Battery state of charge prediction method, apparatus, device, medium, and product

Abstract

This application relates to a method, apparatus, computer device, computer-readable storage medium, and computer program product for predicting the state of charge (SOC) of a battery, applied in the field of big data technology. The method includes: acquiring battery operating data of a target battery; extracting features from the pressure signal in the battery operating data to obtain a pressure feature vector; extracting time-series features from the electrical time-series data in the battery operating data to obtain an electrical time-series feature vector; fusing the pressure feature vector and the electrical time-series feature vector to obtain a fused vector; and predicting the battery charge of the target battery based on the fused vector to obtain the battery SOC. This method can improve the accuracy of battery SOC prediction.

Description

Technical Field

[0001] This application relates to the field of big data technology, and in particular to a method, apparatus, computer device, computer-readable storage medium, and computer program product for predicting the state of charge of a battery. Background Technology

[0002] With the rapid development of new energy technologies, lithium batteries are widely used in electric vehicles, energy storage power stations, portable electronic devices, and many other fields. The state of charge (SOC) of a battery, as a core indicator reflecting the remaining capacity of the battery, directly affects battery safety, driving range estimation, and battery lifespan. Accurate SOC prediction can effectively avoid overcharging and over-discharging, improve energy efficiency, and ensure the stable operation of electrical equipment.

[0003] Traditional battery state-of-charge (SOC) prediction methods typically rely solely on electrical time-series data such as battery voltage and current for estimation. However, during the charging and discharging process of lithium batteries, the voltage signal exhibits a significant flatness in the mid-range, making it susceptible to noise, temperature, and aging effects. This makes it difficult to guarantee prediction accuracy using only electrical time-series data. Simultaneously, batteries undergo physical deformations such as expansion and contraction during charging and discharging. Pressure signals directly reflect changes in the internal lithium-ion concentration of the battery, exhibiting high sensitivity and robustness. However, current technologies have not effectively integrated pressure signals into the battery SOC prediction process and lack lightweight feature extraction and efficient multi-source feature fusion mechanisms for pressure signals.

[0004] Furthermore, existing battery charge prediction models use a relatively simple method to fuse electrical time-series features and pressure physical features, directly fusing the electrical time-series features and pressure features. The fused features contain less information, resulting in poor accuracy in predicting the battery state of charge. Summary of the Invention

[0005] Therefore, it is necessary to provide a battery state of charge prediction method, apparatus, computer equipment, computer-readable storage medium, and computer program product that can improve the prediction accuracy of battery state of charge in response to the above-mentioned technical problems.

[0006] In a first aspect, this application provides a method for predicting the state of charge of a battery, including:

[0007] Obtain the battery operating data of the target battery;

[0008] The pressure signal in the battery operation data is extracted to obtain a pressure feature vector;

[0009] The electrical time-series data in the battery operation data are subjected to time-series feature extraction to obtain an electrical time-series feature vector;

[0010] The pressure feature vector and the electrical time-series feature vector are fused to obtain a fused vector;

[0011] Based on the fusion vector, the battery capacity of the target battery is predicted to obtain the battery state of charge.

[0012] In one embodiment, the step of extracting features from the pressure signal in the battery operating data to obtain a pressure feature vector includes:

[0013] The pressure frequency domain features are obtained by performing spectrum extraction on the pressure signal in the battery operation data;

[0014] The noise in the pressure signal of the battery operation data is smoothed to obtain denoising features;

[0015] The pressure frequency domain features and the denoised features are concatenated to obtain the pressure feature vector.

[0016] In one embodiment, the step of extracting time-series features from the electrical time-series data in the battery operation data to obtain an electrical time-series feature vector includes:

[0017] The electrical time-series data in the battery operation data is segmented to obtain multiple electrical time-series segmented data;

[0018] For each of the aforementioned electrical timing segmentation data, the electrical timing segmentation data is transformed into the hidden layer representation space to obtain transformed data;

[0019] Multiple attention feature extraction is performed on the transformed data corresponding to each of the aforementioned electrical time-series segmentation data to obtain multiple attention feature data;

[0020] The multiple attention feature data are fused to obtain the electrical time-series feature vector.

[0021] In one embodiment, fusing the pressure feature vector and the electrical time-series feature vector to obtain a fused vector includes:

[0022] A first mapping vector is obtained by linearly mapping the pressure feature vector; and a second mapping vector is obtained by linearly mapping the electrical timing feature vector.

[0023] The first mapping vector and the second mapping vector are weighted and fused to obtain a fused vector.

[0024] In one embodiment, predicting the battery capacity of the target battery based on the fusion vector to obtain the battery state of charge includes:

[0025] Based on the linear mapping parameters, the fusion vector is linearly mapped to obtain the third mapping vector;

[0026] The third mapping vector is biased using bias parameters to obtain the battery state of charge.

[0027] In one embodiment, the method further includes:

[0028] Obtain the operating conditions corresponding to the target battery;

[0029] The linear mapping parameters are updated based on the operating conditions.

[0030] Secondly, this application also provides a battery state of charge prediction device, comprising:

[0031] The acquisition module is used to acquire the battery operating data of the target battery;

[0032] The feature extraction module is used to extract features from the pressure signal in the battery operation data to obtain a pressure feature vector; and to extract time-series features from the electrical time-series data in the battery operation data to obtain an electrical time-series feature vector.

[0033] The feature fusion module is used to fuse the pressure feature vector and the electrical time-series feature vector to obtain a fused vector;

[0034] The prediction module predicts the battery capacity of the target battery based on the fusion vector to obtain the battery state of charge.

[0035] Thirdly, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to perform the following steps:

[0036] Obtain the battery operating data of the target battery;

[0037] The pressure signal in the battery operation data is extracted to obtain a pressure feature vector;

[0038] The electrical time-series data in the battery operation data are subjected to time-series feature extraction to obtain an electrical time-series feature vector;

[0039] The pressure feature vector and the electrical time-series feature vector are fused to obtain a fused vector;

[0040] Based on the fusion vector, the battery capacity of the target battery is predicted to obtain the battery state of charge.

[0041] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, performs the following steps:

[0042] Obtain the battery operating data of the target battery;

[0043] The pressure signal in the battery operation data is extracted to obtain a pressure feature vector;

[0044] The electrical time-series data in the battery operation data are subjected to time-series feature extraction to obtain an electrical time-series feature vector;

[0045] The pressure feature vector and the electrical time-series feature vector are fused to obtain a fused vector;

[0046] Based on the fusion vector, the battery capacity of the target battery is predicted to obtain the battery state of charge.

[0047] Fifthly, this application also provides a computer program product, including a computer program that, when executed by a processor, performs the following steps:

[0048] Obtain the battery operating data of the target battery;

[0049] The pressure signal in the battery operation data is extracted to obtain a pressure feature vector;

[0050] The electrical time-series data in the battery operation data are subjected to time-series feature extraction to obtain an electrical time-series feature vector;

[0051] The pressure feature vector and the electrical time-series feature vector are fused to obtain a fused vector;

[0052] Based on the fusion vector, the battery capacity of the target battery is predicted to obtain the battery state of charge.

[0053] The aforementioned battery state of charge prediction method, apparatus, computer equipment, computer-readable storage medium, and computer program product acquire battery operating data of a target battery; extract features from the pressure signal in the battery operating data to obtain a pressure feature vector; extract time-series features from the electrical time-series data in the battery operating data to obtain an electrical time-series feature vector; fuse the pressure feature vector and the electrical time-series feature vector to obtain a fused vector; and predict the battery charge of the target battery based on the fused vector to obtain the battery state of charge.

[0054] Thus, by introducing the highly sensitive physical feature of pressure signal, the deficiency of insufficient prediction accuracy of voltage signal in the middle range can be effectively compensated; by fusing multi-source features, the fused vector can represent both the time-series characteristics of electricity and the pressure characteristics, thereby improving the accuracy of battery state of charge prediction by using the fused vector with rich content. Attached Figure Description

[0055] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the description of the embodiments of this application or related technologies will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0056] Figure 1 This is a diagram illustrating the application environment of a battery state-of-charge prediction method in one embodiment.

[0057] Figure 2 This is a flowchart illustrating a battery state of charge prediction method in one embodiment;

[0058] Figure 3 This is a structural block diagram of a battery state of charge prediction device in one embodiment;

[0059] Figure 4 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0060] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0061] It should be noted that all information and data involved in this application (including but not limited to data used for analysis, stored data, and displayed data) are information and data authorized by the user or fully authorized by all parties, and the acquisition, transmission, storage, use, and processing of related data comply with the relevant provisions of national laws and regulations. Users can refuse content pushed to them (e.g., battery state of charge). In the embodiments of this application, certain existing industry solutions such as software, components, and models may be mentioned. These should be considered exemplary, intended only to illustrate the feasibility of implementing the technical solution of this application, and do not imply that the applicant has already used or necessarily used such solutions.

[0062] Understandably, the state of charge (SOC) of a battery is a crucial function of the battery management system (BMS). Accurately estimating the SOC can effectively prevent or mitigate the negative effects of overcharging or over-discharging, thereby extending battery life. Therefore, after predicting the SOC of the target battery, the SOC is sent to the controller of the vehicle corresponding to the target battery. The vehicle controller can then perform appropriate battery charging and discharging control based on the SOC. This, in turn, extends the vehicle's battery life.

[0063] The battery state-of-charge prediction method provided in this application can be applied to, for example... Figure 1 In the application environment shown, the controller of vehicle 102 communicates with server 104 via a network. A data storage system can store the data that server 104 needs to process. The data storage system can be integrated onto server 104 or placed on a cloud or other network server. Server 104 acquires the battery operating data of the target battery; extracts features from the pressure signal in the battery operating data to obtain a pressure feature vector; extracts time-series features from the electrical time-series data in the battery operating data to obtain an electrical time-series feature vector; fuses the pressure feature vector and the electrical time-series feature vector to obtain a fused vector; and predicts the battery charge of the target battery based on the fused vector to obtain the battery state of charge. Server 104 can be a standalone physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing cloud computing services. Server 104 sends the battery state of charge to the controller of vehicle 102 so that the controller of vehicle 102 can control the charging and discharging of the target battery according to the battery state of charge.

[0064] In one exemplary embodiment, such as Figure 2 As shown, a method for predicting the state of charge of a battery is provided, and this method is applied to... Figure 1 Taking server 104 as an example, the explanation includes the following steps 202 to 210. Wherein:

[0065] Step 202: Obtain the battery operation data of the target battery.

[0066] In step 202, the target battery is a lithium battery. The target battery can be a battery inside the vehicle or a battery inside the terminal, and there is no restriction here.

[0067] The battery operating data in step 202 includes at least one of a pressure signal and an electrical signal that varies over time, and the electrical signal includes at least one of a current signal and a voltage signal.

[0068] Step 204: Extract features from the pressure signal in the battery operation data to obtain a pressure feature vector.

[0069] As an embodiment, step 204 includes: extracting the frequency spectrum of the pressure signal in the battery operation data to obtain pressure frequency domain features; smoothing the noise of the pressure signal in the battery operation data to obtain denoising features; and concatenating the pressure frequency domain features and the denoising features to obtain a pressure feature vector.

[0070] Furthermore, the pressure signal in the battery operation data is subjected to spectral extraction to obtain pressure frequency domain features, including: extracting the pressure signal in the battery operation data by short-time Fourier transform to obtain pressure frequency domain features.

[0071] Optionally, the pressure frequency domain features obtained by extracting the pressure signal from the battery operation data through short-time Fourier transform can be expressed by the following formula:

[0072]

[0073] in, The frequency domain characteristics are those of pressure. This is a pressure signal. The time variable for scanning during actual integration. A window function is used to limit the time range of the transformation. For frequency.

[0074] As one embodiment, the noise of the pressure signal in the battery operation data is smoothed to obtain denoising features, including: filtering the pressure signal in the battery operation data to obtain denoising features.

[0075] The filtering process can be implemented through a filter, which can be a Savitzky-Golay filter (a sliding window smoothing algorithm based on local polynomial least squares fitting).

[0076] As one embodiment, the denoising feature is obtained by filtering the pressure signal in the battery operation data, including: windowing the pressure signal in the battery operation data to obtain multiple window pressure signals, and weighting, normalizing and fusing the multiple pressure window signals to obtain the denoising feature.

[0077] Optionally, the multiple pressure window signals are weighted, normalized, and fused to obtain the denoising features, which can be expressed by the following formula:

[0078]

[0079] in, For denoising features, This is a normalization constant for the window size. For window size, The weighting coefficients are obtained through polynomial fitting.

[0080] Optionally, the pressure frequency domain features and the denoising features can be concatenated to obtain a pressure feature vector, which can be achieved through the LFE (Local Feature Enhancement, a module used to enhance the local feature representation capability of convolutional neural networks (CNN)) component.

[0081] Step 206: Extract time-series features from the electrical time-series data in the battery operation data to obtain an electrical time-series feature vector.

[0082] For example, step 206 includes: segmenting the electrical time-series data in the battery operation data to obtain multiple electrical time-series segmented data; for each of the electrical time-series segmented data, converting the electrical time-series segmented data to the hidden layer representation space to obtain converted data; performing multi-attention feature extraction on the converted data corresponding to each of the electrical time-series segmented data to obtain multiple attention feature data; and fusing the multiple attention feature data to obtain an electrical time-series feature vector.

[0083] As one embodiment, the electrical timing data includes multiple current signals and multiple voltage signals that change over time; the electrical timing data in the battery operation data is segmented to obtain multiple electrical timing segmented data, including: combining current signals and voltage signals of the same timing in the battery operation data to obtain multiple combined signals; sorting the multiple combined signals according to their respective timing to obtain electrical combined data; and segmenting the electrical combined data according to a preset segmentation length to obtain multiple electrical timing segmented data.

[0084] The preset cutting length can be set as needed or based on experience; no restrictions are imposed here.

[0085] As one embodiment, converting the electrical time-series segmented data to the hidden layer representation space to obtain converted data includes: obtaining the weights and biases of the linear layer in the hidden layer; weighting the electrical time-series segmented data using the weights of the linear layer to obtain segmented weighted data; and fusing the biases of the linear layer and the segmented weighted data to obtain converted data.

[0086] Optionally, the bias and split-weighted data of the linear layer can be fused to obtain the transformed data, which can be expressed by the following formula:

[0087]

[0088] in, To transform the data, These are the weights of the linear layers in the hidden layers. Electrical timing data segmentation, This is the bias of the linear layer in the hidden layer.

[0089] Among them, the transformation data corresponding to each of the electrical time-series segmentation data is subjected to multi-attention feature extraction to obtain multiple attention feature data, which can be executed through the Moirai 2.0 structure. Moirai 2.0 includes a multi-layer stacked decoder-type Transformer Block (with masked multi-head attention and feedforward network combination) structure.

[0090] As an embodiment, multi-attention feature extraction is performed on the transformation data corresponding to each of the electrical timing segmentation data to obtain multiple attention feature data, including: for each masked multi-head attention layer, obtaining the query linear projection matrix, key linear projection matrix, and value linear projection matrix of the masked multi-head attention layer; mapping the hidden state sequence of the previous masked multi-head attention layer through the query linear projection matrix to obtain the query state sequence, mapping the hidden state sequence of the previous masked multi-head attention layer through the key linear projection matrix to obtain the key state sequence, mapping the hidden state sequence of the previous masked multi-head attention layer through the value linear projection matrix to obtain the value state sequence; determining the attention weight of the masked multi-head attention layer based on the key state sequence and the query state sequence, and fusing the value state sequence and the attention weight to obtain the attention feature data of the masked multi-head attention layer.

[0091] The attention weights of the masked multi-head attention layer can be expressed by the following formula:

[0092]

[0093] in, To mask the attention weights of the multi-head attention layer, To query the state sequence, For key state sequences, To mask the attention dimension of the multi-head attention layer, For causal masking.

[0094] As one embodiment, fusing the multiple attention feature data to obtain an electrical time-series feature vector includes: fusing the multiple attention feature data through an information fusion matrix to obtain an electrical time-series feature vector.

[0095] Optionally, the output data of the masked multi-head attention layer can be expressed by the following formula:

[0096]

[0097] in, This is the output data of the masked multi-head attention layer. For normalization function, This is the hidden state sequence of the previous attention layer. This is a multi-head attention function.

[0098] Optionally, the output of the feedforward network in the masked multi-head attention layer can be expressed by the following formula:

[0099]

[0100] in, This is the output of the feedforward network for masking the multi-head attention layer. The weights of the second linear layer, For activation function, The weights of the first linear layer, The bias is for the first-level linear transformation. This is the bias for the second-level linear transformation.

[0101] The Moirai 2.0 structure does not participate in training and can be defined as a fixed function.

[0102] Alternatively, when Moirai 2.0 is defined as a fixed function, it can be expressed by the following formula:

[0103]

[0104] in, This is the output data for Moirai 2.0. For the Moirai 2.0 prediction function, It is electrical timing data. Fixed parameters for pre-training.

[0105] The Moirai 2.0 architecture is not used in training; therefore, the parameters in Moirai 2.0 have the following relationship:

[0106]

[0107] Step 208: The pressure feature vector and the electrical timing feature vector are fused to obtain a fused vector.

[0108] As one embodiment, step 208 includes: performing a linear mapping on the pressure feature vector to obtain a first mapping vector; and performing a linear mapping on the electrical timing feature vector to obtain a second mapping vector; and weighting and fusing the first mapping vector and the second mapping vector to obtain a fused vector.

[0109] As one embodiment, linearly mapping the pressure feature vector to obtain a first mapping vector includes: obtaining a first mapping matrix and a first bias term; mapping the pressure feature vector through the first mapping matrix to obtain a pressure mapping feature vector; and fusing the pressure mapping feature vector with the first bias term to obtain the first mapping vector.

[0110] Optionally, the pressure mapping feature vector can be fused with the first bias term to obtain the first mapping vector, which can be expressed by the following formula:

[0111]

[0112] in, For the first mapping vector, This is the first mapping matrix. This is the first bias term.

[0113] As one embodiment, linearly mapping the electrical timing feature vector to obtain a second mapping vector includes: obtaining a second mapping matrix and a second bias term; mapping the electrical timing feature vector through the second mapping matrix to obtain an electrical timing mapped feature vector; and fusing the electrical timing mapped feature vector with the second bias term to obtain the second mapping vector.

[0114] Optionally, the electrical timing mapping feature vector is fused with the second bias term to obtain the second mapping vector, which can be expressed by the following formula:

[0115]

[0116] in, For the second mapping vector, This is the second mapping matrix. This is the second bias term.

[0117] As one embodiment, the first mapping vector and the second mapping vector are weighted and fused to obtain a fused vector, including: obtaining a first gating parameter and a second gating parameter, wherein the sum of the first gating parameter and the second gating parameter is 1; multiplying the first gating parameter by the first mapping vector element by element to obtain a first gating vector; multiplying the second gating parameter by the second mapping vector element by element to obtain a second gating vector; and fusing the first gating vector and the second gating vector to obtain a fused vector.

[0118] Optionally, the first gating vector and the second gating vector can be fused to obtain a fused vector, which can be expressed by the following formula:

[0119]

[0120] in, For the fusion vector, This is an element-wise multiplication function. This is the second gating parameter. This is the first gating parameter.

[0121] Optionally, the first gating parameter and the second gating parameter can be fixed empirical values ​​or dynamically learned values, without any restrictions.

[0122] Step 210: Based on the fusion vector, predict the battery capacity of the target battery to obtain the battery state of charge.

[0123] For example, step 210 includes: performing a linear mapping on the fusion vector according to the linear mapping parameters to obtain a third mapping vector; and performing a biasing process on the third mapping vector using bias parameters to obtain the battery state of charge.

[0124] Optionally, the third mapping vector can be biased using bias parameters to obtain the battery state of charge, which can be expressed by the following formula:

[0125]

[0126] in, The battery is in its state of charge. For linear mapping parameters, This is the bias parameter.

[0127] Optionally, the above method further includes: obtaining the operating conditions corresponding to the target battery; and updating the linear mapping parameters according to the operating conditions.

[0128] As one embodiment, updating the linear mapping parameters according to the operating conditions includes: obtaining a forgetting factor, updating the covariance matrix based on the forgetting factor and the fusion vector; determining the target battery state of charge corresponding to the target battery based on the operating conditions, and updating the linear mapping parameters based on the updated covariance matrix and the target battery state of charge.

[0129] Optionally, the covariance matrix is ​​updated based on the forgetting factor and the fusion vector, which can be expressed by the following formula:

[0130]

[0131] in, The updated covariance matrix, Let covariance be the value at the previous time step. It is a forgetting factor.

[0132] Optionally, the linear mapping parameters can be updated based on the updated covariance matrix and the target battery state of charge, which can be expressed by the following formula:

[0133]

[0134] in, For the updated linear mapping parameters, These are the linear mapping parameters from the previous time step. The target battery state of charge.

[0135] In the aforementioned battery state of charge (SOC) prediction method, battery operating data of the target battery is acquired; features of the pressure signal in the battery operating data are extracted to obtain a pressure feature vector; time-series features of the electrical time-series data in the battery operating data are extracted to obtain an electrical time-series feature vector; the pressure feature vector and the electrical time-series feature vector are fused to obtain a fused vector; and the battery charge of the target battery is predicted based on the fused vector to obtain the battery SOC. Thus, by introducing the highly sensitive physical feature of the pressure signal, the deficiency of insufficient prediction accuracy of the voltage signal in the mid-range is effectively compensated for; through the fusion of multi-source features, the fused vector can represent both the electrical time-series features and the pressure features, thereby improving the accuracy of battery SOC prediction by using a fused vector with rich representation.

[0136] As a detailed embodiment, battery operating data of the target battery is acquired; the pressure signal in the battery operating data is subjected to spectrum extraction to obtain pressure frequency domain features; the noise in the pressure signal in the battery operating data is smoothed to obtain denoising features; the pressure frequency domain features and the denoising features are concatenated to obtain a pressure feature vector; the electrical time-series data in the battery operating data is segmented to obtain multiple electrical time-series segmented data; for each electrical time-series segmented data, the electrical time-series segmented data is transformed to a hidden layer representation space to obtain transformed data; and the corresponding electrical time-series segmented data is processed accordingly. The corresponding transformed data is subjected to multi-attention feature extraction to obtain multiple attention feature data; the multiple attention feature data are fused to obtain an electrical time series feature vector; the pressure feature vector is linearly mapped to obtain a first mapping vector; and the electrical time series feature vector is linearly mapped to obtain a second mapping vector; the first mapping vector and the second mapping vector are weighted and fused to obtain a fused vector; according to the linear mapping parameters, the fused vector is linearly mapped to obtain a third mapping vector; the third mapping vector is biased by the bias parameters to obtain the battery state of charge.

[0137] Thus, by acquiring the battery operating data of the target battery; extracting features from the pressure signal in the battery operating data to obtain a pressure feature vector; extracting time-series features from the electrical time-series data in the battery operating data to obtain an electrical time-series feature vector; fusing the pressure feature vector and the electrical time-series feature vector to obtain a fused vector; and predicting the battery charge of the target battery based on the fused vector to obtain the battery state of charge. In this way, by introducing the highly sensitive physical feature of the pressure signal, the deficiency of insufficient prediction accuracy of the voltage signal in the mid-range is effectively compensated for; through the fusion of multi-source features, the fused vector can represent both the electrical time-series features and the pressure features, thereby improving the accuracy of battery state of charge prediction by using a fused vector with rich representation.

[0138] It should be understood that although the steps in the flowcharts of the above embodiments are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the above embodiments may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0139] Based on the same inventive concept, this application also provides a battery state of charge prediction device for implementing the battery state of charge prediction method described above. The solution provided by this device is similar to the solution described in the above method; therefore, the specific limitations in one or more battery state of charge prediction device embodiments provided below can be found in the limitations of the battery state of charge prediction method described above, and will not be repeated here.

[0140] In one exemplary embodiment, such as Figure 3 As shown, a battery state of charge prediction device 300 is provided, including: an acquisition module 302, a feature extraction module 304, a feature fusion module 306, and a prediction module 308, wherein:

[0141] The acquisition module 302 is used to acquire the battery operating data of the target battery;

[0142] The feature extraction module 304 is used to extract features from the pressure signal in the battery operation data to obtain a pressure feature vector; and to extract time-series features from the electrical time-series data in the battery operation data to obtain an electrical time-series feature vector.

[0143] The feature fusion module 306 is used to fuse the pressure feature vector and the electrical time-series feature vector to obtain a fused vector;

[0144] The prediction module 308 predicts the battery capacity of the target battery based on the fusion vector to obtain the battery state of charge.

[0145] In one embodiment, the feature extraction module 304 is further configured to perform spectrum extraction on the pressure signal in the battery operation data to obtain pressure frequency domain features; perform noise smoothing on the pressure signal in the battery operation data to obtain denoising features; and concatenate the pressure frequency domain features and the denoising features to obtain a pressure feature vector.

[0146] In one embodiment, the feature extraction module 304 is further configured to segment the electrical time-series data in the battery operation data to obtain multiple electrical time-series segmented data; for each of the electrical time-series segmented data, the electrical time-series segmented data is converted to the hidden layer representation space to obtain converted data; multiple attention features are extracted from the converted data corresponding to each of the electrical time-series segmented data to obtain multiple attention feature data; and the multiple attention feature data are fused to obtain an electrical time-series feature vector.

[0147] In one embodiment, the feature fusion module 306 is further configured to perform linear mapping on the pressure feature vector to obtain a first mapping vector; and to perform linear mapping on the electrical timing feature vector to obtain a second mapping vector; and to perform weighted fusion of the first mapping vector and the second mapping vector to obtain a fusion vector.

[0148] In one embodiment, the prediction module 308 is further configured to perform a linear mapping on the fusion vector according to the linear mapping parameters to obtain a third mapping vector; and to perform bias processing on the third mapping vector through the bias parameters to obtain the battery state of charge.

[0149] In one embodiment, the battery state of charge prediction device further includes: an update module, configured to acquire the operating condition corresponding to the target battery; and update the linear mapping parameters according to the operating condition.

[0150] Each module in the aforementioned battery state of charge prediction device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the corresponding operations of each module.

[0151] In one exemplary embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be as follows: Figure 4 As shown, the computer device includes a processor, memory, input / output interfaces, a communication interface, a display unit, and an input device. The processor, memory, and input / output interfaces are connected via a system bus, and the communication interface, display unit, and input device are also connected to the system bus via the input / output interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The input / output interfaces are used for exchanging information between the processor and external devices. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, Near Field Communication (NFC), or other technologies. When executed by the processor, the computer program implements a battery state-of-charge prediction method. The display unit is used to form a visually visible image and can be a display screen, a projection device, or a virtual reality imaging device. The display screen can be an LCD screen or an e-ink screen. The input device of the computer device can be a touch layer covering the display screen, or buttons, trackballs, or touchpads set on the casing of the computer device, or external keyboards, touchpads, or mice, etc.

[0152] Those skilled in the art will understand that Figure 4 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0153] In one exemplary embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method embodiments.

[0154] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps in the above method embodiments.

[0155] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above method embodiments.

[0156] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, artificial intelligence (AI) processors, etc., and are not limited to these.

[0157] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0158] The above embodiments are merely illustrative of several implementation methods of this application, and their descriptions are relatively specific and detailed. However, they should not be construed as limiting the scope of this application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A method for predicting the state of charge of a battery, characterized in that, The method includes: Obtain the battery operating data of the target battery; The pressure signal in the battery operation data is extracted to obtain a pressure feature vector; The electrical time-series data in the battery operation data are subjected to time-series feature extraction to obtain an electrical time-series feature vector; The pressure feature vector and the electrical time-series feature vector are fused to obtain a fused vector; Based on the fusion vector, the battery capacity of the target battery is predicted to obtain the battery state of charge.

2. The method according to claim 1, characterized in that, The step of extracting features from the pressure signal in the battery operating data to obtain a pressure feature vector includes: The pressure frequency domain features are obtained by performing spectrum extraction on the pressure signal in the battery operation data; The noise in the pressure signal of the battery operation data is smoothed to obtain denoising features; The pressure frequency domain features and the denoised features are concatenated to obtain the pressure feature vector.

3. The method according to claim 1, characterized in that, The step of extracting time-series features from the electrical time-series data in the battery operation data to obtain an electrical time-series feature vector includes: The electrical time-series data in the battery operation data is segmented to obtain multiple electrical time-series segmented data; For each of the aforementioned electrical timing segmentation data, the electrical timing segmentation data is transformed into the hidden layer representation space to obtain transformed data; Multiple attention feature extraction is performed on the transformed data corresponding to each of the aforementioned electrical time-series segmentation data to obtain multiple attention feature data; The multiple attention feature data are fused to obtain the electrical time-series feature vector.

4. The method according to claim 1, characterized in that, The step of fusing the pressure feature vector and the electrical time-series feature vector to obtain a fused vector includes: A first mapping vector is obtained by linearly mapping the pressure feature vector; and a second mapping vector is obtained by linearly mapping the electrical timing feature vector. The first mapping vector and the second mapping vector are weighted and fused to obtain a fused vector.

5. The method according to claim 1, characterized in that, The step of predicting the battery capacity of the target battery based on the fusion vector to obtain the battery state of charge includes: Based on the linear mapping parameters, the fusion vector is linearly mapped to obtain the third mapping vector; The third mapping vector is biased using bias parameters to obtain the battery state of charge.

6. The method according to claim 5, characterized in that, The method further includes: Obtain the operating conditions corresponding to the target battery; The linear mapping parameters are updated based on the operating conditions.

7. A battery state of charge prediction device, characterized in that, The device includes: The acquisition module is used to acquire the battery operating data of the target battery; The feature extraction module is used to extract features from the pressure signal in the battery operation data to obtain a pressure feature vector; and to extract time-series features from the electrical time-series data in the battery operation data to obtain an electrical time-series feature vector. The feature fusion module is used to fuse the pressure feature vector and the electrical time-series feature vector to obtain a fused vector; The prediction module predicts the battery capacity of the target battery based on the fusion vector to obtain the battery state of charge.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

Images