Energy storage equipment life prediction method and system, electronic equipment and storage medium

Abstract

The invention provides an energy storage equipment service life prediction method and system, electronic equipment and a storage medium, and belongs to the technical field of equipment health prediction, and the method comprises the steps: obtaining operation data corresponding to target energy storage equipment; a target correlation index is obtained, the target correlation index is a correlation index between the multiple factors and the service life of the energy storage equipment, and the target correlation index is obtained through calculation based on historical operation data of the multiple pieces of energy storage equipment; target nonlinear change features are obtained, the target nonlinear change features are nonlinear change features between a multi-factor association vector extracted based on a causal index and the service life of the energy storage equipment, and the multi-factor association vector is constructed based on data corresponding to multiple factors; and inputting the target correlation index, the target nonlinear change feature and the operation data corresponding to the target energy storage equipment in the target time period into an energy storage equipment life prediction model for equipment life prediction to obtain a life prediction result. The accuracy of predicting the service life of the energy storage equipment can be improved.

Description

Technical Field

[0001] This application belongs to the technical field of equipment health prediction, and more specifically, it relates to a method and system for predicting the life of energy storage equipment, an electronic device, and a storage medium. Background Art

[0002] In the field of energy storage, accurately predicting the life of energy storage equipment is crucial for optimizing system design, ensuring the reliability of power supply, and reducing operation and maintenance costs. In reality, the aging and life of energy storage equipment are not only related to the usage time or frequency but also affected by various complex factors.

[0003] Currently, traditional methods for predicting the life of energy storage equipment have deficiencies. On the one hand, some methods based on empirical formulas are too rigid, with poor adaptability and accuracy. On the other hand, some existing technologies rely on traditional machine learning algorithms such as linear regression or support vector machines to estimate the equipment life. However, these methods are difficult to comprehensively capture the complex coupling effects among multiple factors affecting the equipment life, resulting in low accuracy of prediction results. Summary of the Invention

[0004] The purpose of this application is to provide a method and system for predicting the life of energy storage equipment, an electronic device, and a storage medium to improve the accuracy of predicting the life of energy storage equipment.

[0005] In the first aspect of the embodiments of this application, a method for predicting the life of energy storage equipment is provided, including: Obtaining the operation data corresponding to the target energy storage equipment within the target time period, where the target energy storage equipment is the energy storage equipment to be predicted; Obtaining the target correlation index, where the target correlation index is the correlation index between multiple factors and the life of the energy storage equipment, and the multiple factors include environmental temperature, charge-discharge rate, charge-discharge depth, and usage time, and the target correlation index is calculated based on the historical operation data of multiple energy storage equipment; Obtaining the target non-linear change feature, where the target non-linear change feature is the non-linear change feature between the multi-factor association vector extracted based on the causal index and the life of the energy storage equipment, and the causal index is the causal index between every two of the multiple factors, and the multi-factor association vector is constructed based on the data corresponding to the multiple factors; the data corresponding to the multiple factors includes environmental temperature data, charge-discharge rate data, charge-discharge depth data, and usage time data; Inputting the target correlation index, the target non-linear change feature, and the operation data corresponding to the target energy storage equipment within the target time period into the energy storage equipment life prediction model for equipment life prediction to obtain the life prediction result.

[0006] In the second aspect of the embodiments of this application, a system for predicting the life of energy storage equipment is provided, including: An operating data acquisition module, configured to acquire the operating data corresponding to a target energy storage device within a target time period, where the target energy storage device is an energy storage device to be predicted; A correlation index acquisition module, configured to acquire a target correlation index, where the target correlation index is the correlation index between multiple factors and the life of the energy storage device, and the multiple factors include ambient temperature, charge-discharge rate, charge-discharge depth, and usage time, and the target correlation index is calculated based on the historical operating data of multiple energy storage devices; A non-linear feature acquisition module, configured to acquire a target non-linear change feature, where the target non-linear change feature is the non-linear change feature between the multi-factor association vector extracted based on the causal index and the life of the energy storage device, and the causal index is the causal index between every two of the multiple factors, and the multi-factor association vector is constructed based on the data corresponding to the multiple factors; the data corresponding to the multiple factors includes ambient temperature data, charge-discharge rate data, charge-discharge depth data, and usage time data; A life prediction module, configured to input the target correlation index, the target non-linear change feature, and the operating data corresponding to the target energy storage device within the target time period into an energy storage device life prediction model to perform device life prediction, and obtain a life prediction result.

[0007] In a third aspect of the embodiments of the present application, an electronic device is provided, including a memory, a processor, and a computer program stored in the memory and running on the processor, and when the processor executes the computer program, the steps of the above-mentioned energy storage device life prediction method are implemented.

[0008] In a fourth aspect of the embodiments of the present application, a computer-readable storage medium is provided, where the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the above-mentioned energy storage device life prediction method are implemented.

[0009] The beneficial effects of the energy storage device life prediction method, system, electronic device, and storage medium provided by the embodiments of the present application are as follows: By calculating the correlation index between multiple factors and the life of the energy storage device, the embodiments of the present application can accurately screen out the key factors that have a significant impact on the life, avoid interference from redundant information, and lay a reliable foundation for prediction. Based on the causal index, the embodiments of the present application extract non-linear change features, which can deeply capture the complex interaction and non-linear relationship between multiple factors, and effectively make up for the deficiencies of traditional methods in non-linear modeling. By inputting the correlation index, non-linear change features, and operating data into the prediction model together, the embodiments of the present application can more comprehensively and accurately describe the life of the energy storage device. Compared with traditional methods, the prediction accuracy is significantly improved, which helps to optimize the operation and maintenance strategy of the energy storage device and ensure the stable and efficient operation of the energy storage system. Description of the Drawings

[0010] To more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the accompanying drawings required for use in the embodiments or the description of the prior art. Obviously, the accompanying drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0011] Figure 1 It is a schematic flowchart of a method for predicting the life of an energy storage device provided by an embodiment of the present application; Figure 2 It is a structural block diagram of a system for predicting the life of an energy storage device provided by an embodiment of the present application; Figure 3 It is a schematic block diagram of an electronic device provided by an embodiment of the present application. Detailed implementation manners

[0012] In the following description, specific details such as specific system structures and technologies are presented for the purpose of illustration rather than limitation, so as to thoroughly understand the embodiments of the present application. However, those skilled in the art should clearly understand that the present application can also be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to avoid unnecessary details from interfering with the description of the present application.

[0013] To make the purpose, technical solutions, and advantages of the present application clearer, the following will be described through specific embodiments in conjunction with the accompanying drawings.

[0014] Please refer to Figure 1 , Figure 1 It is a schematic flowchart of a method for predicting the life of an energy storage device provided by an embodiment of the present application. This method can be executed by an electronic device. Specifically, this method may include S101 to S104.

[0015] S101: Obtain the operation data corresponding to the target energy storage device within the target time period, where the target energy storage device is the energy storage device to be predicted.

[0016] In this embodiment, the target time period refers to the time interval during which the operation data to be collected is generated, which is preset. The target time period can be flexibly set according to the prediction requirements. For example, for short-term prediction, data in the recent month can be used, and for long-term prediction, data in the recent year can be used. The target energy storage device is the specific energy storage device for which the life needs to be predicted, and the unique identifier of the device can be determined by parameters such as the device number, model, and installation location. The operation data may include environmental temperature data, charge and discharge rate data, charge and discharge depth data, usage time data, and basic electrical data such as battery voltage, current, and internal impedance.

[0017] In this embodiment, the operation data further includes device operation status information, which is closely related to the device lifespan. Obtaining this data is the basis for lifespan prediction. Only by accurately grasping the operation status of the device within a specific time period can an accurate lifespan prediction result be obtained through the prediction model by combining the correlation index and the non-linear change characteristics.

[0018] Exemplarily, in this embodiment, the target energy storage device can be uniquely identified by the device ID, physical address, or asset code. In this embodiment, a sensor network such as a pre-deployed temperature sensor, current transformer, and voltage transmitter can be used to collect ambient temperature, charge / discharge current / voltage data in real time. This embodiment can interface with the battery management system to obtain internal state data such as charge / discharge depth, state of charge (the percentage of the remaining battery charge to its total capacity), and number of cycles, so as to collect the operation data corresponding to the target energy storage device.

[0019] This embodiment can set a target time period based on the prediction requirement, perform data cleaning on the operation data within the target time period, detect and replace outliers, and obtain accurate and complete operation data.

[0020] S102: Obtain the target correlation index. The target correlation index is the correlation index between multiple factors and the lifespan of the energy storage device. The multiple factors include ambient temperature, charge / discharge rate, charge / discharge depth, and usage time. The target correlation index is calculated based on the historical operation data of multiple energy storage devices.

[0021] In this embodiment, the target correlation index is a quantitative index that measures the degree of tightness of the association between factors such as ambient temperature and charge / discharge rate and the lifespan of the energy storage device. The target correlation index can include the Pearson correlation coefficient between each factor and the lifespan. The historical operation data can include the operation parameters of multiple energy storage devices in the past period of time and their corresponding lifespan indicators.

[0022] In this embodiment, the linear association between factors and lifespan is quantified through the correlation index, which is used to screen out the factors that have a significant impact on lifespan, perform weight allocation, and provide key features for subsequent lifespan prediction. High-correlation factors are important drivers of lifespan changes. For example, the charge / discharge rate is strongly negatively correlated with lifespan, indicating that high rates will accelerate device aging. Incorporating it into the lifespan prediction model can improve the prediction accuracy.

[0023] Exemplarily, in this embodiment, historical operation data such as ambient temperature, charge / discharge rate, charge / discharge depth, and usage time can be obtained from data sources such as the battery management systems and sensors of multiple energy storage devices, and at the same time, lifespan-related data such as the health status and remaining number of cycles of the devices can be obtained.

[0024] In this embodiment, the acquired data is subjected to data cleaning, data integrity is checked, records with excessive missing values are deleted, a small amount of missing data is filled with adjacent values, and abnormal data is identified and removed, such as temperature values or magnification values that exceed the normal range. The data collected in this embodiment is sorted according to the equipment and time sequence to ensure that the operation data and life data of the same equipment correspond one by one in time.

[0025] In this embodiment, the Pearson correlation coefficient analysis method is used to calculate the target correlation index between each factor and the equipment life, providing a calculation basis for the life prediction of the energy storage equipment life prediction model.

[0026] S103: Obtain the target non-linear change characteristics. The target non-linear change characteristics are the non-linear change characteristics between the multi-factor correlation vector extracted based on the causal index and the life of the energy storage equipment. The causal index is the causal index between every two factors among multiple factors, and the multi-factor correlation vector is constructed based on the data corresponding to multiple factors; the data corresponding to multiple factors includes environmental temperature data, charge and discharge magnification data, charge and discharge depth data, and usage time data.

[0027] In this embodiment, the target non-linear change characteristics refer to the non-linear influence mode of the complex interaction between multiple factors on the life of the energy storage equipment, which is different from the linear correlation of a single factor. For example, the target non-linear change characteristics may include: the influence of multiple factors acting together on the life is greater than or less than the sum of their individual actions; the influence of a certain factor on the life is weak within a specific threshold, and after exceeding the threshold, it acts together with other factors to cause a sudden change in life decay; the influence of factors accumulates non-linearly over time; the presence of a certain factor weakens the influence of other factors on the life; the influence of a certain factor on the life only appears after interacting with other factors through a time delay.

[0028] Exemplarily, when a high-temperature environment is superimposed with a high charge and discharge magnification, the battery life decay rate significantly exceeds the linear superposition of the independent effects of the two, showing a non-linear enhancement effect of "1 + 1 > 2". When the charge and discharge depth is lower than 80%, the environmental temperature has little influence on the life, but after exceeding 80%, the increase in temperature will trigger a non-linear inflection point of accelerated life decline. In the case of a long-term low charge and discharge magnification but high-frequency cycling working condition, progressive damage is caused to the battery through time accumulation, and its life decay curve shows a non-linear characteristic of accelerating with the extension of the usage time. Good temperature control measures can inhibit the negative impact of high charge and discharge depth on the battery life, making the decay rate when the two act together lower than that of a single high-depth working condition.

[0029] The causal index is a quantitative indicator for measuring the strength and direction of causal relationships among multiple factors. The causal index can include the F-statistic value of the Granger causality test, the probability value (p-value) of the causal relationship, and the causal effect value in the structural causal model, such as the average treatment effect ATE. In this embodiment, the causal index refers to the causal index between every two factors.

[0030] The multi-factor association vector refers to a multi-dimensional feature vector formed by aggregating the original data of multiple factors and their derived features according to a time window, and is used to characterize the comprehensive effect of multiple factors. The parameters of the multi-factor association vector can include environmental temperature (mean / maximum value), charge-discharge rate (proportion of high-rate), charge-discharge depth (mean / number of times exceeding 80%), usage time, and cross terms constructed based on causal relationships, such as temperature × rate, rate × depth, etc.

[0031] In this embodiment, the life of the energy storage device is affected by the non-linear interaction of multiple factors, rather than the independent action of a single factor. For example, the synergistic effect of high temperature and high rate is more significant in accelerating capacity decay than their individual effects; the time-sequence dependence of charge-discharge depth and temperature, such as deep discharge under long-term high temperature, will cause irreversible aging.

[0032] The causal index of this embodiment is used to identify key causal chains and guide feature construction. For example, through the causal index, factor pairs with strong causal relationships can be screened, such as temperature → rate, p < 0.05, to avoid including pseudo-correlated factors in the prediction model. And, cross terms are constructed only for factor pairs with causal relationships to ensure that non-linear features have physical meanings, rather than blindly combining them.

[0033] The multi-factor association vector of this embodiment can aggregate high-frequency operation data according to a time window, transform it into a low-frequency feature vector, and align it with the life index in the time dimension, which is convenient for capturing long-term cumulative non-linear effects.

[0034] Exemplarily, in this embodiment, the Granger causality test can be performed on each pair of factors, the F-statistic and p-value can be calculated, and then it can be judged whether there is a significant causal relationship through threshold comparison. For factor pairs with a significant causal relationship, the causal direction and causal index value are recorded, such as the magnitude of the F-statistic reflecting the action intensity.

[0035] This embodiment can slide and divide historical data according to a fixed time window, and generate a feature vector for each window. This embodiment calculates feature data such as the average temperature, maximum rate, average depth, and total usage time within the window; this embodiment can also calculate features such as the standard deviation of temperature and the proportion of the duration of high rate within the window to capture the time-sequence change pattern; this embodiment can also construct product terms for factor pairs with significant causal tests to avoid meaningless feature combinations. This embodiment can construct a multi-factor association vector based on the extracted feature data and product terms.

[0036] In this embodiment, the execution order of the combination of S101 with S102 and S103 can be swapped. For example, execute S102, S103, S101, etc. Figure 1 This is only an example of the execution order.

[0037] This embodiment can adopt models such as long short-term memory networks or gradient boosting trees that are good at dealing with non-linear relationships and time series data to implement the specific steps of "extracting the non-linear change characteristics between the multi-factor association vector and the energy storage device life based on the causal index".

[0038] S104: Input the target correlation index, the target non-linear change characteristics, and the operation data corresponding to the target energy storage device within the target time period into the energy storage device life prediction model for device life prediction to obtain the life prediction result.

[0039] Considering that a linear model of a single factor cannot capture the complexity of energy storage device aging, while a pure non-linear model will ignore physical priors. Fusing these two types of features can balance the accuracy and interpretability of the model. Specifically, the target correlation index of this embodiment can provide a direct influence path of a single factor. For example, high rate directly causes life attenuation. The target non-linear change characteristics of this embodiment can be used to supplement the indirect influence of multi-factor interaction. For example, the synergistic acceleration effect of high temperature and high rate. The combination of the two can comprehensively describe the life change characteristics.

[0040] This embodiment trains the model through historical data to learn the mapping relationship from the correlation index, non-linear characteristics, real-time operation data to the device life, so as to achieve accurate evaluation of the current state of the device.

[0041] Exemplarily, the energy storage device life prediction model can adopt a fully connected neural network model. The model is trained through historical data, which can specifically include: training the initial model through multiple training samples to obtain the energy storage device life prediction model. Among them, a training sample includes historical data and the life label corresponding to the historical data (such as health status level, remaining life duration, etc.). The historical data can include historical correlation index, historical non-linear change characteristics, and historical operation data. Further, the historical data is used for energy storage device life prediction through the initial model to obtain the life prediction result. Based on the difference between the life prediction result and the corresponding life label, the initial model is trained to obtain the energy storage device life prediction model.

[0042] Exemplarily, the energy storage device life prediction model of this embodiment can also adopt a multiple linear regression model. The historical correlation index, historical non-linear feature vector, and historical operation data statistic are used as the input feature matrix X, with a dimension of [n×m], where n is the number of samples and m is the number of features. The weight vector W is solved by the least squares method to minimize the mean square error between the predicted life value Y' = X×W and the actual life label Y, so as to further obtain the energy storage device life prediction model.

[0043] Exemplarily, in this embodiment, statistical methods are used to calculate the correlation index between each factor and the device life, and a quantitative value of the association degree between each factor and the life is obtained. Then, according to the absolute value of the correlation index, weight distribution is performed on each factor. The larger the absolute value of the correlation index, the higher the weight of the corresponding factor. The original data of each factor is multiplied by the corresponding weight and then added together to generate weighted comprehensive feature data. In this embodiment, the weighted feature data, non-linear change features, and operation data in the target time period are jointly input into the life prediction model, and the target data is output to obtain the life prediction result.

[0044] Exemplarily, taking the case where the energy storage device life prediction model adopts a fully connected neural network model as an example to introduce the method of predicting the life of an energy storage device. Among them, the fully connected neural network model can include an embedding layer, an attention fusion layer, and an output layer. The embedding layer is used to map the target correlation index, target non-linear features, and operation data corresponding to the target energy storage device to a feature space of a unified dimension; the attention fusion layer is used to use the absolute value of the target correlation index as the attention weight to perform weighted summation on the three types of features to obtain a fused feature; the output layer is used to map the fused feature to a life index and output the life prediction result.

[0045] Specifically, in the actual application process, the processing flow of the energy storage device life prediction model for the target correlation index, the target non-linear change feature, and the operation data corresponding to the target energy storage device in the target time period can include: the embedding layer maps the input data to a 128-dimensional unified feature space; the attention fusion layer uses the absolute value of the target correlation index as the weight to fuse the three types of features, captures the interaction and dependence relationship between each factor, and obtains a fused feature vector; the output layer can include two fully connected layers. The first fully connected layer performs a linear transformation on the fused feature vector to obtain a 64-dimensional intermediate vector, and the second fully connected layer performs a linear transformation on the 64-dimensional intermediate vector to obtain the life prediction result.

[0046] As can be seen from the above, in the embodiments of the present application, by calculating the correlation index between multiple factors and the life of the energy storage device, key factors that have a significant impact on the life can be accurately screened out, avoiding interference from redundant information and laying a reliable foundation for prediction. The embodiments of the present application extract non-linear change characteristics based on the causal index, can deeply capture the complex interaction and non-linear relationship between multiple factors, and effectively make up for the deficiencies of traditional methods in non-linear modeling. The embodiments of the present application jointly input the correlation index, non-linear change characteristics and operation data into the prediction model to achieve a more comprehensive and accurate characterization of the life of the energy storage device. Compared with traditional methods, the prediction accuracy is significantly improved, which helps to optimize the operation and maintenance strategy of the energy storage device and ensure the stable and efficient operation of the energy storage system.

[0047] In an embodiment of the present application, the target correlation index is determined based on the historical operation data of multiple energy storage devices and in the following manner: Extract the data corresponding to multiple factors respectively based on the historical operation data of multiple energy storage devices; Obtain the life index data of multiple energy storage devices, and the life index data includes the battery health state, internal battery impedance, charge and discharge cycle times, and battery capacity attenuation rate of multiple energy storage devices; Align the life index data with the data corresponding to multiple factors respectively in the time dimension; For each factor among the multiple factors, select the first time period set from the target historical period based on the data change rate of the multiple factors; calculate the correlation index between the factor and the device life based on the data corresponding to the factor and the life index data within the first time period set; the target historical period is the data generation period corresponding to the historical operation data; Determine the target correlation index based on the correlation index between each factor and the device life respectively.

[0048] In this embodiment, extracting the data corresponding to the multiple factors respectively based on the historical operation data of multiple energy storage devices specifically includes: for the historical operation data of each energy storage device, extract the environmental temperature data, charge and discharge rate data, charge and discharge depth data, and usage time data.

[0049] In this embodiment, the battery management system can calculate the battery health state based on parameters such as battery capacity and internal resistance through built-in algorithms, trigger full-capacity tests regularly, and generate battery health state percentage data. This embodiment can use the alternating current impedance spectroscopy technique to inject a small alternating current signal when the battery is in a static state, measure the voltage response at different frequencies, and calculate the internal resistance data. This embodiment can set a counter in the battery management system. When the charge and discharge depth exceeds a specific threshold, it is counted as a complete cycle, and the cycle count is accumulated and stored in real time. This embodiment can conduct standard charge and discharge tests regularly, record the difference between the actual available capacity and the initial capacity, and divide it by the test interval time to obtain the capacity attenuation rate. This embodiment can synchronize the above data to the database according to the device number and timestamp to ensure alignment with the operation factor data time for subsequent analysis.

[0050] In this embodiment, for each of the multiple factors, a first time period set is selected from the target historical period based on the data change rate of the multiple factors, specifically including: For each of the multiple factors, calculate a first data change rate based on the data corresponding to this factor, and determine a second time period set from the target historical period based on the first data change rate; the second time period set includes time periods that meet the first condition; the first condition is that the first data change rate corresponding to a certain time period is greater than or equal to the first threshold; Calculate a second data change rate in the second time period set based on the data corresponding to all factors other than this factor among the multiple factors, and select a first time period set from the second time period set; the first time period set includes time periods in the second time period set that meet the second condition; the second condition is that the second data change rate corresponding to a certain time period is less than or equal to the second threshold.

[0051] In this embodiment, the first data change rate is for a single factor (such as ambient temperature), calculating the degree of data fluctuation in each time period of its historical operation data. The parameters can include indicators such as slope and standard deviation that characterize the change amplitude. The second time period set refers to the set of time periods when the first data change rate is greater than or equal to the first threshold, and its parameters include the start and end times of each time period. The first threshold is a critical value set artificially to measure whether the first data change rate is significant. The first thresholds for different factors are different and need to be determined according to experience or experiments. The second data change rate refers to the comprehensive change rate obtained by weighted summing the data change rates of all factors other than a certain factor. The parameters are also indicators characterizing the degree of change, such as weighted standard deviation. The first time period set is selected from the second time period set, and the time periods when the second data change rate is less than or equal to the second threshold are used for subsequent correlation index calculation. The second threshold is a critical value set artificially for screening the first time period set to judge whether the changes of other factors are relatively stable.

[0052] In this embodiment, the lifespan of the energy storage device is affected by multiple factors comprehensively. Excessive data change rates of a single factor will interfere with the true influence of other factors. By setting double thresholds to screen the first time period set, first select the time periods with significant changes in a certain factor, and then select the time periods with relatively stable changes in other factors among these time periods, the mutual interference of multiple factor changes can be reduced, and the calculated correlation index can more accurately reflect the true relationship between this factor and the lifespan.

[0053] Exemplarily, taking the ambient temperature as an example, in this embodiment, the first data change rate of the temperature within each month can be calculated, such as the standard deviation of the temperature within the month. Set the first threshold to 5°C, that is, screen the months with a temperature fluctuation greater than or equal to 5°C) to obtain the second time period set, such as the high-temperature months in summer. In this embodiment, the second data change rate of other factors within the same time period is then calculated, such as the weighted sum of the standard deviation of the rate and the depth fluctuation amplitude. According to the set second threshold, the first time period set is screened as the weeks with relatively small fluctuations in the rate and depth from June to August, such as the rate being less than or equal to 1C per week and the depth being less than or equal to 70%. Analyze the relationship between the temperature and the battery health state within the selected first time period set: the average temperature of this time period is 35°C, and the battery health state drops from 92% to 91%. Combining the historical data under the same conditions, the calculated correlation index between the temperature and the battery health state is -0.8 (strong negative correlation), indicating that high temperature significantly accelerates battery decay.

[0054] Similarly, repeat the above steps for the charge and discharge rate: screen the time periods with a rate change rate greater than or equal to 0.5C / day and stable other factors, and calculate the correlation index between it and the capacity decay rate as 0.7 (strong positive correlation).

[0055] Similarly, perform the corresponding steps for other factors.

[0056] In this embodiment, the lifespan index data is the data collected based on the first frequency, and the data corresponding to multiple factors are the data collected based on the second frequency; the second frequency is greater than the first frequency; In this embodiment, align the lifespan index data and the data corresponding to multiple factors in the time dimension, specifically including: obtaining the forward time window; For each collection of lifespan index data, perform the first operation; Among them, the first operation includes: Obtain the collection timestamp of the lifespan index data collected this time, and extract the data within the forward time window from the data corresponding to multiple factors; Aggregate the data corresponding to multiple factors within the forward time window to obtain the aggregated data sequences corresponding to multiple factors; Align the timestamps of the aggregated data sequences corresponding to multiple factors with the collection timestamp of the lifespan index data.

[0057] In this embodiment, obtaining the forward time window may include: determining the forward time window based on the difference between the acquisition timestamps of any two adjacent lifetime index data.

[0058] In this embodiment, the first frequency refers to the acquisition frequency of the lifetime index data. The second frequency refers to the acquisition frequency of operation factor data such as environmental temperature and charge-discharge rate. The forward time window is a time range delimited forward with each acquisition timestamp of the lifetime index data as a reference, and the parameter includes the window length, such as 1 day or 7 days. For example, if the acquisition frequency of the lifetime index data is once a month, the difference between the adjacent acquisition timestamps of any two lifetime index data is one month, and the forward time window is one month.

[0059] The aggregated data sequence refers to the data sequence formed after statistically aggregating the factor data within the forward time window. The parameters may include statistics such as the mean, maximum value, minimum value, standard deviation, etc. corresponding to multiple factors, as well as other data characteristics.

[0060] Considering that the acquisition frequency of the lifetime index data is low and the acquisition frequency of the operation factor data is high, directly correlating the two is likely to cause information mismatch. In this embodiment, a time range is delimited through the forward time window, and the high-frequency operation factor data is aggregated into a low-frequency data sequence, which is aligned with the acquisition timestamp of the low-frequency lifetime index data. This can not only retain the comprehensive influence information of the operation factors but also enable the two types of data to be accurately matched in the time dimension, providing a reliable data basis for subsequent correlation analysis and lifetime prediction.

[0061] Exemplarily, in a large lithium battery energy storage power station, the lifetime index data such as the battery health state is acquired once a week, while the operation factor data such as environmental temperature and charge-discharge rate is acquired once an hour.

[0062] In this embodiment, the forward time window length can be set to 7 days, that is, each time with the acquisition timestamp of the lifetime index data as a reference, looking back 7 days. For example, if the acquisition time of a certain battery health state data is October 15, 2024, the corresponding data from October 8 to October 14 is extracted from the operation factor data. In this embodiment, for the temperature data within these 7 days, the mean, maximum value, and minimum value can be calculated; for the charge-discharge rate data, the daily average rate, the proportion of the hours when the high rate (greater than 1C) appears, etc. can be calculated to obtain the aggregated data sequence of each factor.

[0063] In this embodiment, the timestamp of the aggregated data sequence of each factor is set to October 14 to achieve alignment with the battery health state data collected this time. Repeat the above steps to complete the time dimension alignment of all lifetime index data and operation factor data, providing complete and accurate data for subsequent calculation of the correlation index and construction of the lifetime prediction model.

[0064] In this embodiment, by screening the first time period set through double thresholds, multi-factor interference can be reduced, and the correlation index can be accurately calculated; by aligning high-frequency and low-frequency data using a forward time window in this embodiment, information mismatch can be avoided. The combination of the two can effectively improve data quality and analysis accuracy, more truly reflect the association between various factors and the life of energy storage devices, and further improve the accuracy of the life prediction model, providing a reliable basis for equipment operation and maintenance.

[0065] In one embodiment of the present application, the calculation process of the causal index between every two factors among multiple factors includes: Construct a first autoregressive model based on the data corresponding to the first factor among the two factors; Construct a second regression model based on the first autoregressive model, the data corresponding to the second factor among the two factors, and the optimal lag order; Calculate the sum of squared residuals of the first autoregressive model, and calculate the sum of squared residuals of the second regression model; Determine the causal index between the two factors based on the sum of squared residuals of the first autoregressive model and the sum of squared residuals of the second regression model.

[0066] In this embodiment, determining the causal index between the two factors based on the sum of squared residuals of the first autoregressive model and the sum of squared residuals of the second regression model includes: Obtain the data sample sizes of the data corresponding to the two factors; Calculate the variance ratio statistic value based on the sum of squared residuals of the first autoregressive model, the sum of squared residuals of the second regression model, the optimal lag order, and the data sample sizes of the data corresponding to the two factors; Determine the causal index between the two factors based on the variance ratio statistic value.

[0067] In this embodiment, the first factor and the second factor are any two factors selected from multiple factors such as ambient temperature and charge-discharge rate that affect the life of energy storage devices.

[0068] The first autoregressive model is constructed based on the historical data of the first factor and is used to describe the law of change of the factor itself over time. The parameters may include the model order, regression coefficients, etc. The second regression model is constructed by adding the data of the second factor on the basis of the first autoregressive model and is used to analyze the influence of the second factor on the first factor. The parameters may include the optimal lag order, and the optimal lag order represents the time delay of the influence of the data of the second factor on the first factor. The sum of squared residuals of the model is an index to measure the error between the predicted value and the actual value of the model. The smaller the value of the sum of squared residuals of the model, the better the fitting effect of the model. The variance ratio statistic value is calculated by integrating the sum of squared residuals of the first and second models, the optimal lag order, and the data sample size, and is an index used to quantify the strength of the causal relationship between two factors.

[0069] In this embodiment, the life of the energy storage device is affected by the interaction of multiple factors. By constructing a first autoregressive model and a second regression model in this embodiment and comparing the changes in the model errors before and after adding the second factor, it can be determined whether the second factor can significantly improve the prediction ability for the first factor. If the error decreases significantly after adding, it indicates that the second factor has a causal effect on the first factor. The variance ratio statistic value and the causal index can quantify the degree of this influence, providing a basis for analyzing the causal relationship between factors.

[0070] Exemplarily, in a large lithium battery energy storage power station, to analyze the causal relationship between the ambient temperature and the charge-discharge rate, the following operations are carried out: This embodiment obtains historical data for nearly one year, where the ambient temperature is collected once per hour and the charge-discharge rate is collected once per minute. The charge-discharge rate data is summarized by hour into an average value to align with the time frequency of the temperature data.

[0071] Taking the ambient temperature as the first factor and the charge-discharge rate as the second factor. First, based on the historical temperature data, multiple autoregressive models are tried to be constructed from the first order to the fifth order. By comparing the differences between the model predicted values and the actual temperature values, it is determined that the prediction error of the third-order autoregressive model is the smallest. Then, the third-order autoregressive model is used as the first autoregressive model, and the sum of squared residuals of the first autoregressive model is calculated, that is, the differences between the predicted temperature and the actual temperature at each time point are squared and then added up.

[0072] Determine the optimal lag order of the charge-discharge rate: Start testing from a lag of 1 hour, gradually increase the lag duration, compare the prediction effects of the second regression model under different lag orders, and finally analyze and determine that the optimal lag order of the charge-discharge rate is 2 hours, that is, the current temperature is affected by the charge-discharge rate 2 hours ago. Based on the 3rd-order autoregressive model, the charge-discharge rate lagging 2 hours is added as an independent variable to construct a second regression model containing the lag data of the charge-discharge rate, and the sum of squared residuals of the second regression model is calculated.

[0073] This embodiment compares the sum of squared residuals of the two models. If the difference between the sum of squared residuals of the second regression model and the sum of squared residuals of the first autoregressive model exceeds the threshold, it indicates that the charge-discharge rate has a causal effect on the ambient temperature. Combining the data sample size and the lag order, this embodiment can further calculate the variance ratio statistic value. If the variance ratio statistic value is greater than the preset threshold, it indicates that the causal relationship between the two is strong. The causal index can be mapped to a specific value according to the size of the statistic value, providing a basis for optimizing the temperature control and charge-discharge strategy of the power station.

[0074] Exemplarily, the calculation process of the variance ratio statistic value is: (Sum of squared residuals of the first model - Sum of squared residuals of the second model) ÷ Sum of squared residuals of the second model × (Data sample size - Number of model parameters).

[0075] In this embodiment, by constructing an autoregressive and regression model, comparing the error changes to accurately judge the causal relationship between factors, and using the variance ratio statistic to quantify the influence intensity, the interaction mechanism of multiple factors of the energy storage device can be effectively analyzed, avoiding the interference of pseudo-correlation, and providing a scientific basis for formulating temperature control, charging and discharging strategies for the power station.

[0076] In an embodiment of the present application, the target non-linear change feature is obtained in the following manner: Construct an interaction term of factors based on the causal index and the multi-factor correlation vector, and splice the interaction term of factors into the multi-factor correlation vector to obtain a spliced vector; Perform attention allocation on each element in the spliced vector based on the causal index to obtain a target vector; Extract the non-linear change feature between the target vector and the life of the energy storage device as the target non-linear change feature.

[0077] In this embodiment, the specific steps of extracting the non-linear change feature between the multi-factor correlation vector and the life of the energy storage device based on the causal index can be implemented by an LSTM model based on the attention mechanism and feature crossing.

[0078] Exemplarily, extracting the non-linear change feature between the multi-factor correlation vector and the life of the energy storage device based on the causal index specifically includes: Input the causal index and the multi-factor correlation vector into an LSTM model based on the attention mechanism and feature crossing to obtain the non-linear change feature between the multi-factor correlation vector and the device life; The LSTM model based on the attention mechanism and feature crossing includes an input layer, a feature crossing layer, an LSTM layer, an attention layer, a fully connected layer and an output layer; The feature crossing layer is used to construct an interaction term of factors based on the causal index, and splice the interaction term of factors into the multi-factor correlation vector to obtain a target feature vector; The LSTM layer is used to extract the temporal dependence features of adjacent time steps of the target feature vector to obtain the hidden state of each time step; The attention layer is used to assign attention weights to the hidden states of each time step based on the causal index, and perform weighted summation on the hidden states of all time steps based on the attention weights to obtain a target vector.

[0079] In this embodiment, the factor interaction term is a new feature formed by combining factors with causal relationships based on the causal index. During specific implementation, all causal indices can be traversed to filter out all factors with causal relationships. For example, if the causal index between the ambient temperature and the charge-discharge rate meets the preset threshold condition, it is determined that if there is a causal relationship, the product or other combination form of the two can be used as the factor interaction term. If there is a causal relationship between the ambient temperature and the charge-discharge depth, the product or other combination form of the two can also be used as the factor interaction term. The spliced vector refers to the new vector formed by adding the factor interaction term to the multi-factor correlation vector. Attention allocation means assigning different weights to each element of the spliced vector according to the magnitude of the causal index to highlight important factors. The target vector is the vector obtained after attention allocation, which centrally reflects the factor characteristics that have an important impact on the device life.

[0080] The LSTM model based on the attention mechanism and feature crossing is a deep learning model that integrates feature crossing and the attention mechanism. The input layer receives the causal index and the multi-factor correlation vector; the feature crossing layer realizes the construction and splicing of the factor interaction term; the LSTM layer extracts the time series dependence features; the attention layer assigns weights; the fully connected layer and the output layer output the non-linear change features.

[0081] In this embodiment, the life of the energy storage device is affected by multi-factor non-linear interaction and time series changes. In this embodiment, the factor interaction term is constructed through the causal index to capture the interaction between factors; in this embodiment, the attention mechanism is used to assign weights according to the causal strength to highlight key factors. The LSTM model is good at processing time series data. Through the cooperation of each layer, it can extract the complex non-linear change features between multi-factors and the device life, provide more accurate input for life prediction, and improve the prediction accuracy.

[0082] Exemplarily, in this embodiment, the calculated causal indices of each factor and the constructed multi-factor correlation vector are input into the input layer of the LSTM model based on the attention mechanism and feature crossing to ensure that the time order and dimension of the data meet the model requirements.

[0083] In the feature crossing layer, the model judges the causal relationship between factors according to the causal index. For factor pairs with causal relationships, factor interaction terms are generated according to the set rules, such as multiplying the data corresponding to the two factors or performing other combination operations. The model splices the generated factor interaction terms after the original multi-factor correlation vector to obtain the target feature vector.

[0084] The target feature vector is input into the LSTM layer, and the model processes the data step by step according to the time steps. The LSTM layer captures the time series dependence features between adjacent time steps of the target feature vector through internal memory units and gating mechanisms, and a hidden state is output at each time step.

[0085] In the attention layer, the model assigns attention weights to the hidden states at each time step according to the causal index. The hidden states corresponding to factors with stronger causal relationships are assigned higher weights. Based on the assigned weights, the model performs a weighted sum of the hidden states at all time steps to obtain the final target vector.

[0086] The target vector is input into the fully connected layer for feature fusion and then output through the output layer to obtain the non - linear change characteristics between the multi - factor correlation vector and the device life.

[0087] In this embodiment, the causal index is used to construct the factor interaction term, which can accurately capture the non - linear relationship between multiple factors; the attention mechanism is used to assign weights in this embodiment, which can highlight the key influencing factors; the LSTM is combined to extract time - series features in this embodiment, which can effectively process the dynamic changes of data. This embodiment can comprehensively mine the complex relationship between multiple factors and the device life, provide high - precision features for the life prediction of energy storage devices, and improve the prediction accuracy and reliability.

[0088] Corresponding to the energy storage device life prediction method in the above - mentioned embodiment, Figure 2 is the structural block diagram of the energy storage device life prediction system provided by an embodiment of the present application. For the sake of convenience of description, only the parts related to the embodiment of the present application are shown. Refer to Figure 2 The energy storage device life prediction system 20 includes: an operating data acquisition module 21, a correlation index acquisition module 22, a non - linear feature acquisition module 23, and a life prediction module 24.

[0089] Among them, the operating data acquisition module 21 is used to acquire the operating data corresponding to the target energy storage device within the target time period, and the target energy storage device is the energy storage device to be predicted; The correlation index acquisition module 22 is used to acquire the target correlation index, and the target correlation index is the correlation index between multiple factors and the life of the energy storage device. The multiple factors include environmental temperature, charge - discharge rate, charge - discharge depth, and usage time. The target correlation index is calculated based on the historical operating data of multiple energy storage devices; The non - linear feature acquisition module 23 is used to acquire the target non - linear change characteristics. The target non - linear change characteristics are the non - linear change characteristics between the multi - factor correlation vector extracted based on the causal index and the life of the energy storage device. The causal index is the causal index between every two factors among the multiple factors, and the multi - factor correlation vector is constructed based on the data corresponding to the multiple factors; the data corresponding to the multiple factors includes environmental temperature data, charge - discharge rate data, charge - discharge depth data, and usage time data; The lifespan prediction module 24 is configured to input the target correlation index, the target non-linear change characteristics, and the operation data corresponding to the target energy storage device within the target time period into the energy storage device lifespan prediction model for device lifespan prediction, so as to obtain the lifespan prediction result.

[0090] In an embodiment of the present application, the correlation index acquisition module 22 is specifically configured to extract the data corresponding to multiple factors respectively based on the historical operation data of multiple energy storage devices; Obtain the lifespan index data of multiple energy storage devices, where the lifespan index data includes the battery health state, the internal impedance of the battery, the charge and discharge cycle times, and the battery capacity attenuation rate of multiple energy storage devices; Align the lifespan index data with the data corresponding to multiple factors respectively in the time dimension; For each of the multiple factors, select the first time period set from the target historical period based on the data change rate of the multiple factors; calculate the correlation index between the factor and the device lifespan based on the data corresponding to the factor within the first time period set and the lifespan index data; the target historical period is the data generation period corresponding to the historical operation data; Determine the target correlation index based on the correlation index between each factor and the device lifespan respectively.

[0091] In an embodiment of the present application, the correlation index acquisition module 22 is specifically further configured to, for each of the multiple factors, calculate the first data change rate based on the data corresponding to the factor, and determine the second time period set from the target historical period based on the first data change rate; the second time period set includes the time periods that meet the first condition; the first condition is that the first data change rate corresponding to a certain time period is greater than or equal to the first threshold; Calculate the second data change rate in the second time period set based on the data corresponding to all factors other than the factor among the multiple factors, and select the first time period set from the second time period set; the first time period set includes the time periods in the second time period set that meet the second condition; the second condition is that the second data change rate corresponding to a certain time period is less than or equal to the second threshold.

[0092] In an embodiment of the present application, the lifespan index data is the data collected based on the first frequency, and the data corresponding to multiple factors respectively is the data collected based on the second frequency; the second frequency is greater than the first frequency; the correlation index acquisition module 22 is specifically further configured to obtain the forward time window; for each collection of lifespan index data, perform the first operation; Among them, the first operation includes: Obtain the collection timestamp of the lifespan index data collected this time, and extract the data within the forward time window from the data corresponding to multiple factors respectively; Aggregate the data corresponding to multiple factors within the forward time window to obtain the aggregated data sequences corresponding to the multiple factors respectively; Align the timestamps of the aggregated data sequences corresponding to the multiple factors with the acquisition timestamps of the life index data.

[0093] In an embodiment of the present application, the non - linear feature acquisition module 23 is specifically configured to construct a first autoregressive model based on the data corresponding to the first factor among the two factors; Construct a second regression model based on the first autoregressive model, the data corresponding to the second factor among the two factors, and the optimal lag order; Calculate the sum of squared model residuals of the first autoregressive model, and calculate the sum of squared model residuals of the second regression model; Determine the causal index between the two factors based on the sum of squared model residuals of the first autoregressive model and the sum of squared model residuals of the second regression model.

[0094] In an embodiment of the present application, the non - linear feature acquisition module 23 is specifically further configured to obtain the data sample sizes of the data corresponding to the two factors; Calculate the variance ratio statistic value based on the sum of squared model residuals of the first autoregressive model, the sum of squared model residuals of the second regression model, the optimal lag order, and the data sample sizes of the data corresponding to the two factors; Determine the causal index between the two factors based on the variance ratio statistic value.

[0095] In an embodiment of the present application, the non - linear feature acquisition module 23 is specifically further configured to construct a factor interaction term based on the causal index and the multi - factor correlation vector, and splice the factor interaction term into the multi - factor correlation vector to obtain the spliced vector; Perform attention allocation on each element in the spliced vector based on the causal index to obtain the target vector; Extract the non - linear change feature between the target vector and the energy storage device life as the target non - linear change feature.

[0096] See Figure 3 , Figure 3 is a schematic block diagram of an electronic device provided in an embodiment of the present application. As Figure 3The electronic device 300 in the present embodiment shown may include: one or more processors 301, one or more input devices 302, one or more output devices 303, and one or more memories 304. The above-mentioned processors 301, input devices 302, output devices 303, and memories 304 communicate with each other through a communication bus 305. The memory 304 is used to store computer programs, and the computer programs include program instructions. The processor 301 is used to execute the program instructions stored in the memory 304. Among them, the processor 301 is configured to call the program instructions to execute the functions of each module in the above system embodiments, for example Figure 2 the functions of the running data acquisition module 21, the correlation index acquisition module 22, the non-linear feature acquisition module 23, and the life prediction module 24 shown.

[0097] It should be understood that in the embodiments of the present application, the so-called processor 301 may be a central processing unit (CPU), and this processor may also be other general-purpose processors, digital signal processors (DSPs), application specific integrated circuits (ASICs), field programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or this processor may also be any conventional processor, etc.

[0098] The input device 302 may include a touchpad, a fingerprint acquisition sensor (for acquiring the fingerprint information and the direction information of the fingerprint of the user), a microphone, etc., and the output device 303 may include a display (such as an LCD), a speaker, etc.

[0099] The memory 304 may include a read-only memory and a random access memory, and provide instructions and data to the processor 301. A part of the memory 304 may also include a non-volatile random access memory. For example, the memory 304 may also store information about the type of energy storage device.

[0100] In specific implementation, the processors 301, input devices 302, and output devices 303 described in the embodiments of the present application may execute the implementation manners described in the embodiments of the energy storage device life prediction method provided in the embodiments of the present application, and may also execute the implementation manner of the electronic device 300 described in the embodiments of the present application, which will not be elaborated here.

[0101] In another embodiment of the present application, a computer-readable storage medium is provided. The computer-readable storage medium stores a computer program, and the computer program includes program instructions. When the program instructions are executed by a processor, all or part of the processes in the method of the above embodiment are implemented. It can also be completed by instructing related hardware through the computer program. The computer program can be stored in a computer-readable storage medium. When the computer program is executed by the processor, the steps of the above method embodiments can be implemented. Among them, the computer program includes computer program code, and the computer program code can be in the form of source code, object code, executable file or some intermediate form, etc. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electrical carrier signal, telecommunication signal, and software distribution medium, etc.

[0102] The computer-readable storage medium can be the internal storage unit of the electronic device in any of the foregoing embodiments, such as the hard disk or memory of the electronic device. The computer-readable storage medium can also be an external storage device of the electronic device, such as a plug-in hard disk equipped on the electronic device, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. Further, the computer-readable storage medium can also include both the internal storage unit and the external storage device of the electronic device. The computer-readable storage medium is used to store the computer program and other programs and data required by the electronic device. The computer-readable storage medium can also be used to temporarily store the data that has been output or will be output.

[0103] Those of ordinary skill in the art can realize that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be implemented by electronic hardware, computer software, or a combination of the two. To clearly illustrate the interchangeability of hardware and software, the components and steps of each example have been generally described according to functions in the above description. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of this application.

[0104] Those skilled in the art can clearly understand that for the convenience and simplicity of description, the specific working processes of the above-described electronic devices and units can refer to the corresponding processes in the foregoing method embodiments and will not be described in detail here.

[0105] In several embodiments provided by the present application, it should be understood that the disclosed electronic devices and methods can be implemented in other ways. For example, the system embodiments described above are merely illustrative. For example, the division of units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the displayed or discussed coupling or direct coupling or communication connection between each other can be an indirect coupling or communication connection through some interfaces or units, or can also be an electrical, mechanical or other form of connection.

[0106] The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or may be distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiments of the present application.

[0107] In addition, each functional unit in various embodiments of the present application can be integrated in a processing unit, or each unit can exist physically alone, or two or more units can be integrated in one unit. The above-mentioned integrated units can be implemented in the form of hardware or in the form of software functional units.

[0108] The above is only the specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present application can easily think of various equivalent modifications or substitutions, and these modifications or substitutions should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A method for predicting the lifespan of an energy storage device, characterized in that, Including: Obtain the operation data corresponding to the target energy storage device within the target time period, where the target energy storage device is the energy storage device to be predicted; Obtain the target correlation index, where the target correlation index is the correlation index between multiple factors and the life of the energy storage device, and the multiple factors include environmental temperature, charge-discharge rate, charge-discharge depth, and usage time, and the target correlation index is calculated based on the historical operation data of multiple energy storage devices; Obtain the target non-linear change feature, where the target non-linear change feature is the non-linear change feature between the multi-factor association vector extracted based on the causal index and the life of the energy storage device, and the causal index is the causal index between every two of the multiple factors, and the multi-factor association vector is constructed based on the data corresponding to the multiple factors; the data corresponding to the multiple factors includes environmental temperature data, charge-discharge rate data, charge-discharge depth data, and usage time data; Input the target correlation index, the target non-linear change feature, and the operation data corresponding to the target energy storage device within the target time period into the energy storage device life prediction model to predict the device life, and obtain the life prediction result.

2. The method for predicting the lifespan of an energy storage device according to claim 1, wherein The target correlation index is determined based on the historical operation data of multiple energy storage devices in the following manner: Extract the data corresponding to the multiple factors respectively based on the historical operation data of multiple energy storage devices; Obtain the life index data of the multiple energy storage devices, where the life index data includes the battery health state, internal battery impedance, charge-discharge cycle times, and battery capacity attenuation rate of the multiple energy storage devices; Align the life index data with the data corresponding to the multiple factors respectively in the time dimension; For each of the multiple factors, select the first time period set from the target historical period based on the data change rate of the multiple factors; calculate the correlation index between this factor and the device life based on the data corresponding to this factor within the first time period set and the life index data; the target historical period is the data generation period corresponding to the historical operation data; Determine the target correlation index based on the correlation index between each factor and the device life respectively.

3. The method for predicting the lifespan of an energy storage device according to claim 2, wherein, For each of the multiple factors, selecting the first time period set from the target historical period based on the data change rate of the multiple factors includes: For each of the multiple factors, calculate the first data change rate based on the data corresponding to this factor, and determine the second time period set from the target historical period based on the first data change rate; the second time period set includes the time periods that meet the first condition; the first condition is that the first data change rate corresponding to a certain time period is greater than or equal to the first threshold; Calculate the second data change rate in the second time period set based on the data corresponding to all factors other than this factor among the multiple factors, and select the first time period set from the second time period set based on the second data change rate; the first time period set includes the time periods in the second time period set that meet the second condition; the second condition is that the second data change rate corresponding to a certain time period is less than or equal to the second threshold.

4. The method for predicting the life of an energy storage device according to claim 2, wherein, The life index data is data collected based on a first frequency, and the data corresponding to each of the multiple factors is data collected based on a second frequency; The second frequency is greater than the first frequency; The aligning the life index data with the data corresponding to each of the multiple factors in the time dimension includes: Obtaining a forward time window; For the life index data collected each time, performing a first operation; Wherein, the first operation includes: Obtaining the collection timestamp of the life index data collected this time, and extracting the data within the forward time window from the data corresponding to each of the multiple factors; Aggregating the data corresponding to each of the multiple factors within the forward time window to obtain an aggregated data sequence corresponding to each of the multiple factors; Aligning the timestamps of the aggregated data sequences corresponding to each of the multiple factors with the collection timestamp of the life index data.

5. The method for predicting the lifespan of an energy storage device according to claim 1, characterized in that, The calculation process of the causal index between every two factors among the multiple factors includes: Constructing a first autoregressive model based on the data corresponding to the first factor among the two factors; Constructing a second regression model based on the first autoregressive model, the data corresponding to the second factor among the two factors, and the optimal lag order; Calculating the sum of squared model residuals of the first autoregressive model, and calculating the sum of squared model residuals of the second regression model; Determining the causal index between the two factors based on the sum of squared model residuals of the first autoregressive model and the sum of squared model residuals of the second regression model.

6. The method for predicting the life of the energy storage device according to claim 5, wherein, The determining the causal index between the two factors based on the sum of squared model residuals of the first autoregressive model and the sum of squared model residuals of the second regression model includes: Obtaining the data sample size of the data corresponding to the two factors; Calculating a variance ratio statistic value based on the sum of squared model residuals of the first autoregressive model, the sum of squared model residuals of the second regression model, the optimal lag order, and the data sample size of the data corresponding to the two factors; Determining the causal index between the two factors based on the variance ratio statistic value.

7. The method for predicting the lifespan of an energy storage device according to claim 1, wherein, The target non-linear change feature is obtained by the following method: Constructing a factor interaction term based on the causal index and the multi-factor correlation vector, and splicing the factor interaction term into the multi-factor correlation vector to obtain a spliced vector; Performing attention allocation on each element in the spliced vector based on the causal index to obtain a target vector; Extracting the non-linear change feature between the target vector and the life of the energy storage device as the target non-linear change feature.

8. A life prediction system for an energy storage device, characterized in that, Including: An operation data acquisition module, configured to acquire operation data corresponding to a target energy storage device within a target time period, where the target energy storage device is an energy storage device to be predicted; A correlation index acquisition module, configured to acquire a target correlation index, where the target correlation index is a correlation index between multiple factors and the life of the energy storage device, the multiple factors include ambient temperature, charge-discharge rate, charge-discharge depth, and usage time, and the target correlation index is calculated based on the historical operation data of multiple energy storage devices; A non-linear feature acquisition module, configured to acquire target non-linear variation features, where the target non-linear variation features are non-linear variation features between a multi-factor correlation vector extracted based on a causal index and the life of an energy storage device, the causal index is the causal index between every two of multiple factors, and the multi-factor correlation vector is constructed based on data corresponding to the multiple factors; the data corresponding to the multiple factors includes environmental temperature data, charge and discharge rate data, charge and discharge depth data, and usage time data; A life prediction module, configured to input the target correlation index, the target non-linear variation features, and the operation data corresponding to a target energy storage device within the target time period into an energy storage device life prediction model to perform device life prediction, and obtain a life prediction result.

9. An electronic device, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images