Method and system for estimating SOC (State of Charge) of full-life-cycle lithium ion battery, terminal and medium

Abstract

The invention relates to the technical field of energy storage lithium ion battery SOC estimation, in particular to a full-life-cycle lithium ion battery SOC estimation method and system, a terminal and a medium, and the method comprises the following steps: processing state parameters of a lithium ion battery into input data; inputting the input data into a one-dimensional convolutional layer, and extracting deep features related to the state of charge and the state of health of the lithium ion battery in the state parameters; taking the output of the pooling layer as the input of the long short-term memory network; and the output result of the long-short-term memory network is input to the attention module, and the attention module outputs the estimation result of the SOC of the lithium ion battery through the full connection layer. According to the method, the Attention-LSTM deep learning model is established, the accurate association between the input data and the SOC is established in the whole life cycle, the capability of capturing different characteristics and relationships in the input data and the long sequence dependence extraction capability of the model are enhanced, and the lithium ion battery SOC estimation value with relatively high accuracy is output.

Description

Technical Field

[0001] The present invention relates to the technical field of SOC estimation of energy storage lithium-ion batteries, and in particular to a method, system, terminal and medium for estimating SOC of a lithium-ion battery throughout its life cycle. Background Art

[0002] With the rapid development of new energy, lithium-ion batteries have been widely used in new energy storage and other fields due to their many advantages such as high energy density, small thermal effect, long service life and low self-discharge rate. Among them, the state of charge (SOC) of lithium-ion batteries indicates the percentage of the current remaining capacity of the battery to its maximum capacity, which is an important basis for evaluating battery performance and formulating battery management strategies. Therefore, accurate SOC estimation can timely update the remaining power of the battery, prevent overcharging and over-discharging of the battery, and improve the battery's efficiency. It is also of great significance to the safe operation of the battery and the extension of the battery life.

[0003] At present, the commonly used SOC estimation methods mainly include the open circuit voltage method, the ampere-hour integration method and the Kalman filter method. Although they can realize the SOC estimation to a certain extent, they all have their own limitations. For example, since the size of the residual power is directly reflected in the value of the open circuit voltage (OCV), the open circuit voltage method based on the nonlinear mapping relationship between OCV and SOC is widely used. However, the open circuit voltage method requires the battery to be stationary for a long time after the load is disconnected, and the accuracy will be affected after high-power discharge. Using the value of the initial SOC, the battery SOC can also be estimated by the ampere-hour integration method. The most obvious disadvantage of the ampere-hour integration method is that only calculating the current will lead to error accumulation, and there is no other parameter for correction. Once the initial value is inaccurate, the estimation result will have a large deviation. In order to overcome these limitations, people have proposed a method based on Kalman filtering. This method uses the established equivalent circuit model to recursively correct the SOC through the state equation and the observation equation. However, since the electrochemical process is dynamic and nonlinear, it is difficult to obtain a sophisticated model to describe all the characteristics of the battery.

[0004] Therefore, due to the complex electrochemical reactions inside lithium-ion batteries and the influence of the external environment, the existing technology still has the problem of insufficient accuracy in battery SOC estimation. Summary of the invention

[0005] In order to solve the technical problem that the prior art still has insufficient accuracy in battery SOC estimation due to the complex electrochemical reactions inside lithium-ion batteries and the influence of the external environment, the present invention provides a full life cycle lithium-ion battery SOC estimation method, and also provides a full life cycle lithium-ion battery SOC estimation system, a terminal and a medium.

[0006] To achieve the above objectives, in a first aspect, a technical solution adopted by a full life cycle lithium-ion battery SOC estimation method in the present invention is: A method for estimating SOC of a lithium-ion battery over its entire life cycle comprises the following steps: Acquire state parameters of the lithium-ion battery and process the state parameters as input data; Construct a one-dimensional convolution layer, input the input data into the one-dimensional convolution layer, extract the deep features related to the state of charge and health state of the lithium-ion battery in the state parameters; use the pooling layer to reduce the deep features to a single vector; Construct a long short-term memory network, which includes a forget gate, an input gate, an output gate, and a cell state. Use the output of the pooling layer as the input of the long short-term memory network to obtain the output result. Construct an attention module and input the output results of the long short-term memory network into the attention module. The attention module outputs the estimated results of the lithium-ion battery state of charge through the fully connected layer.

[0007] As a preferred implementation of a full life cycle lithium-ion battery SOC estimation method, the processing of state parameters as input data includes the following steps: The state parameters include voltage, current, temperature and time, and the state parameters are processed into a four-variable data sequence consisting of voltage, current, temperature and time; Set sequence length and the time interval between adjacent four-variable data series ; Will wait for the estimated time and before this moment Interval The voltage, current and temperature corresponding to the moment The data sequence is taken as input data.

[0008] As a preferred implementation of a full life cycle lithium-ion battery SOC estimation method, the input data is input into a one-dimensional convolution layer to extract deep features related to the state of charge and health state of the lithium-ion battery in the state parameters, including the following calculation process: , where The first elements, represents a nonlinear activation function, represents the weight of the convolution kernel, Indicates that the input data is from Elements to A slice of elements, Represents the bias term.

[0009] As a preferred implementation of a full life cycle lithium-ion battery SOC estimation method, the output of the pooling layer is used as the input of the long short-term memory network to obtain the output result, which includes the following steps: a. Forget gate determines useful information in cell state: calculate forget factor ,

[0010] In the formula, represents the sigmoid activation function, represents the weight matrix of the forget gate, Represents the output at the previous moment, Represents the input at the current moment, Indicates that and These two vectors are concatenated, Represents the bias of the forget gate; b. The input gate determines the information to be retained in the cell state: first calculate the updated ratio ,

[0011] In the formula, represents the weight matrix of the input gate, Represents the bias of the input gate; Then get the current input ,

[0012] In the formula, represents the weight matrix input at the current moment, Indicates the offset input at the current moment; c. Update the current cell state : ; d. Determine the output of the LSTM network: First calculate the output ratio :

[0013] In the formula, represents the weight matrix of the output gate, Represents the bias of the output gate; Then the current cell state Processed and output ratio Multiply to get the output of the hidden layer : .

[0014] As a preferred implementation of a full life cycle lithium-ion battery SOC estimation method, the output result of the long short-term memory network is input into the attention module, and the attention module outputs the estimation result of the lithium-ion battery state of charge through a fully connected layer, including the following steps: Introducing the weight matrix , using three different weight matrices , and Perform linear transformation on the input data to obtain the query vector matrix , key vector matrix Sum vector matrix ; Among them, the key vector matrix Sum vector matrix The initial state is , The output of the long short-term memory network , ;Weight matrix and the query vector matrix The initial state is , The output of the hidden layer of the long short-term memory network , ; Using Linear Layer Calculate the weight matrix related to the number of attention heads , and successively with , and Multiplication:

[0015]

[0016]

[0017]

[0018]

[0019] In the formula, Indicates A query vector matrix, Indicates A key vector matrix, Indicates A matrix of value vectors; Computing attention output :

[0020]

[0021]

[0022] In the formula, represents the scaling factor, express Activation function, represents the attention score, represents the attention output of each head, Represents a splicing operation; Output your attention The input is sent to the fully connected layer, which outputs the estimated state of charge of the lithium-ion battery.

[0023] As a preferred implementation of a full life cycle lithium-ion battery SOC estimation method, the state parameters also include the number of charge and discharge cycles and the charge and discharge state of the lithium-ion battery, the input data includes a training set and a test set, only voltage, current, temperature and time are used in the training set, the data in the training set are used for training during the model training process, and the data in the test set are used for testing during the result testing process, wherein the two state parameters, the number of charge and discharge cycles and the charge and discharge state, are used to respectively verify the estimation accuracy of the state of charge of the lithium-ion battery under different discharge cycle numbers and charge and discharge states.

[0024] As a preferred implementation of a full life cycle lithium-ion battery SOC estimation method, a Bayesian optimization method is used to automatically optimize hyperparameters during model training.

[0025] In the second aspect, the technical solution adopted by a full life cycle lithium-ion battery SOC estimation system in the present invention is: A full life cycle lithium-ion battery SOC estimation system includes a data acquisition and processing module; a one-dimensional convolution module, a long short-term memory network module and an attention module, wherein: The data acquisition and processing module is configured to acquire state parameters of the lithium-ion battery and process the state parameters into input data; The one-dimensional convolution module is configured to construct a one-dimensional convolution layer, input the input data into the one-dimensional convolution layer, extract the deep features related to the state of charge and health state of the lithium-ion battery in the state parameters; and use the pooling layer to reduce the deep features to a single vector; The long short-term memory network module is configured to construct a long short-term memory network, which includes a forget gate, an input gate, an output gate, and a cell state. The output of the pooling layer is used as the input of the long short-term memory network to obtain an output result. The attention module is configured to input the output result of the long short-term memory network into the attention module, and the attention module outputs the estimated result of the state of charge of the lithium-ion battery through the fully connected layer.

[0026] In a third aspect, a technical solution adopted by a terminal in the present invention is: A terminal comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the steps of any one of the above-mentioned methods for estimating the SOC of a lithium-ion battery throughout its life cycle are implemented.

[0027] In a fourth aspect, a technical solution adopted by a medium in the present invention is: A storage medium stores a computer program, which, when executed by a processor, implements the steps of any one of the above-mentioned methods for estimating the SOC of a lithium-ion battery throughout its life cycle.

[0028] The beneficial effects of the present invention include: (1) The key state parameters of lithium-ion batteries, such as voltage, current, and temperature, will change over time, forming time series data. Long short-term memory networks can capture the temporal dependencies of these data, thereby more accurately predicting SOC.

[0029] (2) A data sequence consisting of data at the estimated moment and multiple historical moments in the previous period is used as input data. Compared with the data at a single moment, the data sequence containing multiple historical moments can reflect the degradation of battery characteristics under different SOHs, and an accurate correlation between the input data and SOC can be established over the entire life cycle.

[0030] (3) Multiple state parameters of lithium-ion batteries, such as voltage, current, temperature, and internal resistance, may affect the precise value of SOC. The attention mechanism can focus on the different combinations and changes of these parameters, thereby capturing the battery state information more comprehensively, enhancing the model's ability to capture different features and relationships in the input data sequence and its ability to extract long sequence dependencies, thereby improving the model's performance and generalization ability.

[0031] (4) Deep learning models usually contain a large number of hyperparameters, such as network structure, learning rate, regularization term, etc. The selection of these hyperparameters has a significant impact on model performance. Traditional methods such as grid search and random search are inefficient in hyperparameter tuning, while Bayesian optimization can effectively solve this problem. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] In order to more clearly illustrate the technical solution of the present invention, the accompanying drawings required for use in the description will be briefly introduced below. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0033] Figure 1 A schematic diagram of a flow chart of a method for estimating SOC of a lithium-ion battery throughout its life cycle in a specific embodiment of the present invention; Figure 2 The calculation steps of Bayesian optimization in the specific implementation mode of the present invention; Figure 3 It is a structural schematic diagram of a full life cycle lithium-ion battery SOC estimation system in a specific implementation manner of the present invention. DETAILED DESCRIPTION

[0034] Before describing the technical content of the specific embodiments of the present invention, the professional knowledge involved in the present invention is first introduced: One-dimensional convolutional neural network (1D-CNN) is a deep learning architecture specifically designed to process one-dimensional sequence data. Compared with traditional neural networks, it can effectively extract local features in the sequence by performing convolution operations on sequence data through one-dimensional convolution kernels in the convolution layer. It mainly consists of multiple one-dimensional convolutional layers, pooling layers (optional), and fully connected layers.

[0035] The state of charge (SOC) of a lithium-ion battery refers to the ratio of the current remaining power of the battery to the total capacity of the battery. It is usually expressed as a percentage, ranging from 0% to 100%. When the SOC is 0%, it means that the battery is fully discharged; when the SOC is 100%, it means that the battery is fully charged. Accurately estimating the SOC can provide an important reference for the battery management system, optimize the battery charging and discharging strategy, and extend the battery life.

[0036] The state of health of a lithium-ion battery is an indicator used to measure the degree of degradation of the performance of a lithium-ion battery relative to its initial performance. It is usually expressed as a percentage. For example, the SOH of a brand new battery is 100%. With the use and aging of the battery, the SOH will gradually decrease.

[0037] Long-Short Term Memory (LSTM) is a special type of Recurrent Neural Network (RNN). It is designed to solve the gradient vanishing and gradient exploding problems that occur when traditional RNN processes long sequence data, so that it can effectively model long-term dependencies in long sequence data.

[0038] The Attention Mechanism is a widely used technology in deep learning. It imitates the human visual system's behavior of focusing on certain areas when observing a scene. When processing information, it can automatically focus on the key parts of the input data, thereby more effectively utilizing the important information in the data.

[0039] In order to make the purpose, features and advantages of the present invention more obvious and easy to understand, the technical scheme of the present invention will be clearly and completely described below in conjunction with the drawings in this specific embodiment. Obviously, the embodiments described below are only part of the embodiments of the present invention, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0040] Reference Figure 1 The present embodiment provides a method for estimating the SOC of a lithium-ion battery over its entire life cycle, comprising the following steps: A method for estimating SOC of a lithium-ion battery over its entire life cycle comprises the following steps: Acquire state parameters of the lithium-ion battery and process the state parameters as input data; Construct a one-dimensional convolution layer, input the input data into the one-dimensional convolution layer, extract the deep features related to the state of charge and health state of the lithium-ion battery in the state parameters; use the pooling layer to reduce the deep features to a single vector to perform feature dimensionality reduction; Construct a long short-term memory network, which includes a forget gate, an input gate, an output gate, and a cell state. Use the output of the pooling layer as the input of the long short-term memory network to obtain the output result. Construct an attention module and input the output results of the long short-term memory network into the attention module. The attention module outputs the estimated results of the lithium-ion battery state of charge through the fully connected layer.

[0041] This embodiment constructs an Attention-LSTM deep learning model through the above method, which can output an estimated value of the SOC of a lithium-ion battery with high accuracy.

[0042] In this embodiment, processing the state parameters into input data includes the following steps: The state parameters include voltage, current, temperature and time, and the state parameters are processed into a four-variable data sequence consisting of voltage, current, temperature and time; Set sequence length and the time interval between adjacent four-variable data series ; Will wait for the estimated time and before this moment Interval The voltage, current and temperature corresponding to the moment The data sequence is taken as input data.

[0043] Based on the above processing of the state parameters, in a specific embodiment, the state parameters also include the number of charge and discharge cycles and the charge and discharge state of the lithium-ion battery, the input data includes a training set and a test set, only voltage, current, temperature and time are used in the training set, the data in the training set are used for training in the model training process, and the data in the test set are used for testing in the result testing process, wherein the two state parameters of the charge and discharge cycle number and the charge and discharge state are used to respectively verify the estimation accuracy of the state of charge of the lithium-ion battery under different discharge cycle numbers and charge and discharge states.

[0044] The accuracy of the deep learning model is not only related to the quality of the training data, the rationality of the network structure and other factors, but also closely related to many hyperparameters of the model. By optimizing various hyperparameters, the accuracy of the model can be significantly improved. The parameters of the deep learning model are numerous and their interrelationships are complex, which brings difficulties to manual parameter adjustment. Based on this, this embodiment uses the Bayesian optimization method to automatically optimize the hyperparameters during the model training process.

[0045] Reference Figure 2 The Bayesian optimization is used to optimize the hyperparameters in the Attention-LSTM model established in this embodiment. The calculation steps are as follows: (1) Initialize the hyperparameter combination according to the preset hyperparameter range, substitute it into the probabilistic proxy model, train the Attention-LSTM model, obtain the loss value output by the objective function, and correct the probabilistic proxy model; (2) Use the acquisition function to select the next set of sample points in the modified probabilistic proxy model, train the Attention-LSTM model again, obtain a new objective function, and update the sample set and probabilistic proxy model; (3) If the objective function loss value corresponding to the new sample point meets the requirements, the calculation ends and the currently selected optimal parameters are output; (4) If the objective function loss value corresponding to the new sample point does not meet the requirements, update the sample point to the sample set and repeat step 2.

[0046] This embodiment takes obtaining the minimum root mean square error (RMSE) as the optimization goal, adopts the Gaussian process as the probability proxy model, obtains the optimized parameter values, and then performs prediction and error analysis on the data.

[0047] In this embodiment, the input data is input into the one-dimensional convolution layer to extract the deep features related to the state of charge and health state of the lithium-ion battery in the state parameters, including the following calculation process: , where The first elements, represents a nonlinear activation function, represents the weight of the convolution kernel, Indicates that the input data is from Elements to A slice of elements, Represents the bias term.

[0048] In this embodiment, the output of the pooling layer is used as the input of the long short-term memory network to obtain the output result, which includes the following steps: a. Forget gate determines useful information in cell state: calculate forget factor ,

[0049] In the formula, represents the sigmoid activation function, represents the weight matrix of the forget gate, Represents the output at the previous moment, Represents the input at the current moment, Indicates that and These two vectors are concatenated, Represents the bias of the forget gate; b. The input gate determines the information to be retained in the cell state: first calculate the updated ratio ,

[0050] In the formula, represents the weight matrix of the input gate, Represents the bias of the input gate; Then get the current input ,

[0051] In the formula, represents the weight matrix input at the current moment, Indicates the offset input at the current moment; c. Update the current cell state : ; d. Determine the output of the LSTM network: First calculate the output ratio :

[0052] In the formula, represents the weight matrix of the output gate, Represents the bias of the output gate; Then the current cell state Processed and output ratio Multiply to get the output of the hidden layer : .

[0053] After the output result of the long short-term memory network is obtained through the above steps, further, the output result of the long short-term memory network is input into the attention module, and the attention module outputs the estimated result of the state of charge of the lithium-ion battery through the fully connected layer, including the following steps: Introducing the weight matrix , using three different weight matrices , and Perform linear transformation on the input data to obtain the query vector matrix , key vector matrix Sum vector matrix ; Among them, the key vector matrix Sum vector matrix The initial state is , The output of the long short-term memory network , ;Weight matrix and the query vector matrix The initial state is , The output of the hidden layer of the long short-term memory network , ; Using Linear Layer Calculate the weight matrix related to the number of attention heads , and successively with , and Multiplication:

[0054]

[0055]

[0056]

[0057]

[0058] In the formula, Indicates A query vector matrix, Indicates A key vector matrix, Indicates A matrix of value vectors; Computing attention output :

[0059]

[0060]

[0061] In the formula, represents the scaling factor, express Activation function, represents the attention score, represents the attention output of each head, Represents a splicing operation; Output your attention The input is sent to the fully connected layer, which outputs the estimated state of charge of the lithium-ion battery.

[0062] The key state parameters of lithium-ion batteries, such as voltage, current, and temperature, will change over time, forming time series data. In this embodiment, a long short-term memory network is used to capture the temporal dependency of these data, thereby more accurately predicting SOC.

[0063] This embodiment is based on a data sequence consisting of data at the estimated moment and multiple historical moments in the previous period as input data. Compared with the data at a single moment, the data sequence containing multiple historical moments can reflect the degradation of battery characteristics under different SOHs, and can establish an accurate association between the input data and SOC over the entire life cycle.

[0064] Multiple state parameters of lithium-ion batteries, such as voltage, current, temperature, and internal resistance, may affect the precise value of SOC. In this embodiment, an attention mechanism is used, which can focus on different combinations and changes of these parameters, thereby capturing the battery state information more comprehensively, enhancing the model's ability to capture different features and relationships in the input data sequence and the ability to extract long sequence dependencies, improving the model's representation ability, and improving the model's performance and generalization ability.

[0065] Deep learning models usually contain a large number of hyperparameters, such as network structure, learning rate, regularization term, etc. The selection of these hyperparameters has an important impact on model performance. Traditional methods such as grid search and random search have the problem of low efficiency in hyperparameter tuning, while this embodiment uses Bayesian optimization to effectively solve this problem.

[0066] like Figure 3 As shown below, a full life cycle lithium-ion battery SOC estimation system is provided in an embodiment of the present disclosure. The full life cycle lithium-ion battery SOC estimation system and the full life cycle lithium-ion battery SOC estimation method in the above-mentioned embodiments belong to the same inventive concept. For details not described in detail in the embodiment of the full life cycle lithium-ion battery SOC estimation system, please refer to the embodiment of the above-mentioned full life cycle lithium-ion battery SOC estimation method.

[0067] A full life cycle lithium-ion battery SOC estimation system includes a data acquisition and processing module; a one-dimensional convolution module, a long short-term memory network module and an attention module, wherein: The data acquisition and processing module is configured to acquire state parameters of the lithium-ion battery and process the state parameters into input data; The one-dimensional convolution module is configured to construct a one-dimensional convolution layer, input the input data into the one-dimensional convolution layer, extract the deep features related to the state of charge and health state of the lithium-ion battery in the state parameters; and use the pooling layer to reduce the deep features to a single vector; The long short-term memory network module is configured to construct a long short-term memory network, which includes a forget gate, an input gate, an output gate, and a cell state. The output of the pooling layer is used as the input of the long short-term memory network to obtain an output result. The attention module is configured to input the output result of the long short-term memory network into the attention module, and the attention module outputs the estimated result of the state of charge of the lithium-ion battery through the fully connected layer.

[0068] An embodiment of the present application also proposes a terminal, including a memory, a processor, a communication unit, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the steps of a method for estimating the SOC of a lithium-ion battery throughout its life cycle are implemented; the memory, the processor, and the communication unit communicate via one or more buses.

[0069] The processor may include one or more processing units, for example, the processor may include a central processing unit (CPU), an application processor (AP), a modem processor, a graphics processing unit (GPU), an image signal processor (ISP), a controller, a memory, a video codec, a digital signal processor (DSP), a baseband processor and / or a neural network processing unit (NPU), etc. Different processing units may be independent devices or integrated into one or more processors.

[0070] The processor can be the nerve center and command center of the terminal. The controller can generate operation control signals according to the instruction operation code and timing signal to complete the control of fetching and executing instructions.

[0071] The memory is used to store the execution instructions of the processor, and the memory can be implemented by any type of volatile or non-volatile storage terminal or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk. When the execution instructions in the memory are executed by the processor, the terminal can execute some or all of the steps in the above-mentioned embodiment of the full life cycle lithium-ion battery SOC estimation method.

[0072] The wireless communication function of an electronic device can be realized through an antenna, a wireless communication module, a modem processor, and a baseband processor.

[0073] The wireless communication module can provide wireless communication solutions for electronic devices including wireless LAN, Bluetooth, global navigation satellite system, frequency modulation, short-range wireless communication technology, infrared technology, etc.

[0074] This embodiment further provides a storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps of any one of the methods for estimating the SOC of a lithium-ion battery throughout its life cycle as described above are implemented.

[0075] The above description of the disclosed embodiments enables one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for estimating SOC of a lithium-ion battery over its entire life cycle, characterized in that: The following steps are involved: Acquire state parameters of the lithium-ion battery and process the state parameters as input data; Construct a one-dimensional convolution layer, input the input data into the one-dimensional convolution layer, extract the deep features related to the state of charge and health state of the lithium-ion battery in the state parameters; use the pooling layer to reduce the deep features to a single vector; Construct a long short-term memory network, which includes a forget gate, an input gate, an output gate, and a cell state. Use the output of the pooling layer as the input of the long short-term memory network to obtain the output result. Construct an attention module and input the output results of the long short-term memory network into the attention module. The attention module outputs the estimated results of the lithium-ion battery state of charge through the fully connected layer.

2. A method for estimating SOC of a lithium-ion battery over its entire life cycle according to claim 1, characterized in that: The processing of the state parameters as input data comprises the following steps: The state parameters include voltage, current, temperature and time, and the state parameters are processed into a four-variable data sequence consisting of voltage, current, temperature and time; Set sequence length and the time interval between adjacent four-variable data series ; Will wait for the estimated time and before this moment Interval The voltage, current and temperature corresponding to the moment The data sequence is taken as input data.

3. The method for estimating SOC of a lithium-ion battery during its entire life cycle according to claim 1, characterized in that: The step of inputting the input data into the one-dimensional convolution layer and extracting the deep features related to the state of charge and health state of the lithium-ion battery in the state parameters includes the following calculation process: , where The first elements, represents a nonlinear activation function, represents the weight of the convolution kernel, Indicates that the input data is from Elements to A slice of elements, Represents the bias term.

4. The method for estimating SOC of a lithium-ion battery during its entire life cycle according to claim 1, characterized in that: The step of using the output of the pooling layer as the input of the long short-term memory network to obtain the output result comprises the following steps: a. Forget gate determines useful information in cell state: calculate forget factor , In the formula, represents the sigmoid activation function, represents the weight matrix of the forget gate, Represents the output at the previous moment, Represents the input at the current moment, Indicates that and These two vectors are concatenated, Represents the bias of the forget gate; b. The input gate determines the information to be retained in the cell state: first calculate the updated ratio , In the formula, represents the weight matrix of the input gate, Represents the bias of the input gate; Then get the current input , In the formula, represents the weight matrix input at the current moment, Indicates the offset input at the current moment; c. Update the current cell state : ; d. Determine the output of the LSTM network: First calculate the output ratio : In the formula, represents the weight matrix of the output gate, Represents the bias of the output gate; Then the current cell state Processed and output ratio Multiply to get the output of the hidden layer : .

5. A method for estimating SOC of a lithium-ion battery over its entire life cycle according to claim 4, characterized in that: The step of inputting the output result of the long short-term memory network into the attention module, and the attention module outputting the estimated result of the state of charge of the lithium-ion battery through the fully connected layer, comprises the following steps: Introducing the weight matrix , using three different weight matrices , and Perform linear transformation on the input data to obtain the query vector matrix , key vector matrix Sum vector matrix ; Among them, the key vector matrix Sum vector matrix The initial state is , The output of the long short-term memory network , ;Weight matrix and the query vector matrix The initial state is , The output of the hidden layer of the long short-term memory network , ; Using Linear Layer Calculate the weight matrix related to the number of attention heads , and successively with , and Multiplication: In the formula, Indicates A query vector matrix, Indicates A key vector matrix, Indicates A matrix of value vectors; Computing attention output : In the formula, represents the scaling factor, express Activation function, represents the attention score, represents the attention output of each head, Represents a splicing operation; Output your attention The input is sent to the fully connected layer, which outputs the estimated state of charge of the lithium-ion battery.

6. A method for estimating SOC of a lithium-ion battery over its entire life cycle according to claim 2, characterized in that: The state parameters also include the number of charge and discharge cycles and the charge and discharge state of the lithium-ion battery. The input data includes a training set and a test set. Only voltage, current, temperature and time are used in the training set. The data in the training set are used for training during the model training process, and the data in the test set are used for testing during the result testing process. The two state parameters of the charge and discharge cycle number and the charge and discharge state are used to verify the estimation accuracy of the charge state of lithium-ion batteries under different discharge cycle numbers and charge and discharge states.

7. A method for estimating SOC of a lithium-ion battery over its entire life cycle according to claim 6, characterized in that: During the model training process, the Bayesian optimization method is used to automatically optimize the hyperparameters.

8. A full life cycle lithium-ion battery SOC estimation system, characterized in that: It includes data acquisition and processing module; one-dimensional convolution module, long short-term memory network module and attention module, among which: The data acquisition and processing module is configured to acquire state parameters of the lithium-ion battery and process the state parameters into input data; The one-dimensional convolution module is configured to construct a one-dimensional convolution layer, input the input data into the one-dimensional convolution layer, extract the deep features related to the state of charge and health state of the lithium-ion battery in the state parameters; and use the pooling layer to reduce the deep features to a single vector; The long short-term memory network module is configured to construct a long short-term memory network, which includes a forget gate, an input gate, an output gate, and a cell state. The output of the pooling layer is used as the input of the long short-term memory network to obtain an output result. The attention module is configured to input the output result of the long short-term memory network into the attention module, and the attention module outputs the estimated result of the state of charge of the lithium-ion battery through the fully connected layer.

9. A terminal, characterized in that: The invention comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the steps of a method for estimating the SOC of a lithium-ion battery throughout its life cycle are implemented as described in any one of claims 1 to 7.

10. A storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of a method for estimating SOC of a lithium-ion battery throughout its life cycle are implemented as described in any one of claims 1 to 7.

Images