Method for predicting remaining service life of energy storage battery and related device

Abstract

The invention discloses a method for predicting the remaining service life of an energy storage battery and a related device. The method comprises the following steps of: establishing maximum discharge energy-circulation circle data of a full life cycle of the battery; sG filtering smoothing is carried out, the maximum discharge energy value of the first circle is used as an initial energy value to convert the maximum discharge energy of the single circle of the battery into SOH, standardization processing is carried out, and a time window sliding method is adopted to divide an obtained standardized data set; constructing a time sequence prediction model based on the LSTM, and training the time sequence prediction model based on the LSTM by using the standardized data set X; the method comprises the following steps: collecting discharge energy data of a to-be-tested battery in an actual operation process in an energy storage system in the (1-n) th circle, inputting the discharge energy data into a pre-trained LSTM time sequence prediction model after standardization processing, outputting predicted SOH values of the (n + 1) th circle to the Nth circle, and establishing a complete battery SOH attenuation curve in combination with the data after standardization processing to obtain the remaining service life of the battery.

Description

Technical Field

[0001] The present invention belongs to the technical field of remaining service life prediction of energy storage batteries, and in particular relates to a remaining service life prediction method and related devices for energy storage batteries. Background Art

[0002] In today's rapidly developing energy field, energy storage batteries, especially lithium-ion batteries, have become the core technology for new energy storage due to their high energy density, long cycle life and good electrochemical stability. With their widespread application in electric vehicles, smart grids and portable electronic devices, the remaining service life prediction technology of batteries has become particularly important. The remaining service life (RUL) prediction of batteries can not only prevent potential safety risks, but also has important guiding significance for battery maintenance, replacement strategies and cascade utilization.

[0003] After repeated charge and discharge cycles, the electrochemically active materials inside the energy storage battery will gradually decay, resulting in a decrease in battery capacity and an increase in internal resistance. These changes directly affect the performance and life of the battery. However, the battery aging process is affected by many factors, such as charge and discharge rate, temperature, and depth of discharge, making the aging behavior highly nonlinear and uncertain. In addition, the operating conditions of the battery in actual applications are far more complex and changeable than those under laboratory conditions, which places higher requirements on the accuracy and robustness of RUL prediction.

[0004] At present, the remaining service life prediction methods of energy storage batteries are mainly divided into three categories: model-based prediction methods, data-driven prediction methods, and fusion algorithm-based prediction methods. The model-based method describes the battery aging behavior through mathematical modeling, but it depends on the accuracy of the model parameters, and as the battery aging model parameters change, the prediction accuracy of the late aging period decreases; the data-driven method directly uses historical data to establish a prediction model, but has high data requirements, and the generalization is limited by the diversity of training data; the fusion-based method combines the advantages of the model method and the data-driven method, but the design and implementation are more complex, the computational cost is higher, and more resources may be required to optimize the model performance. Summary of the invention

[0005] The technical problem to be solved by the present invention is to provide a method and related device for predicting the remaining service life of an energy storage battery in view of the deficiencies in the above-mentioned prior art, so as to solve the technical problems of complex feature extraction, high model complexity and high calculation cost.

[0006] The present invention adopts the following technical solutions: A method for predicting the remaining service life of an energy storage battery comprises the following steps: Select lithium-ion battery samples to conduct 100% SOH~80% SOH cycle performance tests to establish the maximum discharge energy-cycle data of the battery's entire life cycle; The obtained maximum discharge energy-cycle number data of the battery's entire life cycle is smoothed by SG filtering, and the maximum discharge energy value of the first cycle is used as the initial energy value to convert the battery's single-cycle maximum discharge energy into SOH and perform standardization to obtain a standardized data set X; The time window sliding method is used to divide the obtained standardized data set X into input sets I With output set O ; Build a time series prediction model based on LSTM and use the standardized data set X to train the time series prediction model based on LSTM; The discharge energy data of the 1st to nth cycles of the battery under test in the actual operation of the energy storage system are collected and input into the pre-trained LSTM time series prediction model after standardization. The predicted SOH values ​​of the n+1th to Nth cycles are output, and a complete battery SOH decay curve is established in combination with the standardized data to obtain the remaining service life of the battery.

[0007] Preferably, the SG filtering and smoothing of the maximum discharge energy-cycle number data of the battery's entire life cycle is specifically performed as follows: Let the window size be n and the window center be , in the window The estimated value of the discharge energy E corresponding to the i-th cycle is fitted by a k-order polynomial. as follows:

[0008] in, represents the discharge energy value of the battery in the i-th cycle, a0, a1, a2, ..., a n Represents the coefficients of the polynomial.

[0009] Preferably, the standardized data set X is specifically:

[0010] in, is the standardized data after transformation.

[0011] Preferably, all SOH data are standardized by normalization and homogenization. The minimum SOH value before conversion is 80% and the maximum value is 100%. The standardized data after conversion is dimensionless. , the size is between 0 and 1.

[0012] Preferably, the converted normalized data is dimensionless for:

[0013] in, is the number of cycles, For the Batteries, For the The battery is in SOH at the time of the second cycle.

[0014] Preferably, the input battery SOH data of the 1st to nth cycles is assumed, the output set is the SOH data of the n+1th cycle, and the input set is I and output set O The calculation is as follows:

[0015]

[0016] in, is the transformed standardized data, Representative Battery The predicted value of the cycle.

[0017] Preferably, the standardized data set X is used to train the LSTM-based time series prediction model as follows: Randomly select 80% of the standardized data set X as the training set and 20% as the validation set. Use the training set to train the LSTM-based time series prediction model. Perform model hyperparameter optimization through multiple cross-validations, freeze all parameters in the LSTM time series prediction model, and output the encapsulated basic model M1. Unfreeze all layers in the basic model M1, train all network layers again with 20% of the battery discharge energy data of the energy storage system to be predicted after being processed in steps S2 and S3, verify with 80% of the corresponding data, and encapsulate the trained basic model M1 again as a new battery remaining service life prediction model M1'; For different types of energy storage batteries, model transfer learning is used for iterative adaptation.

[0018] Preferably, the LSTM-based time series prediction model includes two LSTM layers and a linear fully connected layer. In a single LSTM layer, c represents a neuron, x(t) represents the input layer at time t, and h(t-1) represents the hidden layer at the previous moment.

[0019] Preferably, the iterative adaptation using model transfer learning is specifically as follows: Collect the discharge energy data of the 1st to 300th cycles of the energy storage battery of the model to be predicted, perform SG filtering according to step S2, convert it into SOH data and perform standardization processing; Based on transfer learning, the discharge energy data from the 1st to the 300th cycle is used as the transfer training set to perform adaptive correction and iteration of model parameters to realize the prediction of the remaining service life of this model of battery.

[0020] Preferably, the predicted SOH values ​​of the n+1th to Nth cycles are output until the predicted SOH decays to below 80%, and the standard vector of the battery's entire life cycle is obtained by combining the input data , by de-standardization, the complete discharge energy-cycle number prediction curve is output , extract the first prediction value that reaches 80% , get the remaining battery life .

[0021] In a second aspect, an embodiment of the present invention provides a system for predicting the remaining service life of an energy storage battery, comprising: Data module, select lithium-ion battery samples to conduct 100%SOH~80%SOH cycle performance test, and establish the maximum discharge energy-cycle data set of the battery's entire life cycle; The standardization module performs SG filtering and smoothing on the maximum discharge energy-cycle data of the battery's entire life cycle, uses the maximum discharge energy value of the first cycle as the initial energy value to convert the battery's single-cycle maximum discharge energy into SOH and performs standardization to obtain a standardized data set X; The partitioning module uses the time window sliding method to divide the standardized data set X into input sets I With output set O ; Training module, building a time series prediction model based on LSTM, and using the standardized data set X to train the time series prediction model based on LSTM; The prediction module collects the discharge energy data of the 1st to nth cycles of the battery under test during its actual operation in the energy storage system, inputs the data into the pre-trained LSTM time series prediction model after standardization, outputs the predicted SOH values ​​of the n+1th to Nth cycles, and establishes a complete battery SOH decay curve based on the standardized data to obtain the remaining service life of the battery.

[0022] Preferably, the SG filtering and smoothing of the maximum discharge energy-cycle number data of the battery's entire life cycle is specifically performed as follows: Let the window size be n and the window center be , in the window The estimated value of the discharge energy E corresponding to the i-th cycle is fitted by a k-order polynomial. as follows:

[0023] in, represents the discharge energy value of the battery in the i-th cycle, a0, a1, a2, ..., a n Represents the coefficients of the polynomial.

[0024] Preferably, the standardized data set X is specifically:

[0025] in, is the standardized data after transformation.

[0026] Preferably, all SOH data are standardized by normalization and homogenization. The minimum SOH value before conversion is 80% and the maximum value is 100%. The standardized data after conversion is dimensionless. , the size is between 0 and 1, and the converted standardized data is dimensionless for:

[0027] in, is the number of cycles, For the Batteries, For the The battery is in SOH at the time of the second cycle.

[0028] Preferably, the input battery SOH data of the 1st to nth cycles is assumed, the output set is the SOH data of the n+1th cycle, and the input set is I and output set O The calculation is as follows:

[0029]

[0030] in, is the transformed standardized data, Representative Battery The predicted value of the cycle.

[0031] Preferably, the standardized data set X is used to train the LSTM-based time series prediction model as follows: Randomly select 80% of the standardized data set X as the training set and 20% as the validation set. Use the training set to train the LSTM-based time series prediction model. Perform model hyperparameter optimization through multiple cross-validations, freeze all parameters in the LSTM time series prediction model, and output the encapsulated basic model M1. Unfreeze all layers in the basic model M1, retrain all network layers with 20% of the processed battery discharge energy data of the energy storage system to be predicted, verify with 80% of the corresponding data, and re-encapsulate the trained basic model M1 into a new battery remaining service life prediction model M1'; For different types of energy storage batteries, model transfer learning is used for iterative adaptation.

[0032] Preferably, the LSTM-based time series prediction model includes two LSTM layers and a linear fully connected layer. In a single LSTM layer, c represents a neuron, x(t) represents the input layer at time t, and h(t-1) represents the hidden layer at the previous moment.

[0033] Preferably, the iterative adaptation using model transfer learning is specifically as follows: Collect the discharge energy data of the 1st to 300th cycles of the energy storage battery of the model to be predicted, perform SG filtering, convert it into SOH data and perform standardization processing; Based on transfer learning, the discharge energy data from the 1st to the 300th cycle is used as the transfer training set to perform adaptive correction and iteration of model parameters to realize the prediction of the remaining service life of this model of battery.

[0034] Preferably, the predicted SOH values ​​of the n+1th to Nth cycles are output until the predicted SOH decays to below 80%, and the standard vector of the battery's entire life cycle is obtained by combining the input data , by de-standardization, the complete discharge energy-cycle number prediction curve is output , extract the first prediction value that reaches 80% , get the remaining battery life .

[0035] In a third aspect, a computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above-mentioned method for predicting the remaining service life of an energy storage battery when executing the computer program.

[0036] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium, including a computer program, which, when executed by a processor, implements the steps of the above-mentioned method for predicting the remaining service life of an energy storage battery.

[0037] In a fifth aspect, a chip includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above-mentioned method for predicting the remaining service life of an energy storage battery when executing the computer program.

[0038] In a sixth aspect, an embodiment of the present invention provides an electronic device, including a computer program, which, when executed by the electronic device, implements the steps of the above-mentioned method for predicting the remaining service life of an energy storage battery.

[0039] Compared with the prior art, the present invention has at least the following beneficial effects: A method for predicting the remaining service life of an energy storage battery is based on the SG-LSTM data-driven method. It extracts battery aging characteristics and accurately captures the time series changes and long-term dependencies of the healthy life of the energy storage battery, so as to predict the remaining service life of the energy storage battery through early discharge energy data; Data collection and processing: Through experimental and actual operation data, the relationship between battery discharge energy and SOH is established, and data noise is removed through SG filtering and smoothing, and standardized processing is performed to ensure the validity and operability of the data. LSTM model: LSTM is a neural network suitable for processing time series data and has the ability to remember long-term dependencies. In this method, the discharge energy and SOH value of the battery change over time, so the LSTM model can capture these time-varying dependencies, thereby predicting the future health state (SOH) and remaining service life of the battery. Time series prediction: Through the sliding time window method, the historical discharge data is input into the model, and LSTM can extract rules from past battery behaviors and predict future SOH changes. Combined with actual battery data, the LSTM model can output a predicted SOH curve, and then obtain the remaining service life of the battery. The entire process is designed to accurately obtain the aging characteristics of the battery during discharge and accurately estimate the future health status and remaining life of the battery through a time series prediction model. This can not only improve the efficiency of battery management, but also reduce maintenance costs through early prediction.

[0040] Furthermore, noise suppression: SG filtering can effectively remove random noise from the measured data, which is crucial for battery discharge energy data because these data may be interfered with by a variety of external factors in actual measurements. Trend retention: Unlike the simple moving average method, SG filtering can better retain the local trend of the data through polynomial fitting. This can more accurately reflect the changes in the discharge energy of the battery under different cycle numbers, thereby helping to identify potential degradation of battery performance. Improve data quality: By smoothing the data, a smoother curve shape can be obtained, making data analysis and subsequent model building more robust. Good data quality is the basis for establishing an effective prediction model.

[0041] Furthermore, unify the data range: standardize the SOH data to the range of 0 to 1, so that all data are in the same range, avoiding model instability caused by large differences in data values ​​during training. Improve model training efficiency: Standardized data can accelerate the convergence process of the model, especially when using gradient descent for optimization, a unified data range helps to find the optimal solution faster. Enhance feature importance: Balance feature contribution: Through standardization, the contribution of all features in the model is at the same level, reducing the risk of certain features dominating model learning, and helping to better capture the relative importance of various features on SOH.

[0042] Furthermore, time series prediction: the input set contains historical cycle data, which can capture the trend and pattern of SOH changes over time, and predict the next SOH through these historical data. This time series prediction can more accurately reflect the usage of the battery. The recursiveness of the model: by taking the previous n data as input, the model can learn the recursive relationship between cycles and the dynamic characteristics of SOH changes based on past SOH degradation patterns, thereby improving the prediction accuracy of the SOH of the n+1th cycle and establishing a more accurate prediction model. Avoid overfitting: by dividing the data into input and output sets, training and verification can be effectively performed to reduce the risk of overfitting the model to the training data.

[0043] Furthermore, data set division: the data set is randomly divided into 80% training set and 20% validation set, aiming to ensure that the training and validation processes of the model are independent. This approach helps to evaluate the generalization ability of the model and ensure that the model does not perform poorly on unseen data due to overfitting the training set. Hyperparameter optimization: Through multiple cross-validations, the hyperparameter space can be systematically explored to find the optimal hyperparameter combination (learning rate, number of LSTM network layer nodes, number of model training times), thereby improving the model's predictive ability and stability. Freeze parameters: Freezing all parameters in the model after training means that the structure and weights of the model remain unchanged. This approach ensures that the model's prediction effect is stable in subsequent applications and avoids prediction uncertainty caused by parameter updates; and after freezing the parameters, the model can be quickly applied to new data without the need for training every time, saving computing resources and time.

[0044] Furthermore, capturing long-term dependencies: The LSTM neural network effectively captures long-term dependencies in the input sequence through its special structure (input gate, forget gate, and output gate). The cell state c(t) allows information to be transferred between multiple time steps, which helps the model remember past information. Multi-level feature learning: Using two LSTM layers can extract high-level features. The first LSTM layer captures low-level features, and the second LSTM layer extracts higher-level and complex temporal patterns based on this. Function of the linear fully connected layer: The linear fully connected layer converts the output of the LSTM layer into a predicted value in the target space. It maps the high-dimensional output of the LSTM to a low-dimensional one through a weight matrix, thereby completing the final prediction task.

[0045] Furthermore, based on transfer learning technology, the similar aging characteristics of different types of lithium iron phosphate energy storage batteries are utilized to share knowledge and experience between different energy storage batteries, thereby improving the generalization ability and accuracy of the model in predicting the remaining service life.

[0046] It can be understood that the beneficial effects of the second to sixth aspects mentioned above can be found in the relevant description of the first aspect mentioned above, and will not be repeated here.

[0047] In summary, in the experiment, the present invention conducts a cycle performance test on the battery from 100% SOH to 80% SOH to ensure that the data covers the entire life cycle of the battery, which is helpful to provide a comprehensive performance evaluation; SG filtering is used for data smoothing to reduce the impact of noise on data analysis, better retain local trends of data, and ensure the accuracy and stability of model training; the LSTM model can effectively handle long-term dependencies in time series data and is suitable for the prediction of battery SOH; combining the prediction results and standardized data, it can intuitively display the SOH decay trend of the battery, providing data support and decision-making basis for subsequent battery maintenance and replacement.

[0048] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 It is a schematic diagram of the process of the present invention; Figure 2 The discharge energy-cycle number curve of some batteries after filtering and the filtering effect diagram; Figure 3 It is a schematic diagram of the discharge energy-cycle number curve after segmentation; Figure 4 This is a grid architecture diagram of the LSTM model of the present invention; Figure 5 It is a complete prediction curve of discharge energy-cycle number; Figure 6A schematic diagram of a computer device provided by an embodiment of the present invention; Figure 7 A block diagram of a chip provided according to an embodiment of the present invention; Figure 8 It is a capacity decay curve diagram of MIT block 60 of the present invention; Fig. 9 This is a diagram of the model results of the present invention, where (a) is the result directly predicted by the trained model, and (b) is the prediction result after transfer learning. DETAILED DESCRIPTION

[0050] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0051] In the description of the present invention, it should be understood that the terms “include” and “comprises” indicate the presence of described features, wholes, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components and / or collections thereof.

[0052] It should also be understood that the terms used in the present specification are only for the purpose of describing specific embodiments and are not intended to limit the present invention. As used in the present specification and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.

[0053] It should be further understood that the term "and / or" used in the present specification and the appended claims refers to any combination of one or more of the associated listed items and all possible combinations, and includes these combinations. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in the present invention generally indicates that the associated objects are in an "or" relationship.

[0054] It should be understood that, although the terms first, second, third, etc. may be used to describe preset ranges, etc. in the embodiments of the present invention, these preset ranges should not be limited to these terms. These terms are only used to distinguish preset ranges from each other. For example, without departing from the scope of the embodiments of the present invention, the first preset range may also be referred to as the second preset range, and similarly, the second preset range may also be referred to as the first preset range.

[0055] The word "if" as used herein may be interpreted as "at the time of" or "when" or "in response to determining" or "in response to detecting", depending on the context. Similarly, the phrases "if it is determined" or "if (stated condition or event) is detected" may be interpreted as "when it is determined" or "in response to determining" or "when detecting (stated condition or event)" or "in response to detecting (stated condition or event)", depending on the context.

[0056] Various structural schematic diagrams of the embodiments disclosed in the present invention are shown in the accompanying drawings. These figures are not drawn to scale, and some details are magnified and some details may be omitted for the purpose of clear expression. The shapes of various regions and layers shown in the figures and the relative sizes and positional relationships therebetween are only exemplary, and may deviate in practice due to manufacturing tolerances or technical limitations, and those skilled in the art may additionally design regions / layers with different shapes, sizes, and relative positions according to actual needs.

[0057] The present invention provides a method for predicting the remaining service life of an energy storage battery. The method is based on the energy-cycle number data of the energy storage battery and uses SG-LSTM data-driven model training to input early battery data and output the health state (SOH) of the battery in future cycles. The number of cycles at which the SOH decays to 80% is predicted to be the RUL of the battery. Furthermore, the model can be applied to multiple types of lithium iron phosphate energy storage batteries through transfer learning correction of a small amount of data.

[0058] See also Figure 1 The present invention provides a method for predicting the remaining service life of an energy storage battery, including basic model construction and model migration iteration. Through the basic model construction, the early discharge energy data of the battery is used to predict the future service life of the battery; through the model migration iteration, a small amount of new battery data is input to correct the model, so as to realize the accurate prediction of the service life of the model between different types of energy storage batteries. The specific steps are as follows: S1. Select a certain type of lithium-ion battery sample, conduct 100%SOH~80%SOH cycle performance test according to the national standard GB / T 36276-2023, and establish a data set; Among them, the battery discharge energy-cycle number curve of a certain type of lithium iron phosphate battery in the whole life cycle is obtained. The cycle test method should conduct a cycle charge and discharge test on the battery at 0%SOE-100%SOE according to the rated charge and discharge power. Compared with charging, the battery discharge energy-cycle number, which is more related to the aging of energy storage batteries, is collected to establish a data set.

[0059] S2. Perform SG filtering and smoothing on the discharge energy-cycle number curve in the whole life cycle, use the discharge energy value of the first cycle as the initial energy value to convert the battery discharge energy into SOH and perform standardization, use the time window sliding method to divide the data, and establish a standardized data set X; the time window size is set to 150, and each time window contains continuous standardized discharge SOH data; if the battery life is N, then divide the 1st to 150th, 2nd to 151st, ..., N-150th to N-1st cycles, a total of N-150 time windows; The specific SG filtering and smoothing of the discharge energy-cycle number curve of this model battery in the whole life cycle is as follows: S201, the charging voltage-energy curve of each cycle is smoothed by SG filtering according to the following formula. The battery discharge energy-cycle number curve after filtering is monotonous, continuous and smooth. The partial battery discharge energy-cycle number curve after filtering and the filtering effect are as follows: Figure 2 As shown; SG filtering and smoothing: Specifically: Let the window size be n and the window center be , in the window The k-order polynomial fitting is used internally, as follows:

[0060] in, represents the discharge energy value of the battery in the i-th cycle, represents the estimated value of the discharge energy E corresponding to the i-th cycle, a0, a1, a2, ..., a n Represents the coefficients of the polynomial, and the residual between the estimated value and the actual data point is minimized through the least squares method.

[0061] S202, taking the filtered discharge energy of the first cycle as the standard, calculating the SOH of the battery in each cycle, converting the discharge energy-cycle number curve into a SOH-cycle number curve; performing standardization on the battery's SOH-cycle number, and finally obtaining a standard data set X.

[0062] Standardization includes normalization and homogenization: All SOH data were standardized by normalization and homogenization. The minimum SOH value before conversion was 80% and the maximum value was 100%. The standardized data after conversion was dimensionless and ranged from 0 to 1, as follows: (i = 1,2,3...., j=1,2,3....) Among them, i is the number of cycles and j is the jth battery.

[0063] The standard data set X is as follows:

[0064] S3, the sample data divided by the time window sliding method is used as the input set I, and the data of the next cycle of the time window is used as the output set O. Assuming that the SOH data of the 1st to 150th cycles of the input battery is used, the output set is the SOH of the 151st cycle; The standard data set X is processed by the time window sliding method to obtain the input set I With output set O , specifically, As input set I ,by As output set O , represents the predicted value of the jth battery at the i+150th cycle, the input set I and output set O The calculation is as follows:

[0065]

[0066] See also Figure 3 , which is a schematic diagram of the discharge energy-cycle number curve after segmentation.

[0067] S4, the input set obtained in step S3 I As input, the output set O As output, the mapping relationship is trained based on the LSTM time series prediction model; See also Figure 4 The entire LSTM time series prediction model includes two LSTM layers and a linear fully connected layer. In a single LSTM unit, c represents a neuron, and the superscript (t) represents the time t; x(t) represents the input layer at time t, and h(t-1) represents the hidden layer at the previous time, i.e., time t-1.

[0068] Randomly select 80% of the standardized data set X as the training set and 20% as the validation set. Perform multiple cross-validations to optimize hyperparameters such as model loss, number of trainings, learning rate, and hidden layer parameters. Freeze all parameters in the model and output the encapsulated basic model M1.

[0069] For different types of lithium iron phosphate energy storage batteries, the present invention proposes to use model transfer learning for iterative adaptation (Transfer-Step), and the specific steps are as follows: Collect the discharge energy data of the 1st to 300th cycles of the energy storage battery of the model to be predicted, perform SG filtering according to step S2, convert it into SOH data and perform standardization processing; Based on transfer learning, the discharge energy data of the 1st to 300th cycles are used as the transfer training set to perform adaptive correction iteration of model parameters, so as to achieve accurate prediction of the remaining service life of the same step S5 function for this model of battery.

[0070] All layers in the basic model M1 are unfrozen, and all network layers are trained again with 20% of the data. 80% of the data is used as verification, and the trained basic model M1 is packaged again as a new battery remaining service life prediction model M1'.

[0071] When data from other types of batteries are input, the model transfer learning steps are repeated. By accumulating a large amount of battery charging data and utilizing the similar aging characteristics of different types of lithium iron phosphate energy storage batteries, knowledge and experience are shared between different energy storage batteries to achieve adaptive updates of the SG-LSTM model, enabling the model to achieve accurate remaining service life predictions on different types of energy storage batteries.

[0072] S5. Collect the discharge energy data of the 1st to 150th cycles of the battery model in the actual operation of the energy storage system, and input the data into the LSTM time series prediction model pre-trained in step S4 after standardization in step S2, and output the predicted SOH values ​​of the 151st to Nth cycles, until the predicted SOH decays to below 80%. Combined with the input, establish a complete battery SOH decay curve, extract the cycle number at which the SOH of the energy storage battery decays to 80% for the first time, and obtain the RUL of the battery.

[0073] Take the discharge energy data of the 1st to 150th cycle of this type of battery in actual operation, normalize it with S2 and input it into the trained battery remaining service life prediction model M, output the battery's future SOH prediction value, and combine the input data to obtain the standard vector of the battery's entire life cycle And by de-standardization, the complete prediction curve of discharge energy-cycle number is output ,like Figure 5 As shown; Take the first prediction value that reaches 80% , then the RUL of the battery is .

[0074] In another embodiment of the present invention, a system for predicting the remaining useful life of an energy storage battery is provided. The system can be used to implement the above-mentioned method for predicting the remaining useful life of an energy storage battery. Specifically, the system for predicting the remaining useful life of an energy storage battery includes a data module, a standardization module, a division module, a training module and a prediction module.

[0075] Among them, the data module selects lithium-ion battery samples to conduct 100%SOH~80%SOH cycle performance tests to establish the maximum discharge energy-cycle data set of the battery's entire life cycle; The standardization module performs SG filtering and smoothing on the maximum discharge energy-cycle data of the battery's entire life cycle, uses the maximum discharge energy value of the first cycle as the initial energy value to convert the battery's single-cycle maximum discharge energy into SOH and performs standardization to obtain a standardized data set X; The partitioning module uses the time window sliding method to divide the standardized data set X into input sets I With output set O ; Training module, building a time series prediction model based on LSTM, and using the standardized data set X to train the time series prediction model based on LSTM; The prediction module collects the discharge energy data of the 1st to nth cycles of the battery under test during its actual operation in the energy storage system, inputs the data into the pre-trained LSTM time series prediction model after standardization, outputs the predicted SOH values ​​of the n+1th to Nth cycles, and establishes a complete battery SOH decay curve based on the standardized data to obtain the remaining service life of the battery.

[0076] In another embodiment of the present invention, a terminal device is provided, which includes a processor and a memory, wherein the memory is used to store a computer program, the computer program includes program instructions, and the processor is used to execute the program instructions stored in the computer storage medium. The processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. It is the computing core and control core of the terminal, which is suitable for implementing one or more instructions, and is specifically suitable for loading and executing one or more instructions to implement the corresponding method flow or corresponding function; the processor described in the embodiment of the present invention can be used for the operation of the remaining service life prediction method of the energy storage battery, including: Lithium-ion battery samples were selected for 100% SOH~80% SOH cycle performance tests to establish the maximum discharge energy-cycle data of the battery's entire life cycle; the obtained maximum discharge energy-cycle data of the battery's entire life cycle was smoothed by SG filtering, and the maximum discharge energy value of the first cycle was used as the initial energy value to convert the battery's single-cycle maximum discharge energy into SOH and perform normalization to obtain a standardized data set X; the obtained standardized data set X was divided into an input set using a time window sliding method I With output setO ; Build a time series prediction model based on LSTM, and use the standardized data set X to train the time series prediction model based on LSTM; collect the discharge energy data of the 1st to nth cycles of the battery under test in the actual operation of the energy storage system, input it into the pre-trained LSTM time series prediction model after standardization, output the predicted SOH value of the n+1th to Nth cycles, and establish a complete battery SOH decay curve based on the standardized data to obtain the remaining service life of the battery.

[0077] See also Figure 6 , the terminal device is a computer device. The computer device 60 of this embodiment includes: a processor 61, a memory 62, and a computer program 63 stored in the memory 62 and executable on the processor 61. When the computer program 63 is executed by the processor 61, the method for calculating the fluid composition in the reservoir transformation wellbore in the embodiment is implemented. To avoid repetition, it is not described one by one here. Alternatively, when the computer program 63 is executed by the processor 61, the functions of each model / unit in the remaining service life prediction system of the energy storage battery in the embodiment are implemented. To avoid repetition, it is not described one by one here.

[0078] The computer device 60 may be a computing device such as a desktop computer, a notebook, a PDA, or a cloud server. The computer device 60 may include, but is not limited to, a processor 61 and a memory 62. Those skilled in the art will appreciate that Figure 6 This is only an example of the computer device 60 and does not constitute a limitation of the computer device 60. It may include more or fewer components than shown in the figure, or a combination of certain components, or different components. For example, the computer device may also include input and output devices, network access devices, buses, etc.

[0079] The processor 61 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc.

[0080] The memory 62 may be an internal storage unit of the computer device 60, such as a hard disk or memory of the computer device 60. The memory 62 may also be an external storage device of the computer device 60, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc., equipped on the computer device 60.

[0081] Furthermore, the memory 62 may include both an internal storage unit of the computer device 60 and an external storage device. The memory 62 is used to store computer programs and other programs and data required by the computer device. The memory 62 may also be used to temporarily store data that has been output or is to be output.

[0082] See also Figure 7 , the terminal device is a chip, and the chip 600 of this embodiment includes a processor 622, which may be one or more, and a memory 632 for storing a computer program executable by the processor 622. The computer program stored in the memory 632 may include one or more modules, each corresponding to a set of instructions. In addition, the processor 622 may be configured to execute the computer program to execute the above-mentioned method for predicting the remaining service life of the energy storage battery.

[0083] In addition, the chip 600 may further include a power supply component 626 and a communication component 650, wherein the power supply component 626 may be configured to perform power management of the chip 600, and the communication component 650 may be configured to implement communication, for example, wired or wireless communication, of the chip 600. In addition, the chip 600 may further include an input / output interface 658. The chip 600 may operate based on an operating system stored in the memory 632.

[0084] In another embodiment of the present invention, the present invention also provides a storage medium, specifically a computer-readable storage medium, which is a memory device in a terminal device for storing programs and data. It can be understood that the computer-readable storage medium here can include both the built-in storage medium in the terminal device and the extended storage medium supported by the terminal device. The computer-readable storage medium provides a storage space, which stores the operating system of the terminal. In addition, one or more instructions suitable for being loaded and executed by the processor are also stored in the storage space, and these instructions can be one or more computer programs. It should be noted that the computer-readable storage medium here can be a high-speed RAM memory, or a non-volatile memory (Non-VolatileMemory), such as at least one disk memory.

[0085] The processor may load and execute one or more instructions stored in a computer-readable storage medium to implement the corresponding steps of the method for predicting the remaining service life of an energy storage battery in the above embodiment; the processor may load and execute the following steps: Lithium-ion battery samples were selected for 100% SOH~80% SOH cycle performance tests to establish the maximum discharge energy-cycle data of the battery's entire life cycle; the obtained maximum discharge energy-cycle data of the battery's entire life cycle was smoothed by SG filtering, and the maximum discharge energy value of the first cycle was used as the initial energy value to convert the battery's single-cycle maximum discharge energy into SOH and perform normalization to obtain a standardized data set X; the obtained standardized data set X was divided into an input set using a time window sliding method I With output set O ; Build a time series prediction model based on LSTM, and use the standardized data set X to train the time series prediction model based on LSTM; collect the discharge energy data of the 1st to nth cycles of the battery under test in the actual operation of the energy storage system, input it into the pre-trained LSTM time series prediction model after standardization, output the predicted SOH value of the n+1th to Nth cycles, and establish a complete battery SOH decay curve based on the standardized data to obtain the remaining service life of the battery.

[0086] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of the embodiments. The components of the embodiments of the present invention described and shown in the drawings here can usually be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the claimed invention, but merely represents selected embodiments of the present invention. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0087] The proposed method and related device for predicting the remaining service life of energy storage batteries have been verified on the open source MIT-Stanford commercial lithium-ion battery dataset. The dataset includes 124 commercial LiFePO4 / graphite batteries that were cycled under different fast charging conditions but the same discharge conditions (4C constant current).

[0088] Fast charging strategies include multi-stage constant current charging and constant voltage charging. The format of multi-stage constant current charging is "C1(Q1)-C2", where C1 and C2 are the charging current rates of the first and second steps respectively, and Q1 is the SOC point of current switching.

[0089] The second step ends when the battery is charged to 80% SOC, after which the battery is charged with 1C current, constant current and constant voltage mode. All batteries are cycled in a room with an ambient temperature of 30°C until their capacity drops below 80% of the nominal value. The nominal capacity of the battery is 1.1Ah, and the upper and lower voltage limits are 3.6V and 2.0V respectively.

[0090] See also Figure 8 and Fig. 9 , 60 batteries were selected, with a lifespan of 480 to 1200 cycles. The discharge data of the battery with 100% SOH to 80% SOH were selected to establish a discharge energy dataset. Subsequently, the discharge energy dataset was trained through steps S2, S3, and S4 to obtain an LSTM time series model M, which was used to predict the remaining service life of the battery.

[0091] On the test set, the average root mean square error (RMSE) of model M is 5.3%. For the prediction of new batteries, the RMSE is 1.6%; after transfer learning, the prediction RMSE is further reduced to 1.0%.

[0092] In summary, the present invention provides a method and a related device for predicting the remaining service life of an energy storage battery, which have the following significant effects: 1. High-precision prediction By establishing a complete battery discharge energy dataset and applying the LSTM time series prediction model, the remaining service life of the battery can be accurately predicted, thus providing reliable data support for battery management.

[0093] 2. Full life cycle assessment The method covers the entire life cycle of lithium-ion batteries from 100% SOH to 80% SOH, ensuring the integrity and representativeness of the data, thereby improving the reliability of the prediction results.

[0094] 3. Noise suppression The SG filtering technology is used to smooth the data, which effectively reduces the noise interference in the experimental data and improves the accuracy of subsequent analysis and model training.

[0095] 4. Standardization and consistency By standardizing the discharge energy data, the performance differences between different batteries are eliminated, the applicability and generalization ability of the model are enhanced, and the prediction results are more consistent.

[0096] 5. Dynamic responsiveness By using the time window sliding method, the model can effectively capture the dynamic characteristics of battery performance changing over time, improve the timeliness of the prediction, and adapt to changes in the battery usage environment.

[0097] 6. Strong practicality By applying actual operating data, the prediction results are closer to actual usage, which enhances the practicality of the method and provides a scientific basis for subsequent battery maintenance and replacement.

[0098] 7. Visualize SOH decay curve By constructing the battery SOH decay curve, the changing trend of battery performance can be intuitively displayed, helping users better understand the health status and usage of the battery.

[0099] 8. Promote intelligent management The implementation of the present invention provides important technical support for an intelligent battery management system, which can improve the efficiency and safety of battery use and extend the overall service life of the battery.

[0100] 9. Wide application prospects It is suitable for a variety of energy storage scenarios, such as electric vehicles, renewable energy storage, etc., and has broad market application prospects.

[0101] In summary, the present invention effectively improves the ability to predict the remaining service life of energy storage batteries and provides an innovative solution for battery management and its application fields.

[0102] The technicians in the relevant field can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example for illustration. In practical applications, the above-mentioned function allocation can be completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiment can be integrated in a processing unit, or each unit can exist physically separately, or two or more units can be integrated in one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of this application. The specific working process of the units and modules in the above-mentioned system can refer to the corresponding process in the aforementioned method embodiment, which will not be repeated here.

[0103] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0104] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed in the present invention can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0105] In the embodiments provided by the present invention, it should be understood that the disclosed devices / terminals and methods can be implemented in other ways. For example, the device / terminal embodiments described above are only schematic. For example, the division of the modules or units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

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

[0107] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0108] If the integrated module / unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present invention implements all or part of the processes in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and the computer program can implement the steps of the above-mentioned various method embodiments when executed by the processor. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, U disk, mobile hard disk, disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electric carrier signal, telecommunication signal and software distribution medium, etc. It should be noted that the content contained in the computer-readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media do not include electric carrier signals and telecommunication signals.

[0109] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices, and computer program products according to the embodiments of the present application. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of the processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0110] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0111] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0112] The above contents are only for explaining the technical idea of ​​the present invention and cannot be used to limit the protection scope of the present invention. Any changes made on the basis of the technical solution in accordance with the technical idea proposed by the present invention shall fall within the protection scope of the claims of the present invention.

Claims

1. A method for predicting the remaining service life of an energy storage battery, characterized in that: The following steps are involved: Select lithium-ion battery samples to conduct 100% SOH~80% SOH cycle performance tests to establish the maximum discharge energy-cycle data of the battery's entire life cycle; The maximum discharge energy-cycle number data of the battery's entire life cycle is smoothed by SG filtering. The maximum discharge energy value of the first cycle is used as the initial energy value to convert the battery's single-cycle maximum discharge energy into SOH and perform standardization to obtain a standardized data set X. The time window sliding method is used to divide the standardized data set X into input sets I With output set O ; Build a time series prediction model based on LSTM and use the standardized data set X to train the time series prediction model based on LSTM; The discharge energy data of the 1st to nth cycles of the battery under test in the actual operation of the energy storage system are collected and input into the pre-trained LSTM time series prediction model after standardization. The predicted SOH values ​​of the n+1th to Nth cycles are output, and a complete battery SOH decay curve is established in combination with the standardized data to obtain the remaining service life of the battery.

2. The method for predicting the remaining service life of an energy storage battery according to claim 1, characterized in that: The SG filtering and smoothing of the maximum discharge energy-cycle number data of the battery's entire life cycle is as follows: Let the window size be n and the window center be , in the window The estimated value of the discharge energy E corresponding to the i-th cycle is fitted by a k-order polynomial. as follows: in, represents the discharge energy value of the battery in the i-th cycle, a0, a1, a2, ..., a n Represents the coefficients of the polynomial.

3. The method for predicting the remaining service life of an energy storage battery according to claim 1, characterized in that: The standardized data set X is specifically: in, is the standardized data after transformation.

4. The method for predicting the remaining service life of an energy storage battery according to claim 3, characterized in that: All SOH data were standardized by normalization. The minimum SOH value before conversion was 80% and the maximum value was 100%. The standardized data after conversion was dimensionless. , the size is between 0 and 1.

5. The method for predicting the remaining service life of an energy storage battery according to claim 4, characterized in that: The converted standardized data is dimensionless for: in, is the number of cycles, For the Batteries, For the The battery is in SOH at the time of the second cycle.

6. The method for predicting the remaining service life of an energy storage battery according to claim 1, characterized in that Input the battery's 1st to nth cycle SOH data, and the output set is the SOH data of the n+1th cycle. I and output set O The calculation is as follows: in, is the transformed standardized data, Representative Battery The predicted value of the cycle.

7. The method for predicting the remaining service life of an energy storage battery according to claim 1, characterized in that: The specific training of the LSTM-based time series prediction model using the standardized dataset X is as follows: Randomly select 80% of the standardized data set X as the training set and 20% as the validation set. Use the training set to train the LSTM-based time series prediction model. Perform model hyperparameter optimization through multiple cross-validations, freeze all parameters in the LSTM time series prediction model, and output the encapsulated basic model M1. Unfreeze all layers in the basic model M1, retrain all network layers with 20% of the processed battery discharge energy data of the energy storage system to be predicted, verify with 80% of the corresponding data, and re-encapsulate the trained basic model M1 into a new battery remaining service life prediction model M1'; For different types of energy storage batteries, model transfer learning is used for iterative adaptation.

8. The method for predicting the remaining service life of an energy storage battery according to claim 7, characterized in that: The LSTM-based time series prediction model includes two LSTM layers and a linear fully connected layer. In a single LSTM layer, c represents a neuron, x(t) represents the input layer at time t, and h(t-1) represents the hidden layer at the previous moment.

9. The method for predicting the remaining service life of an energy storage battery according to claim 7, characterized in that: The specific steps of iterative adaptation using model transfer learning are as follows: Collect the discharge energy data of the 1st to 300th cycles of the energy storage battery of the model to be predicted, perform SG filtering according to step S2, convert it into SOH data and perform standardization processing; Based on transfer learning, the discharge energy data from the 1st to the 300th cycle is used as the transfer training set to perform adaptive correction and iteration of model parameters to realize the prediction of the remaining service life of this model of battery.

10. The method for predicting the remaining service life of an energy storage battery according to claim 1, characterized in that: Output the predicted SOH value from the n+1th to the Nth cycle until the predicted SOH decays to below 80%. Combined with the input data, the standard vector of the battery's entire life cycle is obtained. , by de-standardization, the complete discharge energy-cycle number prediction curve is output , extract the first prediction value that reaches 80% , get the remaining battery life .

11. A system for predicting the remaining service life of an energy storage battery, characterized in that: include: Data module, select lithium-ion battery samples to conduct 100%SOH~80%SOH cycle performance test, and establish the maximum discharge energy-cycle data set of the battery's entire life cycle; The standardization module performs SG filtering and smoothing on the maximum discharge energy-cycle data of the battery's entire life cycle, uses the maximum discharge energy value of the first cycle as the initial energy value to convert the battery's single-cycle maximum discharge energy into SOH and performs standardization to obtain a standardized data set X; The partitioning module uses the time window sliding method to divide the standardized data set X into input sets I With output set O ; Training module, building a time series prediction model based on LSTM, and using the standardized data set X to train the time series prediction model based on LSTM; The prediction module collects the discharge energy data of the 1st to nth cycles of the battery under test during its actual operation in the energy storage system, inputs the data into the pre-trained LSTM time series prediction model after standardization, outputs the predicted SOH values ​​of the n+1th to Nth cycles, and establishes a complete battery SOH decay curve based on the standardized data to obtain the remaining service life of the battery.

12. The energy storage battery remaining service life prediction system according to claim 11, characterized in that: The SG filtering and smoothing of the maximum discharge energy-cycle number data of the battery's entire life cycle is as follows: Let the window size be n and the window center be , in the window The estimated value of the discharge energy E corresponding to the i-th cycle is fitted by a k-order polynomial. as follows: in, represents the discharge energy value of the battery in the i-th cycle, a0, a1, a2, ..., a n Represents the coefficients of the polynomial.

13. The energy storage battery remaining service life prediction system according to claim 11, characterized in that: The standardized data set X is specifically: in, is the standardized data after transformation.

14. The energy storage battery remaining service life prediction system according to claim 13, characterized in that: All SOH data were standardized by normalization. The minimum SOH value before conversion was 80% and the maximum value was 100%. The standardized data after conversion was dimensionless. , the size is between 0 and 1, and the converted standardized data is dimensionless for: in, is the number of cycles, For the Batteries, For the The battery is in SOH at the time of the second cycle.

15. The energy storage battery remaining service life prediction system according to claim 11, characterized in that Input the battery's 1st to nth cycle SOH data, and the output set is the SOH data of the n+1th cycle. I and output set O The calculation is as follows: in, is the transformed standardized data, Representative Battery The predicted value of the cycle.

16. The energy storage battery remaining service life prediction system according to claim 11, characterized in that: The specific training of the LSTM-based time series prediction model using the standardized dataset X is as follows: Randomly select 80% of the standardized data set X as the training set and 20% as the validation set. Use the training set to train the LSTM-based time series prediction model. Perform model hyperparameter optimization through multiple cross-validations, freeze all parameters in the LSTM time series prediction model, and output the encapsulated basic model M1. Unfreeze all layers in the basic model M1, retrain all network layers with 20% of the processed battery discharge energy data of the energy storage system to be predicted, verify with 80% of the corresponding data, and re-encapsulate the trained basic model M1 into a new battery remaining service life prediction model M1'; For different types of energy storage batteries, model transfer learning is used for iterative adaptation.

17. The energy storage battery remaining service life prediction system according to claim 16, characterized in that: The LSTM-based time series prediction model includes two LSTM layers and a linear fully connected layer. In a single LSTM layer, c represents a neuron, x(t) represents the input layer at time t, and h(t-1) represents the hidden layer at the previous moment.

18. The energy storage battery remaining service life prediction system according to claim 16, characterized in that: The specific steps of iterative adaptation using model transfer learning are as follows: Collect the discharge energy data of the 1st to 300th cycles of the energy storage battery of the model to be predicted, perform SG filtering, convert it into SOH data and perform standardization processing; Based on transfer learning, the discharge energy data from the 1st to the 300th cycle is used as the transfer training set to perform adaptive correction and iteration of model parameters to realize the prediction of the remaining service life of this model of battery.

19. The energy storage battery remaining service life prediction system according to claim 11, characterized in that: Output the predicted SOH value from the n+1th to the Nth cycle until the predicted SOH decays to below 80%. Combined with the input data, the standard vector of the battery's entire life cycle is obtained. , by de-standardization, the complete discharge energy-cycle number prediction curve is output , extract the first prediction value that reaches 80% , get the remaining battery life .

20. A computer-readable storage medium storing one or more programs, characterized in that: The one or more programs include instructions, which, when executed by a computing device, cause the computing device to perform the method of any one of claims 1 to 10.

21. A computing device, characterized in that include: One or more processors, a memory and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, and the one or more programs include steps for executing the method according to any one of claims 1 to 10.

Images