Battery SOC prediction method and system, electronic equipment and storage medium

Abstract

The invention provides a battery SOC prediction method and system, electronic equipment and a storage medium, and the method comprises the steps: obtaining battery data of a to-be-predicted battery in a used time period, carrying out the short-time Fourier transform STFT of the battery data, and obtaining a time frequency domain feature; processing the time frequency domain features through a trained convolutional neural network (CNN), and obtaining a predicted voltage at the next moment of the used time period; and processing the predicted voltage through an unscented Kalman filter (UKF) to obtain a predicted SOC as a final prediction result. The trained neural network CNN is fused into the UKF, fusion calculation of data is achieved, and the SOC of the electric equipment in the low-temperature and high-pressure deep sea environment can be accurately predicted.

Description

Technical Field

[0001] The present application belongs to the field of artificial intelligence, and in particular relates to a method, system, electronic device and storage medium for battery SOC prediction. Background Art

[0002] The state of charge (SOC) prediction of lithium-ion batteries in underwater vehicles is a key technology in battery management systems. Especially in deep-sea exploration and underwater operations, accurate SOC prediction is directly related to the safety and reliability of the vehicle. Lithium-ion batteries are widely used in underwater vehicles due to their high energy density, long life and excellent cycle performance. However, in extreme deep-sea environments, accurate SOC prediction faces many challenges.

[0003] Although traditional SOC prediction methods such as the ampere-hour integration method and the open circuit voltage method have a certain application basis, their accuracy and reliability are often limited in the complex environment of deep sea low temperature and high pressure. The ampere-hour integration method relies on current measurement, which is affected by current measurement errors and battery aging, and is prone to deviations in SOC prediction. In practical applications, the open circuit voltage method requires the battery to be left to stand still after the discharge is completed to measure the voltage, which is not feasible for real-time monitoring. In addition, these methods cannot fully consider the dynamic effects of factors such as temperature changes, discharge rates, and internal chemical reactions of the battery on SOC, resulting in a decrease in the accuracy of SOC prediction in complex environments.

[0004] In recent years, with the development of machine learning and data-driven technologies, researchers have begun to explore more advanced SOC prediction methods. For example, the neural network (NN) method aims to process multi-dimensional battery data through deep learning and predict SOC in real time. These methods can integrate multiple input information such as battery voltage, current, temperature, etc., and take into account the influence of environmental factors, thereby improving the accuracy of SOC prediction.

[0005] Nevertheless, there are still some limitations in the current research. In the deep-sea environment, changes in temperature and other complex conditions have a significant impact on battery performance, and existing SOC prediction methods often fail to fully cope with these environmental changes. In addition, the complexity and uncertainty of deep-sea operations make real-time data acquisition and processing difficult, resulting in challenges to the reliability of prediction results. Summary of the invention

[0006] The main purpose of the embodiments of the present invention is to provide a method, system, electronic device and storage medium for battery SOC prediction, which can accurately predict the SOC of electrical equipment in a low-temperature, high-pressure deep-sea environment.

[0007] In a first aspect, a method for predicting battery SOC is provided, the method comprising:

[0008] Obtaining battery data of the battery to be predicted for the time period in use, and performing short-time Fourier transform (STFT) on the battery data to obtain time-frequency domain features, wherein the battery data includes: voltage data for the time period in use, current data for the time period in use, and temperature data for the time period in use;

[0009] Processing the time-frequency domain features by means of a trained convolutional neural network (CNN) to obtain a predicted voltage at the next moment of the used time period;

[0010] The predicted voltage is processed by an unscented Kalman filter UKF to obtain a predicted SOC as a final prediction result.

[0011] In a possible implementation, the processing of the time-frequency domain features by a trained convolutional neural network (CNN) to obtain a predicted voltage at the next moment of the used time period includes:

[0012] Training a preset initial CNN using the time-frequency domain features to obtain the trained CNN;

[0013] The time-frequency domain features are processed by the trained CNN to obtain the predicted voltage.

[0014] In another possible implementation, the processing of the predicted voltage by an unscented Kalman filter UKF to obtain a predicted SOC as a final prediction result includes:

[0015] Setting the state transfer equation and observation equation of the UKF;

[0016] Processing the predicted voltage according to the state transfer equation to obtain intermediate data;

[0017] The intermediate data is updated according to the observation equation to obtain the predicted SOC.

[0018] In another possible implementation, the formula of the transfer state equation is: SOC k is the SOC at the current moment, I k is the current at the current moment, C is the capacity of the battery to be predicted, t is the time interval, and the state transfer equation is calculated by the ampere-hour integration method.

[0019] In another possible implementation, the observation equation is: k =H·SOC k +∈, H is the measurement matrix, ∈ is the observation noise.

[0020] In a second aspect, a system for battery SOC prediction is provided, the system comprising:

[0021] A battery data acquisition module is used to acquire battery data of the battery to be predicted during the period of use, and perform short-time Fourier transform (STFT) on the battery data to acquire time-frequency domain features, wherein the battery data includes: voltage data during the period of use, current data during the period of use, and temperature data during the period of use;

[0022] A predicted voltage acquisition module, used to process the time-frequency domain features through a trained convolutional neural network CNN to obtain the predicted voltage at the next moment of the used time period;

[0023] The predicted SOC acquisition module is used to process the predicted voltage through an unscented Kalman filter UKF to obtain a predicted SOC as a final prediction result.

[0024] In a possible implementation, the predicted voltage acquisition module includes:

[0025] A trained CNN acquisition unit, used to train a preset initial CNN through the time-frequency domain features to acquire the trained CNN;

[0026] A predicted voltage acquisition unit is used to process the time-frequency domain features through the trained CNN to obtain the predicted voltage.

[0027] In another possible implementation, the predicted SOC acquisition module includes:

[0028] A setting unit, used for setting the state transfer equation and observation equation of the UKF;

[0029] An intermediate data acquisition unit, used for processing the predicted voltage according to the state transfer equation to acquire intermediate data;

[0030] A predicted SOC acquisition unit is used to update the intermediate data according to the observation equation to obtain the predicted SOC.

[0031] In a third aspect, an electronic device is provided, comprising 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 method for battery SOC prediction as provided in the first aspect is implemented.

[0032] In a fourth aspect, a non-transitory computer-readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the method for battery SOC prediction as provided in the first aspect is implemented.

[0033] The beneficial effect of the present application is that the trained neural network CNN is integrated into UKF to realize data fusion calculation, which can accurately predict the SOC of electrical equipment in a low-temperature and high-pressure deep-sea environment. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in describing the embodiments of the present application are briefly introduced below.

[0035] Figure 1 A flowchart of a method for battery SOC prediction provided by one embodiment of the present invention;

[0036] Figure 2 A flowchart of steps for obtaining a first predicted SOC provided by one embodiment of the present invention;

[0037] Figure 3 A flowchart of steps for obtaining a second predicted SOC provided by one embodiment of the present invention;

[0038] Figure 4 A structural diagram of a battery SOC prediction system provided by an embodiment of the present invention;

[0039] Figure 5 A schematic diagram of the physical structure of an electronic device provided by the present invention.

[0040] Specific implementation method

[0041] The embodiments of the present application are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar modules or modules with the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present application, and cannot be interpreted as limiting the present invention.

[0042] It will be understood by those skilled in the art that, unless expressly stated, the singular forms "a", "an", "said" and "the" used herein may also include plural forms. It should be further understood that the term "comprising" used in the specification of the present application refers to the presence of the features, integers, steps, operations, modules and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, modules, components and / or groups thereof. It should be understood that when we refer to a module as being "connected" or "coupled" to another module, it may be directly connected or coupled to the other modules, or there may be intermediate modules. In addition, the "connection" or "coupling" used herein may include wireless connection or wireless coupling. The term "and / or" used herein includes all or any modules and all combinations of one or more associated listed items.

[0043] In order to make the objectives, technical solutions and advantages of the present application clearer, the implementation method of the present application will be further described in detail below in conjunction with the accompanying drawings.

[0044] The technical solution of the present application and how to solve the above-mentioned technical problems are described in detail below with specific embodiments. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments. The embodiments of the present application will be described below in conjunction with the accompanying drawings.

[0045] like Figure 1 FIG. 1 is a flow chart of a method for battery SOC prediction provided by an embodiment of the present invention, the method comprising:

[0046] Step S101, obtaining battery data of a battery to be predicted for a time period in use, and performing a short-time Fourier transform (STFT) on the battery data to obtain time-frequency domain features, wherein the battery data includes: voltage data for the time period in use, current data for the time period in use, and temperature data for the time period in use;

[0047] Step S102, processing the time-frequency domain features by means of a trained convolutional neural network CNN to obtain a predicted voltage at the next moment of the used time period;

[0048] Step S103, processing the predicted voltage by using an unscented Kalman filter UKF to obtain a predicted SOC as a final prediction result.

[0049] In step S101, the battery data of the used time period of the battery to be tested is the basic information for battery SOC prediction, and the basic information includes but is not limited to: voltage data of the used time period, current data of the used time period, and temperature data of the used time period. The basic information is subjected to short-time Fourier transform STFT processing to obtain time-frequency domain characteristics. Among them, the function of STFT (short-time Fourier transform) is to retain the frequency analysis characteristics of Fourier transform. The frequency distribution in each time window after segmentation has a clear physical meaning, which helps to interpret the time-varying frequency characteristics of the signal and filter out the noise interference in the high-frequency band, so as to effectively capture the data of the battery under complex conditions such as low temperature and high pressure, and adapt to the dynamic changes of the battery under different working conditions.

[0050] It should be noted that the formula of the STFT is:

[0051]

[0052] Among them, e -j2πfτis a complex exponential, representing the frequency component of the Fourier transform of frequency f, w(t-τ) is the window function, x(t) is the input signal, and τ is the integral variable, which is used to represent the translation of the window function w in the entire time domain. This formula is the integral process of STFT, which obtains the characteristics of the frequency change of the signal over time by performing Fourier transform on the local signal at each time point.

[0053] like Figure 2 The flowchart of the steps of obtaining the predicted voltage provided by one embodiment of the present invention includes:

[0054] Step S201, training a preset initial CNN using the time-frequency domain features to obtain the trained CNN;

[0055] Step S202: Process the time-frequency domain features through the trained CNN to obtain the predicted voltage.

[0056] Among them, the output formula of the trained CNN is:

[0057] y=f(W·x+b)

[0058] Among them, y is the output, W is the weight matrix, x is the input data, b is the bias, and f is the activation function. After training with battery data, CNN can effectively extract features from the input data and generate a prediction of the voltage at the next moment.

[0059] like Figure 3 The flowchart of the steps of obtaining the predicted SOC provided by one embodiment of the present invention includes:

[0060] Step S301, setting the state transfer equation and observation equation of the UKF;

[0061] Step S302, processing the predicted voltage according to the state transfer equation to obtain intermediate data;

[0062] Step S303: Update the intermediate data according to the observation equation to obtain the predicted SOC.

[0063] The formula of the UKF state transition equation is:

[0064]

[0065] SOC k is the SOC at the current moment, I k is the current at the current moment, C is the capacity of the battery to be predicted, and t is the time interval. The state transfer equation is calculated using the ampere-hour integration method.

[0066] The formula of the observation equation of UKF is:

[0067] V k =H·SOC k +∈

[0068] H is the measurement matrix and ∈ is the observation noise.

[0069] The formula for predicting SOC in the intermediate state is:

[0070]

[0071] The formula for predicting SOC is:

[0072]

[0073] in, It represents the predicted value of the state at the kth moment based on the state at the k-1th moment. and control input u k The state prediction, H represents the observation matrix, P k|k-1 is the prediction covariance, R is the observation noise covariance, z k is the measured value at the current moment.

[0074] It should be pointed out that the trained convolutional neural network CNN is an integral part of UKF.

[0075] In an embodiment of the present invention, the battery data of the battery to be predicted for the time period in use is obtained, and the battery data is subjected to a short-time Fourier transform (STFT) to obtain time-frequency domain features, wherein the battery data includes: voltage data for the time period in use, current data for the time period in use, and temperature data for the time period in use; the time-frequency domain features are processed by a trained convolutional neural network (CNN) to obtain the predicted voltage at the next moment of the time period in use; the predicted voltage is processed by an unscented Kalman filter (UKF) to obtain the predicted SOC as the final prediction result. The trained neural network CNN is integrated into the UKF to realize data fusion calculation, which can accurately predict the SOC of electrical equipment in a low-temperature, high-pressure deep-sea environment.

[0076] For example:

[0077] First, the input time domain signal x(t) is divided into overlapping window segments. Set the window function w(t). Common window functions include Hamming window and Hanning window. These window functions help reduce spectrum leakage. Each window segment x w (t) can be expressed as:

[0078] x w (t) = x(t)·w(t-τ)

[0079] Here, T is the center position of the window. For each window segment, a Fourier transform is applied to obtain the corresponding frequency domain representation:

[0080]

[0081] Next, the calculated frequency domain representation X w (f) A time-frequency spectrum will be formed, where each column corresponds to a time window and each row corresponds to a frequency. The output of the entire short-time Fourier transform can be expressed as a two-dimensional matrix:

[0082]

[0083] Where T is the sampling interval and n is the window index.

[0084] Through short-time Fourier transform, the local features of the signal in the frequency domain can be extracted. This process provides rich feature information for the subsequent convolutional neural network (CNN) input, allowing the network to better capture the relationship between battery voltage and SOC.

[0085] Step 2, the input data preparation stage involves the battery voltage V(t), current I(t), temperature T(t) and their features after short-time Fourier transform (STFT) processing. Assume that we represent the input data as a matrix X, which contains observations of multiple time steps, which can be expressed as:

[0086]

[0087] When building a convolutional neural network, assuming that we define L convolutional layers and K pooling layers, the output feature map of the network can be expressed as:

[0088] F l (x) = σ(W l *x+b l )

[0089] Among them, F l (x) is the output feature map of the lth layer, W l is the convolution kernel, * indicates the convolution operation, b l is the bias term, σ is the activation function, and the ReLU activation function is used here:

[0090] In the pooling layer, the commonly used maximum pooling operation can be expressed as:

[0091] P k (F l )=max(F l (x)

[0092] Next, the model is trained using mean square error (MSE) as the loss function, defined as:

[0093]

[0094] Among them, y i is the real next step battery terminal voltage, is the battery terminal voltage predicted by the model, and N is the number of samples. In order to optimize the model parameters, the gradient descent method is applied to update the weights and biases. Specifically, for each parameter 0 (weight or bias), the update formula is:

[0095]

[0096] Among them, α is the learning rate, is the gradient of the loss function with respect to the parameters.

[0097] Finally, the trained model will generate a set of weights and biases that can be used to predict new input data. By passing the input data X into the trained CNN model, the battery terminal voltage at the next moment (t+1) is output:

[0098]

[0099] Unscented Kalman Filter (UKF) is an improved Kalman filter suitable for state prediction of nonlinear systems. The biggest difference between UKF and traditional Kalman filter is that it uses unscented transform (UT) to select a set of special sampling points (Sigma Points) for state prediction and update, thereby providing the best prediction of the system state. First, UKF uses the dynamic model of the system and the state prediction of the previous moment to predict the state at the current moment. Then, the difference between the predicted value and the actual observed value is used to correct and update the state prediction.

[0100] The operation process of the unscented Kalman filter is as follows:

[0101] Define a matrix χ containing 2n+1 sigma vectors based on the mean and covariance of the random vector x.

[0102]

[0103] Intermediate parameter Y i can be derived from the vector,

[0104] Y i =g(χ i ) i=0,…,2n

[0105] Mean of y and covariance P y Nonlinear transformation and weighting coefficients through sigma points The function is calculated.

[0106]

[0107] The state space equation of UKF is as follows:

[0108] x k+1 =f(x k ,u k )+w k

[0109] y k =h(x k ,u k )+ν k

[0110] Among them, x k is the state variable, u k is the input variable, y k is the output vector. f(.) and h(.) are the state transfer equation and observation equation, w k and ν k are process noise and measurement noise.

[0111] Q k and R K Given by:

[0112] Q k =E[w k w k T ]

[0113] R K =E[ν k ν k T ]

[0114] The calculation process of UKF algorithm is as follows:

[0115] Initialization:

[0116]

[0117] Predicted state variables:

[0118]

[0119] Update measurements:

[0120]

[0121] Calculate the Kalman gain, state variables, and error covariance:

[0122]

[0123] In the new framework, the following equations are used as the state space equation and the observation equation respectively:

[0124]

[0125] U t,k =h(U t,T,k-1 ,T L,k ,i L,k ,stft L,k )=CNN+R k

[0126] Where η is the charge and discharge efficiency, which is defined as 1 here; Δt is the time interval between adjacent sampling moments; subscripts k and k-1 represent the moment; i L,k is the battery current at time k; C a is the nominal battery capacity of the dataset; stft L,k It represents the time-frequency characteristics after short-time Fourier transform. The state equation is the discretized expression of the ampere-hour integration method, which performs idealized calculation of SOC. CNN represents the process of predicting the terminal voltage using the matrixed CNN network.

[0127] The entire SOC prediction process is as follows: the sensors in the UUV's front-end battery management system (BMS) collect battery terminal voltage, current, and temperature information. During testing, the corresponding currents under different working conditions can be input into the battery management system. The battery management system transmits the collected battery voltage, current, and temperature data to the CPU through CAN communication. After the STFT converter extracts the time-frequency features, they are sent to the observer model. The CNN network predicts the next terminal voltage. After initialization, state variable update, and measurement value correction in the UKF, the state variables are continuously updated to finally obtain an accurate SOC prediction.

[0128] like Figure 4 FIG. 1 is a structural diagram of a system for battery SOC prediction provided by an embodiment of the present invention, wherein the system comprises:

[0129] The battery data acquisition module 401 is used to acquire the battery data of the battery to be predicted during the period of use, and perform short-time Fourier transform (STFT) on the battery data to acquire time-frequency domain features, wherein the battery data includes: voltage data during the period of use, current data during the period of use, and temperature data during the period of use;

[0130] A predicted voltage acquisition module 402 is used to process the time-frequency domain features through a trained convolutional neural network CNN to obtain a predicted voltage at the next moment of the used time period;

[0131] The predicted SOC acquisition module 403 is used to process the predicted voltage through an unscented Kalman filter UKF to obtain a predicted SOC as a final prediction result.

[0132] The predicted voltage acquisition module 402 includes:

[0133] A trained CNN acquisition unit, used to train a preset initial CNN through the time-frequency domain features to acquire the trained CNN;

[0134] A predicted voltage acquisition unit is used to process the time-frequency domain features through the trained CNN to obtain the predicted voltage.

[0135] The predicted SOC acquisition module 403 includes:

[0136] A setting unit, used for setting the state transfer equation and observation equation of the UKF;

[0137] An intermediate data acquisition unit, used for processing the predicted voltage according to the state transfer equation to acquire intermediate data;

[0138] A predicted SOC acquisition unit is used to update the intermediate data according to the observation equation to obtain the predicted SOC.

[0139] Figure 5 An example of a physical structure diagram of an electronic device is shown in FIG. Figure 5 As shown, the electronic device may include: a processor (processor) 501, a communication interface (Communications Interface) 502, a memory (memory) 503 and a communication bus 504, wherein the processor, the communication interface and the memory complete mutual communication through the communication bus. The processor may call the logic instructions in the memory to execute the method for battery SOC prediction, the method comprising: obtaining the battery data of the used time period of the battery to be predicted, and performing a short-time Fourier transform STFT on the battery data to obtain time-frequency domain features, the battery data comprising: voltage data of the used time period, current data of the used time period, and temperature data of the used time period; processing the time-frequency domain features by a trained convolutional neural network CNN to obtain the predicted voltage at the next moment of the used time period; processing the predicted voltage by an unscented Kalman filter UKF to obtain the predicted SOC as the final prediction result.

[0140] In addition, the logic instructions in the above-mentioned memory can be implemented in the form of software functional units and can be stored in a computer-readable storage medium when sold or used as an independent product. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0141] On the other hand, an embodiment of the present invention further provides a computer program product, which includes a computer program stored on a non-transitory computer-readable storage medium, and the computer program includes program instructions. When the program instructions are executed by a computer, the computer can execute the battery SOC prediction method provided by the above-mentioned method embodiments, the method including: obtaining battery data of a battery to be predicted for a period of time that has been used, and performing a short-time Fourier transform (STFT) on the battery data to obtain time-frequency domain characteristics, the battery data including: voltage data for the period of time that has been used, current data for the period of time that has been used, and temperature data for the period of time that has been used; processing the time-frequency domain characteristics through a trained convolutional neural network CNN to obtain a predicted voltage at the next moment of the period of time that has been used; processing the predicted voltage through an unscented Kalman filter UKF to obtain a predicted SOC as the final prediction result.

[0142] On the other hand, an embodiment of the present invention further provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to execute the battery SOC prediction method provided in the above-mentioned embodiments, the method comprising: obtaining battery data of a battery to be predicted for a period of time that has been used, and performing a short-time Fourier transform (STFT) on the battery data to obtain time-frequency domain features, the battery data comprising: voltage data for the period of time that has been used, current data for the period of time that has been used, and temperature data for the period of time that has been used; processing the time-frequency domain features through a trained convolutional neural network (CNN) to obtain a predicted voltage at the next moment of the period of time that has been used; processing the predicted voltage through an unscented Kalman filter (UKF) to obtain a predicted SOC as a final prediction result.

[0143] It should be understood that, although the steps in the flowchart of the accompanying drawings are displayed in sequence as indicated by the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least a part of the steps in the flowchart of the accompanying drawings may include multiple sub-steps or multiple stages, and these sub-steps or stages are not necessarily executed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be executed in turn or alternately with other steps or at least a part of the sub-steps or stages of other steps.

[0144] The above is only a partial implementation 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 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 battery SOC prediction, characterized in that: The method comprises: Obtaining battery data of the battery to be predicted for the time period in use, and performing short-time Fourier transform (STFT) on the battery data to obtain time-frequency domain features, wherein the battery data includes: voltage data for the time period in use, current data for the time period in use, and temperature data for the time period in use; Processing the time-frequency domain features by means of a trained convolutional neural network (CNN) to obtain a predicted voltage at the next moment of the used time period; The predicted voltage is processed by an unscented Kalman filter UKF to obtain a predicted SOC as a final prediction result.

2. The method according to claim 1, characterized in that The step of processing the time-frequency domain features by using a trained convolutional neural network (CNN) to obtain a predicted voltage at the next moment of the used time period includes: Training a preset initial CNN using the time-frequency domain features to obtain the trained CNN; The time-frequency domain features are processed by the trained CNN to obtain the predicted voltage.

3. The method according to claim 1, characterized in that The step of processing the predicted voltage by using an unscented Kalman filter UKF to obtain a predicted SOC as a final prediction result includes: Setting the state transfer equation and observation equation of the UKF; Processing the predicted voltage according to the state transfer equation to obtain intermediate data; The intermediate data is updated according to the observation equation to obtain the predicted SOC.

4. The method according to claim 3, characterized in that The formula of the transfer state equation is: SOC k is the SOC at the current moment, I k is the current at the current moment, C is the capacity of the battery to be predicted, t is the time interval, and the state transfer equation is calculated by the ampere-hour integration method.

5. The method according to claim 3, characterized in that The observation equation is: V k =H·SOC k +∈, H is the measurement matrix, ∈ is the observation noise.

6. A battery SOC prediction system, characterized in that: The system comprises: A battery data acquisition module is used to acquire battery data of the battery to be predicted during the period of use, and perform short-time Fourier transform (STFT) on the battery data to acquire time-frequency domain features, wherein the battery data includes: voltage data during the period of use, current data during the period of use, and temperature data during the period of use; A predicted voltage acquisition module, used to process the time-frequency domain features through a trained convolutional neural network CNN to obtain the predicted voltage at the next moment of the used time period; The predicted SOC acquisition module is used to process the predicted voltage through an unscented Kalman filter UKF to obtain a predicted SOC as a final prediction result.

7. The system according to claim 6, characterized in that The predicted voltage acquisition module comprises: A trained CNN acquisition unit, used to train a preset initial CNN through the time-frequency domain features to acquire the trained CNN; A predicted voltage acquisition unit is used to process the time-frequency domain features through the trained CNN to obtain the predicted voltage.

8. The system according to claim 6, characterized in that The predicted SOC acquisition module includes: A setting unit, used for setting the state transfer equation and observation equation of the UKF; An intermediate data acquisition unit, used for processing the predicted voltage according to the state transfer equation to acquire intermediate data; A predicted SOC acquisition unit is used to update the intermediate data according to the observation equation to obtain the predicted SOC.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the method for battery SOC prediction according to any one of claims 1 to 5 is implemented.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for battery SOC prediction according to any one of claims 1 to 5 is implemented.

Images