A method, device, equipment and medium for predicting remaining life of lithium-ion battery

Abstract

The present invention discloses a method, device, equipment and medium for predicting the remaining life of a lithium-ion battery. The method comprises the following steps: obtaining the battery capacity of a target lithium-ion battery in multiple rounds of historical charge and discharge cycles; generating input samples and sample labels corresponding to each round of historical charge and discharge cycles according to the battery capacity; constructing a capacity prediction model based on SENet, an autoencoder and a GPU; training the capacity prediction model through input samples and sample labels; constructing input samples corresponding to the first round of future charge and discharge cycles and inputting the trained capacity prediction model, iteratively obtaining the battery capacity of multiple rounds of future charge and discharge cycles; determining the remaining life of the target lithium-ion battery according to the battery capacity of multiple rounds of future charge and discharge cycles; the present invention can accurately predict the remaining life of a lithium-ion battery.

Description

Technical Field

[0001] The invention relates to a method, device, equipment and medium for predicting the remaining life of a lithium ion battery, and belongs to the technical field of lithium ion batteries. Background Art

[0002] A large number of lithium-ion batteries are integrated into energy storage power stations. The safety and reliability of these batteries have always been a very important issue during their use. During the long-term charge and discharge cycle of lithium-ion batteries in use, irreversible physical and chemical processes will occur inside them, including the loss of active materials such as internal active lithium. The loss of active lithium and harsh external working conditions (such as high or low temperature environments) will seriously affect the internal electrochemical reactions of the battery, resulting in increased internal resistance and continuous decrease in capacity. This phenomenon is called battery degradation. Battery degradation will eventually lead to a series of safety and reliability issues such as reduced battery performance and shortened RUL, and may even lead to reduced performance of electrical equipment or system failures, causing fire and explosion accidents.

[0003] Due to the complex electrochemical characteristics and harsh working conditions, the relevant parameters of lithium batteries have highly dynamic and nonlinear characteristics. In order to ensure that lithium batteries for energy storage have the required working performance, the battery status in use must be accurately estimated. In order to avoid sudden battery failure and improve battery utilization, it is necessary to predict the RUL (remaining service life) of the battery in advance. RUL prediction aims to use historical and current data to predict future battery degradation behavior and provide early warning before the battery fails. It is generally believed that when the capacity of a lithium battery drops to 80% of the initial value, it reaches the end of its service life. If it continues to be used at this time, a series of accidents may occur. Therefore, accurate prediction of the battery's RUL can evaluate the reliability of the battery and is an important part of fault prediction and health management. Predicting the RUL based on the operating status data of the energy storage battery and then formulating a targeted maintenance plan can save a lot of manpower and financial costs, while ensuring safety during operation and reducing the risk of accidents. Summary of the invention

[0004] The purpose of the present invention is to overcome the deficiencies in the prior art and provide a method, device, equipment and medium for predicting the remaining life of a lithium-ion battery to solve the technical problem of predicting the remaining life of a lithium-ion battery, thereby ensuring the safety of the battery during operation and reducing the risk of accidents.

[0005] To achieve the above object, the present invention is implemented by adopting the following technical solutions:

[0006] In a first aspect, the present invention provides a method for predicting the remaining life of a lithium-ion battery, comprising:

[0007] Obtain the battery capacity of the target lithium-ion battery in multiple rounds of historical charge and discharge cycles;

[0008] Generate input samples and sample labels corresponding to each round of historical charge and discharge cycles based on battery capacity;

[0009] Build a capacity prediction model based on SENet, autoencoder and GRU;

[0010] The capacity prediction model is trained by inputting samples and sample labels;

[0011] Construct the input samples corresponding to the first round of future charge and discharge cycles and input them into the trained capacity prediction model, iteratively obtaining the battery capacity of multiple rounds of future charge and discharge cycles;

[0012] Determine the remaining life of the target lithium-ion battery based on the battery capacity over multiple future charge and discharge cycles.

[0013] Optionally, generating input samples and sample labels corresponding to each round of historical charge and discharge cycles according to the battery capacity includes:

[0014] The battery capacity Q of the kth round of historical charge and discharge cycle k , as the sample label corresponding to the kth round of historical charge and discharge cycles;

[0015] The battery capacity Q of the 1st to (k-1)th round of historical charge and discharge cycles 1～(k-1) , as the input sample corresponding to the kth round of historical charge and discharge cycles, Q 1～(k-1) = {Q1,Q2,…,Q k-1};

[0016] Among them, 2≤k≤K, K is the total number of historical charge and discharge cycles.

[0017] Optionally, the capacity prediction model includes a first SENet model, a second SENet model, a first autoencoder, a second autoencoder, a GRU model, and a Liner linear layer;

[0018] The input ends of the first SENet model and the first autoencoder are both used as the input ends of the capacity prediction model, the output ends of the first SENet model and the first autoencoder are added and connected to the input end of the second autoencoder, the output end of the first autoencoder is connected to the input end of the second autoencoder, the output ends of the second autoencoder and the second SENet model are added and connected to the input end of the GRU model, the output end of the GRU model is connected to the input end of the Liner linear layer, and the output end of the Liner linear layer serves as the output end of the capacity prediction model.

[0019] Optionally, the training of the capacity prediction model by inputting samples and sample labels includes:

[0020] Repeat the following steps until the capacity prediction model training is complete:

[0021] Feeding the input samples into the capacity prediction model;

[0022] Perform local feature extraction and global feature extraction on the input sample through the first SENet model and the first autoencoder respectively;

[0023] The local features and global features extracted by the first SENet model and the first autoencoder are fused, and local features are extracted from the fusion results by the second SENet model;

[0024] Performing global vital sign extraction on the global features extracted by the first autoencoder through the second autoencoder;

[0025] The local features and global features extracted by the second SENet model and the second autoencoder are fused and predicted by the GRU model;

[0026] Output the prediction results through the Liner linear layer;

[0027] Calculate the model loss according to the prediction results and sample labels, and update the weight parameters of the first SENet model, the second SENet model, the first autoencoder, the second autoencoder, and the GRU model by back propagation of gradient descent according to the model loss;

[0028] The completion of the capacity prediction model training includes the number of repeated executions being greater than a preset maximum number or the model loss being less than a preset loss threshold.

[0029] Optionally, the input sample corresponding to the first round of future charge-discharge cycles is: the battery capacity Q of the 1st to Kth rounds of historical charge-discharge cycles 1～K ;

[0030] The iterative acquisition of battery capacity for multiple rounds of future charge and discharge cycles includes:

[0031] Initialize t=1 and repeat the following operations:

[0032] Judge Q t ≤Q0 is true, Q0 is the preset battery capacity threshold:

[0033] If true, then output the battery capacity Q of the 1st to tth future charge and discharge cycles 1～t ;

[0034] If not, the input sample corresponding to the tth round of future charge and discharge cycle is input into the trained capacity prediction model to obtain the battery capacity Q after the t+1th round of future charge and discharge training. t+1 ;

[0035] Construct the input sample corresponding to the t+1th round of future charge and discharge cycle: the battery capacity of the 1st to Kth round of historical charge and discharge cycles and the 1st to tth round of future charge and discharge cycles {Q 1～K ,Q 1～t};

[0036] Let t=t+1.

[0037] Optionally, determining the remaining life of the target lithium-ion battery includes:

[0038] According to the output of the battery capacity Q of the 1st to tth round of future charge and discharge cycles 1～t , determine the remaining life of the target lithium-ion battery as t rounds of charge and discharge cycles.

[0039] In a second aspect, the present invention provides a device for predicting the remaining life of a lithium-ion battery, the device comprising:

[0040] A historical data module is used to obtain the battery capacity of the target lithium-ion battery in multiple rounds of historical charge and discharge cycles;

[0041] A sample generation module is used to generate input samples and sample labels corresponding to each round of historical charge and discharge cycles according to the battery capacity;

[0042] Model building module, used to build capacity prediction models based on SENet, autoencoder and GRU;

[0043] A model training module is used to train the capacity prediction model by inputting samples and sample labels;

[0044] The future data module is used to construct the input samples corresponding to the first round of future charge and discharge cycles and input them into the trained capacity prediction model to iteratively obtain the battery capacity of multiple rounds of future charge and discharge cycles;

[0045] The life statistics module is used to determine the remaining life of the target lithium-ion battery based on the battery capacity of multiple rounds of future charge and discharge cycles.

[0046] Optionally, the capacity prediction model includes a first SENet model, a second SENet model, a first autoencoder, a second autoencoder, a GRU model, and a Liner linear layer;

[0047] The input ends of the first SENet model and the first autoencoder are both used as the input ends of the capacity prediction model, the output ends of the first SENet model and the first autoencoder are added and connected to the input end of the second autoencoder, the output end of the first autoencoder is connected to the input end of the second autoencoder, the output ends of the second autoencoder and the second SENet model are added and connected to the input end of the GRU model, the output end of the GRU model is connected to the input end of the Liner linear layer, and the output end of the Liner linear layer serves as the output end of the capacity prediction model.

[0048] In a third aspect, the present invention provides an electronic device, including a processor and a storage medium;

[0049] The storage medium is used to store instructions;

[0050] The processor is used to operate according to the instructions to execute the steps of the above method.

[0051] In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, which implements the steps of the above method when executed by a processor.

[0052] Compared with the prior art, the present invention has the following beneficial effects:

[0053] The present invention provides a method, device, equipment and medium for predicting the remaining life of a lithium-ion battery. The method does not need to analyze the complex characteristics inside the battery to establish a mechanism model, but is based on data-driven, digging out the evolution law of the battery capacity of the lithium-ion battery in operation, so as to build a data model to predict the remaining service life of the battery; it solves a series of shortcomings of the existing technology based on physical model prediction, such as high complexity, difficulty in solving, unsuitable for drastic changes in working conditions, and many input parameters; at the same time, the method of the present invention integrates multiple algorithms to make up for a series of problems existing in a single model. At the same time, the coupling of SENet and autoencoder makes up for the short board of the difficulty in extracting features of lithium-ion battery data under different conditions and different batches, and gives full play to the advantages of different models to achieve higher prediction accuracy, stability and generalization ability; at the same time, the use of GRU can save computing resources to a large extent, improve computing efficiency, and meet the real-time requirements of energy storage battery life prediction; the device, equipment and medium of the present invention, using the method of the present invention, can achieve the same technical effect. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 is a flow chart of a method for predicting the remaining life of a lithium-ion battery provided in Example 1 of the present invention;

[0055] Figure 2is a schematic diagram of the structure of a capacity prediction model provided in the first embodiment of the present invention;

[0056] Figure 3 is a training flow chart of a capacity prediction model provided in the first embodiment of the present invention;

[0057] Figure 4 is a flow chart of iteratively obtaining the battery capacity of multiple rounds of future charge and discharge cycles provided by the first embodiment of the present invention;

[0058] Figure 5 It is a schematic diagram of comparing the battery capacities of a certain type of battery provided in Example 1 of the present invention after multiple rounds of charge and discharge cycle tests. DETAILED DESCRIPTION

[0059] The present invention will be further described below in conjunction with the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and cannot be used to limit the protection scope of the present invention.

[0060] Embodiment 1:

[0061] like Figure 1 As shown, an embodiment of the present invention provides a method for predicting the remaining life of a lithium-ion battery, comprising the following steps:

[0062] 1. Obtain the battery capacity of the target lithium-ion battery over multiple rounds of historical charge and discharge cycles.

[0063] 2. Generate input samples and sample labels corresponding to each round of historical charge and discharge cycles based on the battery capacity; specifically:

[0064] The battery capacity Q of the kth round of historical charge and discharge cycle k , as the sample label corresponding to the kth round of historical charge and discharge cycles;

[0065] The battery capacity Q of the 1st to (k-1)th round of historical charge and discharge cycles 1～(k-1) , as the input sample corresponding to the kth round of historical charge and discharge cycles, Q 1～(k-1) = {Q1,Q2,…,Q k-1};

[0066] Among them, 2≤k≤K, K is the total number of historical charge and discharge cycles.

[0067] 3. Build a capacity prediction model based on SENet, autoencoder and GRU;

[0068] like Figure 2 As shown, the capacity prediction model provided by the embodiment of the present invention includes a first SENet model, a second SENet model, a first autoencoder, a second autoencoder, a GRU model and a Liner linear layer;

[0069] The input ends of the first SENet model and the first autoencoder are both used as the input ends of the capacity prediction model, the output ends of the first SENet model and the first autoencoder are added and connected to the input end of the second autoencoder, the output end of the first autoencoder is connected to the input end of the second autoencoder, the output ends of the second autoencoder and the second SENet model are added and connected to the input end of the GRU model, the output end of the GRU model is connected to the input end of the Liner linear layer, and the output end of the Liner linear layer serves as the output end of the capacity prediction model.

[0070] SENet (Squeeze-and-Excitation Networks) can use a new "feature recalibration" strategy for input data, automatically obtain the importance of each feature channel through learning, and then strengthen useful features according to this importance and suppress features that are not very useful for the task objectives. Using SENet, the features of important channels in the input data can be strengthened and the features of unimportant channels can be weakened, thereby achieving effective extraction of local features, which has considerable advantages in extracting lithium battery features under actual working conditions.

[0071] Auto-Encoder is an unsupervised learning mode. It is based on the back-propagation algorithm and optimization method. It uses input data as supervision to guide the auto-encoder neural network to learn a mapping relationship. Based on the encoder, the multi-dimensional input X is encoded into a low-dimensional hidden variable h, thereby forcing the neural network to learn the most informative features; then the decoder is used to restore the hidden variable h of the hidden layer to the initial dimension. Based on these two steps, the reconstruction output of the input data is completed, and the deep fusion of the inherent multi-features of these influencing factors is realized. The advantages of one feature are used to make up for the disadvantages of another feature, so that each feature is more effectively integrated, and the global feature extraction is completed, which greatly enhances the robustness of the final remaining service life prediction model. From the data processing process of Auto-Encoder, it can be seen that it is a nonlinear dimensionality reduction with strong generalization. It can effectively solve the problem of unclear mechanism of action between various influencing factors and the remaining service life RUL of lithium batteries, and efficiently realize the extraction of global features.

[0072] GRU (Gate Recurrent Unit) is a type of recurrent neural network and can be regarded as a variant of LSTM. It can reduce the gradient vanishing problem while retaining long-term sequence information. Compared with LSTM, GRU can achieve comparable training results, but its structure can greatly improve training efficiency and save computing resources, which is quite practical for the actual application scenarios of energy storage power stations.

[0073] 4. Train the capacity prediction model by inputting samples and sample labels;

[0074] like Figure 3 As shown, specifically including:

[0075] Repeat the following steps until the capacity prediction model training is complete:

[0076] S101, inputting the input sample into the capacity prediction model;

[0077] S102, performing local feature extraction and global feature extraction on the input sample respectively through the first SENet model and the first autoencoder;

[0078] S103, fusing the local features and global features extracted by the first SENet model and the first autoencoder, and performing local feature extraction on the fusion result through the second SENet model;

[0079] S104, extracting global vital signs from the global features extracted by the first autoencoder using the second autoencoder;

[0080] S105, fusing the local features and global features extracted by the second SENet model and the second autoencoder, and performing prediction through the GRU model;

[0081] S106, output the prediction result through the Liner linear layer;

[0082] S107, calculating the model loss according to the prediction results and the sample labels, and updating the weight parameters of the first SENet model, the second SENet model, the first autoencoder, the second autoencoder, and the GRU model by back propagation of gradient descent according to the model loss;

[0083] The completion of the capacity prediction model training includes the number of repeated executions being greater than a preset maximum number or the model loss being less than a preset loss threshold.

[0084] 5. Construct the input samples corresponding to the first round of future charge and discharge cycles and input them into the trained capacity prediction model to iteratively obtain the battery capacity of multiple rounds of future charge and discharge cycles;

[0085] like Figure 4 As shown, specifically including:

[0086] The input sample corresponding to the first round of future charge and discharge cycles is: The battery capacity Q of the 1st to Kth round of historical charge and discharge cycles 1～K ;

[0087] Iteratively obtaining the battery capacity for multiple rounds of future charge and discharge cycles includes:

[0088] Initialize t=1 and repeat the following operations:

[0089] Judge Qt ≤Q0 is true, Q0 is the preset battery capacity threshold:

[0090] If true, then output the battery capacity Q of the 1st to tth round of future charge and discharge cycles 1～t ;

[0091] If it is not true, the input sample corresponding to the tth round of future charge and discharge cycle is input into the trained capacity prediction model to obtain the battery capacity Q after the t+1th round of future charge and discharge training. t+1 ;

[0092] Construct the input sample corresponding to the t+1th round of future charge and discharge cycle: the battery capacity of the 1st to Kth round of historical charge and discharge cycles and the 1st to tth round of future charge and discharge cycles {Q 1～K ,Q 1～t};

[0093] Let t=t+1.

[0094] 6. Determine the remaining life of the target lithium-ion battery based on the battery capacity of multiple future charge and discharge cycles; specifically including:

[0095] According to the output of the battery capacity Q of the 1st to tth round of future charge and discharge cycles 1～t , determine the remaining life of the target lithium-ion battery as t rounds of charge and discharge cycles.

[0096] like Figure 5 As described, a certain model of battery is used to perform multiple rounds of charge and discharge cycle tests, and a comparison chart of the actual battery capacity (True Value) in each round and the predicted battery capacity (Predicted Value) predicted by the method of this embodiment is statistically compared. It can be observed that the predicted battery capacity predicted by the method of this embodiment has a higher accuracy rate.

[0097] Embodiment 2:

[0098] An embodiment of the present invention provides a device for predicting the remaining life of a lithium-ion battery, the device comprising:

[0099] A historical data module is used to obtain the battery capacity of the target lithium-ion battery in multiple rounds of historical charge and discharge cycles;

[0100] A sample generation module is used to generate input samples and sample labels corresponding to each round of historical charge and discharge cycles according to the battery capacity;

[0101] Model building module, used to build capacity prediction models based on SENet, autoencoder and GRU;

[0102] A model training module is used to train the capacity prediction model by inputting samples and sample labels;

[0103] The future data module is used to construct the input samples corresponding to the first round of future charge and discharge cycles and input them into the trained capacity prediction model to iteratively obtain the battery capacity of multiple rounds of future charge and discharge cycles;

[0104] The life statistics module is used to determine the remaining life of the target lithium-ion battery based on the battery capacity of multiple rounds of future charge and discharge cycles.

[0105] The capacity prediction model includes a first SENet model, a second SENet model, a first autoencoder, a second autoencoder, a GRU model, and a Liner linear layer;

[0106] The input ends of the first SENet model and the first autoencoder are both used as the input ends of the capacity prediction model, the output ends of the first SENet model and the first autoencoder are added and connected to the input end of the second autoencoder, the output end of the first autoencoder is connected to the input end of the second autoencoder, the output ends of the second autoencoder and the second SENet model are added and connected to the input end of the GRU model, the output end of the GRU model is connected to the input end of the Liner linear layer, and the output end of the Liner linear layer serves as the output end of the capacity prediction model.

[0107] Embodiment three:

[0108] An embodiment of the present invention provides an electronic device, including a processor and a storage medium;

[0109] The storage medium is used to store instructions;

[0110] The processor is used to operate according to the instructions to execute the steps of the above method.

[0111] Embodiment 4:

[0112] An embodiment of the present invention provides a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the steps of the above method are implemented.

[0113] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0114] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of 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 produce 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.

[0115] 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.

[0116] 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.

[0117] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the technical principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A method for predicting the remaining life of a lithium-ion battery, characterized in that: include: Obtain the battery capacity of the target lithium-ion battery in multiple rounds of historical charge and discharge cycles; Generate input samples and sample labels corresponding to each round of historical charge and discharge cycles based on battery capacity; Construct a capacity prediction model based on SENet, autoencoder and GRU; the capacity prediction model includes a first SENet model, a second SENet model, a first autoencoder, a second autoencoder, a GRU model and a Liner linear layer; the input ends of the first SENet model and the first autoencoder are both used as the input ends of the capacity prediction model, the output ends of the first SENet model and the first autoencoder are added and connected to the input end of the second autoencoder, the output end of the first autoencoder is connected to the input end of the second autoencoder, the output ends of the second autoencoder and the second SENet model are added and connected to the input end of the GRU model, the output end of the GRU model is connected to the input end of the Liner linear layer, and the output end of the Liner linear layer is used as the output end of the capacity prediction model; The capacity prediction model is trained by inputting samples and sample labels; Construct the input samples corresponding to the first round of future charge and discharge cycles and input them into the trained capacity prediction model, iteratively obtaining the battery capacity of multiple rounds of future charge and discharge cycles; Determine the remaining life of the target lithium-ion battery based on the battery capacity over multiple future charge and discharge cycles.

2. The method for predicting the remaining life of a lithium-ion battery according to claim 1, characterized in that: The step of generating input samples and sample labels corresponding to each round of historical charge and discharge cycles according to the battery capacity includes: The battery capacity Q of the kth round of historical charge and discharge cycle k , as the sample label corresponding to the kth round of historical charge and discharge cycles; The battery capacity Q of the 1st to (k-1)th round of historical charge and discharge cycles 1～(k-1) , as the input sample corresponding to the kth round of historical charge and discharge cycles, Q 1～(k-1) = {Q1,Q2,…,Q k-1 }; Among them, 2≤k≤K, K is the total number of historical charge and discharge cycles.

3. The method for predicting the remaining life of a lithium-ion battery according to claim 1, characterized in that: The training of the capacity prediction model by inputting samples and sample labels includes: Repeat the following steps until the capacity prediction model training is complete: Feeding the input samples into the capacity prediction model; Perform local feature extraction and global feature extraction on the input sample through the first SENet model and the first autoencoder respectively; The local features and global features extracted by the first SENet model and the first autoencoder are fused, and local features are extracted from the fusion results by the second SENet model; Performing global vital sign extraction on the global features extracted by the first autoencoder through the second autoencoder; The local features and global features extracted by the second SENet model and the second autoencoder are fused and predicted by the GRU model; Output the prediction results through the Liner linear layer; Calculate the model loss according to the prediction results and sample labels, and update the weight parameters of the first SENet model, the second SENet model, the first autoencoder, the second autoencoder, and the GRU model by back propagation of gradient descent according to the model loss; The completion of the capacity prediction model training includes the number of repeated executions being greater than a preset maximum number or the model loss being less than a preset loss threshold.

4. The method for predicting the remaining life of a lithium-ion battery according to claim 2, characterized in that: The input samples corresponding to the first round of future charge and discharge cycles are: the battery capacity Q of the 1st to Kth rounds of historical charge and discharge cycles 1～K ; The iterative acquisition of battery capacity for multiple rounds of future charge and discharge cycles includes: Initialize t=1 and repeat the following operations: Judge Q t ≤Q0 is true, Q0 is the preset battery capacity threshold: If true, then output the battery capacity Q of the 1st to tth future charge and discharge cycles 1～t ; If not, the input sample corresponding to the tth round of future charge and discharge cycle is input into the trained capacity prediction model to obtain the battery capacity Q after the t+1th round of future charge and discharge training. t+1 ; Construct the input sample corresponding to the t+1th round of future charge and discharge cycle: the battery capacity of the 1st to Kth round of historical charge and discharge cycles and the 1st to tth round of future charge and discharge cycles {Q 1～K ,Q 1～t }; Let t=t+1.

5. The method for predicting the remaining life of a lithium-ion battery according to claim 4, characterized in that: Determining the remaining life of the target lithium-ion battery includes: According to the output of the battery capacity Q of the 1st to tth round of future charge and discharge cycles 1～t , determine the remaining life of the target lithium-ion battery as t rounds of charge and discharge cycles.

6. A lithium-ion battery remaining life prediction device, characterized in that: The device comprises: A historical data module is used to obtain the battery capacity of the target lithium-ion battery in multiple rounds of historical charge and discharge cycles; A sample generation module is used to generate input samples and sample labels corresponding to each round of historical charge and discharge cycles according to the battery capacity; A model construction module, for constructing a capacity prediction model based on SENet, autoencoder and GRU; the capacity prediction model includes a first SENet model, a second SENet model, a first autoencoder, a second autoencoder, a GRU model and a Liner linear layer; the input ends of the first SENet model and the first autoencoder are both used as the input ends of the capacity prediction model, the output ends of the first SENet model and the first autoencoder are added and connected to the input end of the second autoencoder, the output end of the first autoencoder is connected to the input end of the second autoencoder, the output ends of the second autoencoder and the second SENet model are added and connected to the input end of the GRU model, the output end of the GRU model is connected to the input end of the Liner linear layer, and the output end of the Liner linear layer is used as the output end of the capacity prediction model; A model training module is used to train the capacity prediction model by inputting samples and sample labels; The future data module is used to construct the input samples corresponding to the first round of future charge and discharge cycles and input them into the trained capacity prediction model to iteratively obtain the battery capacity of multiple rounds of future charge and discharge cycles; The life statistics module is used to determine the remaining life of the target lithium-ion battery based on the battery capacity of multiple rounds of future charge and discharge cycles.

7. An electronic device, characterized in that: including processor and storage medium; The storage medium is used to store instructions; The processor is configured to operate according to the instructions to execute the steps of the method according to any one of claims 1-5.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps of the method according to any one of claims 1 to 5 are implemented.

Images