time t - 2. The slope of the internal resistance value may be, for example, the difference in the internal resistance value between time t - 1 and time t - 2, or a value obtained by dividing the difference by the number of days between time t - 1 and time t - 2. Further, these values (such as the value obtained by dividing the difference by the number of days between time t - 1 and time t - 2) may be multiplied by a predetermined scaling factor.

[0026] The measurement data corresponding to time t - 1 includes the internal resistance value (mΩ) of the target battery at time t - 1 and the slope of the internal resistance value between time t and time t - 1. The slope of the internal resistance value may be, for example, the difference in the internal resistance value between time t and time t - 1, or a value obtained by dividing the difference by the number of days between time t and time t - 1. Further, these values may be multiplied by a predetermined scaling factor.

[0027] By adding the slope of the internal resistance value to each of the measurement data at time t-2 and time t-1, the transition of the degree of change in the internal resistance value can be explicitly given to the machine learning model. That is, unlike the case where only the internal resistance value is simply included in the learning data, in this embodiment, the occurrence of an abnormality in the storage battery can be predicted by explicitly taking into account the transition of the degree of change in the internal resistance value (i.e., the accelerating increase in the internal resistance value).

[0028] The measurement data corresponding to time t includes the internal resistance value (mΩ) of the target storage battery at time t. The data indicating the usage status includes information on the usage period, and the information on the usage period indicates the number of days elapsed since the day when the storage battery was first charged (the number of usage days since the charging day). It has been found that the number of days elapsed since the start of use of the storage battery is correlated with the occurrence of an abnormality.

[0029] By using such learning data, by predicting the trend of deterioration of the storage battery by utilizing the results of regular inspections performed in the past, it becomes possible to implement necessary countermeasures for the storage battery before the occurrence of an abnormality. In addition, the occurrence of an abnormality in a storage battery in which the internal resistance value changes rapidly between time t (the current inspection period) and time t+1 (the next inspection period) can be predicted by explicitly taking into account the transition of the degree of change in the internal resistance value.

[0030] In this embodiment, as an example, the case of using xgboost, which is a gradient boosting method based on the decision tree algorithm, as the machine learning model will be described. However, the machine learning model may be configured using other machine learning models. For example, it may be configured using a deep neural network. In addition to the case of using a deep neural network having an intermediate layer with a simple configuration, for example, a neural network using an RNN (recurrent neural network) such as LSTM, which can more appropriately take into account the characteristics of the temporal change of time series data, may be used. Also, a deep neural network having an attention mechanism (also referred to as an attention mechanism) that can provide an output focusing on characteristic data in time series data may be used.

[0031] The learning data may include further information, as shown in FIG. 4. It may include other measured values such as the voltage (V) of the storage battery at each of time t-2, time t-1, and time t. Further, the information regarding the usage status may include information on the connection position in addition to the information on the usage period. The information on the connection position is the connection position (order from the beginning) of the target storage battery among the storage battery groups connected in series. As described above, a plurality of storage batteries are connected in series and arranged in the storage battery group. The storage batteries are arranged two-dimensionally or three-dimensionally, for example, in a rack or a box. Therefore, when the storage battery discharges during a power outage or a periodic inspection, the deterioration of the storage battery is affected by the temperature change according to the arrangement in the box. However, the timing of measuring the storage battery in the periodic health check may not be during the discharge of the storage battery, as shown in FIG. 8, for example. For example, it is assumed that the discharge of the storage battery occurs during a very limited time in the inspection work for operation confirmation or during the time when power supply is required due to a power outage. Therefore, the timing of measurement and recording in the periodic inspection does not always enable measurement of a temperature that affects the storage battery. Therefore, in the present embodiment, the connection position of the storage battery (or the arrangement of the storage batteries arranged two-dimensionally or three-dimensionally) is used as an input parameter of the learning model. Alternatively, instead of or in addition to the connection position of the storage battery, the discharge time may be used. The discharge time is the cumulative time during which the storage battery has discharged. Thereby, the influence received by the operation of the storage battery can be reflected in the detection of the deterioration prediction.

[0032] In the present embodiment, a case will be described as an example in which the communication device 210 transmits information regarding the measurement result, information regarding the usage status, and correct data to the information processing device 100, and the information processing device 100 generates learning data. However, the communication device 210 may generate the learning data according to the present embodiment and transmit the generated learning data to the information processing device 100. In this case, the information processing device 100 may store the acquired learning data in the learning data DB522.

[0033] (Configuration of Information Processing Device) Next, with reference to FIG. 5, a functional configuration example of the information processing apparatus 100 will be described. Each of the functional blocks described with reference to the following figures may be integrated or separated, and the functions to be described may be realized by other blocks. Also, what is described as hardware may be realized by software, and vice versa.

[0034] The communication unit 501 includes a communication circuit or communication module that communicates with the communication device 210 used by the maintenance person in charge of the storage battery via a network. The communication unit 501 communicates with a plurality of communication devices 210.

[0035] The control unit 502 includes a CPU 510 which is a central processing unit and a RAM 511. The CPU 510 may be composed of one or more CPUs. By executing the computer program stored in the recording unit 504, the control unit 502 executes the abnormal sign detection process of the storage battery described later and controls the operations of each part of the information processing apparatus 100. The control unit 502 may further include an arithmetic unit (for example, GPU) or dedicated hardware for executing statistical processing such as machine learning at a higher speed.

[0036] The RAM 511 is a volatile storage medium such as a DRAM, and temporarily stores parameters, processing results, etc. for the control unit 502 to execute a computer program.

[0037] The power supply unit 503 is a circuit or module for providing power for each part of the information processing apparatus 100 to operate. The power supply unit 503 may be further configured to include a battery.

[0038] The recording unit 504 includes a non-volatile recording medium such as a hard disk or a semiconductor memory, and records setting values, calculation results, etc. necessary for the operation of the information processing apparatus 100. Also, the recording unit 504 stores information regarding the measurement results received from the communication device 210 and information regarding the usage status in the measurement data DB 521. Further, the recording unit 504 stores the above-described learning data in the learning data DB 522.

[0039] The battery data receiving unit 512 receives information regarding measurement results and usage status information input by the maintenance person in charge of the battery from the communication device 210 and records them in the measurement data DB 521 of the recording unit 504. The information regarding measurement results and usage status information stored in the measurement data DB 521 is configured as shown in, for example, FIG. 7A.

[0040] The data shown in FIG. 7A includes a battery identification ID 701, information regarding measurement results for each time and usage status information, and usage status information 704 that is invariant with respect to time. The information regarding measurement results for each time and usage status information includes, for example, information regarding measurement results and usage status information 702 measured at time t-n, information regarding measurement results and usage status information measured at time t-n-1, ··· information regarding measurement results and usage status information 703 measured at time t.

[0041] The battery identification ID 701 is the same as that described above in FIGS. 3 and 4. The internal resistance value and voltage value in the information regarding measurement results for each time are the same as the internal resistance value and voltage value described above in FIG. 4. The date is the date when data such as the internal resistance value was measured. The usage period and discharge time in the usage status information represent the cumulative number of days up to the time point at the corresponding time (for example, t-n). The usage status information 704 includes connection position information, and this connection position information is the same as the connection position information shown in FIG. 4.

[0042] Referring to FIG. 5 again, the learning data generation unit 513 generates the above-described learning data using data of a specific storage battery (e.g., specified by a maintenance person in charge of the storage battery or an administrator of the abnormality prediction detection system 10) among the data stored in the measurement data DB 521. For example, the learning data generation unit 513 calculates the slope of the internal resistance value at time t - 1 from the date and the internal resistance value at time t - 1 and the date and the internal resistance value at time t - 1 in the measurement data DB 521. Also, based on the information on the usage status at time t and the information 704 on the usage status that does not change over time, the information on the usage status shown in FIG. 3 or FIG. 4 is generated. The specific storage battery is a storage battery in which an abnormality has actually occurred or has been confirmed to be operating normally. According to the inventor's experiment, the learning data can detect an abnormality prediction well even when it is composed of, for example, data of three regular inspections (times t - 2, t - 1, t). However, it is not limited to this example, and for example, data of more regular inspections such as five times may be used.

[0043] The learning data generation unit 513 adds the determination information of the learning data (i.e., correct answer data indicating "OK" or "NG") to the data of the storage battery identification ID of the specific storage battery. The learning data generation unit 513 receives, for example, the correct answer data from the maintenance person in charge of the storage battery or the administrator of the storage battery abnormality prediction detection system 10 via the communication device 210 and adds it to the learning data. The learning data generation unit 513 stores the generated learning data in the learning data DB 522.

[0044] The measurement data acquisition unit 514 generates information regarding measurement data and usage status for a specific storage battery from the measurement data DB 521 in order to execute the abnormal sign detection process in the estimation stage described later. The storage battery targeted by the measurement data acquisition unit 514 is the storage battery for which the presence or absence of an abnormal sign is to be determined. The measurement data acquisition unit 514 performs the same process as the process of generating data other than "judgment" among the processes described for the learning data generation unit 513, and generates a data set to be input to the learning model for abnormal sign detection. As shown in FIG. 7B, the data set generated by the measurement data acquisition unit 514 includes the internal resistance value and the slope of the internal resistance value at each of time t-2 and time t-1, the internal resistance value at time t, and information on the usage period. Note that the example of the data set shown in FIG. 7B is in a format corresponding to the learning data shown in FIG. 3. Also, when generating a data set in a format corresponding to the learning data shown in FIG. 4, the generated data set includes the internal resistance value, the slope of the internal resistance value, and the voltage at each of time t-2 and time t-1, the internal resistance value and the voltage at time t, and information on the usage period, connection position, and discharge time.

[0045] The preprocessing unit 515 performs common preprocessing on the training data and the data shown in FIG. 7B as a pre-step of inputting the data into the learning model. This preprocessing reduces the influence of the variation in the characteristics of the storage battery on the learning and estimation results of the learning model when the storage battery is manufactured by multiple manufacturers and models and the characteristics of the storage battery vary depending on the manufacturer and model. The preprocessing may include, for example, a process of normalizing each parameter of information regarding measurement results and information regarding usage conditions. The normalization may scale each parameter so that each parameter falls within the range of 0 to 1 (using, for example, the maximum value for each manufacturer or model). Alternatively, the preprocessing unit 515 may use, for example, the value obtained by dividing the measurement results such as the voltage value and the internal resistance value by the device specification information. For example, by dividing the voltage value of the measurement result by the standard voltage that varies for each storage battery product, the ratio to the standard voltage is calculated, and the calculated ratio to the standard voltage is input into the learning model. The standard voltage varies depending on the difference in the number of series cells in the product. Also, the preprocessing unit 515 may divide the internal resistance value by the standard internal resistance value that varies for each product to calculate the ratio to the standard internal resistance, and use this ratio. The standard internal resistance value varies depending on the difference due to the number of series cells and the terminal structure in the product. Also, the usage period of the storage battery may be divided by the average life that varies for each product to calculate the ratio to the average life, and this ratio may be used. Note that the information such as the maximum value used for normalization, the standard voltage for each manufacturer and model, the standard internal resistance value, and the average life may be recorded in advance in the recording unit 504.

[0046] The model processing unit 516 inputs the learning data shown in FIG. 3 or FIG. 4 to train the internal machine learning model. In addition, using the trained machine learning model, it inputs the input data shown in FIG. 7B and executes the abnormal sign detection process for the storage battery. The machine learning model used by the model processing unit 516 in this embodiment functions as a classifier that outputs the prediction result in two values (OK or NG). As described above, in this embodiment, as the learning model, for example, as a machine learning model, xgboost, which is a gradient boosting method based on the decision tree algorithm, is used. However, the machine learning model may be configured using, for example, a deep neural network. The abnormal information providing unit 517 transmits the result of the abnormal sign detection process in the estimation stage by the model processing unit 516 to the communication device 210. The abnormal information providing unit 517 transmits the result of the abnormal sign detection process to the communication device 210 as information on the abnormal sign notification screen, which will be described later with reference to FIG. 10, for example.

[0047] (Configuration of Communication Device) Next, with reference to FIG. 6, a functional configuration example of the communication device 210 will be described. In this embodiment, as an example of the communication device, the case of using a personal computer will be described as an example, but the communication device may be other electronic devices such as a smartphone or a tablet terminal. Each of the functional blocks described with reference to the following figures may be integrated or separated, and the functions to be described may be realized by other blocks. Also, what is described as hardware may be realized by software, and vice versa.

[0048] The communication unit 601 includes a communication circuit and the like, and connects to the Internet via wireless communication such as a wireless LAN, or connects to a network via wireless communication for mobile phones, etc., to communicate with the information processing device 100.

[0049] The control unit 602 includes a CPU 610 and a RAM 611, and controls the operations of each part in the communication device 210 by the CPU 610 executing a computer program recorded in the recording unit 607, for example.

[0050] The operation unit 603 includes at least one of a keyboard, a mouse, or a touch panel, and can perform operations on the GUIs for various operations displayed on the display unit 606 (for example, input of information regarding measurement results and usage status, and assignment of correct data). The power supply unit 604 supplies power to each part of the communication device 210. The imaging device 305 is, for example, a camera mechanism including an imaging element, and performs imaging according to an instruction from the control unit 602.

[0051] The display unit 606 includes a display device such as an LCD or an OLED. The display unit 606 displays, according to an instruction from the control unit 602, a GUI for input of information regarding measurement results and usage status and assignment of correct data, a GUI of an abnormal sign notification screen, and the like.

[0052] The recording unit 607 includes a non-volatile memory such as a semiconductor memory, and holds programs and set values executed by the control unit 602. The computer programs held in the recording unit 607 include an operating system and various applications for realizing the various functions of the communication device 210.

[0053] (A series of operations of the abnormal sign detection process in the learning stage) Next, a series of operations of the abnormal sign detection process in the learning stage, which is executed in the information processing apparatus 100, will be described with reference to FIG. 9A. Further, this process is realized by the CPU 510 of the control unit 502 executing a computer program recorded in the recording unit 504. In the following description, for the sake of simplicity of explanation, the processing entity of each step is collectively described as the control unit 502, but each part functioning in the control unit 502 executes corresponding processing according to the processing content. Note that this process is a process of training a machine learning model in the model processing unit 516, and the training of the machine learning model is not yet completed.

[0054] In S401, the control unit 502 receives information regarding the measurement result and information regarding the usage status from the communication device 210. As described above, the control unit 502 stores the information regarding the measurement result and the information regarding the usage status in the measurement data DB521 in the format shown in FIG. 7A.

[0055] In S402, the control unit 502 acquires information necessary for generating learning data from the communication device 210. The information necessary for generating learning data is the correct answer data shown in FIG. 3 or FIG. 4.

[0056] In S404, the control unit 502 generates the learning data shown in FIG. 3 or FIG. 4 based on the data shown in FIG. 7A (information regarding the measurement result and information regarding the usage status) stored in the measurement data DB521 and the correct answer data acquired in S402. The generation of the learning data is executed as described above for the learning data generation unit 513.

[0057] In S405, the control unit 502 performs learning on the learning model. The control unit 502 causes the machine learning model to learn using the generated learning data. Since known learning methods can be used for the learning methods of decision trees and deep neural networks for the learning of the machine learning model, detailed description thereof is omitted. When the control unit 502 finishes the learning of the machine learning model using the learning data, the control unit 502 ends the series of operations of the abnormal sign detection process in the learning stage.

[0058] (Series of operations of the abnormal sign detection process in the estimation stage) Next, a series of operations of the abnormal sign detection process in the estimation stage, which is executed in the information processing apparatus 100, will be described with reference to FIG. 9B. Further, this process is realized by the CPU 510 of the control unit 502 executing a computer program recorded in the recording unit 504. In the following description, for the sake of simplicity, the processing entity of each step will be collectively described as the control unit 502. However, each part functioning within the control unit 502 executes the corresponding process according to the processing content. Note that this process is a process of determining the presence or absence of an abnormal sign from the data of the newly input battery using the learned machine learning model in the model processing unit 516. In the process shown in FIG. 9B, the case of performing the process on one specific battery is described as an example. However, the processes of S921 and S922 may be repeated to perform abnormal sign detection on a plurality of batteries.

[0059] In S921, the control unit 502 generates the data shown in FIG. 7B from the data shown in FIG. 7A stored in the measurement data DB 521 for a specific battery that is the target of the abnormal sign detection process. Note that after generating the data shown in FIG. 7B, the control unit 502 may apply the preprocessing by the above-described preprocessing unit 515 to the generated data.

[0060] In S922, the control unit 502 executes the abnormal sign detection process. The control unit 502 determines the presence or absence of an abnormal sign for a specific battery to be processed using the learned machine learning model. The control unit 502 inputs the data generated in S921 into the learned machine learning model and obtains the output (that is, a binary value of OK or NG) from the machine learning model.

[0061] In S923, the control unit 502 transmits information on the battery for which an abnormal sign has been detected (for example, the battery identification ID and the transition of the degree of change in the internal resistance value) to the communication device based on the determination result obtained from the machine learning model. When the control unit 502 transmits the battery information to the communication device 210, this series of processes ends.

[0062] Note that the processes of S921 to S922 may be repeatedly executed for a plurality of storage batteries, and information on the abnormal signs of the plurality of storage batteries as shown in FIG. 10 (also referred to as an abnormal sign notification screen) may be transmitted to the communication device 210. On the abnormal sign notification screen 1000 shown in FIG. 10, the storage battery identification ID of one or more storage batteries in which abnormal signs are detected and the information 1002 of the storage battery corresponding to the storage battery identification ID are displayed. The information 1002 of the storage battery may be configured to display at least any one of the characteristics of the storage battery indicating an abnormal sign, for example, the transition 1003 of the slope of the internal resistance value, the usage period 1004, and the connection position 1005. In this way, it is possible to grasp at once the storage batteries estimated to have abnormal signs at the current time (time t), and to overview the characteristics of each storage battery, and it is possible to easily confirm the validity of the estimation result.

[0063] As described above, in the present embodiment, a dataset for a specific storage battery including information on measurement results and information on usage status for the specific storage battery recorded at each regular inspection performed at a predetermined time interval is acquired and input into a learned machine learning model, thereby determining the presence or absence of an abnormal sign of the specific storage battery. At this time, the information on the measurement results for the specific storage battery includes the degree of change in the internal resistance value of the specific storage battery at each predetermined time interval, and the transition of the degree of change in the internal provided value is explicitly given to the machine learning model. By doing so, it becomes possible to accurately predict the abnormal signs of the storage battery using the data of the storage battery measured during the regular inspection of the storage battery system. According to the present embodiment, even with a small amount of data for about three regular inspections, it is possible to accurately predict the abnormal signs of the storage battery by giving the transition of the degree of change in the internal provided value, so that the configuration in the storage battery system can be made simple (without requiring a sensor for constantly monitoring the characteristics of the storage battery).

[0064] In the above-described embodiment, as information regarding the measurement results of the storage battery, the internal resistance value, the slope of the internal resistance value, and the voltage were given as examples. However, the information regarding the measurement results may further include other information. For example, it may include information such as at least any one of the temperature of the storage battery obtained by regular inspection, the number of discharge cycles, the discharge time, or the discharge capacity.

[0065] Also, in the above-described embodiment, as the abnormal sign detection process, the presence or absence of an abnormal sign is determined for each individual storage battery. However, in addition to the above-described determination of the abnormal sign for each individual storage battery, the model processing unit 516 may use a machine learning algorithm such as clustering or SVM (Support Vector Machine) to determine a storage battery that deviates from the class of storage batteries determined to be normal among a plurality of storage batteries as a storage battery indicating an abnormal sign.

[0066] For example, as shown in FIG. 11, when considering a space having information regarding the measurement results of each storage battery and information regarding the usage status as components of a vector, the data of a plurality of storage batteries are distributed at various positions within the space (for example, 1103 and 1104). By applying a machine learning algorithm such as clustering or SVM to the data of these storage batteries, the storage battery 1103 included in the normal storage battery class 1102 and the other storage battery 1104 are distinguished. Then, the distinguished storage battery 1104 is determined as a storage battery having an abnormal sign.

[0067] The control unit 502 may finally detect the battery as a battery having an abnormal sign when the battery detected with an abnormal sign by the process of S922 and the battery 1104 that does not belong to the class of normal batteries match by executing a plurality of machine learning algorithms by the model processing unit 516. In this way, maintenance can be performed on the battery with a high probability of having an abnormal sign determined by a plurality of machine learning algorithms. On the other hand, if it is determined that there is an abnormal sign in any one of the machine learning algorithms by executing a plurality of machine learning algorithms by the model processing unit 516, the control unit 502 may finally detect the battery as a battery having an abnormal sign. In this case, maintenance can be widely performed on the batteries that may have an abnormal sign.

[0068] The invention is not limited to the above embodiments, and various modifications and changes are possible within the scope of the gist of the invention.

Explanation of Signs

[0069] 100… Information processing device, 210… Communication device, 513… Learning data generation unit, 514… Measurement data acquisition unit, 515… Preprocessing unit, 516… Model processing unit

Claims

1. A program that causes a computer to function as each means of an information processing device for detecting an abnormal sign related to a plurality of storage batteries constituting an uninterruptible power supply device, wherein the information processing device includes: an acquisition means for acquiring a data set for a specific storage battery, including information regarding measurement results for the specific storage battery recorded for each regular inspection performed at a predetermined time interval and information regarding the usage status of the specific storage battery; a determination means for determining the presence or absence of an abnormal sign of the specific storage battery by inputting the data set for the specific storage battery into a pre-learned first learning model, and the information regarding the measurement results for the specific storage battery includes the degree of change in the internal resistance value of the specific storage battery for each of the predetermined time intervals, and the information regarding the usage status of the specific storage battery includes the number of days of use since the charging date. A program characterized by this.

2. The program according to claim 1, wherein the information regarding the usage status of the specific storage battery further includes the connection position of the specific storage battery among the plurality of storage batteries connected in series.

3. The program according to claim 1 or 2, wherein the information regarding the usage status of the specific storage battery further includes the arrangement of the specific storage battery among the plurality of storage batteries arranged in a two-dimensional or three-dimensional manner.

4. The program according to any one of claims 1 to 3, wherein the information regarding the measurement results for the specific storage battery further includes at least any one of the number of discharges, the discharge time, or the discharge capacity.

5. The determination means is further capable of determining the presence or absence of an abnormal sign of the plurality of storage batteries by inputting the data set for the plurality of storage batteries into a pre-learned second learning model that determines a storage battery with a different distribution from the distribution of the information of the data sets of the other storage batteries among the plurality of storage batteries as a storage battery with an abnormal sign. The program according to any one of claims 1 to 4.

6. The program according to claim 5, further comprising a detection means for detecting at least any one of the storage batteries determined to have an abnormal sign by the first learning model and the storage batteries determined to be storage batteries with an abnormal sign by the second learning model as storage batteries with an abnormal sign.

7. The information regarding the measurement result is based on the measurement result measured when the plurality of storage batteries are not discharging, The program according to any one of claims 1 to 6, characterized in that.

8. The plurality of storage batteries include storage batteries from different manufacturers, The information regarding the measurement result is characterized in that numerical values normalized in advance for each storage battery from a different manufacturer are used, The program according to any one of claims 1 to 7.

9. An information processing method executed by an information processing device for detecting an abnormal sign related to a plurality of storage batteries constituting an uninterruptible power supply device, An acquisition step of acquiring a data set for a specific storage battery including information regarding a measurement result for the specific storage battery and information regarding the usage status of the specific storage battery, which are recorded for each periodic inspection performed at a predetermined time interval; A determination step of determining the presence or absence of an abnormal sign of the specific storage battery by inputting the data set for the specific storage battery into a first pre-trained learning model, including. The information regarding the measurement result for the specific storage battery includes the degree of change in the internal resistance value of the specific storage battery for each of the predetermined time intervals, and the information regarding the usage status of the specific storage battery includes the number of days of use since the charging date, An information processing method characterized by that.

10. A data structure of data for a specific storage battery used in an information processing device for detecting an abnormal sign related to a plurality of storage batteries constituting an uninterruptible power supply device, Information regarding the measurement result for the specific storage battery, which is recorded for each periodic inspection performed at a predetermined time interval, and Information regarding the usage status of the specific storage battery, including, The data for the specific storage battery is used in the information processing device in a process of inputting into a first pre-trained learning model to determine the presence or absence of an abnormal sign of the specific storage battery, The information regarding the measurement result for the specific storage battery includes the degree of change in the internal resistance value of the specific storage battery for each of the predetermined time intervals, and the information regarding the usage status of the specific storage battery includes the number of days of use since the charging date, A data structure characterized by that.

Images