Battery health state assessment method, device, equipment, medium and program product

Abstract

The invention provides a battery health state assessment method which can be applied to the technical field of artificial intelligence. The battery health state evaluation method comprises the following steps: in response to an obtained charging and discharging voltage curve of a target battery, calculating the differential of voltage along with the change of charging capacity according to the charging and discharging voltage curve, and obtaining an incremental voltage curve; the differential of the charging capacity along with voltage change is calculated, and an incremental capacity curve is obtained; extracting a peak height, a peak position and a peak area in the incremental voltage curve, and a valley depth, a valley position and a valley area in the incremental capacity curve as health factors representing the target battery; and outputting the health state of the target battery based on the health factor of the target battery and a pre-trained support vector regression prediction model. The invention further provides a battery health state evaluation device and equipment, a storage medium and a program product.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence, and more specifically to a method, apparatus, device, medium, and program product for assessing battery health status. Background Technology

[0002] Banks' core data centers, regional data centers, core transaction systems, and key branch outlets are all equipped with large-scale uninterruptible power supply (UPS) systems. These UPS systems extensively utilize lithium-ion battery packs as backup power. Accurately predicting when the UPS's lithium-ion battery packs will reach the end of their lifespan or experience significant capacity degradation can prevent catastrophic consequences such as data center downtime, transaction interruptions, and data loss due to battery failure during grid faults, thus ensuring the stability and business continuity of the financial system. Existing battery health assessment methods are inaccurate or their models are complex and require optimization; therefore, more accurate assessments of battery health are of great significance. Summary of the Invention

[0003] In view of the above problems, this application provides battery health status assessment methods, apparatus, devices, media, and procedures to improve the accuracy of assessment.

[0004] According to a first aspect of this application, a battery health status assessment method is provided, comprising: in response to acquiring the charge-discharge voltage curve of a target battery, calculating the derivative of voltage change with charging capacity based on the charge-discharge voltage curve to obtain an incremental voltage curve; and calculating the derivative of charging capacity change with voltage to obtain an incremental capacity curve; extracting the peak height, peak position, and peak area from the incremental voltage curve, and the valley depth, valley position, and valley area from the incremental capacity curve as health factors characterizing the target battery; and outputting the health status of the target battery based on the health factors of the target battery and a pre-trained support vector regression prediction model; wherein the support vector regression prediction model is obtained by fusing training based on the health factors and corresponding battery capacities of similar samples of the target battery; and the battery capacity reflects the health status of the sample.

[0005] According to an embodiment of this application, the method further includes: calculating the voltage difference between the target battery and the reference battery at the same state point based on the charge-discharge voltage curve, and obtaining the difference voltage curve; extracting the characteristic point voltage position and / or characteristic point height of the difference voltage curve, and incorporating them into the health factor.

[0006] According to an embodiment of this application, the training method of the support vector regression prediction model includes: generating N random numbers based on a preset initial random value and a perturbation value, according to a preset maximum number of chaotic iterations N, wherein, in each generation process, if the current random number is less than 0.5, the generated random number is twice the current random number plus the perturbation value; if the current random number is greater than or equal to 0.5, the generated random number is twice the difference between 1 and the current random number plus the perturbation value; N is a positive integer; converting the N random numbers into the actual position of the whale according to preset upper and lower bounds of the parameters; with the goal of reducing the prediction error of the support vector regression prediction model, the parameter space of the support vector regression prediction model is searched using the whale optimization algorithm based on the actual position of the whale, and the parameters that meet the expected target are determined.

[0007] According to an embodiment of this application, the whale optimization algorithm is used to search the parameter space of a support vector regression prediction model, including: calculating the ratio of the current iteration number to the maximum iteration number; calculating the cosine value based on the ratio and adding 1 to obtain the dynamic convergence factor of the whale optimization algorithm; and determining the search range of the current whale pod in the parameter space based on the dynamic convergence factor.

[0008] According to an embodiment of this application, the whale optimization algorithm is used to search the parameter space of a support vector regression prediction model, including: calculating the ratio of the current iteration number to the maximum iteration number; inputting the difference between the ratio and the midpoint value of the search progress into a sigmoid function to obtain the output result of the sigmoid function; adding a preset minimum weight value to the product of a preset weight range and the output result of the sigmoid function; and determining the position of the whale in the current search based on the dynamic weight value.

[0009] According to an embodiment of this application, the method of using the whale optimization algorithm to search the parameter space of the support vector regression prediction model further includes: triggering a local search if the dynamic convergence factor is less than a preset convergence factor threshold and / or the current iteration number is greater than a local iteration number threshold; the current search parameter solution is the previous search parameter solution minus the product of the local search step size and the gradient of the objective function; if the current search parameter solution makes the objective function value smaller than the previous search parameter solution, then the current search parameter solution is accepted.

[0010] A second aspect of this application provides a battery health status assessment device, comprising: a data acquisition module, configured to, in response to acquiring the charge-discharge voltage curve of a target battery, calculate the derivative of voltage change with charging capacity based on the charge-discharge voltage curve to obtain an incremental voltage curve; and calculate the derivative of charging capacity change with voltage to obtain an incremental capacity curve; a factor extraction module, configured to extract the peak height, peak position, and peak area from the incremental voltage curve, and the valley depth, valley position, and valley area from the incremental capacity curve, as health factors characterizing the target battery; and a status prediction module, configured to output the health status of the target battery based on the health factors of the target battery and a pre-trained support vector regression prediction model; wherein the support vector regression prediction model is obtained by fusing and training the health factors and corresponding battery capacities of similar samples of the target battery; and the battery capacity reflects the health status of the sample.

[0011] A third aspect of this application provides an electronic device comprising: one or more processors; and a memory for storing one or more computer programs, wherein the one or more processors execute the one or more computer programs to implement the steps of the method described above.

[0012] A fourth aspect of this application also provides a computer-readable storage medium having a computer program or instructions stored thereon, which, when executed by a processor, implement the steps of the above-described method.

[0013] The fifth aspect of this application also provides a computer program product, including a computer program or instructions that, when executed by a processor, implement the steps of the above-described method. Attached Figure Description

[0014] The above-mentioned contents, other objects, features and advantages of this application will become clearer from the following description of embodiments with reference to the accompanying drawings, in which:

[0015] Figure 1 The illustration schematically depicts application scenarios of the battery health status assessment method, apparatus, device, medium, and program products according to embodiments of this application;

[0016] Figure 2 A flowchart illustrating a battery health status assessment method according to an embodiment of this application is shown schematically.

[0017] Figure 3 The schematic diagram illustrates the steps of the whale optimization algorithm according to an embodiment of this application;

[0018] Figure 4 A schematic diagram illustrating the structure of a battery health status assessment device according to an embodiment of this application is shown; and

[0019] Figure 5 A block diagram schematically illustrates an electronic device suitable for implementing a battery health status assessment method according to an embodiment of this application. Detailed Implementation

[0020] The embodiments of this application will now be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of this application. In the following detailed description, numerous specific details are set forth to provide a thorough understanding of the embodiments of this application for ease of explanation. However, it will be apparent that one or more embodiments may be implemented without these specific details. Furthermore, descriptions of well-known structures and technologies are omitted in the following description to avoid unnecessarily obscuring the concepts of this application.

[0021] The terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of this application. The terms “comprising,” “including,” etc., as used herein indicate the presence of features, steps, operations, and / or components, but do not exclude the presence or addition of one or more other features, steps, operations, or components.

[0022] All terms used herein (including technical and scientific terms) have the meanings commonly understood by those skilled in the art, unless otherwise defined. It should be noted that the terms used herein are to be interpreted in a manner consistent with the context of this specification, and not in an idealized or overly rigid way.

[0023] When using expressions such as "at least one of A, B and C", they should generally be interpreted in accordance with the meaning that is commonly understood by those skilled in the art (e.g., "a system having at least one of A, B and C" should include, but is not limited to, a system having A alone, a system having B alone, a system having C alone, a system having A and B, a system having A and C, a system having B and C, and / or a system having A, B and C, etc.).

[0024] Battery health is an indicator that measures the degree of degradation of a battery's current performance relative to its brand-new state, reflecting the battery's aging condition and remaining lifespan. Battery health is typically quantified based on capacity decay or internal resistance increase, and is one of the core parameters in a battery management system, directly impacting device range, safety, and maintenance strategies.

[0025] Experimental battery health assessment methods rely on experimental techniques to measure key parameters such as battery voltage and current, and then assess the battery's internal resistance and capacity to infer its health status. This technology typically encompasses various methods, including ampere-hour integration, impedance spectroscopy, and internal resistance measurement. While experimental methods can fundamentally reveal the root causes of lithium-ion battery degradation, they have limitations. For example, they are usually only applicable to specific types of batteries, and the applicability of experimental results is limited for different battery models due to significant differences. Furthermore, experimental methods often need to be used in conjunction with other methods for assessing battery health.

[0026] Model-based battery health assessment methods are mainly divided into two categories: physical modeling methods and mathematical modeling methods. Physical models, also known as "white-box models," characterize the ion diffusion and migration processes within the battery, as well as the electrochemical reactions occurring at the positive and negative electrodes, by constructing a battery state model. Mathematical modeling methods, on the other hand, establish mathematical formulas related to battery health or depict function graphs reflecting health information. They then extract feature information from these expressions through mathematical analysis or data analysis techniques, thereby revealing the battery's health status. Modeling methods offer high accuracy in explaining the complex electrochemical processes within batteries. However, developing an accurate model requires a deep understanding of the battery's electrochemical characteristics. Furthermore, the constructed models are often very complex, requiring the identification of numerous model parameters. The parameter identification process is often limited by the convergence speed of optimization and filtering algorithms. Moreover, model construction is also influenced by external factors such as experimental conditions and battery type, making the accurate identification of all key parameters extremely challenging.

[0027] This application provides a battery health status assessment method that employs a data-driven strategy to evaluate the health status of lithium batteries. The data-driven health status assessment technology monitors key parameters such as current and voltage during the battery's charging and discharging process to identify characteristics closely related to health status degradation. It then extracts potential patterns from the data sequence to construct a health status assessment model, thereby accurately predicting the health status of the lithium battery. Unlike model-dependent assessment methods, the data-driven method does not require in-depth understanding of the battery's internal structure or the complex electrochemical reaction mechanisms. It can construct a health status assessment model solely based on charging and discharging data, exhibiting strong practicality, adaptability, and a wide range of applications.

[0028] Figure 1 The illustration schematically depicts application scenarios of the battery health status assessment method, apparatus, device, medium, and program products according to embodiments of this application.

[0029] like Figure 1As shown, application scenario 100 according to this embodiment may include a first terminal device 101, a second terminal device 102, a third terminal device 103, a network 104, and a server 105. The network 104 serves as a medium for providing a communication link between the first terminal device 101, the second terminal device 102, the third terminal device 103, and the server 105. The network 104 may include various connection types, such as wired or wireless communication links, or fiber optic cables, etc.

[0030] Users can use the first terminal device 101, the second terminal device 102, and the third terminal device 103 to interact with the server 105 via the network 104 to receive or send messages, etc. Various communication client applications can be installed on the first terminal device 101, the second terminal device 102, and the third terminal device 103, such as shopping applications, web browser applications, search applications, instant messaging tools, email clients, social media platform software, etc. (for example only).

[0031] The first terminal device 101, the second terminal device 102, and the third terminal device 103 can be various electronic devices with displays and support web browsing, including but not limited to smartphones, tablets, laptops, and desktop computers.

[0032] Server 105 can be a server that provides various services, such as a backend management server that supports websites browsed by users using the first terminal device 101, the second terminal device 102, and the third terminal device 103 (this is just an example). The backend management server can analyze and process data such as received user requests, and feed back the processing results (such as web pages, information, or data obtained or generated according to user requests) to the terminal devices.

[0033] It should be noted that the battery health status assessment method provided in this application embodiment can generally be executed by server 105. Correspondingly, the battery health status assessment device provided in this application embodiment can generally be located in server 105. The battery health status assessment method provided in this application embodiment can also be executed by a server or server cluster that is different from server 105 and capable of communicating with the first terminal device 101, the second terminal device 102, the third terminal device 103, and / or server 105. Correspondingly, the battery health status assessment device provided in this application embodiment can also be located in a server or server cluster that is different from server 105 and capable of communicating with the first terminal device 101, the second terminal device 102, the third terminal device 103, and / or server 105.

[0034] It should be understood that Figure 1The number of terminal devices, networks, and servers shown is merely illustrative. Depending on implementation needs, any number of terminal devices, networks, and servers can be included.

[0035] The following will be based on Figure 1 The described scene, through Figures 2-3 A battery health status assessment method according to embodiments of this application will be described in detail.

[0036] Figure 2 A flowchart illustrating a battery health status assessment method according to an embodiment of this application is shown schematically.

[0037] like Figure 2 As shown, the battery health status assessment in this embodiment includes operations S210 to S230.

[0038] In operation S210, in response to acquiring the charge and discharge voltage curve of the target battery, the differential of voltage change with charging capacity is calculated based on the charge and discharge voltage curve to obtain the incremental voltage curve; and the differential of charging capacity change with voltage is calculated to obtain the incremental capacity curve.

[0039] Extracting the health factors or health status of lithium batteries is a core task in battery management systems and predictive maintenance. The aim is to identify characteristic quantities that are strongly correlated with the battery's internal aging mechanisms and are easily measured online or offline. This embodiment can obtain the charge-discharge voltage curve of the target battery by directly reading data collected by the battery management system or through other indirect methods.

[0040] In operation S220, the peak height, peak position, and peak area in the incremental voltage curve, as well as the valley depth, valley position, and valley area in the incremental capacity curve, are extracted as health factors characterizing the target battery.

[0041] In this embodiment, the incremental voltage curve is defined as the differential of voltage change with charging capacity. The peak height, peak position, and peak area are extracted from this curve as health factors characterizing the battery's health status, for example, denoted as H1, H2, and H3, respectively. The incremental capacity curve is defined as the differential of charging capacity change with voltage. The valley depth, valley position, and valley area are extracted from this curve as health factors characterizing the battery's health status, for example, denoted as H4, H5, and H6, respectively.

[0042] For example, the peak height of the incremental voltage curve of one battery is 0.5V / Ah, the position is 1Ah, and the peak area is 0.3; the peak height of the incremental voltage curve of another battery is 0.2V / Ah, the position is 1.2Ah, and the area is 0.1.

[0043] Because different health factors exhibit varying trends, directly comparing the correlation between health factors and battery capacity is difficult. To quantify this correlation, statistical methods for measuring the correlation of data sequences, namely the product-moment correlation coefficient and the rank correlation coefficient, were used. Both correlation coefficients have a range of [-1, 1], with larger absolute values ​​indicating stronger correlations. The calculation results show that in this embodiment, the correlation coefficient between the indirect health factor and capacity of each battery exceeds 0.8, indicating that these health factors can effectively track the degradation process of lithium batteries.

[0044] In operation S230, based on the health factors of the target battery and a pre-trained support vector regression prediction model, the health status of the target battery is output. The support vector regression prediction model is trained by fusing the health factors and corresponding battery capacities of similar samples of the target battery; the battery capacity reflects the health status of the sample.

[0045] The pre-trained support vector regression prediction model in this embodiment can be trained as follows: Sample data from 100 batteries of the same model are collected in advance. The sample data includes the health factors of each battery at different aging stages and the actual capacity at the corresponding stages. Different aging stages can correspond to different cycle counts, such as 100 cycles, 200 cycles, etc. For a battery with a capacity of 2 amp-hours, after 500 cycles, the capacity becomes 1.4 amp-hours. Based on the State of Health (SOH) reflected by the capacity, it can be expressed as SOH = 1.4 / 2 = 70%.

[0046] In this embodiment, during the training phase, the battery capacity and the obtained fused health factor are preprocessed, and the processed data is divided into a training set and a test set. The fused health factor serves as the model input, and the corresponding battery capacity serves as the model output. This data is used to train a support vector regression prediction model, allowing the model to learn the relationship between the health factor and capacity (health status). The obtained optimal parameters are then input into the support vector regression prediction model to complete the model construction.

[0047] When assessing battery health status, target battery data can be input into a support vector regression prediction model to evaluate the lithium battery's health status. When a new target battery is available, its health factors can be extracted and input into the trained support vector regression prediction model to directly output its health status.

[0048] Health factors should not only fully encompass aging information but also possess characteristics such as low redundancy and low dimensionality. To reduce the computational complexity of subsequent health status estimation and remaining life expectancy prediction, this embodiment preferably employs kernel principal component analysis to reduce the dimensionality of health factors, thereby constructing a fused health factor.

[0049] Some embodiments of this application utilize a general or improved whale optimization algorithm. Parameter initialization is performed by setting the initial population size, search dimension, maximum number of iterations, and upper and lower bounds of the parameter search range. A fitness function is set, defined as the root mean square value between the predicted capacity value and the actual capacity value obtained through three-fold cross-validation of the training set. The optimal parameters are searched by calculating the fitness value of each individual whale based on the fitness function, saving the current optimal search individual and its fitness value, and iterating and comparing until the maximum number of iterations is reached, at which point the optimal parameters are output. This approach solves the core challenges in lithium battery life prediction, such as difficulty in model parameter optimization, insufficient prediction accuracy, and poor model robustness, significantly improving the accuracy and reliability of predictions.

[0050] This embodiment simultaneously derives incremental voltage and capacity curves from the target battery's charge / discharge voltage curves, capturing both voltage-related and capacity-related characteristics during battery aging. Compared to single-curve extraction, this provides a more comprehensive coverage of multi-dimensional aging mechanisms, avoiding feature bias. The incremental voltage curve captures dynamic voltage changes caused by internal resistance and polarization, while the incremental capacity curve captures cumulative capacity decay caused by active material loss. Increased internal resistance accelerates capacity decay, which in turn affects the voltage plateau, creating a causal relationship. Collaborative analysis improves prediction accuracy, providing dual early warning protection for both early-stage internal resistance and late-stage capacity. The support vector regression prediction model is trained based on the fusion of health factors and corresponding capacities of battery samples, ultimately achieving accurate and reliable output of the target battery's health status, balancing practicality and prediction accuracy.

[0051] In some embodiments of this application, the battery health status assessment further includes: calculating the voltage difference between the target battery and the reference battery at the same state point based on the charge-discharge voltage curve, and obtaining the difference voltage curve; extracting the characteristic point voltage position and / or characteristic point height of the difference voltage curve and incorporating them into the health factor.

[0052] The reference battery can be a new battery or another selected standard battery.

[0053] The same state point refers to the same charging capacity or the same state of charge. Since the target battery and the reference battery are at the same charge level, the voltage difference reflects the difference in their health status. In this embodiment, the selected state points can include 0.2Ah, 0.4Ah, ..., 2.0Ah, a total of 10 evenly distributed capacity points.

[0054] For example, if the target battery is an aged battery that has undergone 500 cycles, after obtaining the charge-discharge voltage curve of the target battery, the voltage difference between the target battery and the reference battery at the same state point is calculated. For each capacity point, the voltage difference ΔV is obtained by subtracting the reference battery voltage from the target battery voltage. For example, at 0.2Ah, the target battery voltage is 3.1V - the reference battery voltage is 3.2V = -0.1V; at 0.4Ah, the target battery voltage is 3.2V - the reference battery voltage is 3.3V = -0.1V; at 1.2Ah, the target battery voltage is 3.5V - the reference battery voltage is 3.7V = -0.2V; at 2.0Ah, the target battery voltage is 4.1V - the reference battery voltage is 4.2V = -0.1V. Plotting the charging capacity on the horizontal axis and the voltage difference on the vertical axis, connecting these points yields the voltage difference curve. This curve shows a significant trough at 1.2Ah (ΔV = -0.2V).

[0055] The characteristic points of the differential voltage curve directly reflect the internal changes caused by battery aging, such as increased internal resistance and enhanced polarization. The voltage position of the characteristic point indicates the point of greatest capacity difference. New batteries show no significant difference at this point, while aged batteries exhibit a significant voltage difference at this position due to internal material degradation, such as 1.2 Ah in the example above. Positional shift or stability reflects the degree of aging; the height of the characteristic point refers to the absolute value of the voltage difference at that position (e.g., 0.2V). The more cycles and the more severe the aging, the larger this value becomes, directly quantifying the voltage deviation caused by aging. Extracting these characteristic points from the differential voltage curve can serve as health factors to predict the health status of the target battery.

[0056] This embodiment obtains the difference voltage curve by calculating the voltage difference between the target battery and the reference battery at the same state point, and extracts the voltage position and / or height of the feature point as a health factor. This can accurately capture the voltage characteristic differences caused by internal resistance growth, polarization effect and other factors during battery aging, enrich the dimension of the health factor in the characterization of battery health status, and provide more effective feature support for the accurate prediction of the subsequent lithium battery health status and / or remaining life.

[0057] In the training process of a support vector regression prediction model, the goal of training is to find a function that can predict the target value as accurately as possible within a predetermined error tolerance, while maintaining the smoothness of the model as much as possible.

[0058] Assuming a given dataset ,in For the i-th input feature vector, Let n be the corresponding output vector and n be the number of all samples. The nonlinear mapping can be defined as follows.

[0059]

[0060] In the formula, For input data, It is a non-mapping function. The weights determine the orientation of the hyperplane. The intercept determines the distance between the hyperplane and the origin.

[0061] It is generally believed that and They should be completely consistent, at which point the error is zero, but the support vector regression prediction model allows... and There is The deviation indicates that when and The absolute value of the difference between them is greater than The loss is only factored in at the time of training; the training process of a support vector regression prediction model is essentially about finding the optimal solution. and Make Get as close as possible ,at this time and Solving this problem can be viewed as a convex quadratic programming problem. This is equivalent to solving an optimization problem.

[0062]

[0063]

[0064] In the formula, Γ(ω) represents the objective function, which is a function of the model parameter ω. The core of training is to find the ω that minimizes Γ(ω). It is a regularization term. It is an experience-based risk item. This is the loss function, where C represents the penalty parameter. Its main function is to seek a balance between the complexity of the regression model and the goodness of fit to the samples. If the value of C is large, the model will pay more attention to outliers in the samples, which may lead to overfitting, increasing the complexity of the support vector regression prediction model and reducing computational efficiency. Conversely, if the value of C is small, the model will not fit the data samples tightly enough, and the support vector regression prediction model may underfit.

[0065] Introducing slack variables and The formula can be rewritten as a constrained quadratic optimization problem, as shown below.

[0066]

[0067] in, The objective function is denoted as . This is a penalty term for out-of-bounds errors. This represents the relaxation amount when the predicted value of the i-th sample is lower than the true value and exceeds ε. This represents the slack amount by which the predicted value of the i-th sample is higher than the true value and exceeds ε. st represents the constraint condition. Represents the true value Compared with the predicted value The difference. This represents a function that maps to a higher-dimensional space.

[0068] Introducing Lagrange multipliers and By utilizing the duality principle, the regression problem is transformed into a problem of minimizing the objective function. The transformed objective function takes the form shown below.

[0069]

[0070] Based on Mercer's theorem and the Karush-Kuhn-Tucker (KKT) conditions, the solution to the above convex quadratic programming problem can be obtained, and the optimal regression function estimate can be obtained as follows.

[0071]

[0072] In the formula: subscript Indicates optimality. This is the kernel function.

[0073] This paper selects the RBF (Radial Basis Function) kernel function, which has the advantage of mapping data to a higher-dimensional space to address nonlinear problems in the original space. Furthermore, the RBF kernel function has only one hyperparameter, simplifying the parameter optimization process. The expression for this kernel function is as follows.

[0074]

[0075] In the formula, The width factor of RBF. For kernel parameters.

[0076] In this model, the penalty factor and kernel parameters The parameters of the support vector regression prediction model have a significant impact on its accuracy; therefore, it is necessary to select appropriate optimization methods to optimize the model's parameters. and To achieve accurate estimation of the health status of lithium batteries.

[0077] The whale optimization algorithm simulates the unique hunting behavior of humpback whales: in the early stages of foraging, individual whales move towards other whales to form an encirclement; then, they swim towards the surface in a spiraling ascent, using bubbles to drive away prey and constructing a bubble net to tightly surround it; finally, they capture the prey under the bubble net. Inspired by the hunting behavior of humpback whales, the traditional whale optimization algorithm includes three main stages: random prey search, prey encirclement, and spiral foraging.

[0078] When a whale spots prey, the humpback whale gradually approaches and surrounds it. In the algorithm, this is represented by candidate solutions moving closer to the current optimal solution. Assuming the current optimal solution is the prey's location, other candidate solutions (individual whales) update their positions using the following formula, gradually reducing the distance to the optimal solution.

[0079]

[0080] in, Let t be the position of a certain whale in generation t (candidate solution); For the t-th generation global optimal solution; A and C are coefficient vectors: A=2a*r1-a, C=2*r2 (r1,r2 are random numbers in [0,1], a is a linearly decreasing parameter, decreasing from 2 to 0).

[0081] When a whale fails to find prey, it will randomly search new areas, which in the algorithm manifests as moving away from the current non-optimal solution and exploring unknown space. When |A|≥1, it does not rely on the current optimal solution but randomly selects an individual. For reference, the updated formula is as follows.

[0082]

[0083] This mechanism enhances the algorithm's global exploration capabilities and prevents it from getting trapped in local optima too early.

[0084] During the spiral predation phase, humpback whales create a spiral-shaped bubble net beneath their prey to trap it. The encirclement is then tightened by decreasing the value of 'a', gradually shrinking the range of 'A' from [-2,2] to [-0,0], forcing the whales to converge towards the optimal solution. The position is then updated using a spiral-shaped movement pattern simulating the whale's spiral swimming path around its prey. Two mechanisms are chosen by a random probability p: if p < 0.5, the encirclement tightening is performed; if p ≥ 0.5, the spiral update is performed.

[0085] Figure 3 The diagram illustrates the steps of the whale optimization algorithm according to an embodiment of this application.

[0086] like Figure 3 As shown, the general steps of the whale optimization algorithm include: population initialization, fitness calculation, iterative optimization, updating the optimal solution, and termination.

[0087] Population initialization: N individual whales (candidate solutions) are randomly generated in the solution space, and each individual corresponds to a set of parameters of the problem to be optimized.

[0088] Calculate fitness, evaluate the fitness value (objective function value) of each individual, and record the current best individual X. * (The solution with optimal fitness).

[0089] Iterative optimization, updating parameter a ( ), where t is the current iteration number. (Maximum number of iterations); generate random numbers, calculate A and C. Select a position update strategy based on |A| and p.

[0090] Update the optimal solution. If the newly generated individual has better fitness, update the global optimal solution X. * .

[0091] Terminate by checking if the termination condition has been met; if so, if the maximum number of iterations has been reached. Output the global optimal solution X * If not, continue with iterative optimization.

[0092] The embodiments of this application utilize the whale optimization algorithm to search for the optimal parameters that minimize the prediction error of the support vector regression prediction model within a preset model parameter space. However, the general whale optimization algorithm itself has some shortcomings, such as uneven initial population distribution, low convergence accuracy, and a tendency to get trapped in local optima. Therefore, some embodiments of this application improve certain steps in the whale optimization algorithm.

[0093] In some embodiments of this application, the training method of the support vector regression prediction model includes: generating N random numbers according to a preset initial random value and perturbation value, and according to a preset maximum number of chaotic iterations N, wherein, in each generation process, if the current random number is less than 0.5, the generated random number is twice the current random number plus the perturbation value; if the current random number is greater than or equal to 0.5, the generated random number is twice the difference between 1 and the current random number plus the perturbation value; converting the N random numbers into the actual position of the whale according to a preset upper bound and lower bound of the parameters; with the goal of reducing the prediction error of the support vector regression prediction model, searching the parameter space of the support vector regression prediction model using the whale optimization algorithm according to the actual position of the whale, and determining the parameters that meet the expected goal.

[0094] Because lithium batteries have a limited number of cycles, the data sample is limited. In this case, the quality of the initial solution is crucial to finding the global optimum or near-global optimum.

[0095] The method for generating N random numbers in this embodiment can be expressed as follows.

[0096]

[0097]

[0098] in: Let N represent a random number uniformly distributed on [0,1], N represent the maximum number of chaotic iterations, and ∈ represent a small perturbation.

[0099] Based on the chaotic variable λ obtained above, N random numbers are converted into the actual positions of the whales, and the chaotic initialization population is shown in the following formula.

[0100]

[0101] In the formula: This represents the coordinate of the j-th dimension in the i-th search agent group. Let λ represent the coordinates of the j-th dimension of λ after internal random sorting. Let lb represent the lower bound of the parameter search, and ub represent the upper bound of the parameter search.

[0102] In one embodiment, suppose we need to optimize the parameters of a support vector regression prediction model: a penalty factor C (lower bound 0.1, upper bound 100), kernel parameters... (Lower bound is 0.01, upper bound is 10). Five chaotic random numbers are generated iteratively according to the rules: [0.3, 0.61, 0.79, 0.43, 0.87]. Based on the upper and lower bounds of the parameters, the chaotic random numbers are mapped to the actual parameter space, and the positions of the five whales are (30.07, 3.007), (61.04, 6.10), (78.92, 7.89), (43.06, 4.31), and (86.01, 8.70).

[0103] The parameters of the 5 whales (C, Substitute the values ​​into the support vector regression prediction model and calculate the initial fitness. Calculate the prediction error using the training data; for example, the error is 2.3% for the first whale, 3.1% for the second, 1.9% for the third, 2.5% for the fourth, and 2.8% for the fifth. Record the current optimal position as the third whale.

[0104] The whale's position is updated using a prey-surrounding, spiral attack, and prey-searching mechanism. For example, if the second whale (error 3.1%) moves closer to the third whale (error 1.9%), the new position can be adjusted to (70, 7.5), and the error is recalculated to 2.2%. After 100 iterations, the optimal parameter C is found to be 75.2. The error was 7.2. At this point, the prediction error of the support vector regression model dropped to 1.5%, meeting the expected target and reaching the termination condition.

[0105] The initial population generated in this embodiment exhibits both uniformity and randomness in its traversal. This means that the initial whale positions (i.e., candidate solutions for model parameters) can more evenly and broadly cover the entire search space, avoiding excessive concentration of the population in certain regions of the parameter space. This reduces the risk of the algorithm getting stuck in suboptimal regions at the beginning and lays a better foundation for subsequent searches. In multiple runs of the algorithm, this embodiment produces a more consistent initial population distribution than purely random initialization, reducing the algorithm's oversensitivity to initial conditions, improving the stability of model training, and indirectly enhancing generalization ability. It has a synergistic effect of overcoming small sample sizes and avoiding local optima.

[0106] This embodiment generates T uniformly distributed random numbers according to a segmentation rule by preset initial random values ​​and perturbation values, and combines the upper and lower bounds of the parameters to convert them into the actual position of the whale. Then, with the goal of reducing the prediction error of the support vector regression prediction model, the whale optimization algorithm is used to search the parameter space. This can solve the problem of uneven population distribution in traditional pseudo-random initialization, improve the diversity of the initial population, and provide a high-quality and diversified starting point for finding the optimal parameters of the support vector regression prediction model and reducing its prediction error. It can also cope with small samples, avoid initial bias, and improve the starting point of global exploration.

[0107] In some embodiments of this application, the whale optimization algorithm is used to search the parameter space of the support vector regression prediction model, including: calculating the ratio of the current iteration number to the maximum iteration number; calculating the cosine value based on the ratio and adding 1 to obtain the dynamic convergence factor of the whale optimization algorithm; and determining the range of the current whale pod's search in the parameter space based on the dynamic convergence factor.

[0108] Because the capacity decay of lithium batteries is highly nonlinear, especially the capacity drop that occurs later in the process, the linear decreasing parameter (or convergence factor) 'a' has a linear, fixed rate of change between global and local searches, making it difficult to flexibly adapt to this nonlinear optimization requirement.

[0109] This embodiment employs a nonlinear dynamic convergence factor and a cosine decreasing strategy to balance the global and local search capabilities of the algorithm, thereby improving the convergence accuracy. The dynamic convergence factor 'a' in this embodiment can be expressed as follows.

[0110]

[0111] In the formula: Indicates the number of iterations. This indicates the maximum number of iterations.

[0112] In the early stages of iteration (e.g., t=0~3), a(t) is approximately 1.5~2, focusing on global search. In the middle stages of iteration (e.g., t=4~6), a(t) is approximately 0.6~1.4, balancing global exploration and local development, gradually focusing on the potential optimal region. In the later stages of iteration (e.g., t=7~10), a(t) is approximately 0~0.5, focusing on local development, conducting a fine-grained search near the optimal region, and improving parameter accuracy.

[0113] In this embodiment, the cosine function changes slowly in the initial and final stages, and rapidly in the middle stage, allowing for more precise stage control. The slow initial change gives the algorithm ample time for global exploration across a broad space, preventing premature contraction, which is crucial for finding the potential optimal region for the parameters of the lithium battery support vector regression prediction model. The rapid change in the middle stage allows the algorithm to quickly increase the intensity of local development after roughly locating a promising region, focusing on fine-tuning the search. The slow change in the later stage, approaching the optimal solution, allows for more precise fine-tuning, which is particularly important for capturing the optimal parameter combination of the support vector regression prediction model to accurately fit the final stage of battery capacity degradation. The decreasing dynamic convergence factor strategy in this embodiment more effectively balances global exploration and local development, avoiding premature convergence to suboptimal solutions or insufficient development in the later stages that may occur with linear strategies. This significantly increases the probability of finding high-precision support vector regression prediction model parameters, improves convergence performance, effectively controls the transition rhythm between global exploration and local development, matches the nonlinear degradation characteristics of the battery, and avoids premature convergence or insufficient development.

[0114] In some embodiments of this application, the whale optimization algorithm is used to search the parameter space of the support vector regression prediction model, including: calculating the ratio of the current iteration number to the maximum iteration number; inputting the difference between the ratio and the midpoint value of the search progress into a sigmoid function to obtain the output result of the sigmoid function; adding the preset minimum weight value to the product of the preset weight value range and the output result of the sigmoid function to obtain a dynamic weight value; and determining the position of the whale in the current search based on the dynamic weight value.

[0115] The formula for updating the whale's position in a typical whale optimization algorithm using an adaptive weight mechanism is as follows.

[0116]

[0117] In the formula, This indicates the current updated position of the whale. This indicates the randomly selected location of the whale. This represents the spatial distance between the current individual whale and a randomly selected individual whale. This represents the spatial distance between the current whale individual and the best whale individual. Indicates the number of iterations. For the coefficient vector, Indicates the distance between the whale and its prey. It is a constant used to define the shape of the logarithmic spiral. It is a random number belonging to the range [0,1]. The distance between a whale and its prey is determined when... When a whale is closest to its prey, The time indicates when a whale is furthest from its prey.

[0118] Because the fitness terrain of lithium battery data is often rugged, with multiple local optima, a fixed-weight strategy cannot flexibly address the needs of different search stages and individual states. To better adapt to complex optimization terrain, this embodiment introduces adaptive weights. The adaptive weights in this embodiment can be expressed as follows.

[0119]

[0120] In the formula: Indicates the number of iterations. Indicates the maximum number of iterations. This represents the minimum weight. This represents the maximum weight, and k represents a parameter controlling the rate of weight decrease. Initially, when... When approaching 0, near This makes the search agent more random, which is beneficial for global exploration; in the mid-game, as... The increase in weight As the value gradually decreases, the randomness of the agent weakens, and it begins to focus more closely on the vicinity of the potential optimal solution; in the later stages, when near , near Search agents exhibit minimal randomness, which is beneficial for localized development.

[0121] The formula for updating the whale position in the whale optimization algorithm of this embodiment can be expressed as follows.

[0122] ,

[0123]

[0124]

[0125] For example, in one embodiment, the maximum number of iterations is 20. The weight values ​​range from a minimum of 0.2 to a maximum of 0.8, with a midpoint of 0.5 for the search progress. The current search is the 5th iteration, and the calculated dynamic weight is approximately 0.463. This weight determines the specific whale position, which is closer to the random solution, emphasizing global exploration. In the 18th search, a larger dynamic weight of 0.559 is obtained, indicating that the whale position is closer to the current optimal solution, emphasizing local development, fine-tuning parameters, and improving accuracy.

[0126] This embodiment dynamically adjusts the search strategy during optimization based on weights derived from iteration counts or individual fitness changes. Early-stage individuals with poor fitness are given greater weights to encourage exploration of new regions. Later-stage individuals with good fitness are given less weight to encourage in-depth development near their current location. This allows for smarter resource allocation, enabling bold exploration when needed and focused development when refined, thus accelerating convergence and increasing the probability of finding a better solution. When the algorithm shows signs of getting stuck in local optima, adaptive weights can automatically increase the exploration component, helping the population escape local traps. It dynamically adjusts individual search behavior, enhancing the ability to escape local optima and dynamically balancing exploration and development.

[0127] In some embodiments of this application, the method of using the whale optimization algorithm to search the parameter space of the support vector regression prediction model further includes: triggering a local search if the dynamic convergence factor is less than a preset convergence factor threshold and / or the current iteration number is greater than a local iteration number threshold; the current search parameter solution is the previous search parameter solution minus the product of the local search step size and the gradient of the objective function; if the current search parameter solution makes the objective function value smaller than the previous search parameter solution, then the current search parameter solution is accepted.

[0128] The performance of support vector regression prediction models depends on parameters C and Extremely sensitive. Even if the whale optimization algorithm finds the vicinity of the global optimum, it still requires a very fine local search to find the exact optimal combination of parameters. The local search capability of the standard whale optimization algorithm (especially the spiral update) is sometimes not strong enough.

[0129] To address parameter sensitivity and subsequent fine-tuning, this embodiment employs gradient descent. Local search is triggered when the algorithm's convergence factor 'a' falls below a certain threshold or the number of iterations reaches a certain proportion. The local search trigger condition can be expressed as follows.

[0130]

[0131] in It is the convergence factor threshold that triggers local search. It represents the proportion of iterations that trigger local search. It is the maximum number of iterations. This represents the threshold for the number of local iterations. The local search strategy can be expressed as follows.

[0132] )

[0133] in This represents the solution currently being searched locally. It is the step size of the local search. It is the objective function The gradient.

[0134] The evaluation method after local search is: if )< Accepting the new interpretation: = Otherwise, the original solution remains unchanged.

[0135] This embodiment employs a local search strategy. After identifying promising parameter regions in the whale optimization algorithm, a more intensive and systematic fine-grained search is performed within those regions, significantly improving local search accuracy. This is crucial for capturing subtle patterns of lithium battery capacity degradation at the end of the lifespan and making accurate predictions of remaining usable lifetime. By sampling and evaluating the neighborhood of the current optimal solution multiple times, this embodiment's local search strategy can smooth out the influence of lithium battery data noise to some extent, finding more robust parameter combinations in noisy environments and overcoming the impact of noise. This embodiment ensures that the solution converged by the algorithm is indeed a local optimum in its region, rather than a point close to optimum but still with room for improvement, thereby maximizing the prediction accuracy of the support vector regression prediction model and improving the reliability of the final solution.

[0136] Based on the above-described battery health status assessment method, this application also provides a battery health status assessment device. The following will be combined with... Figure 4 The device is described in detail.

[0137] Figure 4 A schematic block diagram of a battery health status assessment device according to an embodiment of this application is shown.

[0138] like Figure 4 As shown, the battery health status assessment device 400 of this embodiment includes a data acquisition module 410, a factor extraction module 420, and a status prediction module 430.

[0139] The data acquisition module 410 is used to, in response to acquiring the charging and discharging voltage curve of the target battery, calculate the differential of voltage change with charging capacity based on the charging and discharging voltage curve to obtain the incremental voltage curve; and calculate the differential of charging capacity change with voltage to obtain the incremental capacity curve.

[0140] The factor extraction module 420 is used to extract the peak height, peak position and peak area in the incremental voltage curve, and the valley depth, valley position and valley area in the incremental capacity curve, as health factors characterizing the target battery.

[0141] The state prediction module 430 is used to output the health status of the target battery based on the health factors of the target battery and a pre-trained support vector regression prediction model. The support vector regression prediction model is obtained by fusing the health factors and corresponding battery capacities of similar samples of the target battery. The battery capacity reflects the health status of the sample.

[0142] According to an embodiment of this application, the data acquisition module 410 is further configured to calculate the voltage difference between the target battery and the reference battery at the same state point based on the charge-discharge voltage curve, and obtain a difference voltage curve. The factor extraction module 420 is further configured to extract the characteristic point voltage position and / or characteristic point height of the difference voltage curve and incorporate them into a health factor.

[0143] According to an embodiment of this application, the state prediction module 430 is further configured to generate N random numbers based on a preset initial random value and a perturbation value, and according to a preset maximum number of chaotic iterations N. In each generation process, if the current random number is less than 0.5, the generated random number is twice the current random number plus the perturbation value; if the current random number is greater than or equal to 0.5, the generated random number is twice the difference between 1 and the current random number plus the perturbation value; N is a positive integer; based on preset upper and lower bounds of the parameters, the N random numbers are converted into the actual position of the whale; with the goal of reducing the prediction error of the support vector regression prediction model, based on the actual position of the whale, the whale optimization algorithm is used to search the parameter space of the support vector regression prediction model to determine parameters that meet the expected target.

[0144] According to an embodiment of this application, the state prediction module 430 is further configured to calculate the ratio of the current iteration number to the maximum iteration number; calculate the cosine value based on the ratio and add 1 to obtain the dynamic convergence factor of the whale optimization algorithm; and determine the search range of the current whale pod in the parameter space based on the dynamic convergence factor.

[0145] According to an embodiment of this application, the state prediction module 430 is further configured to calculate the ratio of the current iteration number to the maximum iteration number; input the S-shaped function based on the difference between the ratio and the midpoint value of the search progress, and obtain the output result of the S-shaped function; add the preset minimum weight value to the product of the preset weight value range and the output result of the S-shaped function, and obtain the dynamic weight value; and determine the position of the whale in the current search based on the dynamic weight value.

[0146] According to an embodiment of this application, the state prediction module 430 is further configured to trigger a local search if the dynamic convergence factor is less than a preset convergence factor threshold and / or the current iteration number is greater than a local iteration number threshold; the current search parameter solution is the previous search parameter solution minus the product of the local search step size and the gradient of the objective function; if the current search parameter solution makes the objective function value smaller than the previous search parameter solution, then the current search parameter solution is accepted.

[0147] According to embodiments of this application, any multiple modules among the data acquisition module 410, factor extraction module 420, and state prediction module 430 can be merged into one module, or any one of these modules can be split into multiple modules. Alternatively, at least some of the functions of one or more of these modules can be combined with at least some of the functions of other modules and implemented in one module. According to embodiments of this application, at least one of the data acquisition module 410, factor extraction module 420, and state prediction module 430 can be at least partially implemented as hardware circuitry, such as a field-programmable gate array (FPGA), a programmable logic array (PLA), a system-on-a-chip, a system-on-a-substrate, a system-on-package, an application-specific integrated circuit (ASIC), or any other reasonable means of integrating or packaging circuitry, or implemented in software, hardware, or firmware, or in any appropriate combination of any of these three implementation methods. Alternatively, at least one of the data acquisition module 410, factor extraction module 420, and state prediction module 430 can be at least partially implemented as a computer program module, which can perform corresponding functions when the computer program module is run.

[0148] Figure 5 A block diagram schematically illustrates an electronic device suitable for implementing a battery health status assessment method according to an embodiment of this application.

[0149] like Figure 5 As shown, an electronic device 500 according to an embodiment of this application includes a processor 501, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 502 or a program loaded from a storage portion 508 into a random access memory (RAM) 503. The processor 501 may include, for example, a general-purpose microprocessor (e.g., a CPU), an instruction set processor and / or an associated chipset and / or a special-purpose microprocessor (e.g., an application-specific integrated circuit (ASIC)), etc. The processor 501 may also include onboard memory for caching purposes. The processor 501 may include a single processing unit or multiple processing units for performing different actions of the method flow according to an embodiment of this application.

[0150] RAM 503 stores various programs and data required for the operation of electronic device 500. Processor 501, ROM 502, and RAM 503 are interconnected via bus 504. Processor 501 executes various operations of the method flow according to embodiments of this application by executing programs in ROM 502 and / or RAM 503. It should be noted that programs may also be stored in one or more memories other than ROM 502 and RAM 503. Processor 501 may also execute various operations of the method flow according to embodiments of this application by executing programs stored in one or more memories.

[0151] According to embodiments of this application, the electronic device 500 may further include an input / output (I / O) interface 505, which is also connected to a bus 504. The electronic device 500 may also include one or more of the following components connected to the input / output (I / O) interface 505: an input section 506 including a keyboard, mouse, etc.; an output section 507 including a cathode ray tube (CRT), liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 508 including a hard disk, etc.; and a communication section 509 including a network interface card such as a LAN card, modem, etc. The communication section 509 performs communication processing via a network such as the Internet. A drive 510 is also connected to the input / output (I / O) interface 505 as needed. A removable medium 511, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., is installed on the drive 510 as needed so that computer programs read from it can be installed into the storage section 508 as needed.

[0152] This application also provides a computer-readable storage medium, which may be included in the device / apparatus / system described in the above embodiments; or it may exist independently and not assembled into the device / apparatus / system. The computer-readable storage medium carries one or more programs, which, when executed, implement the method according to the embodiments of this application.

[0153] According to embodiments of this application, the computer-readable storage medium can be a non-volatile computer-readable storage medium, such as including but not limited to: portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this application, the computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. For example, according to embodiments of this application, the computer-readable storage medium may include ROM 502 and / or RAM 503 and / or one or more memories other than ROM 502 and RAM 503 described above.

[0154] Embodiments of this application also include a computer program product comprising a computer program containing program code for performing the methods shown in the flowchart. When the computer program product is run on a computer system, the program code enables the computer system to implement the battery health status assessment method provided in the embodiments of this application.

[0155] When the computer program is executed by the processor 501, it performs the functions defined in the system / apparatus of this application embodiment. According to the embodiments of this application, the systems, apparatuses, modules, units, etc., described above can be implemented by computer program modules.

[0156] In one embodiment, the computer program may rely on a tangible storage medium such as an optical storage device or a magnetic storage device. In another embodiment, the computer program may also be transmitted and distributed in the form of signals over a network medium, and may be downloaded and installed via the communication section 509, and / or installed from a removable medium 511. The program code contained in the computer program can be transmitted using any suitable network medium, including but not limited to: wireless, wired, etc., or any suitable combination thereof.

[0157] In such an embodiment, the computer program can be downloaded and installed from a network via communication section 509, and / or installed from removable medium 511. When the computer program is executed by processor 501, it performs the functions defined in the system of this application embodiment. According to embodiments of this application, the systems, devices, apparatuses, modules, units, etc., described above can be implemented by computer program modules.

[0158] According to embodiments of this application, program code for executing the computer programs provided in the embodiments of this application can be written in any combination of one or more programming languages. Specifically, these computational programs can be implemented using high-level procedural and / or object-oriented programming languages, and / or assembly / machine languages. Programming languages ​​include, but are not limited to, languages ​​such as Java, C++, Python, "C", or similar programming languages. The program code can be executed entirely on the user's computing device, partially on the user's device, partially on a remote computing device, or entirely on a remote computing device or server. In cases involving remote computing devices, the remote computing device can be connected to the user's computing device via any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computing device (e.g., via the Internet using an Internet service provider).

[0159] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram or flowchart, and combinations of blocks in a block diagram or flowchart, may be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0160] Those skilled in the art will understand that the features described in the various embodiments of this application can be combined and / or combined in various ways, even if such combinations or combinations are not explicitly described in this application. In particular, the features described in the various embodiments of this application can be combined and / or combined in various ways without departing from the spirit and teachings of this application. All such combinations and / or combinations fall within the scope of this application.

Claims

1. A method for assessing battery health status, characterized in that, include: In response to obtaining the charge and discharge voltage curve of the target battery, the differential of the voltage change with the charging capacity is calculated based on the charge and discharge voltage curve to obtain the incremental voltage curve. And calculate the differential of the charging capacity as a function of voltage to obtain the incremental capacity curve; The peak height, peak position, and peak area in the incremental voltage curve, and the valley depth, valley position, and valley area in the incremental capacity curve are extracted as health factors characterizing the target battery. Based on the health factors of the target battery and the pre-trained support vector regression prediction model, the health status of the target battery is output; The support vector regression prediction model is obtained by fusing health factors and corresponding battery capacities of similar samples of the target battery; the battery capacity reflects the health status of the sample.

2. The method according to claim 1, characterized in that, Also includes: Based on the charge-discharge voltage curves, the voltage difference between the target battery and the reference battery at the same state point is calculated to obtain the difference voltage curve. Extract the characteristic point voltage location and / or characteristic point height of the differential voltage curve and incorporate them into the health factor.

3. The method according to claim 1, characterized in that, The training methods for the support vector regression prediction model include: Based on a preset initial random value and a perturbation value, N random numbers are generated according to a preset maximum number of chaotic iterations N. In each generation process, if the current random number is less than 0.5, the generated random number is twice the current random number plus the perturbation value; if the current random number is greater than or equal to 0.5, the generated random number is twice the difference between 1 and the current random number plus the perturbation value; N is a positive integer. Based on the preset upper and lower bounds of the parameters, the N random numbers are converted into the actual location of the whale; With the goal of reducing the prediction error of the support vector regression prediction model, the whale optimization algorithm is used to iteratively search the parameter space of the support vector regression prediction model based on the actual location of the whale, and determine the parameters that meet the expected goal.

4. The method according to claim 3, characterized in that, The iterative search of the parameter space of the support vector regression prediction model using the whale optimization algorithm includes: Calculate the ratio of the current iteration number to the maximum iteration number; Calculate the cosine value based on the ratio, and add 1 to obtain the dynamic convergence factor of the whale optimization algorithm; The current range of the whale pod's search in the parameter space is determined based on the dynamic convergence factor.

5. The method according to claim 3, characterized in that, The iterative search of the parameter space of the support vector regression prediction model using the whale optimization algorithm includes: Calculate the ratio of the current iteration number to the maximum iteration number; Based on the difference between the ratio and the midpoint value of the search progress, input the S-shaped function to obtain the output result of the S-shaped function; The dynamic weight value is obtained by adding the product of the preset weight range and the output result of the S-shaped function to the preset minimum weight value. The location of the whale in the current search is determined based on the dynamic weight value.

6. The method according to claim 4, characterized in that, The iterative search of the parameter space of the support vector regression prediction model using the whale optimization algorithm also includes: If the dynamic convergence factor is less than the preset convergence factor threshold, and / or the current iteration number is greater than the local iteration number threshold, a local search is triggered; The current search parameter solution is the previous search parameter solution minus the product of the local search step size and the gradient of the objective function; If the current search parameter solution makes the objective function value smaller compared to the previous search parameter solution, then the current search parameter solution is accepted.

7. A battery health status assessment device, characterized in that, include: The data acquisition module is used to, in response to acquiring the charging and discharging voltage curve of the target battery, calculate the differential of the voltage change with the charging capacity based on the charging and discharging voltage curve to obtain the incremental voltage curve; and calculate the differential of the charging capacity change with the voltage to obtain the incremental capacity curve. The factor extraction module is used to extract the peak height, peak position and peak area in the incremental voltage curve, and the valley depth, valley position and valley area in the incremental capacity curve, as health factors characterizing the target battery. as well as The state prediction module is used to output the health state of the target battery based on the health factors of the target battery and a pre-trained support vector regression prediction model. The support vector regression prediction model is obtained by fusing health factors and corresponding battery capacities of similar samples of the target battery; the battery capacity reflects the health status of the sample.

8. An electronic device, comprising: One or more processors; Memory, used to store one or more computer programs. The characteristic feature is that the one or more processors execute the one or more computer programs to implement the steps of the method according to any one of claims 1 to 6.

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

10. A computer program product, comprising a computer program or instructions, characterized in that, When the computer program or instructions are executed by a processor, they implement the steps of the method according to any one of claims 1 to 6.

Images