Storage battery pack parameter estimation method, device, equipment, medium and product
By collecting voltage and current data of lead-acid battery packs, and optimizing neural network models using preprocessing and genetic algorithms, the problems of complex and inaccurate parameter estimation in the prior art are solved, and efficient and accurate parameter estimation is achieved.
Patent Information
- Application Number
- CN202510306615.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-03-14
- Publication Date
- 2025-07-25
AI Technical Summary
The existing lead-acid battery pack parameter estimation method requires a lot of time and manpower, cannot accurately reflect the complex dynamic response of the battery, and the construction of an equivalent circuit model is complicated.
The voltage and current data of the lead-acid battery pack during discharge is collected, the characteristic data is extracted through pre-processing strategies, and the parameter estimation is performed using a neural network model optimized by genetic algorithms, including the estimation of battery capacity, single-unit voltage and group voltage.
Simple and high-precision parameter estimation is realized, reducing experimental requirements and improving estimation accuracy.
Smart Images
Figure CN120370159A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of artificial intelligence, and particularly to a method, device, equipment, medium and product for estimating parameters of a battery pack. Background Art
[0002] Lead-acid batteries are used as backup power sources in substations to store electrical energy and respond to grid abnormalities, etc., and play a key role in maintaining the safe and stable operation of substations. During the repeated charge and discharge processes of lead-acid batteries, chemical energy and electrical energy are converted into each other, and the active substances are continuously consumed, and the battery life and power supply capacity are also affected. In order to determine whether the battery meets the operation and maintenance requirements, it is necessary to confirm the relevant parameters reflecting the battery life, such as capacity, internal resistance, etc., so as to accurately analyze and understand the battery health status.
[0003] Existing methods for estimating battery pack parameters require a large amount of time and manpower for experiments or constructing complex equivalent circuit models, and cannot accurately reflect the complex dynamic response of the battery. Summary of the Invention
[0004] This application provides a method, device, equipment, medium and product for estimating parameters of a battery pack, which solves the defect that existing methods for estimating battery pack parameters require a large amount of time and manpower for experiments or constructing complex equivalent circuit models and cannot accurately reflect the complex dynamic response of the battery.
[0005] This application provides a method for estimating parameters of a battery pack, including: Collecting voltage data and current data of each single battery in the battery pack during the discharge process, the discharge process includes a voltage steep drop and recovery stage and a normal discharge stage, and the voltage data includes first voltage data corresponding to the voltage steep drop and recovery stage and second voltage data corresponding to the normal discharge stage; Preprocessing the first voltage data based on the preprocessing strategy corresponding to the voltage steep drop and recovery stage to obtain first voltage characteristic data, and preprocessing the second voltage data based on the preprocessing strategy corresponding to the normal discharge stage to obtain second voltage characteristic data; Inputting the current data, the first voltage characteristic data and the second voltage characteristic data into a pre-trained battery parameter estimation model to obtain the estimated parameters of the battery pack output by the battery parameter estimation model, and the estimated parameters of the battery pack include battery capacity, single cell voltage and group voltage; Wherein, the battery parameter estimation model is obtained by training a neural network model with current data samples, first voltage characteristic data samples, second voltage characteristic data samples and battery pack estimated parameter samples, and optimizing the trained neural network model based on a genetic algorithm.
[0006] As an embodiment, the preprocessing strategy corresponding to the voltage dip and recovery stage includes extracting features of the first voltage data in the voltage dimension and the time dimension, and the preprocessing strategy corresponding to the normal discharge stage includes interpolating and fitting the second voltage data. Correspondingly, preprocessing the first voltage data based on the preprocessing strategy corresponding to the voltage dip and recovery stage to obtain first voltage feature data, and preprocessing the second voltage data based on the preprocessing strategy corresponding to the normal discharge stage to obtain second voltage feature data, including: In the voltage dimension, extract the difference between the discharge starting voltage and the valley voltage, the difference between the peak voltage and the valley voltage, and the valley voltage from the first voltage data; In the time dimension, extract the difference in time points between the discharge starting voltage and the valley voltage, the difference in time points between the peak voltage and the valley voltage, and the time point of the valley voltage from the first voltage data; Take the difference between the discharge starting voltage and the valley voltage, the difference between the peak voltage and the valley voltage, the valley voltage, the difference in time points between the discharge starting voltage and the valley voltage, the difference in time points between the peak voltage and the valley voltage, and the time point of the valley voltage as the first voltage feature data; Based on the pre-constructed fitting function corresponding to the second voltage data, take the coefficients of the fitting function as the second voltage feature data.
[0007] As an embodiment, the neural network model includes an input layer, a hidden layer, and an output layer, and the genetic algorithm is used to optimize the weights and the center width vectors of the Gaussian basis functions between the hidden layer and the output layer.
[0008] As an embodiment, optimizing the trained neural network model based on the genetic algorithm includes: According to the initial Gaussian basis function width vectors and weight parameters of each neuron in the hidden layer of the neural network model, obtain the chromosome of the genetic algorithm; According to the error of the neural network model, determine the fitness function of the genetic factor of the genetic algorithm; Perform genetic operation optimization on the population composed of chromosomes according to the fitness function until the preset fitness or the preset number of iterations is reached, and obtain the optimal solution of the genetic algorithm; Optimize the weights and the center width vectors of the Gaussian basis functions between the hidden layer and the output layer of the neural network model according to the optimal solution of the genetic algorithm.
[0009] As an embodiment, it further includes: During the training process, the accuracy of the neural network model is evaluated based on the mean absolute error and / or the mean absolute percentage error, and the neural network model is adjusted according to the evaluation results.
[0010] As an embodiment, collecting the voltage data and current data of each single battery in the battery pack during discharge includes: Collecting the voltage data and current data of each single battery in the battery pack during discharge based on a battery discharger.
[0011] This application also provides a device for estimating battery pack parameters, and the device includes: A data acquisition module, configured to collect the voltage data and current data of each single battery in the battery pack during discharge. The discharge process includes a voltage steep drop and recovery stage and a normal discharge stage. The voltage data includes first voltage data corresponding to the voltage steep drop and recovery stage and second voltage data corresponding to the normal discharge stage; A preprocessing module, configured to preprocess the first voltage data based on a preprocessing strategy corresponding to the voltage steep drop and recovery stage to obtain first voltage feature data, and preprocess the second voltage data based on a preprocessing strategy corresponding to the normal discharge stage to obtain second voltage feature data; A parameter estimation module, configured to input the current data, the first voltage feature data, and the second voltage feature data into a pre-trained battery parameter estimation model to obtain the estimated parameters of the battery pack output by the battery parameter estimation model. The estimated parameters of the battery pack include battery capacity, single cell voltage, and group voltage; Wherein, the battery parameter estimation model is obtained by training a neural network model with current data samples, first voltage feature data samples, and second voltage feature data samples, and optimizing the trained neural network model based on a genetic algorithm.
[0012] This application also provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, the method for estimating battery pack parameters as described in any one of the above is implemented.
[0013] This application also provides a non-transitory computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the method for estimating battery pack parameters as described in any one of the above is implemented.
[0014] This application also provides a computer program product, including a computer program. When the computer program is executed by a processor, the method for estimating battery pack parameters as described in any one of the above is implemented.
[0015] The battery pack parameter estimation method, device, equipment, medium and product provided by this application only need to collect the discharge data of the battery pack during the discharge process, and then the battery parameter estimation model trained by the artificial intelligence algorithm can be used for parameter estimation. There is no need for a complex discharge test design, and the implementation is convenient. The battery parameter estimation model is optimized by the genetic algorithm, which greatly improves the accuracy of parameter estimation. Description of the Drawings
[0016] In order to more clearly illustrate the technical solutions in this application or the prior art, the following will briefly introduce the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings in the following description are some embodiments of this application. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.
[0017] Figure 1 It is one of the flow diagrams of the battery pack parameter estimation method provided by this application.
[0018] Figure 2 It is the voltage change curve diagram of the battery pack discharge process provided by this application.
[0019] Figure 3 It is the schematic diagram of different stages of the discharge process provided by this application.
[0020] Figure 4 It is the schematic diagram of feature extraction in the voltage steep drop and recovery stage provided by this application.
[0021] Figure 5 It is the schematic diagram of feature extraction in the normal discharge stage provided by this application.
[0022] Figure 6 It is the schematic diagram of the structure of the neural network model provided by this application.
[0023] Figure 7 It is the flow diagram of optimizing the model based on the genetic algorithm provided by this application.
[0024] Figure 8 It is the second flow diagram of the battery pack parameter estimation method provided by this application.
[0025] Figure 9 It is the schematic diagram of the structure of the device for the battery pack parameter estimation method provided by this application.
[0026] Figure 10 It is the schematic diagram of the structure of the electronic device provided by this application. Detailed Embodiments
[0027] To make the objectives, technical solutions, and advantages of this application clearer, the following will clearly and completely describe the technical solutions in this application in conjunction with the accompanying drawings in this application. Obviously, the described embodiments are part of the embodiments of this application, rather than all the embodiments. Based on the embodiments in this application, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of this application.
[0028] It should be noted that all actions of obtaining signals, information, or data in this application are carried out on the premise of complying with the corresponding data protection regulations and policies of the location and with the authorization given by the owner of the corresponding device.
[0029] Lead-acid batteries usually work in a series structure with 24 cells in a group and also have a common discharge current and discharge time during the discharge test. To accurately analyze and understand the health status of the battery and replace the battery in time when the battery performance is too low, it is necessary to extract and analyze the parameters of the battery. For example, the remaining capacity of a lead-acid battery is an important parameter in the battery performance index and has a crucial impact on the output performance and service life of the battery. Existing methods for estimating the remaining capacity of batteries include the ampere-hour integration method, the equivalent circuit method, the pulse test method, etc. The ampere-hour integration method takes a long time and requires a lot of time and manpower. The construction of the equivalent circuit model in the equivalent circuit method is complex and cannot accurately reflect the complex dynamic response of the battery. The pulse test method requires designing multiple currents for alternating discharges and has high requirements for experiments. Therefore, it is necessary to provide a simple and accurate method for estimating the parameters of a lead-acid battery pack.
[0030] The battery pack parameter estimation method, device, equipment, medium, and product provided by this application have high accuracy and are easy to implement. The following will be described in detail in conjunction with the accompanying drawings.
[0031] Figure 1 is one of the flow schematic diagrams of the battery pack parameter estimation method provided by this application. As Figure 1 shown, this application provides a battery pack parameter estimation method, including steps S100 - S300.
[0032] Step S100, collect the voltage data and current data of each single cell in the battery pack during the discharge process. The discharge process includes a voltage steep drop and recovery stage and a normal discharge stage. The voltage data includes first voltage data corresponding to the voltage steep drop and recovery stage and second voltage data corresponding to the normal discharge stage.
[0033] As an embodiment, collecting the voltage data and current data of each single battery in the battery pack during discharge includes: collecting the voltage data and current data of each single battery in the battery pack during discharge based on a battery discharger. The battery pack is obtained by connecting a number of single batteries in series. In this embodiment of the present application, 24 single batteries are taken as an example for illustration. The discharge mode of the battery pack is a constant voltage and constant current mode, and a battery discharger is used to record the discharge data.
[0034] The battery discharger can collect a large amount of discharge data of single batteries in real time, providing a data basis for data analysis and processing. The voltage data includes the voltage value and the time point corresponding to the voltage value. According to the voltage data, a voltage change curve can be plotted, as Figure 2 shown; the current data includes the current value and the time point corresponding to the current value. According to the current data, a current change curve can be plotted. The voltage change curve and the current change curve contain rich information related to the battery state, and features that can characterize the actual capacity of the battery can be extracted from them.
[0035] Step S200, preprocessing the first voltage data based on the preprocessing strategy corresponding to the voltage steep drop and recovery stage to obtain first voltage feature data, and preprocessing the second voltage data based on the preprocessing strategy corresponding to the normal discharge stage to obtain second voltage feature data.
[0036] As Figure 3 shown, during the constant current discharge process, the terminal voltage of the battery pack will rapidly drop in the early stage of discharge, then rise back to the platform voltage, and then enter the normal discharge stage. In this application, based on the maximum value of the voltage recovery as the demarcation basis, the discharge curve is divided into a voltage steep drop and recovery stage and a normal discharge stage. The voltage steep drop and recovery stage has complex non-linearity, and the voltage change amplitude and time are selected as feature parameters. In the normal discharge stage, polynomial fitting is used, and the function coefficients are used as feature parameters.
[0037] Step S300, inputting the current data, the first voltage feature data, and the second voltage feature data into a pre-trained battery parameter estimation model to obtain the estimated parameters of the battery pack output by the battery parameter estimation model. The estimated parameters of the battery pack include battery capacity, single cell voltage, and pack voltage.
[0038] Among them, the battery parameter estimation model is obtained by training a neural network model with current data samples, first voltage feature data samples, second voltage feature data samples, and battery pack estimated parameter samples, and optimizing the trained neural network model based on a genetic algorithm.
[0039] Artificial intelligence algorithms only need to reasonably select inputs and outputs, without the need for an exact formula to describe the relationship between the inputs and outputs. The relationship between the inputs and outputs can be determined during the training process of the neural network model. Therefore, in this application, the neural network model is trained based on the collected current data samples, first voltage feature data samples, second voltage feature data samples, and battery pack estimation parameter samples, enabling the neural network model to learn the mapping relationship between the input samples and the output samples. After training and optimization, it is used as a mapping model to estimate the battery parameters of the battery pack. Among them, there is a mapping relationship between the current data samples and the battery capacity samples, and there are mapping relationships between the first voltage feature data samples, second voltage feature data samples and the single-cell voltage samples, group voltage samples.
[0040] It can be understood that in this application, only the discharge data of the battery pack during the discharge process needs to be collected, and then the battery parameter estimation model trained by the artificial intelligence algorithm can be used for parameter estimation. There is no need for a complex discharge test design, and the implementation is convenient. The battery parameter estimation model has been optimized by the genetic algorithm, greatly improving the accuracy of parameter estimation.
[0041] As Figure 4 and Figure 5 shown, based on the above embodiments, as an optional embodiment, the preprocessing strategy corresponding to the voltage steep drop and recovery stage includes extracting features of the first voltage data in the voltage dimension and time dimension, and the preprocessing strategy corresponding to the normal discharge stage includes interpolation fitting of the second voltage data. Correspondingly, preprocessing the first voltage data based on the preprocessing strategy corresponding to the voltage steep drop and recovery stage to obtain first voltage feature data, and preprocessing the second voltage data based on the preprocessing strategy corresponding to the normal discharge stage to obtain second voltage feature data, including steps S210 - step S240.
[0042] Step S210, in the voltage dimension, extract the difference between the discharge starting voltage and the valley voltage, the difference between the peak voltage and the valley voltage, and the valley voltage from the first voltage data. It should be noted that the execution order of step S210 can be interchanged with steps S220 and S240.
[0043] In the voltage dimension, extract the difference between the discharge starting voltage and the valley voltage (Δ U 1), the difference between the peak voltage and the valley voltage (Δ U 2), and the valley voltage ( V b ) of three features.
[0044] Step S220: Extract, in the time dimension, the difference between the time points at which the discharge start voltage and the valley voltage are located, the difference between the time points at which the peak voltage and the valley voltage are located, and the time point at which the valley voltage is located from the first voltage data.
[0045] In the time dimension, extract the difference between the time points at which the discharge start voltage and the valley voltage are located (Δ T 1), the difference between the time points at which the peak voltage and the valley voltage are located (Δ T 2), and the time point at which the valley voltage is located ( T b ) as three features.
[0046] Step S230: Use the difference between the discharge start voltage and the valley voltage, the difference between the peak voltage and the valley voltage, the valley voltage, the difference between the time points at which the discharge start voltage and the valley voltage are located, the difference between the time points at which the peak voltage and the valley voltage are located, and the time point at which the valley voltage is located as the first voltage feature data.
[0047] Step S240: Based on the pre-constructed fitting function corresponding to the second voltage data, use the coefficients of the fitting function as the second voltage feature data.
[0048] Preprocess the normal discharge stage, use a polynomial interpolation function to fit the voltage change curve in the normal discharge stage, construct a fitting function corresponding to the second voltage data, and use the coefficients of the polynomial interpolation function as feature parameters, thereby realizing the extraction of key features from the voltage change curve of normal discharge. For example, use a fourth-order polynomial function for fitting, and use the five function coefficients of the fourth-order polynomial interpolation function as the second voltage feature data. The expression of the fourth-order polynomial interpolation function is as follows: ; where is the fourth-order polynomial interpolation function, x is the time parameter, - are the coefficients of the fourth-order polynomial interpolation function.
[0049] The extraction steps of the first voltage feature data sample and the second voltage feature data sample are the same as above and will not be elaborated here. The battery pack estimation parameter sample is obtained by actual measurement of the battery pack. Use the method of random sampling for the preprocessed data set to ensure the randomness of the data, and divide it into a training set according to a ratio of 70% - 80%, and the remaining 20% - 30% as the test set for model training and testing.
[0050] It can be understood that the present application performs feature extraction in stages and dimensions, which is beneficial to improving the accuracy of parameter estimation.
[0051] Such asFigure 6 As shown in the above embodiment, as an alternative embodiment, the neural network model includes an input layer, a hidden layer, and an output layer, and the genetic algorithm is used to optimize the weights between the hidden layer and the output layer and the center width vector of the Gaussian basis function.
[0052] In the embodiment of the present application, the neural network model is composed of a Radial Basis Function (RBF) neural network. The RBF neural network is a special three-layer feedforward neural network, including a simple topological structure of an input layer, a hidden layer, and an output layer, and uses the radial basis function as the activation function. The genetic algorithm (GA) optimizes the neural network model, and a battery parameter extraction model of the GA genetic algorithm - RBF neural network can be constructed. Taking the preprocessed features as the network input and the battery capacity, single-cell voltage, and group voltage as the output, the mapping of input and output is realized. The GA genetic algorithm optimizes the parameters of the RBF neural network, including the weights ω i between the hidden layer and the output layer and the center width vector σ i of the Gaussian basis function, so as to improve the efficiency and training accuracy of the neural network. The input of the GA - RBF model is the voltage drop and recovery and the characteristics of the normal discharge stage of each single battery, and the output is the battery capacity, single-cell voltage, and group voltage.
[0053] In the RBF neural network, the input layer vector is represented by , where m represents the number of input nodes, and the output layer vector is represented by . The hidden layer uses a Gaussian kernel. The RBF neural network has high requirements for parameter optimization. In order to achieve the global approximation of the RBF neural network, the genetic algorithm is used to optimize it. The genetic algorithm performs processes such as crossover, selection, and mutation in the form of a population, and obtains the individual with the optimal fitness as the optimal solution of the target value after multiple generations of genetic evolution, and has strong search ability in the global range.
[0054] As Figure 7 shown, as an alternative embodiment, optimizing the trained neural network model based on the genetic algorithm includes the following steps.
[0055] According to the initial Gaussian basis function width vector and weight parameters of each neuron in the hidden layer of the neural network model, the chromosome of the genetic algorithm is obtained. Regarding the initial Gaussian basis function width vector and weight parameters of the RBF neural network as chromosomes, the chromosomes are randomly converted into binary hybrid codes using gray binary coding, and there is only one-bit change in the coding conversion when performing crossover and mutation operations. Assuming that there are P neurons in the hidden layer, that is, the scale of the initial population, the chromosomes of each generation of the new population are encoded as .
[0056] Determine the fitness function of the genetic factor of the genetic algorithm according to the error of the neural network model.
[0057] The error expression output by the RBF neural network is as follows: where, is the actual output, is the predicted output of the RBF neural network, n is the number of output layers of the RBF neural network, represents the error of the RBF neural network.
[0058] In the embodiment of the present application, the fitness function of the genetic factor is defined as the exponent of the product of a constant and the error , when the error gradually decreases, the fitness gradually converges to a minimum value, and the obtained chromosome is more excellent.
[0059] Perform genetic operation optimization on the population composed of chromosomes according to the fitness function until the preset fitness or the preset number of iterations is reached, and obtain the optimal solution of the genetic algorithm.
[0060] Select the best from the population, and the probability of randomly selecting genetic individuals is as follows: represents the i fitness function value of the individual.
[0061] Perform crossover and mutation on the parental and filial generation individuals to obtain a new generation of offspring, where the crossover probability and mutation probability will change continuously with the fitness of the offspring population.
[0062] Optimize the weights between the hidden layer and the output layer of the neural network model and the center width vector of the Gaussian basis function according to the optimal solution of the genetic algorithm.
[0063] It can be understood that in the present application, the weights between the hidden layer and the output layer and the center width vector of the Gaussian basis function are optimized by the genetic algorithm, so that the accuracy of the neural network model is higher, and thus the accuracy of parameter estimation is improved.
[0064] As Figure 8 shown, on the basis of the above embodiment, as an optional embodiment, the method for estimating the parameters of the battery pack provided by the present application further includes the following steps.
[0065] During the training process, evaluate the accuracy of the neural network model based on the mean absolute error and / or the mean absolute percentage error, and adjust the neural network model according to the evaluation result.
[0066] In this application, the evaluation indicators mean absolute error (MAE) and mean absolute percentage error (MAPE) are used to evaluate the accuracy of battery parameter extraction. MAE can avoid the problem of mutual cancellation when calculating errors and can accurately reflect the magnitude of the actual prediction error; MAPE is commonly used to measure the accuracy of predictions. The closer it is to 0, the smaller the prediction error. The calculation formulas for MAE and MAPE are as follows: Among them, is the mean absolute error value of the neural network model, is the mean absolute percentage error value of the neural network model.
[0067] It can be understood that this application evaluates the accuracy of the neural network model through the mean absolute error and / or the mean absolute percentage error, which can be further optimized during the training process of the neural network model or during the application process of the battery parameter estimation model constituted by it, so as to improve the accuracy of parameter estimation.
[0068] The battery pack parameter estimation method and device provided by this application will be described below. The battery pack parameter estimation method and device described below can be mutually referred to the battery pack parameter estimation method described above.
[0069] Figure 9 is a schematic structural diagram of the battery pack parameter estimation method and device provided by this application. As Figure 9 shown, this application also provides a battery pack parameter estimation method and device, including the following modules: A data acquisition module 910, configured to acquire voltage data and current data of each single battery in the battery pack during the discharge process. The discharge process includes a voltage steep drop and recovery stage and a normal discharge stage. The voltage data includes first voltage data corresponding to the voltage steep drop and recovery stage and second voltage data corresponding to the normal discharge stage; A preprocessing module 920, configured to preprocess the first voltage data based on the preprocessing strategy corresponding to the voltage steep drop and recovery stage to obtain first voltage characteristic data, and preprocess the second voltage data based on the preprocessing strategy corresponding to the normal discharge stage to obtain second voltage characteristic data; A parameter estimation module 930, configured to input the current data, the first voltage characteristic data, and the second voltage characteristic data into a pre-trained battery parameter estimation model to obtain the battery pack estimation parameters output by the battery parameter estimation model. The battery pack estimation parameters include battery capacity, single cell voltage, and pack voltage; Among them, the battery parameter estimation model is obtained by training a neural network model with current data samples, first voltage feature data samples, and second voltage feature data samples, and then optimizing the trained neural network model based on a genetic algorithm.
[0070] As an embodiment, the preprocessing strategy corresponding to the voltage dip and recovery stage includes extracting features from the first voltage data in the voltage dimension and the time dimension, and the preprocessing strategy corresponding to the normal discharge stage includes interpolation fitting of the second voltage data. Correspondingly, the preprocessing module 920 is further configured to: In the voltage dimension, extract the difference between the discharge start voltage and the valley voltage, the difference between the peak voltage and the valley voltage, and the valley voltage from the first voltage data; In the time dimension, extract the difference in time points between the discharge start voltage and the valley voltage, the difference in time points between the peak voltage and the valley voltage, and the time point at which the valley voltage occurs from the first voltage data; Take the difference between the discharge start voltage and the valley voltage, the difference between the peak voltage and the valley voltage, the valley voltage, the difference in time points between the discharge start voltage and the valley voltage, the difference in time points between the peak voltage and the valley voltage, and the time point at which the valley voltage occurs as the first voltage feature data; Based on a pre-constructed fitting function corresponding to the second voltage data, take the coefficients of the fitting function as the second voltage feature data.
[0071] As an embodiment, the neural network model includes an input layer, a hidden layer, and an output layer, and the genetic algorithm is used to optimize the weights between the hidden layer and the output layer and the center width vector of the Gaussian basis function.
[0072] As an embodiment, optimizing the trained neural network model based on a genetic algorithm includes: According to the initial Gaussian basis function width vector and weight parameters of each neuron in the hidden layer of the neural network model, obtain the chromosome of the genetic algorithm; According to the error of the neural network model, determine the fitness function of the genetic factor of the genetic algorithm; Perform genetic operation optimization on the population composed of chromosomes according to the fitness function until the preset fitness or the preset number of iterations is reached, and obtain the optimal solution of the genetic algorithm; Optimize the weights between the hidden layer and the output layer of the neural network model and the center width vector of the Gaussian basis function according to the optimal solution of the genetic algorithm.
[0073] As an embodiment, it further includes: An error comparison module is used to evaluate the accuracy of the neural network model based on the mean absolute error and / or the mean absolute percentage error during the training process, and adjust the neural network model according to the evaluation result.
[0074] As an embodiment, the data acquisition module is further configured to: Collect voltage data and current data of each single battery in the battery pack during the discharge process based on a battery discharger.
[0075] The battery pack parameter estimation method and device provided in this application are used to implement the battery pack parameter estimation method provided in any of the above embodiments, and have corresponding technical effects to the battery pack parameter estimation method, which will not be elaborated here.
[0076] Figure 10 An example of the physical structure diagram of an electronic device is shown as Figure 10 shown. The electronic device may include: a processor 1010, a communication interface 1020, a memory 1030, and a communication bus 1040. Among them, the processor 1010, the communication interface 1020, and the memory 1030 communicate with each other through the communication bus 1040. The processor 1010 can call the logical instructions in the memory 1030 to execute the battery pack parameter estimation method, which includes: collecting voltage data and current data of each single battery in the battery pack during the discharge process. The discharge process includes a voltage steep drop and recovery stage and a normal discharge stage. The voltage data includes first voltage data corresponding to the voltage steep drop and recovery stage and second voltage data corresponding to the normal discharge stage; preprocessing the first voltage data based on the preprocessing strategy corresponding to the voltage steep drop and recovery stage to obtain first voltage feature data, and preprocessing the second voltage data based on the preprocessing strategy corresponding to the normal discharge stage to obtain second voltage feature data; inputting the current data, the first voltage feature data, and the second voltage feature data into a pre-trained battery parameter estimation model to obtain the battery pack estimation parameters output by the battery parameter estimation model. The battery pack estimation parameters include battery capacity, single cell voltage, and group voltage; wherein, the battery parameter estimation model is trained by using current data samples, first voltage feature data samples, second voltage feature data samples, and battery pack estimation parameter samples for a neural network model, and is obtained after optimizing the trained neural network model based on a genetic algorithm.
[0077] In addition, when the logical instructions in the above-mentioned memory 1030 are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present application, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present application. The aforementioned storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memories (ROMs), random access memories (RAMs), magnetic disks, or optical discs that can store program codes.
[0078] On the other hand, the present application also provides a computer program product. The computer program product includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the battery pack parameter estimation method provided by the above-mentioned various methods. The method includes: collecting voltage data and current data of each single battery in the battery pack during the discharge process. The discharge process includes a voltage steep drop and recovery stage and a normal discharge stage. The voltage data includes first voltage data corresponding to the voltage steep drop and recovery stage and second voltage data corresponding to the normal discharge stage; preprocessing the first voltage data based on the preprocessing strategy corresponding to the voltage steep drop and recovery stage to obtain first voltage characteristic data, and preprocessing the second voltage data based on the preprocessing strategy corresponding to the normal discharge stage to obtain second voltage characteristic data; inputting the current data, the first voltage characteristic data, and the second voltage characteristic data into a pre-trained battery parameter estimation model to obtain the battery pack estimation parameters output by the battery parameter estimation model. The battery pack estimation parameters include battery capacity, single cell voltage, and pack voltage; wherein, the battery parameter estimation model is trained by a neural network model with current data samples, first voltage characteristic data samples, second voltage characteristic data samples, and battery pack estimation parameter samples, and is obtained by optimizing the trained neural network model based on a genetic algorithm.
[0079] In another aspect, the present application also provides a non-transitory computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it implements the battery pack parameter estimation method provided by the above-mentioned various methods. The method includes: collecting voltage data and current data of each single battery in the battery pack during the discharge process, where the discharge process includes a voltage steep drop and recovery stage and a normal discharge stage, and the voltage data includes first voltage data corresponding to the voltage steep drop and recovery stage and second voltage data corresponding to the normal discharge stage; preprocessing the first voltage data based on the preprocessing strategy corresponding to the voltage steep drop and recovery stage to obtain first voltage characteristic data, and preprocessing the second voltage data based on the preprocessing strategy corresponding to the normal discharge stage to obtain second voltage characteristic data; inputting the current data, the first voltage characteristic data, and the second voltage characteristic data into a pre-trained battery parameter estimation model to obtain the battery pack estimation parameters output by the battery parameter estimation model, where the battery pack estimation parameters include battery capacity, single cell voltage, and pack voltage; among them, the battery parameter estimation model is trained by a neural network model with current data samples, first voltage characteristic data samples, second voltage characteristic data samples, and battery pack estimation parameter samples, and is obtained by optimizing the trained neural network model based on a genetic algorithm.
[0080] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. A person of ordinary skill in the art can understand and implement it without creative effort.
[0081] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on such an understanding, the above technical solution, in essence, or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., including several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.
[0082] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the various embodiments of the present application.
Claims
1. A method for estimating battery pack parameters, characterized in that, Including: Collecting voltage data and current data of each single battery in the battery pack during the discharge process, the discharge process including a voltage steep drop and recovery stage and a normal discharge stage, the voltage data including first voltage data corresponding to the voltage steep drop and recovery stage and second voltage data corresponding to the normal discharge stage; Preprocessing the first voltage data based on the preprocessing strategy corresponding to the voltage steep drop and recovery stage to obtain first voltage feature data, and preprocessing the second voltage data based on the preprocessing strategy corresponding to the normal discharge stage to obtain second voltage feature data; Inputting the current data, the first voltage feature data, and the second voltage feature data into a pre-trained battery parameter estimation model to obtain the estimated parameters of the battery pack output by the battery parameter estimation model, the estimated parameters of the battery pack including battery capacity, single cell voltage, and pack voltage; Wherein, the battery parameter estimation model is obtained by training a neural network model with current data samples, first voltage feature data samples, second voltage feature data samples, and battery pack estimated parameter samples, and optimizing the trained neural network model based on a genetic algorithm.
2. The method for estimating the parameters of the battery pack according to claim 1, wherein The preprocessing strategy corresponding to the voltage steep drop and recovery stage includes extracting features from the first voltage data in the voltage dimension and the time dimension, and the preprocessing strategy corresponding to the normal discharge stage includes interpolation fitting of the second voltage data. Correspondingly, preprocessing the first voltage data based on the preprocessing strategy corresponding to the voltage steep drop and recovery stage to obtain first voltage feature data, and preprocessing the second voltage data based on the preprocessing strategy corresponding to the normal discharge stage to obtain second voltage feature data, including: In the voltage dimension, extracting the difference between the discharge starting voltage and the valley voltage, the difference between the peak voltage and the valley voltage, and the valley voltage from the first voltage data; In the time dimension, extracting the difference in time points between the discharge starting voltage and the valley voltage, the difference in time points between the peak voltage and the valley voltage, and the time point at which the valley voltage is located from the first voltage data; Taking the difference between the discharge starting voltage and the valley voltage, the difference between the peak voltage and the valley voltage, the valley voltage, the difference in time points between the discharge starting voltage and the valley voltage, the difference in time points between the peak voltage and the valley voltage, and the time point at which the valley voltage is located as the first voltage feature data; Based on a pre-constructed fitting function corresponding to the second voltage data, taking the coefficients of the fitting function as the second voltage feature data.
3. The method for estimating battery pack parameters according to claim 1, wherein The neural network model includes an input layer, a hidden layer, and an output layer, and the genetic algorithm is used to optimize the weights and the center width vector of the Gaussian basis function between the hidden layer and the output layer.
4. The method for estimating battery pack parameters according to claim 3, characterized in that, Optimizing the trained neural network model based on a genetic algorithm includes: Obtaining the chromosome of the genetic algorithm according to the initial Gaussian basis function width vector and weight parameters of each neuron in the hidden layer of the neural network model; Determining the fitness function of the genetic factor of the genetic algorithm according to the error of the neural network model; Genetic operations optimization is performed on the population composed of chromosomes according to the fitness function until a preset fitness or a preset number of iterations is reached, and the optimal solution of the genetic algorithm is obtained; Based on the optimal solution of the genetic algorithm, the weights between the hidden layer and the output layer of the neural network model and the center width vector of the Gaussian basis function are optimized.
5. The method for estimating battery pack parameters according to any one of claims 1-4, characterized in that, It also includes: During the training process, the accuracy of the neural network model is evaluated based on the mean absolute error and / or the mean absolute percentage error, and the neural network model is adjusted according to the evaluation results.
6. The method for estimating the parameters of the battery pack according to any one of claims 1-4, characterized in that The acquisition of the voltage data and current data of each single battery in the battery pack during discharge includes: Based on a battery discharger, the voltage data and current data of each single battery in the battery pack during discharge are acquired.
7. A method and device for estimating battery pack parameters, characterized in that, It includes: A data acquisition module, configured to acquire the voltage data and current data of each single battery in the battery pack during discharge. The discharge process includes a voltage steep drop and recovery stage and a normal discharge stage. The voltage data includes first voltage data corresponding to the voltage steep drop and recovery stage and second voltage data corresponding to the normal discharge stage; A preprocessing module, configured to preprocess the first voltage data based on the preprocessing strategy corresponding to the voltage steep drop and recovery stage to obtain first voltage feature data, and preprocess the second voltage data based on the preprocessing strategy corresponding to the normal discharge stage to obtain second voltage feature data; A parameter estimation module, configured to input the current data, the first voltage feature data, and the second voltage feature data into a pre-trained battery parameter estimation model to obtain the estimated parameters of the battery pack output by the battery parameter estimation model. The estimated parameters of the battery pack include battery capacity, single cell voltage, and pack voltage; Wherein, the battery parameter estimation model is obtained by training a neural network model with current data samples, first voltage feature data samples, and second voltage feature data samples and optimizing the trained neural network model based on a genetic algorithm.
8. An electronic device, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the battery pack parameter estimation method according to any one of claims 1 to 6.
9. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the battery pack parameter estimation method according to any one of claims 1 to 6.
10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the battery pack parameter estimation method according to any one of claims 1 to 6.