Method and system for detecting battery anomalies using an artificial neural network battery model based on field data

Abstract

A method and system for detecting abnormal battery behavior using field data, according to the present invention, provide a method and system for sequentially extracting real-time field data from a functioning battery, and comparing the predicted values ​​obtained by predicting the cell voltage and temperature from the extracted field data with the cell voltage and temperature values ​​from the real-time field data, thereby detecting the presence or absence of abnormal battery behavior based on a deviation thereof.

Description

Method and system for detecting battery anomalies using an artificial neural network battery model based on field data Technology sector The present invention relates to a method and system for detecting battery malfunctions using a battery model. Specifically, the present invention relates to a method and system for detecting battery malfunctions using an artificial intelligence neural network battery model that utilizes field data instead of a conventional chemical / electrical equivalent circuit model to construct the battery model. Background of the invention Traditionally, general lithium-ion battery modeling was implemented using chemical composition or the equivalent electrical circuit. In the case of this battery modeling, the modeling is primarily based on cell test data, including charge / discharge data. However, internal cell test data differs from real-world field environment data. Furthermore, even if an internal cell test is performed under various conditions, it may not be possible to obtain sufficient data due to limitations inherent in realistic experiments, such as the number of cell samples, the module / frame unit, the degree of deterioration, temperature conditions, test equipment, and time. With reference to the above, patent document 1 proposes a system for receiving measured initial characteristic data from a battery, training an artificial intelligence neural network, predicting long-term characteristic data from them and determining its reliability, and patent document 2 discloses a method for estimating the battery state in which information about the physical quantity of a battery is fed into a battery training model and estimation information is obtained from the battery training model. It should be noted that these conventional technologies are still insufficient to implement a battery status estimation method based on real field data and to determine whether the battery is malfunctioning or not. US patent 2020 / 355749 discloses a secondary battery control system that performs anomaly detection while predicting various parameters. US 2017 / 184680 discloses a sensor management system that includes a data collector configured to collect various types of data from a plurality of sensors and an estimator configured to estimate other types of data based on two or more of the collected data types. US document 6285163 discloses means for estimating the state of charge of a battery, capable of accurately estimating a SOC even if the battery repeats the charge / discharge in short cycles. The above related technique includes the following documents. Patent Document 1: Korean Patent Publication Open to Public Inspection No. 10-2009-0020448 Patent Document 2: Korean Patent Publication Open to Public Inspection No. 10-2018-0057226 Explanation of the Invention TECHNICAL PROBLEM To address the problems described above, the present invention provides a method and system for detecting battery cell malfunctions using field data. However, the use of field data presents a challenge, as there are inherent limitations in the types of data collection devices available, resulting in low accuracy. To solve this problem, the purpose is to provide a method and system for detecting battery malfunctions in which battery modeling is implemented using AI training to model a battery using different field data and that uses battery modeling. TECHNICAL SOLUTION To solve the problems described above, according to the present invention, a battery malfunction behavior detection system is provided as defined in the appended claims. According to the present invention, a method for detecting battery malfunction behavior is also provided. ADVANTAGEOUS EFFECTS According to the present invention, it is possible to predict the temperature value and the battery voltage value in real time using battery field data and compare the predicted battery temperature value and voltage value with the actual temperature value and voltage value to detect in real time whether the battery is malfunctioning or not. Brief description of the drawings The following drawings attached to this descriptive memorandum are intended to illustrate a preferred embodiment of the present invention and serve to better understand the technical idea of ​​the present invention together with the detailed description of the invention described above, so the present invention should not be interpreted as being limited only to the matters described in the drawings. Figure 1 is a diagram illustrating an example of the respective field data measured and calculated in the present invention. Figure 2 is a diagram illustrating an example of extracting training data by preprocessing field data acquired during a given section or for a predetermined time, from the field data in Figure 1. Figure 3 is a diagram illustrating a sequence of a battery malfunction behavior detection method according to the present invention. Figure 4 illustrates a graph showing the field data values ​​of cell temperature and the predicted values ​​of cell temperature over time. Figure 5 is a block diagram of a battery malfunction behavior detection system according to the present invention. Preferred embodiment of the invention The present invention trains the prediction model using field data to predict voltage and temperature at a later time. The types of field data used in the present invention are as follows. Field data In the present invention, "field data" refers to the following data measured in real time from a battery in operation and the following data calculated from them. Field data are time-series data of information about the battery's state, calculated during battery operation at predetermined time intervals. The types of field data used in the present invention are as follows. (1) Chassis data Rack data refers to data from a battery rack. Rack data includes rack current, ambient temperature, and fan on / off information. This data is typically measured and calculated by a rack battery management system (BMS). Rack current can be expressed as Rack.Current, ambient temperature as Ambient.Temp, and fan on / off information as the Fan_ON / OFF indicator value, and are measured at predetermined time intervals. The rack current, ambient temperature (Ambient Temp.), and Fan.ON values ​​of the fan on / off information at time t, measured in a predetermined cycle, can be expressed as Rack.Current(t), Ambient.Temp(t), and Fan.ON(t). In this case, the Rack.Current(t) and Ambient.Temp(t) values ​​are measured from the appropriate sensors, and Fan.ON(t) can have a value of "1" when the fan is running and "0" when the fan is not running. For example, when measured at 1-second intervals, the rack current, ambient temperature, and fan on / off information (if the fan is running) at 3 seconds after actuation can be expressed respectively as Rack.Current(3), Ambient.Temp(3), and Fan.ON(3). (2) Module data Module data refers to the data for each of a plurality of battery modules included in the battery rack. Module data includes a State of Charge (SOC) value for each module. When there are i modules in the battery rack, each module is denoted as Mk (k = 1, 2, 3, ...i), and the SOC value of each module (SOC M) is expressed as SOC.Mk. For example, the SOC value of the second module is expressed as SOC.M2. Module data is typically measured and calculated using a module-level battery management system (BMS). Similar to rack data, the SOC.Mk value can be measured at predetermined time intervals, and the SOC.Mk value at time t is indicated as SOC.Mk(t). (3) Cell data Cell data refers to the data for each individual cell within a battery module. This data includes a cell voltage value, Cell V, representing the voltage of each cell. The cell voltage value is denoted as Cell.V.MkCj (j = 1, 2, 3, ...m), where each module comprises m cells. For example, the voltage value of the third cell in the second module is denoted as Cell_V_M2C3. Cell data is typically measured and calculated by a cell-level battery management system (BMS) or a module-level battery management system (BMS) when a cell-level BMS is unavailable. The Cell.V.MkCj value can also be measured at predetermined time intervals, and the Cell.V.MkCj value at time t is indicated as Cell.V.MkCj (t) . Cell data may include cell temperature. The j-th cell temperature of the k-th module at time t is denoted as Cell.T.MkCj (t) . (4) Calculated value The present invention also uses an SOH value, SOH MkCj, as one of the calculated values. The SOH value of each cell is calculated by the cell battery management system (BMS), the module battery management system (BMS), or the rack battery management system (BMS), and the SOH of a specific cell is indicated as SOH.MkCj (k = 1, 2, 3, ...m). The value of SOH.MkCj can also be measured at predetermined time intervals, and the value of SOH.MkCj at time t is indicated as SOH.MkCj (t) . 2. Method for detecting battery malfunction behavior The method for detecting battery malfunction behavior according to the present invention will be described with reference to Figure 3. 2-1. Calculation and preprocessing of field data (S100) This process is a procedure for measuring, calculating, and preprocessing the field data described above. (1) Field Data Calculation Process (S110) The field data calculation process is a process of measuring and calculating real-time battery status information data from a functioning battery. The measurement and calculation of field data are performed in real time from the operating battery in a predetermined time unit, and Figure 1 illustrates an example of each field data measured and calculated in the present invention. Field data measurement and calculation can be performed on each cell battery management system (BMS), module battery management system (BMS), or rack battery management system (BMS), and the field data is transmitted to the 100 field data calculation unit. The field data calculation unit can be configured to be integrated into the rack BMS. Among the field data shown in Figure 1 and described above, the field data obtained in a certain section or during a predetermined time are preprocessed and extracted as training data, as shown in Figure 2. (2) Field data preprocessing process for cell voltage prediction (S120) The field data preprocessing process for cell voltage prediction is a field data preprocessing process for cell voltage prediction that extracts data for cell voltage prediction from the calculated field data. In the present invention, a cell voltage is predicted using a cell voltage prediction model that predicts the cell voltage from field data. As shown in Figure 2, the rack current, ambient temperature, SOC, SOH, and cell voltage from the previous cycle are used as input values ​​for the cell voltage prediction. To this end, in the preprocessing process, the Rack.Current(t), Ambient.Temp(t), Fan.ON(t), SOC.Mk(t), SOH.MkCj(t), and Cell.V.MkCj(t) values, which are the rack current, ambient temperature, SOC, SOH, and cell voltage values, are periodically extracted from the field data calculated in real time on a predetermined cycle. Figure 2 illustrates a data format for extracting the Rack.Current(t), Ambient.Temp(t), Fan.ON(t), and SOC.Mk(t) values ​​for each time cycle, and the corresponding SOH_M1C2(t) and Cell_V_M1C2(t) values ​​from cells 1 and 2 of module 1. (3) Field data preprocessing process for cell temperature prediction (S130) Preprocessing field data for cell temperature prediction is a data extraction process to predict cell temperature from calculated field data. In the present invention, a cell temperature is predicted using a cell temperature prediction model that predicts the cell temperature from field data. As data for the cell temperature prediction, the cell temperature of the next cycle is predicted using Cell.T.MkCj(t), which is the current cell temperature value, the ambient temperature Ambient.Temp(t), and Fan.ON(t). To do this, in the preprocessing process, the Cell.T.MkCj (t) , Ambient.Temp (t) and Fan.ON (t) values ​​are periodically extracted from the field data calculated in real time on a predetermined cycle. 2-2. Prediction model generation process (S200) (1) Cell voltage prediction model generation process (S210) A cell voltage prediction model is generated for cell voltage prediction by performing supervised learning by inputting Rack.Current(t), Ambient.Temp(t), Fan.ON(t), SOC.Mk(t), SOH.MkCj(t), and Cell.V.MkCj(t) into a predefined artificial intelligence neural network and using its output value as Cell.V.MkCj(t+1). The cell voltage prediction model is generated using a predefined artificial intelligence neural network. An artificial intelligence neural network, known as an AI neural network, can be used, and the AI ​​neural network that has been trained with predetermined training data is stored as the cell 210 voltage prediction model in Figure 5. As training data for the cell 210 voltage prediction model, the values ​​Rack.Current(t), Ambient.Temp(t), Fan.ON(t), SOC.Mk(t), SOH.MkCj(t), Cell.V.MkCj(t) for a predetermined time section preprocessed through the field data preprocessing process (S120) for cell voltage prediction are used as input values ​​for the neural network and Cell.V.MkCj(t+1), which is the cell voltage measurement value of the next cycle, is used as the output value of the same, and the cell voltage prediction model is generated by performing supervised learning on the artificial intelligence neural network so that when the values ​​Rack.Current(t), Ambient.Temp(t), Fan.ON(t), SOC.Mk(t), SOH.MkCj(t), Cell.V.MkCj(t) for a predetermined time section are entered into the artificial intelligence neural network, the output value becomes Cell.V.MkCj(t+1), which is the cell voltage measurement value of the next cycle. The AI ​​neural network trained in this way is stored as the cell voltage prediction model 210, which receives the values ​​Rack.Current(t), Ambient.Temp(t), Fan.ON(t), SOC.Mk(t), SOH.MkCj(t), Cell.V.MkCj(t), and calculates Cell.Vpredic.MkCj(t+1), which is the cell voltage prediction value for the next cycle. The values ​​Rack.Current(t), Ambient. Temp(t), Fan.ON(t), SOC.Mk(t), SOH.MkCj(t), Cell.V.MkCj(t), Cell.V.MkCj(t+1), which are the training data used in the cell voltage prediction model generation process, may be field data measured at a different time than the data used in a cell voltage prediction process (S310) below, and may be sample data obtained before battery operation or by operating sample batteries. In another embodiment, the cell voltage prediction model can be generated using Rack.Current(t), Ambient.Temp(t), Fan.ON(t), SOC.Mk(t), SOH.MkCj(t), Cell.V.MkCj(t), Cell.V.MkCj(t+1) for a predetermined time section as training data for the 210 cell voltage prediction model, not as field data. In other words, the cell voltage prediction model is generated by training it with values ​​corresponding to the cell voltage prediction data calculated from a standard battery, rather than from a battery in operation for which field data is calculated as training data. The cell voltage prediction model then receives data to predict the cell voltage of the battery in operation and outputs a cell voltage prediction value for the next cycle. (2) Cell temperature prediction model generation process (S220) The present invention also includes a cell temperature prediction model. The cell temperature prediction model 220 for cell temperature prediction is generated in the same way as the cell voltage prediction model generation process, using preprocessed data as training data. Specifically, supervised learning is performed by periodically inputting the values ​​Cell.T.MkCj(t), Ambient.Temp(t), and Fan.ON(t) into a predetermined artificial intelligence neural network and using the network's output value as Cell.T.MkCj(t+1). The cell temperature prediction model is generated by performing supervised learning on the artificial intelligence neural network, such that the values ​​Cell.T.MkCj(t), Ambient.Temp(t), and Fan.ON(t) are used.ON(t) is periodically introduced for a predetermined time as input data to the artificial intelligence neural network and the output value becomes Cell.T.MkCj(t+1), which is the cell temperature measurement value of the next cycle. The artificial intelligence neural network trained in this way is stored as the cell temperature prediction model 220, which receives the values ​​Cell.T.MkCj (t) , Ambient.Temp (t) , Fan.ON (t) , and calculates Cell.Tpredic.MkCj (t+1) , which is the cell temperature prediction value for the next cycle. The values ​​Cell.T.MkCj (t) , Ambient.Temp (t) , Fan.ON (t) , Cell.T.MkCj (t+1) , which are the training data used in the cell temperature prediction model generation process, may be field data measured at a different time than the data used in the cell temperature prediction process (S320) below, and may be sample data obtained before battery operation or by operating sample batteries. In another embodiment, the cell temperature prediction model can be generated using the Cell.T.MkCj(t), Ambient.Temp(t), Fan.ON(t), Cell.T.MkCj(t+1) values ​​for a predetermined time section calculated using a standard battery in a laboratory as training data, not as field data. That is, the cell temperature prediction model 220 is generated by training it with values ​​corresponding to the data to predict the cell temperature calculated from the standard battery instead of a working battery for which field data are calculated as training data, and it receives data to predict the cell temperature of the working battery and outputs a cell temperature prediction value for the next cycle. (3) Prediction model update process (S230) Another realization is a process of retraining and updating the cell voltage prediction model and the cell temperature prediction model by adding field data for a predetermined period of a section of normal operating time of the battery in operation, for which the field data are calculated, as new training data. In the process of generating the cell voltage prediction model, the 210 cell voltage prediction model can be updated by inputting the values ​​Rack.Current(t), Ambient.Temp(t), Fan.ON(t), SOC.Mk(t), SOH.MkCj(t), Cell.V.MkCj(t), and Cell.V.MkCj(t+1) as new training data over a predetermined time calculated while the battery is not exhibiting malfunctioning behavior. In this case, since the training model is updated using real field data, it is possible to obtain a more accurate prediction value for the actual battery deployed in the field. In the process of generating the cell temperature prediction model, the cell temperature prediction model 220 can be updated by inputting the values ​​Cell.T.MkCj(t), Ambient.Temp(t), Fan.ON(t), and Cell.T.MkCj(t+1) for a predetermined time calculated while the battery is not exhibiting malfunctioning behavior, as new learning data. In this case, since the training model is updated using real field data (S230), it is possible to obtain a more accurate prediction value for the actual battery deployed in the field. 2-3. Real-time prediction process (S300) The real-time prediction process is a procedure for predicting cell voltage and cell temperature using the cell voltage prediction model and the cell temperature prediction model for which training has been completed. (1) Cell voltage prediction process (S310) The cell voltage prediction process is a procedure for inputting the preprocessed and extracted values ​​Rack.Current(t), Ambient.Temp(t), Fan.ON(t), SOC.Mk(t), SOH.MkCj(t), Cell.V.MkCj(t) from the field data preprocessing process (S120) for cell voltage prediction into the generated or updated cell voltage prediction model 210 and calculating Cell.Vpredic.MkCj(t+1), which is a cell voltage prediction value for the next cycle. (2) Cell temperature prediction process (S320) In the cell temperature prediction process, the Cell.T.MkCj (t) , Ambient.Temp (t) , Fan.ON (t) values ​​extracted in the preprocessing process (S220) are introduced into the generated or updated cell temperature prediction model 220 and Cell.Tpredic.MkCj (t+1) is calculated, which is the cell temperature prediction value for the next cycle. 2-4. Battery Malfunction Behavior Detection Process (S400) (1) Cell voltage malfunction behavior detection process (S410) This cell voltage malfunction behavior detection process is a case detection process where the Cell.Vpredic.MKCj (t+1) value, which is a prediction value generated by the cell voltage prediction model 210 through the cell voltage prediction process (S310), is compared to the Cell.V.MkCj (t+1) value from field data calculated from the battery, and a difference greater than or equal to a predetermined reference value occurs. When a difference greater than or equal to the predetermined reference value occurs, a cell voltage malfunction behavior detection signal is generated. (2) Cell temperature malfunction behavior detection process (S420) This cell temperature malfunction detection process detects a case where the Cell.Tpredic.MkCj(t+1) value, a prediction value generated by the Cell Temperature Prediction Model 220 through the Cell Temperature Prediction Process (S320), is compared to the Cell.T.MkCj(t+1) value in field data calculated from the battery. If a difference greater than or equal to a predetermined reference value occurs, a cell temperature malfunction detection signal is generated. When describing an example of detection with reference to Figure 4, the Cell.T.MkCj (t+1) values, which are field data, are referred to as "test data," and the Cell.Tpredic.MkCj (t+1) values, which are prediction values, are referred to as "simulation." As illustrated in (a) of Figure 4, when there is no deviation greater than or equal to a predetermined reference value between the field data and the prediction value, it is determined to be normal behavior. As illustrated in (b) of Figure 4, when there is a deviation greater than or equal to the predetermined reference value between the field data and the prediction value, it is determined and detected as battery malfunction behavior. 2-5. Diagnostic and alarm issuance process (S500) The diagnostic and alarm issuance process is a procedure to perform a diagnostic procedure or generate an alarm externally when there is a difference between the Cell.Vpredic.MkCj (t+1) , Cell.Tpredic.MkCj (t+1) , and Cell.V.MkCj (t+1) , Cell.T.MkCj (t+1) , which is greater than or equal to a reference value. In the present invention, the diagnostic procedure is not limited to a specially defined procedure, and when the deviation greater than or equal to the reference value occurs more than a predetermined number of times during a predetermined time section, an alarm generation signal may be issued or a control signal may be issued to block a battery charging or discharging operation. 3. Battery malfunction behavior detection system With reference to Figure 5, a battery malfunction behavior detection system according to the present invention will be described. 3-1. Field data calculation unit 100 The field data calculation unit 100 receives real-time measurement values ​​from the operating battery, calculates field data values ​​from the received data, and outputs the calculated values. Measurement and calculation of field data can be performed on each cell battery management system (BMS), module battery management system (BMS), or rack battery management system (BMS), and the field data is transmitted to the field data calculation unit 100. The field data calculation unit can be configured for integration into the rack BMS. The measurement values ​​received from the battery are values ​​received from the respective sensors installed on the battery and include measurement values ​​of normal battery status information. The calculated field data are the field data of the present invention described above.The field data processing unit transmits the field data to the field data preprocessing unit 200. 3-2. Field Data Preprocessing Unit 200 The field data preprocessing unit 200 is configured to include a cell voltage prediction field data preprocessing unit 210 that extracts field data for cell voltage prediction and a cell temperature prediction field data preprocessing unit 220 that extracts field data for cell temperature prediction, from the field data calculated by the field data calculation unit 210, and performs the field data preprocessing process for cell voltage prediction (S120) and the field data preprocessing process for cell temperature prediction (S130). The field data preprocessing unit 200 transmits training data and field data to the artificial intelligence neural network unit 300. In this case, the training data, as indicated by the arrow's trajectory shown by the dotted line in Figure 5, can be provided to the AI ​​neural network unit 300 by calculating the data corresponding to each element that constitutes the field data of a standard battery, whose normal quality has been verified in a laboratory, over a predetermined period of time. That is, the cell voltage prediction model 320 that constitutes the AI ​​neural network unit is trained using values ​​corresponding to the field data to predict the cell voltage calculated from the standard battery as training data, and the cell temperature prediction model 330 can be trained using values ​​corresponding to the field data to predict the cell temperature calculated from the standard battery as training data. In another embodiment, a configuration can be adopted in which training data is provided to the AI ​​neural network unit 300 by extracting field data over a certain period of time as training data during battery operation. Meanwhile, regardless of which of the two types of data is used as training data, the 200 field data preprocessing unit can provide the battery field data in the normal operating section as training data for the pre-trained AI neural networks 320 and 330 to be trained. 3-3. Artificial intelligence neural network unit 300 (1) Neural network training unit 310 The neural network training unit 310 receives the training data described above from the field data preprocessing unit 200 and trains the artificial intelligence neural network. The artificial intelligence neural network is trained as the cell voltage prediction model 320 and the cell temperature prediction model 330, and in the neural network training unit, the prediction models are trained to predict cell voltage and cell temperature, respectively, and the results are stored. The neural network training unit 310 can also be controlled to retrain the artificial intelligence neural network by adding field data for a predetermined period of the normal battery operating section as new training data in a predetermined cycle or under the control of the control unit 500 to update the prediction models 320 and 330. For example, while prediction models 320 and 330 are trained and operated on training data provided by the standard battery and generate prediction values ​​upon receiving field data, by retraining prediction models 320 and 330 by adding field data in a section during which the battery malfunction behavior detection unit 400 described below does not detect malfunction behavior, i.e., in a normal section, to the training data, prediction models 320 and 330 are trained to reflect the field data generated from the actual field battery operating state, thereby improving the prediction accuracy of prediction models 320 and 330. Artificial intelligence neural networks that have been trained with training data or that have been retrained and updated by adding field data are stored as the elda 320 voltage prediction model and the cell 330 temperature prediction model, respectively. (2) Cell voltage prediction model 320 The 320 cell voltage prediction model is a trained artificial intelligence neural network that receives field data for cell voltage prediction from the 200 field data preprocessing unit and predicts the cell voltage for the next cycle. The field data for cell voltage prediction can be the values ​​Rack.Current(t), Ambient.Temp(t), Fan.ON(t), SOC.Mk(t), SOH.MkCj(t), Cell.V.MkCj(t), and Cell.Vpredic.MkCj(t+1), which is a cell voltage prediction value calculated from these received values. (3) Cell temperature prediction model 330 The 330 cell temperature prediction model is a trained artificial intelligence neural network that receives field data for cell temperature prediction from the 200 field data preprocessing unit and predicts the cell temperature for the next cycle. The field data for cell temperature prediction can be the values ​​Cell.T.MkCj(t), Ambient.Temp(t), Fan.ON(t), and Cell.Tpredic.MkCj(t+1), which is a cell temperature prediction value calculated from these received values. 3-4. Battery Malfunction Behavior Detection Unit 400 The Battery Malfunction Behavior Detection Unit 400 compares the predicted cell voltage and cell temperature values ​​calculated by the Cell Voltage Prediction Model 320 and the Cell Temperature Prediction Model 330 of the Artificial Intelligence Neural Network Unit 300 with the cell voltage and cell temperature from field data, detects battery malfunction behavior, and transmits a malfunction behavior detection signal to a control unit. The malfunction behavior detection signal may include at least one or both of a cell voltage malfunction behavior signal and a cell temperature malfunction behavior signal, which will be described later. The 400 battery malfunction behavior detection unit receives the field data issued by the field data calculation unit and the prediction value issued by the 300 artificial intelligence neural network unit. (1) Cell voltage malfunction behavior determination unit 410 The cell voltage malfunction behavior determination unit 410 compares the Cell.Vpredic.MkCj (t+1) value described above with the Cell.V.MkCj (t+1) value from the field data, and determines that it is cell voltage malfunction behavior when the deviation is greater than or equal to a predetermined interval, and transmits the cell voltage malfunction behavior signal to the control unit. (2) Cell temperature malfunction behavior determination unit 420 The 420 cell temperature malfunction behavior determination unit compares the Cell.Tpredic.MkCj (t+1) value described above with the Cell.T.MkCj (t+1) value from the field data, determines that it is cell temperature malfunction behavior when the deviation is greater than or equal to the predetermined interval, and transmits the cell temperature malfunction behavior signal to the control unit. 3-5. Control Unit 500 Control unit 500 receives the malfunction detection signal from malfunction detection unit 400 and performs the diagnostic and alarm issuance process (S500) described above. The control unit can be connected to the field data calculation unit, the field data preprocessing unit, the prediction model, and the malfunction detection unit to control their respective configurations. In another embodiment, all the configurations of the field data calculation unit, the field data preprocessing unit, the prediction model, and the malfunction detection unit can be physically integrated to form a single control unit. In this case, the integrated control unit can be implemented by connecting it directly to the battery rack BMS or by integrating it into the rack BMS.

Claims

1. A battery malfunction behavior detection system comprising: a field data calculation unit (100) configured to receive real-time measurement values ​​from a battery in operation and to calculate and generate field data from the real-time measurement values; a field data preprocessing unit (200) configured to extract the first field data to predict a cell voltage and the second field data to predict a cell temperature from the field data; an artificial intelligence neural network unit (300) configured to generate a cell voltage prediction value from the first field data and to generate a cell temperature prediction value from the second field data; and a battery malfunction behavior detection unit (400) configured to compare the cell voltage prediction value and thecell temperature prediction value with the first field data and the second field data, respectively, and to determine that it exhibits malfunctioning behavior when the deviation is greater than or equal to a predetermined interval; wherein the first field data comprises time series values ​​of rack current, ambient temperature, fan on / off information, module SOC, cell SOH, and cell voltage for each battery cell and battery module that constitute the battery, and the second field data comprises time series values ​​of cell temperature, ambient temperature, and fan on / off information for each battery cell and battery module that constitute the battery; wherein the artificial intelligence neural network unit (300) is configured to comprehend: a cell voltage prediction model that is trained using valuescorresponding to field data to predict the cell voltage calculated from a standard battery instead of the battery for which the field data are calculated, as training data, wherein the cell voltage prediction model is configured to receive the first field data to predict the cell voltage and to generate the cell voltage prediction value for the next cycle, and a cell temperature prediction model that is trained using values ​​corresponding to the field data to predict the cell temperature calculated from the standard battery instead of the battery for which the field data are calculated, as training data, wherein the cell temperature model is configured to receive the second field data to predict the cell temperature and to generate the cell temperature prediction value for the next cycle; wherein the neural network unit ofThe artificial intelligence system (300) further comprises a neural network training unit configured to retrain and update a cell voltage prediction model and a cell temperature prediction model by adding field data from a predetermined period of normal battery operation as new training data.

2. The system of claim 1, wherein the artificial intelligence neural network unit (300) is configured to receive the first field data to predict the cell voltage in order to generate a cell voltage prediction value for a subsequent cycle, and to receive the second field data to predict the cell temperature in order to generate a cell temperature prediction value for a subsequent cycle.

3. A method for detecting battery malfunction behavior carried out by the system according to any of claims 1 and 2.comprising the method: a field data calculation process (S110) for measuring and calculating real-time battery state information data from a functioning battery; a field data preprocessing process including a field data preprocessing process for cell voltage prediction (S120) consisting of extracting data for cell voltage prediction from the calculated field data; a prediction model update process (S230) for retraining and updating a cell voltage prediction model and a cell temperature prediction model by adding field data from a predetermined period of a normal operating section of the functioning battery as new training data; a real-time prediction process (S300) including a cell voltage prediction process (S310) consisting of inputting the data for cell voltage prediction into theupdated cell voltage prediction model and calculate a cell voltage prediction value for the next cycle; and a battery malfunction behavior detection process (S400) comprising a cell voltage malfunction behavior detection process (S410) consisting of comparing the cell voltage prediction value with a cell voltage value from field data and generating a cell voltage malfunction behavior detection signal when the deviation is greater than or equal to a predetermined interval; wherein the field data preprocessing process further comprises a field data preprocessing process for cell temperature prediction (S130) consisting of extracting data to predict cell temperature from the calculated field data; the real-time prediction process (S300) further comprises a temperature prediction processcell (S320) comprising inputting cell temperature prediction data into the updated cell temperature prediction model and calculating a cell temperature prediction value for the next cycle; the battery malfunction behavior detection process (S400) further comprising a cell temperature malfunction behavior detection process (S420) comprising comparing the cell temperature prediction value with the cell temperature value from field data and generating a cell temperature malfunction behavior detection signal when the deviation between the cell temperature prediction value and the cell temperature value is greater than or equal to a predetermined interval; the cell voltage prediction model being trained using data corresponding to values ​​to predict the cell voltage calculated from a batteryThe standard battery, instead of the operating battery for which field data is calculated as training data, is updated and retrained to receive data to predict the cell voltage of the operating battery and is configured to generate the cell voltage prediction value for the next cycle. The cell temperature prediction model is trained using data corresponding to predicting the cell temperature calculated from a standard battery, instead of the operating battery for which field data is calculated as training data. It is updated and retrained to receive data to predict the cell temperature and is configured to generate the cell temperature prediction value for the next cycle. The data for predicting cell voltage are time series values ​​of rack current, ambient temperature, and other information.The data for predicting cell temperature are time series values ​​of cell temperature, ambient temperature, and fan on / off information for each battery cell and each battery module that make up the battery.

Images