Battery health state prediction method based on XGBoost-LightGBM combination model

Abstract

The invention provides a battery health state prediction method based on an XGBoost-LightGBM combination model, and the method comprises the steps: obtaining state parameters of a to-be-predicted battery, and generating a battery IC curve according to the state parameters, the state parameters including the discharge temperature peak value corresponding time and the battery equal voltage drop discharge time; extracting a peak value, a peak value voltage and an under-peak area in the battery IC curve; and determining a target SOH value of the to-be-predicted battery according to the discharge temperature peak value corresponding time, the battery equal voltage drop discharge time, the peak value, the peak value voltage and the under-peak area. Therefore, the peak value, the peak voltage and the under-peak area in the battery IC curve can be fully combined in the prediction process to indicate the battery performance, so that the predicted battery health state precision is higher, and the prediction precision is further improved in combination with a data-driven prediction method.

Description

Technical Field

[0001] The present disclosure relates to the technical field of energy storage battery status prediction, and in particular to a battery health status prediction method based on an XGBoost-LightGBM combined model. Background Art

[0002] With the rapid development of electric vehicles and renewable energy, the health status prediction of batteries, as core components of energy storage and power, has become increasingly important. When the maximum available capacity of lithium batteries drops to 70% to 80% of the factory rated capacity, there will be obvious performance and safety failures, affecting the safety of equipment use. The degree of battery aging is described by the battery health state (SOH).

[0003] In the related technologies, the battery health status prediction method mainly relies on physical models and statistical methods. These methods have limitations when dealing with complex nonlinear relationships. The model method requires the establishment of a battery equivalent circuit model or electrochemical model, but as the battery parameters continue to change, the equivalent model needs to be continuously updated, and there are problems such as difficulty in parameter identification and poor robustness. Summary of the invention

[0004] The present disclosure aims to solve one of the technical problems in the related art at least to some extent.

[0005] To this end, the purpose of the present invention is to propose a battery health status prediction method, device, computer equipment and storage medium based on the XGBoost-LightGBM combined model, thereby being able to fully combine the peak value, peak voltage and area under the peak in the battery IC curve to indicate the battery performance during the prediction process, thereby making the predicted battery health status more accurate, and further improving the prediction accuracy by combining with data-driven prediction methods.

[0006] To achieve the above objectives, the battery health status prediction method based on the XGBoost-LightGBM combined model proposed in the first aspect of the present disclosure includes:

[0007] Acquire state parameters of the battery to be predicted, and generate a battery IC curve according to the state parameters, wherein the state parameters include: discharge temperature peak corresponding time and battery equal voltage drop discharge time;

[0008] Extracting the peak value, peak voltage and area under the peak in the IC curve of the battery;

[0009] The target SOH value of the battery to be predicted is determined according to the corresponding time of the discharge temperature peak, the battery constant voltage drop discharge time, the peak value, the peak voltage and the area under the peak.

[0010] Optionally, in some embodiments, determining the target SOH value of the battery to be predicted according to the discharge temperature peak corresponding time, the battery constant voltage drop discharge time, the peak value, the peak voltage and the area under the peak includes:

[0011] Input the discharge temperature peak corresponding time, the battery equal voltage drop discharge time, the peak value, the peak voltage and the area under the peak into the pre-trained XGBoost-LightGBM combination model, wherein the XGBoost-LightGBM combination model is composed of a parallel XGBoost model and a LightGBM model;

[0012] The target SOH value of the battery to be predicted is determined based on the XGBoost-LightGBM combined model.

[0013] Optionally, in some embodiments, determining the target SOH value of the battery to be predicted based on the XGBoost-LightGBM combined model includes:

[0014] Determine a first predicted SOH value output by the XGBoost model and a second predicted SOH value output by the LightGBM model;

[0015] The first predicted SOH value and the second predicted SOH value are fused based on a preset weighting coefficient to obtain the target SOH value, wherein the weighting coefficient is selected based on cross-validation.

[0016] Optionally, in some embodiments, the XGBoost model and the LightGBM model are trained separately.

[0017] Optionally, in some embodiments, the training parameters of the XGBoost model and the LightGBM model are calculated using a sparrow search algorithm; wherein,

[0018] The training parameters include: maximum number of iterations, iteration step length, maximum depth of the tree and minimum loss function drop value required for node splitting.

[0019] Optionally, in some embodiments, in the sparrow search algorithm, the position of the discoverer is updated based on the global optimal value and the historical optimal value.

[0020] Optionally, in some embodiments, updating the position of the discoverer based on the global optimal value and the historical optimal value includes:

[0021]

[0022] Where t represents the current iteration number in the sparrow search algorithm, represents the position of the i-th discoverer in the j-th dimension in the t+1-th generation, represents the position of the i-th discoverer in the j-th dimension in the t-th generation, iter max is the maximum number of iterations in the sparrow search algorithm, α is a random number selected between 0 and 1, Q is a random number that obeys the normal distribution, L is an all-1 matrix, P2 is the alarm value, ST is the safety threshold, w is the inertia weight, c1 and c2 are two different learning factors, r1 and r2 are two random numbers between [0,1], is the individual best position of the tth iteration, is the global optimal position of the tth iteration.

[0023] Optionally, in some embodiments, the calculation formula corresponding to the inertia weight is:

[0024]

[0025] Optionally, in some embodiments, the sparrow search algorithm uses the determination coefficient of the model as the objective function.

[0026] To achieve the above-mentioned purpose, the battery health status prediction device based on the XGBoost-LightGBM combined model proposed in the second aspect of the present disclosure includes:

[0027] An acquisition module is used to acquire the state parameters of the battery to be predicted, and generate a battery IC curve according to the state parameters, wherein the state parameters include: the corresponding time of the discharge temperature peak and the battery equal voltage drop discharge time;

[0028] An extraction module, used to extract the peak value, peak voltage and area under the peak in the battery IC curve;

[0029] A determination module is used to determine the target SOH value of the battery to be predicted according to the corresponding time of the discharge temperature peak, the battery constant voltage drop discharge time, the peak value, the peak voltage and the area under the peak.

[0030] The computer device proposed in the third aspect embodiment of the present disclosure includes: a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, the battery health status prediction method based on the XGBoost-LightGBM combined model proposed in the first aspect embodiment of the present disclosure is implemented.

[0031] The fourth aspect embodiment of the present disclosure proposes a non-temporary computer-readable storage medium, on which a computer program is stored. When the program is executed by a processor, it implements the battery health status prediction method based on the XGBoost-LightGBM combined model proposed in the first aspect embodiment of the present disclosure.

[0032] The fifth aspect embodiment of the present disclosure proposes a computer program product. When the instructions in the computer program product are executed by a processor, the battery health status prediction method based on the XGBoost-LightGBM combined model proposed in the first aspect embodiment of the present disclosure is executed.

[0033] The battery health status prediction method, device, computer equipment and storage medium based on the XGBoost-LightGBM combined model provided by the present disclosure obtains the state parameters of the battery to be predicted and generates a battery IC curve according to the state parameters, wherein the state parameters include: the corresponding time of the discharge temperature peak and the battery isovoltage drop discharge time; extracting the peak value, peak voltage and area under the peak in the battery IC curve; determining the target SOH value of the battery to be predicted according to the corresponding time of the discharge temperature peak, the battery isovoltage drop discharge time, peak value, peak voltage and area under the peak. As a result, the peak value, peak voltage and area under the peak in the battery IC curve can be fully combined in the prediction process to indicate the battery performance, so that the predicted battery health status is more accurate, and the prediction accuracy is further improved by combining the data-driven prediction method.

[0034] Additional aspects and advantages of the present disclosure will be given in part in the following description and in part will be obvious from the following description or learned through practice of the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] The above and / or additional aspects and advantages of the present disclosure will become apparent and easily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:

[0036] Figure 1 It is a flowchart of a battery health status prediction method based on an XGBoost-LightGBM combined model proposed in an embodiment of the present disclosure;

[0037] Figure 2 It is a flowchart of a battery health status prediction method based on an XGBoost-LightGBM combined model proposed in another embodiment of the present disclosure;

[0038] Figure 3 It is a schematic diagram of a battery health status prediction method based on the XGBoost-LightGBM combined model proposed in the present disclosure;

[0039] Figure 4It is a structural schematic diagram of a battery health status prediction device based on an XGBoost-LightGBM combined model proposed in an embodiment of the present disclosure;

[0040] Figure 5 A block diagram of an exemplary computer device suitable for implementing embodiments of the present disclosure is shown. DETAILED DESCRIPTION

[0041] Embodiments of the present disclosure are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present disclosure, and are not to be construed as limitations of the present disclosure. On the contrary, the embodiments of the present disclosure include all changes, modifications, and equivalents that fall within the spirit and connotation of the appended claims.

[0042] Figure 1 It is a flowchart of a battery health status prediction method based on the XGBoost-LightGBM combined model proposed in an embodiment of the present disclosure.

[0043] Among them, it should be noted that the executor of the battery health status prediction method based on the XGBoost-LightGBM combination model of this embodiment is a battery health status prediction device based on the XGBoost-LightGBM combination model, which can be implemented by software and / or hardware. The device can be configured in a computer device, and the computer device may include but is not limited to a terminal, a server, etc. For example, the terminal may be a mobile phone, a handheld computer, etc.

[0044] like Figure 1 As shown, the battery health status prediction method based on the XGBoost-LightGBM combined model includes:

[0045] S101: Acquire state parameters of a battery to be predicted, and generate a battery IC curve according to the state parameters, wherein the state parameters include: a discharge temperature peak corresponding time and a battery constant voltage drop discharge time.

[0046] The battery to be predicted may refer to a battery for which battery health status prediction is to be performed in the embodiment of the present disclosure. The model of the battery to be predicted is not limited in the embodiment of the present disclosure.

[0047] The state parameter may refer to a relevant parameter reflecting the health state of the battery to be predicted.

[0048] Among them, the battery IC curve (or the battery charge and discharge current-voltage curve) refers to the voltage change of the battery under different current loads. It reveals the performance of the battery at different charge and discharge rates.

[0049] The discharge temperature peak corresponding time refers to the time corresponding to the highest temperature point reached by the battery as the battery temperature changes with time during the battery discharge process.

[0050] Among them, the battery constant voltage drop discharge time refers to the time required for the battery to discharge from a high potential (such as a fully charged state) to a low potential (such as a set discharge termination voltage) at a constant current when the external conditions (such as temperature) remain unchanged during the battery discharge process.

[0051] In the disclosed embodiment, when the state parameters of the battery to be predicted are obtained and the battery IC curve is generated according to the state parameters, a reliable analysis object can be provided for the subsequent extraction of peak value, peak voltage and area under peak.

[0052] S102: Extract the peak value, peak voltage and area under the peak in the battery IC curve.

[0053] It is understandable that there may be differences in the battery IC curves under different health states. Therefore, in the embodiment of the present disclosure, the peak value, peak voltage and area under the peak in the battery IC curve can be extracted as reference indicators for subsequent battery health prediction.

[0054] S103: Determine the target SOH value of the battery to be predicted according to the corresponding time of the discharge temperature peak, the battery constant voltage drop discharge time, the peak value, the peak voltage and the area under the peak.

[0055] Among them, the SOH value (State of Health) refers to a measure of the current health status and remaining usage capacity of the battery. It is used to describe the performance level of the battery relative to its ideal state or the state of a new battery, reflecting changes such as the battery's capacity decay and internal resistance increase. The SOH value is usually expressed as a percentage, 100% represents the performance of the battery in a new state, and an SOH below 100% indicates that the battery has aged or lost a certain amount of power. The target SOH value refers to the SOH value finally determined based on the method proposed in this disclosure.

[0056] In this embodiment, the state parameters of the battery to be predicted are obtained, and the battery IC curve is generated according to the state parameters, wherein the state parameters include: the corresponding time of the discharge temperature peak and the battery isovoltage drop discharge time; the peak value, peak voltage and area under the peak in the battery IC curve are extracted; the target SOH value of the battery to be predicted is determined according to the corresponding time of the discharge temperature peak, the battery isovoltage drop discharge time, peak value, peak voltage and area under the peak. As a result, the peak value, peak voltage and area under the peak in the battery IC curve can be fully combined in the prediction process to indicate the battery performance, so that the predicted battery health status is more accurate, and the prediction accuracy is further improved by combining the data-driven prediction method.

[0057] Figure 2 It is a flowchart of a battery health status prediction method based on the XGBoost-LightGBM combination model proposed in another embodiment of the present disclosure.

[0058] like Figure 2 As shown, the battery health status prediction method based on the XGBoost-LightGBM combined model includes:

[0059] S201: Acquire state parameters of the battery to be predicted, and generate a battery IC curve according to the state parameters, wherein the state parameters include: discharge temperature peak corresponding time and battery constant voltage drop discharge time.

[0060] S202: Extract the peak value, peak voltage and area under the peak in the battery IC curve.

[0061] The description of S201 and S202 can be specifically referred to the above embodiment, which will not be repeated here.

[0062] S203: Input the discharge temperature peak corresponding time, battery equal-voltage drop discharge time, peak value, peak voltage and area under the peak into the pre-trained XGBoost-LightGBM combination model, wherein the XGBoost-LightGBM combination model is composed of a parallel XGBoost model and a LightGBM model.

[0063] That is to say, in the embodiments of the present disclosure, the XGBoost model and the LightGBM model can be pre-trained, and the two can be combined to obtain an XGBoost-LightGBM combined model as a battery health prediction model.

[0064] Optionally, in some embodiments, the XGBoost model and the LightGBM model are trained separately. Thus, the independence of the obtained XGBoost model and the LightGBM model can be effectively improved, thereby ensuring the practicality and reliability of the prediction results obtained by the XGBoost model and the LightGBM model.

[0065] Optionally, in some embodiments, the training parameters of the XGBoost model and the LightGBM model are calculated using a sparrow search algorithm; wherein the training parameters include: the maximum number of iterations, the iteration step, the maximum depth of the tree, and the minimum loss function drop value required for node splitting. Thus, the applicability of the obtained training parameters to the XGBoost model and the LightGBM model can be guaranteed based on the sparrow search algorithm.

[0066] Optionally, in some embodiments, in the sparrow search algorithm, the position of the finder is updated based on the global optimal value and the historical optimal value. Thus, the finder position can be quickly and accurately updated based on the global optimal value and the historical optimal value.

[0067] Optionally, in some embodiments, when updating the position of the discoverer based on the global optimal value and the historical optimal value, it may be:

[0068]

[0069] Among them, r represents the current iteration number in the sparrow search algorithm, represents the position of the i-th discoverer in the j-th dimension in the t+1-th generation, represents the position of the i-th discoverer in the j-th dimension in the t-th generation, iter max is the maximum number of iterations in the sparrow search algorithm, α is a random number selected between 0 and 1, Q is a random number that obeys the normal distribution, L is an all-1 matrix, P2 is the alarm value, ST is the safety threshold, w is the inertia weight, c1 and c2 are two different learning factors, r1 and r2 are two random numbers between [0,1], is the individual best position of the tth iteration, is the global optimal position of the tth iteration.

[0070] Optionally, in some embodiments, the calculation formula corresponding to the inertia weight is:

[0071]

[0072] Optionally, in some embodiments, the determination coefficient of the model is used as the objective function in the sparrow search algorithm.

[0073] S204: Determine the target SOH value of the battery to be predicted based on the XGBoost-LightGBM combined model.

[0074] Optionally, in some embodiments, when determining the target SOH value of the battery to be predicted based on the XGBoost-LightGBM combined model, the first predicted SOH value output by the XGBoost model and the second predicted SOH value output by the LightGBM model may be determined; the first predicted SOH value and the second predicted SOH value are fused based on a preset weighting coefficient to obtain the target SOH value, wherein the weighting coefficient is selected based on cross-validation. Thus, the fusion processing of the first predicted SOH value and the second predicted SOH value can be realized based on the weighting coefficient to ensure the indication accuracy of the obtained target SOH value.

[0075] Among them, the first predicted SOH value may refer to the predicted SOH value obtained by processing the discharge temperature peak corresponding time, battery equal-voltage drop discharge time, peak value, peak voltage and area under the peak based on the XGBoost model.

[0076] Among them, the second predicted SOH value may refer to the predicted SOH value obtained by processing the corresponding time of the discharge temperature peak, the battery equal-voltage drop discharge time, the peak value, the peak voltage and the area under the peak based on the LightGBM model.

[0077] That is to say, in the embodiment of the present disclosure, after extracting the peak value, peak voltage and area under the peak in the battery IC curve, the corresponding time of the discharge temperature peak, the battery isovoltage drop discharge time, peak value, peak voltage and area under the peak can be input into the pre-trained XGBoost-LightGBM combined model, wherein the XGBoost-LightGBM combined model is composed of a parallel XGBoost model and a LightGBM model; the target SOH value of the battery to be predicted is determined based on the XGBoost-LightGBM combined model. Therefore, the accurate prediction of the target SOH value can be achieved based on the XGBoost-LightGBM combined model, which can effectively improve the prediction efficiency.

[0078] In this embodiment, the corresponding time of the discharge temperature peak, the battery isovoltage drop discharge time, the peak value, the peak voltage and the area under the peak are input into the pre-trained XGBoost-LightGBM combined model, wherein the XGBoost-LightGBM combined model is composed of a parallel XGBoost model and a LightGBM model; the target SOH value of the battery to be predicted is determined based on the XGBoost-LightGBM combined model. Therefore, the target SOH value can be accurately predicted based on the XGBoost-LightGBM combined model, which can effectively improve the prediction efficiency.

[0079] It is understandable that as the battery health status prediction method based on machine learning has gradually become a research hotspot, the gradient boosting tree algorithm has attracted much attention due to its excellent prediction performance. XGBoost and LightGBM, as two advanced gradient boosting tree algorithms, have demonstrated excellent performance in many fields. However, the prediction results of a single algorithm may be limited by data characteristics and model parameter settings. Therefore, through the fusion and complementarity between different data-driven methods, the combined model method can improve the accuracy of SOH evaluation and is expected to further improve the prediction accuracy.

[0080] The method of the present invention includes selecting influencing factors of health status, data preprocessing, model training, combined model construction, model evaluation, model online prediction, etc.

[0081] Based on the above embodiments, the present invention discloses a battery health status prediction method based on the XGBoost-LightGBM combined model, comprising the following steps: obtaining the state parameters of the battery, and generating the IC curve of the battery according to the state parameters; wherein the state parameters include the corresponding time of the discharge temperature peak and the battery equal-voltage drop discharge time; extracting the peak value, peak voltage and area under the peak in the IC curve; determining the SOH value of the battery according to the corresponding time of the discharge temperature peak, the battery equal-voltage drop discharge time, peak value, peak voltage and area under the peak.

[0082] The present invention selects the discharge temperature peak corresponding time, battery equal voltage drop discharge time, and peak value, peak voltage and area under peak in IC curve that best reflect the performance of the battery from several parameters of the battery, so as to make the predicted battery health status more accurate. Combined with the data-driven prediction method, the prediction accuracy can be further improved.

[0083] Regarding the factors affecting the health status, the battery charging and discharging process is accompanied by the insertion and deinsertion of lithium ions, and the chemical reactions in the battery cause the battery capacity to decay and performance to decline. In this invention, SOH is used to characterize the degree of battery aging, which is defined as the ratio of the maximum available capacity in the current state to the rated capacity, and is expressed by the internal resistance and capacity change trend:

[0084]

[0085] Among them, SOH is the battery health state; C t is the maximum available capacity in the current state; C0 is the nominal capacity (i.e. rated capacity); R t is the internal resistance of the battery; R end is the internal resistance of the battery that has reached the end of its service life; R0 is the nominal resistance.

[0086] The battery life decay rate is affected by many factors, which are manifested as changes in capacity, voltage rise, voltage drop, surface temperature, etc. The primary quantities and their variant data that can be directly measured, such as voltage, current, temperature, etc., are selected, and the battery capacity is used as the metric for battery SOH. The health characteristics of the battery during the charging and discharging stages are extracted as indirect health status influencing factors.

[0087] The change in battery surface temperature is an important parameter that characterizes the battery degradation trend. The discharge temperature of the battery rises nonlinearly at different usage cycles (one cycle refers to one charge and one discharge), and the change trend is consistent.

[0088] The change trend of the discharge temperature peak is obvious. It increases with the increase of cycle number and the time of occurrence is advanced. Therefore, the change trend of the discharge temperature peak obviously reflects the battery aging trend. Therefore, the corresponding time of the discharge temperature peak can be selected as the influencing factor.

[0089] As the number of cycles increases, the time it takes to reach the cut-off voltage is shortened, so the battery constant voltage drop discharge time under constant current discharge can also be selected as an influencing factor. Specifically, the cut-off voltage refers to the minimum operating voltage value at which the battery should not continue to discharge when the voltage drops during battery discharge. The cut-off voltage standards are different for different battery types.

[0090] In addition, by analyzing the peak position, height, area and distance between different peaks of the IC (capacity increment) curve, the degradation process, aging mechanism characteristics and health status changes of the lithium battery can be explored. In summary, the peak value, peak voltage and area under the peak of the IC (capacity increment) curve are selected as the final influencing factors in the present invention.

[0091] Next, the corresponding time of the discharge temperature peak, the battery equal-voltage drop discharge time, peak value, peak voltage and area under the peak are input into the XGBoost-LightGBM combined model; wherein the XGBoost-LightGBM combined model is composed of a parallel XGBoost model and a LightGBM model; the SOH value of the battery is determined based on the XGBoost-LightGBM combined model.

[0092] After selecting the final influencing factors, data preprocessing is required. The battery data set is cleaned, normalized, and feature engineered to ensure data quality and the validity of model input, and the training set and test set are divided.

[0093] After obtaining the data set, the model is trained based on the obtained data set. In the present invention, the training processes of the XGBoost model and the LightGBM model do not interfere with each other, and they are trained separately.

[0094] XGBoost model and LightGBM model training: select the training parameters of the XGBoost model and LightGBM model, such as the maximum number of iterations, iteration step, maximum depth of the tree, and the minimum loss function drop value required for node splitting. According to verification, the above training parameters have a great influence on the model output results. Therefore, in order to reduce the amount of calculation, in the present invention, only the above training parameters are calculated by a pre-designed method, and other training parameters are obtained according to the initial settings of the model. Use the training data set for model training and save the trained model.

[0095] Regarding the calculation method of the model parameters, the improved sparrow search algorithm is selected for calculation in the embodiment of the present invention. The sparrow search algorithm simulates the foraging and anti-predation behavior of sparrows by continuously updating the individual positions. Compared with the traditional algorithm, the sparrow search algorithm has a simple structure, is easy to implement, has fewer control parameters, and has a strong local search capability. The algorithm performs better than traditional algorithms such as particle swarm algorithm and ant colony algorithm on single-peak and multi-peak benchmark functions. In the sparrow search algorithm, individuals are divided into discoverers, followers and guards, and each individual position corresponds to a solution. According to the algorithm settings, the proportion of guards in the population is 10% to 20%, while the discoverers and followers are dynamically changing, that is, one individual becomes a discoverer, which necessarily means that another individual will become a follower. According to the division of labor, the discoverer mainly provides foraging direction and area for the entire population, the follower follows the discoverer to forage, and the guard is responsible for monitoring the foraging area. In the foraging process, the acquisition of resources is completed by continuously updating the positions of the three.

[0096] For the parameters that need to be optimized for the XGBoost model and the LightGBM model, the position of each sparrow represents the optimized solution, and x1 is the position of the first sparrow (the model parameter group obtained in the first iteration), that is, x1 = [x 1，1 , x 1，2 , x 1，3 , x 1，4 ], x 1，1 represents the maximum number of iterations mentioned above, x 1，2 represents the iteration step, x 1，3 Indicates the maximum depth of the tree, x 1，4 Represents the minimum loss function drop value required for node splitting.

[0097] In the usual sparrow search algorithm, the finder's position is updated as follows:

[0098]

[0099] Where t represents the current iteration number, represents the position of the i-th discoverer in the j-th dimension in the t+1-th generation; represents the position of the i-th discoverer in the j-th dimension in the t-th generation, iter max is the maximum number of iterations in the sparrow algorithm, α represents a random number selected between 0 and 1, Q is a random number that obeys a normal distribution, L is an all-1 matrix (a 1*4-dimensional all-1 matrix in this embodiment), P2 is the alarm value, and ST represents the safety threshold.

[0100] In the present invention, the global optimal value and the individual historical optimal value are introduced to improve the position of the discoverer, so that it is not only affected by the global optimal individual position, but also by the individual historical optimal position, thereby improving the information exchange ability between sparrow populations, thereby improving the search speed and optimization accuracy of the algorithm; that is, the position of the discoverer is updated based on the global optimal value and the historical optimal value.

[0101] Specifically, the location update method of the improved discoverer is as follows:

[0102]

[0103] Where w is the inertia weight, c1 and c2 are two different learning factors, r1 and r2 are two random numbers between [0,1]. is the individual best position of the tth iteration, is the global optimal position of the tth iteration.

[0104] In the early stage of optimization, the weight factor is large, the algorithm's global exploration performance is strong and the search range is wide; in the later stage of optimization, the weight factor decreases rapidly, the algorithm's local development ability is strong, and it is conducive to rapid convergence.

[0105] The follower position update formula is as follows:

[0106]

[0107] in, represents the position of the i-th follower in the t+1th generation in the j-th dimension, is the position of the individual with the worst fitness in the tth iteration; A represents a 1*4 matrix, and each element in the matrix is ​​randomly preset to -1 or 1. + =A T (AA T ) -1 , n is the number of initialized sparrows (i.e. the number of individuals).

[0108] The formula for updating the position of the sentinel is as follows:

[0109]

[0110] in, represents the position of the i-th sentinel in the j-th dimension in the t+1-th generation, represents the position of the i-th sentinel in the t-th generation in the j-th dimension, β is the control step length, and it obeys a normal distribution with a mean of 0 and a variance of 1; k is selected between [-1, 1], and ε is a constant; f i represents the fitness value of the current i-th individual; f g and f w Respectively represent the fitness values ​​of the current global optimal and worst individuals.

[0111] In the present invention, R 2 Optimize the (coefficient of determination) objective function:

[0112]

[0113] in, represents the predicted value (i.e., the predicted SOH) obtained by predicting the i-th input data using the model trained by the model parameters obtained by the sparrow search algorithm; i is the actual SOH value corresponding to the i-th input data; It is the average of the actual SOH values ​​corresponding to several input data.

[0114] Next, a combined model is constructed based on the two trained models. In the combined model, the prediction results of the XGBoost model and the LightGBM model are fused using the weighted average method, and the prediction results are adjusted by setting different weight coefficients for the two models.

[0115] Specifically, the optimal weight combination of each model (i.e., the weight coefficients of the XGBoost model and the LightGBM model) is selected through cross-validation to maximize the prediction performance of the combined model. For example, the prediction result of the XGBoost model is 0.8, the prediction result of the LightGBM model is 0.6, and the actual SOH value of the battery is 0.75. The optional weight coefficients of the XGBoost model and the LightGBM model are designed to be several optional values, such as 0.2, 0.4, 0.6, and 0.8.

[0116] When selecting a weighting coefficient, select different weighting coefficients for verification:

[0117] 0.8*0.2+0.6*0.8=0.64;

[0118] 0.8*0.4+0.6*0.6=0.68;

[0119] 0.8*0.6+0.6*0.4=0.72;

[0120] 0.8*0.8+0.6*0.2=0.76;

[0121] Since 0.76 is closest to 0.75, the weight coefficient assigned to the XGBoost model is 0.8, and the weight coefficient assigned to the LightGBM model is 0.2.

[0122] In addition, the combined model can be evaluated using the validation data set to calculate the mean square error (MSE), root mean square error (RMSE) and coefficient of determination (R2 ) and other indicators to evaluate the prediction accuracy and stability of the model.

[0123] When the prediction accuracy and stability of the combined model meet the evaluation criteria, online prediction can be performed. In practical applications, the operating data of the battery is collected in real time and preprocessed, and the preprocessed data is input into the trained combined model. The model is optimized through parameter tuning, and the health status prediction results of the battery are predicted and output, and corresponding management and maintenance decisions are made based on the prediction results.

[0124] In actual application, it is also possible to design regular training for the combined model to effectively improve the recognition accuracy of the combined model in the long term.

[0125] In summary, the present invention provides a battery health state prediction method based on the XGBoost-LightGBM combined model. This method improves the accuracy and stability of battery health state prediction by integrating learning technology and combining the advantages of two efficient gradient boosting algorithms. This method has broad application prospects in battery management systems and can effectively extend battery life and improve system safety.

[0126] The present invention combines two efficient gradient boosting algorithms, XGBoost and LightGBM, to improve the accuracy and stability of battery health status prediction through model fusion strategy, ensures the generalization ability and prediction performance of the model on different data sets through cross-validation and parameter tuning, realizes real-time prediction of battery health status, provides reliable decision support for battery management system, improves the safety and reliability of the system, and effectively copes with the complex working environment and nonlinear characteristics of the battery. It has broad application prospects and important practical significance.

[0127] Figure 4 It is a structural diagram of a battery health status prediction device based on the XGBoost-LightGBM combined model proposed in an embodiment of the present disclosure.

[0128] like Figure 4 As shown, the battery health status prediction device 40 based on the XGBoost-LightGBM combined model includes:

[0129] The acquisition module 401 is used to acquire the state parameters of the battery to be predicted, and generate the battery IC curve according to the state parameters, wherein the state parameters include: the corresponding time of the discharge temperature peak and the battery constant voltage drop discharge time;

[0130] Extraction module 402, used to extract the peak value, peak voltage and area under the peak in the battery IC curve;

[0131] The determination module 403 is used to determine the target SOH value of the battery to be predicted according to the corresponding time of the discharge temperature peak, the battery constant voltage drop discharge time, the peak value, the peak voltage and the area under the peak.

[0132] It should be noted that the aforementioned explanation of the battery health status prediction method based on the XGBoost-LightGBM combination model is also applicable to the battery health status prediction device based on the XGBoost-LightGBM combination model of this embodiment, and will not be repeated here.

[0133] In this embodiment, the state parameters of the battery to be predicted are obtained, and the battery IC curve is generated according to the state parameters, wherein the state parameters include: the corresponding time of the discharge temperature peak and the battery isovoltage drop discharge time; the peak value, peak voltage and area under the peak in the battery IC curve are extracted; the target SOH value of the battery to be predicted is determined according to the corresponding time of the discharge temperature peak, the battery isovoltage drop discharge time, peak value, peak voltage and area under the peak. As a result, the peak value, peak voltage and area under the peak in the battery IC curve can be fully combined in the prediction process to indicate the battery performance, so that the predicted battery health status is more accurate, and the prediction accuracy is further improved by combining the data-driven prediction method.

[0134] Figure 5 A block diagram of an exemplary computer device suitable for implementing embodiments of the present disclosure is shown. Figure 5 The computer device 12 shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present disclosure.

[0135] like Figure 5 As shown, the computer device 12 is in the form of a general-purpose computing device. The components of the computer device 12 may include, but are not limited to: one or more processors or processing units 16, a system memory 28, and a bus 18 that connects various system components (including the system memory 28 and the processing unit 16).

[0136] The bus 18 represents one or more of several types of bus structures, including a memory bus or memory controller, a peripheral bus, a graphics acceleration port, a processor or a local bus using any of a variety of bus structures. For example, these architectures include but are not limited to Industry Standard Architecture (ISA) bus, Micro Channel Architecture (MAC) bus, Enhanced ISA bus, Video Electronics Standards Association (VESA) local bus and Peripheral Component Interconnection (PCI) bus.

[0137] The computer device 12 typically includes a variety of computer system readable media. These media can be any available media that can be accessed by the computer device 12, including volatile and non-volatile media, removable and non-removable media.

[0138] The memory 28 may include computer system readable media in the form of volatile memory, such as random access memory (RAM) 30 and / or cache memory 32. The computer device 12 may further include other removable / non-removable, volatile / non-volatile computer system storage media. By way of example only, the storage system 34 may be used to read and write non-removable, non-volatile magnetic media ( Figure 5 Not shown, often called a "hard drive").

[0139] although Figure 5 Not shown, a disk drive for reading and writing a removable non-volatile disk (e.g., a "floppy disk"), and an optical disk drive for reading and writing a removable non-volatile optical disk (e.g., a compact disc read only memory (Compact Disc Read Only Memory; hereinafter referred to as: CD-ROM), a digital versatile disc read only memory (Digital Video Disc Read Only Memory; hereinafter referred to as: DVD-ROM) or other optical media) may be provided. In these cases, each drive may be connected to the bus 18 via one or more data medium interfaces. The memory 28 may include at least one program product having a set (e.g., at least one) of program modules that are configured to perform the functions of the various embodiments of the present disclosure.

[0140] A program / utility 40 having a set (at least one) of program modules 42 may be stored, for example, in the memory 28, such program modules 42 including but not limited to an operating system, one or more application programs, other program modules, and program data, each of which or some combination may include an implementation of a network environment. The program modules 42 generally perform the functions and / or methods of the embodiments described in the present disclosure.

[0141] The computer device 12 may also communicate with one or more external devices 14 (e.g., keyboard, pointing device, display 24, etc.), one or more devices that enable a human body to interact with the computer device 12, and / or any device that enables the computer device 12 to communicate with one or more other computing devices (e.g., network card, modem, etc.). Such communication may be performed via an input / output (I / O) interface 22. In addition, the computer device 12 may also communicate with one or more networks (e.g., a local area network (LAN), a wide area network (WAN), and / or a public network, such as the Internet) via a network adapter 20. As shown, the network adapter 20 communicates with other modules of the computer device 12 via a bus 18. It should be understood that, although not shown in the figure, other hardware and / or software modules may be used in conjunction with the computer device 12, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0142] The processing unit 16 executes various functional applications and data processing by running programs stored in the system memory 28, such as implementing the battery health status prediction method based on the XGBoost-LightGBM combined model mentioned in the aforementioned embodiment.

[0143] In order to implement the above-mentioned embodiments, the present disclosure also proposes a non-temporary computer-readable storage medium, on which a computer program is stored. When the program is executed by a processor, it implements the battery health status prediction method based on the XGBoost-LightGBM combined model proposed in the aforementioned embodiments of the present disclosure.

[0144] In order to implement the above embodiments, the present disclosure also proposes a computer program product. When the instruction processor in the computer program product is executed, it executes the battery health status prediction method based on the XGBoost-LightGBM combination model proposed in the aforementioned embodiments of the present disclosure.

[0145] Those skilled in the art will readily appreciate other embodiments of the present disclosure after considering the specification and practicing the invention disclosed herein. The present disclosure is intended to cover any variations, uses, or adaptations of the present disclosure that follow the general principles of the present disclosure and include common knowledge or customary techniques in the art that are not disclosed in the present disclosure. The description and examples are to be considered exemplary only, and the true scope and spirit of the present disclosure are indicated by the following claims.

[0146] It should be understood that the present disclosure is not limited to the exact structures that have been described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present disclosure is limited only by the appended claims.

[0147] It should be noted that, in the description of the present disclosure, the terms "first", "second", etc. are only used for descriptive purposes and cannot be understood as indicating or implying relative importance. In addition, in the description of the present disclosure, unless otherwise specified, the meaning of "plurality" is two or more.

[0148] Any process or method description in a flowchart or otherwise described herein may be understood to represent a module, segment or portion of code that includes one or more executable instructions for implementing the steps of a specific logical function or process, and the scope of the preferred embodiments of the present disclosure includes alternative implementations in which functions may not be performed in the order shown or discussed, including performing functions in a substantially simultaneous manner or in the reverse order depending on the functions involved, which should be understood by those skilled in the art to which the embodiments of the present disclosure belong.

[0149] It should be understood that the various parts of the present disclosure can be implemented in hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, multiple steps or methods can be implemented in software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0150] A person skilled in the art may understand that all or part of the steps in the method for implementing the above-mentioned embodiment may be completed by instructing related hardware through a program, and the program may be stored in a computer-readable storage medium, which, when executed, includes one or a combination of the steps of the method embodiment.

[0151] In addition, each functional unit in each embodiment of the present disclosure may be integrated into a processing module, or each unit may exist physically separately, or two or more units may be integrated into one module. The above-mentioned integrated module may be implemented in the form of hardware or in the form of a software functional module. If the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it may also be stored in a computer-readable storage medium.

[0152] The storage medium mentioned above can be a read-only memory, a magnetic disk or an optical disk, etc.

[0153] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present disclosure. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0154] Although the embodiments of the present disclosure have been shown and described above, it is to be understood that the above embodiments are exemplary and are not to be construed as limitations of the present disclosure. A person skilled in the art may change, modify, replace and vary the above embodiments within the scope of the present disclosure.

Claims

1. A battery health status prediction method based on XGBoost-LightGBM combined model, characterized in that: include: Acquire state parameters of the battery to be predicted, and generate a battery IC curve according to the state parameters, wherein the state parameters include: discharge temperature peak corresponding time and battery equal voltage drop discharge time; Extracting the peak value, peak voltage and area under the peak in the IC curve of the battery; The target SOH value of the battery to be predicted is determined according to the corresponding time of the discharge temperature peak, the battery constant voltage drop discharge time, the peak value, the peak voltage and the area under the peak.

2. The method according to claim 1, characterized in that The determining the target SOH value of the battery to be predicted according to the discharge temperature peak corresponding time, the battery constant voltage drop discharge time, the peak value, the peak voltage and the area under the peak includes: Input the discharge temperature peak corresponding time, the battery equal voltage drop discharge time, the peak value, the peak voltage and the area under the peak into the pre-trained XGBoost-LightGBM combination model, wherein the XGBoost-LightGBM combination model is composed of a parallel XGBoost model and a LightGBM model; The target SOH value of the battery to be predicted is determined based on the XGBoost-LightGBM combined model.

3. The method according to claim 2, characterized in that The determining the target SOH value of the battery to be predicted based on the XGBoost-LightGBM combined model includes: Determine a first predicted SOH value output by the XGBoost model and a second predicted SOH value output by the LightGBM model; The first predicted SOH value and the second predicted SOH value are fused based on a preset weighting coefficient to obtain the target SOH value, wherein the weighting coefficient is selected based on cross-validation.

4. The method according to claim 2, characterized in that The XGBoost model and the LightGBM model are both trained separately.

5. The method according to claim 2, characterized in that The training parameters of the XGBoost model and the LightGBM model are calculated using the sparrow search algorithm; The training parameters include: maximum number of iterations, iteration step length, maximum depth of the tree and minimum loss function drop value required for node splitting.

6. The method according to claim 5, characterized in that In the sparrow search algorithm, the position of the finder is updated based on the global optimal value and the historical optimal value.

7. The method according to claim 6, characterized in that The updating of the position of the discoverer based on the global optimal value and the historical optimal value includes: Where t represents the current iteration number in the sparrow search algorithm, represents the position of the i-th discoverer in the j-th dimension in the t+1-th generation, represents the position of the i-th discoverer in the j-th dimension in the t-th generation, iter max is the maximum number of iterations in the sparrow search algorithm, α is a random number selected between 0 and 1, Q is a random number that obeys the normal distribution, L is an all-1 matrix, P2 is the alarm value, ST is the safety threshold, w is the inertia weight, c1 and c2 are two different learning factors, r1 and r2 are two random numbers between [0,1], is the individual best position of the tth iteration, is the global optimal position of the tth iteration.

8. The method according to claim 7, characterized in that The calculation formula corresponding to the inertia weight is:

9. The method according to claim 7, characterized in that The sparrow search algorithm adopts the determination coefficient of the model as the objective function.

10. A battery health status prediction device based on the XGBoost-LightGBM combined model, characterized in that: include: An acquisition module is used to acquire the state parameters of the battery to be predicted, and generate a battery IC curve according to the state parameters, wherein the state parameters include: the corresponding time of the discharge temperature peak and the battery equal voltage drop discharge time; An extraction module, used to extract the peak value, peak voltage and area under the peak in the battery IC curve; A determination module is used to determine the target SOH value of the battery to be predicted according to the corresponding time of the discharge temperature peak, the battery constant voltage drop discharge time, the peak value, the peak voltage and the area under the peak.

11. A computer device, characterized in that: include: at least one processor; as well as a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 9.

12. A non-transitory computer-readable storage medium storing computer instructions, characterized in that: in, The computer instructions are used to cause the computer to execute the method according to any one of claims 1 to 9.

13. A computer program product, characterized in that The invention comprises a computer program which, when executed by a processor, implements the steps of the method according to any one of claims 1 to 9.

Images