Lithium battery internal resistance real-time estimation method and system based on adaptive wavelet denoising

Abstract

The application provides a lithium battery internal resistance real-time estimation method and system based on adaptive wavelet denoising, relates to the technical field of battery management systems, and solves the technical problems of low estimation accuracy, poor real-time performance and high computing power load caused by the use of fixed frequency sampling and fixed parameter denoising in the prior art. The method comprises the following steps: obtaining a charging current signal, determining whether the charging is in a current climbing stage based on the charging current signal; if the charging is in the current climbing stage, collecting voltage signals and current signals based on a preset sampling frequency; performing adaptive wavelet denoising on the voltage signals and the current signals to obtain denoised signals; performing linear fitting on the denoised signals to obtain an initial value of the internal resistance; inputting the initial value of the internal resistance into a first-order RC equivalent circuit model, and performing dynamic estimation by using a recursive least square method to obtain an internal resistance estimation value. The application is used in the lithium battery internal resistance estimation process.

Description

Technical Field

[0001] This application relates to the field of battery management system technology, and in particular to a method and system for real-time estimation of the internal resistance of lithium batteries based on adaptive wavelet denoising. Background Technology

[0002] As a core component of electric vehicles and energy storage systems, the accurate and real-time estimation of the internal resistance of lithium batteries is crucial for the optimized control, lifespan extension, and safety assurance of battery management systems (BMS). Existing methods for estimating internal resistance mainly include experimental measurement methods, model-based methods, and data-driven methods. Experimental measurement methods, such as hybrid pulse power characteristic testing and electrochemical impedance spectroscopy, can provide benchmark data, but they rely on offline operation and specific equipment, which cannot meet the real-time monitoring requirements under dynamic operating conditions. Model-based methods, such as extended Kalman filtering and traditional least squares methods, have a certain degree of real-time performance, but are limited by the fixed structure of the equivalent circuit model and the fixed settings of the filter parameters. When faced with dynamic operating conditions where multiple factors such as temperature and charge / discharge rate are coupled, model mismatch is prone to occur, leading to increased estimation errors. Data-driven methods can handle nonlinear mappings, but they have high requirements for the quantity and quality of sample data, and the computational delay caused by complex network structures makes it difficult to guarantee real-time response. In signal processing, traditional wavelet denoising methods, due to their fixed wavelet basis functions, decomposition levels, and threshold strategies, cannot adapt to the transient signal characteristics of lithium battery charging current ramp-up phase, which are characterized by "strong abrupt changes, noisy and highly sensitive" signals. During high-rate charging, insufficient decomposition can lead to residual noise, while in low-temperature environments, excessively high thresholds may filter out useful ohmic signals, resulting in insufficient accuracy in extracting the initial internal resistance value and consequently causing accumulated errors in subsequent dynamic estimation. Furthermore, the full-cycle high-frequency sampling mode leads to computational redundancy, further restricting the deployment feasibility of embedded platforms. Summary of the Invention

[0003] This application provides a real-time estimation method and system for lithium battery internal resistance based on adaptive wavelet denoising, which solves the technical problems of low estimation accuracy, poor real-time performance and high computing load caused by the use of fixed frequency sampling and fixed parameter denoising throughout the process in the prior art.

[0004] To achieve the above objectives, this application adopts the following technical solution: Firstly, a real-time estimation method for the internal resistance of lithium batteries based on adaptive wavelet denoising is provided, including: Acquire the charging current signal and determine whether the charging is in the current ramp-up phase based on the charging current signal; If the charging process is in the current ramp-up phase, voltage and current signals are collected based on a preset sampling frequency. Adaptive wavelet denoising is performed on the voltage and current signals to obtain the denoised signal. Linear fitting is performed on the denoised signal to obtain the initial value of the internal resistance; The initial value of the internal resistance is input into the first-order RC equivalent circuit model, and the internal resistance of the battery is estimated by dynamic estimation through recursive least squares method.

[0005] Based on the above technical solutions, the real-time estimation method for lithium battery internal resistance based on adaptive wavelet denoising provided in this application achieves real-time and high-precision calculation of lithium battery internal resistance by fusing adaptive wavelet denoising and recursive least squares estimation. First, local high-frequency acquisition is triggered by dynamic identification during the charging current ramp-up phase, acquiring voltage and current data within the key signal window. This avoids redundant calculations caused by full-cycle sampling, significantly improving the efficiency of data acquisition and processing. Second, an adaptive wavelet denoising strategy is used to specifically process transient signals, removing noise while retaining effective ohmic response characteristics, thus obtaining a more accurate initial value of internal resistance. Finally, by combining a first-order RC equivalent circuit model and a recursive least squares algorithm, dynamic parameter tracking is achieved starting from the initial internal resistance value. This ensures that the method maintains good real-time performance and robustness under complex operating conditions, providing a reliable basis for state assessment and control decisions of the battery management system.

[0006] In conjunction with the first aspect above, in one possible implementation, determining whether charging is in the current ramp-up phase based on the charging current signal includes: Based on charging current signal The rate of change of current was calculated. ;in, t For time; rate of change of current and charging current signal Compared with the preset rate of change threshold and target charging current Compare; when > and < If so, it is determined that the charging is in the current ramp-up phase.

[0007] In conjunction with the first aspect above, in one possible implementation, the adaptive wavelet denoising of the voltage and current signals includes: Wavelet transform is used to decompose voltage and current signals to obtain low-frequency approximate components and high-frequency detail components. High-frequency detail components are threshold-filtered based on a threshold strategy to obtain a high-frequency filtered signal; The denoised signal is obtained by performing inverse wavelet transform on the high-frequency filtered signal and the low-frequency approximate component.

[0008] In conjunction with the first aspect above, in one possible implementation, the wavelet decomposition process includes adaptive selection of wavelet basis functions and dynamic determination of the number of decomposition levels. The adaptive selection of the wavelet basis function involves constructing a wavelet basis evaluation index; wherein the expression for the wavelet basis evaluation index is: In the formula, The signal-to-noise ratio after signal reconstruction. The mean absolute error at the signal abrupt change point. To balance the weighting coefficients; calculate the evaluation scores of multiple wavelet basis functions in the preset wavelet basis library based on the wavelet basis evaluation index; select the wavelet basis function corresponding to the maximum evaluation score; The dynamic determination of the number of decomposition layers: the rate of change of current. The number of decomposition levels is obtained by comparing the result with a preset multi-level decomposition threshold.

[0009] In conjunction with the first aspect above, in one possible implementation, the threshold filtering of high-frequency detail components based on a threshold strategy includes: Through formula Peak factor was calculated ;in, These are the wavelet decomposition coefficients corresponding to the high-frequency detail components. N is the length of the wavelet decomposition coefficients; Peak factor With the preset peak threshold Compare; when > At that time, through the formula Calculate the hard threshold The hard threshold formula is obtained; the wavelet decomposition coefficients are filtered using the hard threshold formula to obtain the high-frequency decomposition coefficients; where, The standard deviation of noise. The expression for the hard threshold formula is: ; when ≤ At that time, through the formula Calculate soft threshold The soft thresholding formula is obtained; the wavelet decomposition coefficients are filtered using the soft thresholding formula to obtain the high-frequency decomposition coefficients; wherein, the expression of the soft thresholding formula is: ; The high-frequency filtered signal is obtained by filtering the high-frequency detail components based on the high-frequency decomposition coefficients.

[0010] In conjunction with the first aspect described above, one possible implementation further includes dynamically adjusting the threshold strategy; wherein, the dynamic adjustment of the threshold strategy includes: Obtain the charging temperature T; A temperature correction coefficient is constructed based on the charging temperature T. ; Based on temperature correction factor The soft and hard thresholds are adjusted to obtain the adjusted threshold strategy; wherein, the soft and hard thresholds include the soft threshold. and hard threshold .

[0011] In conjunction with the first aspect above, in one possible implementation, the method for obtaining the battery internal resistance estimate includes: Based on a first-order RC equivalent circuit model, a state vector and a covariance matrix are constructed; wherein, the expression for the state vector is: In the formula, k For discrete time series indexes, K is an integer. The internal resistance of the battery to be estimated is... The open-circuit voltage to be estimated; The state vector is initialized based on the initial value of the internal resistance; The estimated internal resistance of the battery is obtained by iteratively updating the state vector and covariance matrix using the recursive least squares method.

[0012] In conjunction with the first aspect above, in one possible implementation, the iterative update of the state vector and covariance matrix using the recursive least squares method includes: Get k Battery terminal voltage at time and charging / discharging current And construct the input vector ; Based on input vector and state vector The residuals were calculated. ; Based on input vector Covariance Matrix Calculate the gain ;in, This is a forgetting factor used to dynamically balance the weights of old and new data. Based on residuals and gain For the state vector Covariance Matrix The battery internal resistance is estimated by iterative correction.

[0013] In conjunction with the first aspect above, in one possible implementation, the state vector... Covariance Matrix The correction formula for iterative correction is: ; in, I Let Q be the identity matrix of order Q, where Q is a positive integer.

[0014] Secondly, a real-time lithium battery internal resistance estimation system based on adaptive wavelet denoising is provided, including a signal acquisition and judgment module, a local high-frequency acquisition module, an adaptive wavelet denoising module, an internal resistance initial value calculation module, and a dynamic internal resistance estimation module. The signal acquisition and determination module is used to acquire the charging current signal and determine whether the charging is in the current ramp-up stage based on the charging current signal. The local high-frequency acquisition module is used to acquire voltage and current signals based on a preset sampling frequency when the charging is in the current ramp-up phase. The adaptive wavelet denoising module is used to perform adaptive wavelet denoising on voltage and current signals to obtain denoised signals. The internal resistance initial value calculation module is used to perform linear fitting on the denoised signal to obtain the internal resistance initial value. The dynamic internal resistance estimation module is used to input the initial value of the internal resistance into the first-order RC equivalent circuit model, and perform dynamic estimation by recursive least squares method to obtain the estimated value of the battery internal resistance.

[0015] This application provides a real-time estimation method and system for lithium battery internal resistance based on adaptive wavelet denoising. By designing an adaptive wavelet denoising framework and integrating recursive least squares (RLS), it effectively solves the technical bottleneck of traditional internal resistance estimation methods. In the signal preprocessing stage, adaptive wavelet denoising accurately matches the signal characteristics during the charging current ramp-up stage by dynamically selecting wavelet basis functions, adjusting the number of decomposition layers, and employing layer threshold strategies. This achieves efficient noise removal while ensuring lossless preservation of useful transient responses, thus improving the estimation accuracy of the initial internal resistance value from the source. Combined with a local high-frequency acquisition strategy, data acquisition is initiated only during the window period when the transient response is significant, significantly reducing computational power consumption. In the dynamic estimation stage, the RLS algorithm is based on a first-order RC equivalent circuit model and uses high-precision initial values ​​for iterative optimization. By balancing the weights of new and old data through a forgetting factor mechanism, it achieves real-time tracking and adaptive updating of internal resistance parameters under all operating conditions, effectively handling complex scenarios such as temperature fluctuations and charge / discharge switching. The overall approach balances estimation accuracy and computational efficiency. Its modular design enhances its engineering practicality and can be directly integrated into the BMS embedded platform, providing reliable parameter support for battery health status assessment. At the same time, it verifies the flexible adaptability value of wavelet analysis theory in nonlinear transient signal processing.

[0016] It should be understood that the descriptions of technical features, technical solutions, beneficial effects, or similar language in this application do not imply that all features and advantages can be achieved in any single embodiment. Rather, it is understood that the description of a feature or beneficial effect means that a specific technical feature, technical solution, or beneficial effect is included in at least one embodiment. Therefore, the descriptions of technical features, technical solutions, or beneficial effects in this specification do not necessarily refer to the same embodiment. Furthermore, the technical features, technical solutions, and beneficial effects described in this embodiment can be combined in any suitable manner. Those skilled in the art will understand that embodiments can be implemented without one or more specific technical features, technical solutions, or beneficial effects of a particular embodiment. In other embodiments, additional technical features and beneficial effects may be identified in specific embodiments that do not embody all embodiments. Attached Figure Description

[0017] Figure 1 A flowchart illustrating a real-time estimation method for the internal resistance of a lithium battery based on adaptive wavelet denoising, provided for an embodiment of this application; Figure 2 A schematic diagram of high-frequency acquisition of the internal resistance voltage signal of each cell in the battery pack provided in the embodiments of this application; Figure 3 A schematic diagram showing the comparison between the original voltage signal and the denoised voltage signal provided in the embodiments of this application; Figure 4 A schematic diagram of a first-order RC equivalent circuit model provided for an embodiment of this application; Figure 5 This is a schematic diagram of the overall method flow provided for the embodiments of this application; Figure 6 A schematic diagram illustrating the estimated internal resistance of each cell in the battery pack provided in this embodiment of the application; Figure 7 This is a system architecture diagram of a real-time lithium battery internal resistance estimation system based on adaptive wavelet denoising, provided for an embodiment of this application. Detailed Implementation

[0018] In the description of this application, unless otherwise stated, " / " means "or," for example, A / B can mean A or B. The "and / or" in this document is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, and B alone. Furthermore, "at least one" means one or more, and "multiple" means two or more. The terms "first," "second," etc., do not limit the quantity or order of execution, and "first," "second," etc., do not necessarily imply differences.

[0019] It should be noted that, in this application, the terms "exemplary" or "for example" are used to indicate that something is being described as an example, illustration, or illustration. Any embodiment or design described as "exemplary" or "for example" in this application should not be construed as being more preferred or advantageous than other embodiments or design solutions. Specifically, the use of terms such as "exemplary" or "for example" is intended to present the relevant concepts in a concrete manner.

[0020] To address the technical problems of low estimation accuracy, poor real-time performance, and high computational load caused by the use of fixed-frequency sampling and fixed-parameter denoising throughout the process in existing technologies, this application provides a real-time estimation method for lithium battery internal resistance based on adaptive wavelet denoising. This method includes: Acquire the charging current signal and determine whether the charging is in the current ramp-up phase based on the charging current signal; If the charging process is in the current ramp-up phase, voltage and current signals are collected based on a preset sampling frequency. Adaptive wavelet denoising is performed on the voltage and current signals to obtain the denoised signal. Linear fitting is performed on the denoised signal to obtain the initial value of the internal resistance; The initial value of the internal resistance is input into the first-order RC equivalent circuit model, and the internal resistance of the battery is estimated by dynamic estimation through recursive least squares method.

[0021] Based on this, the technical problems of low estimation accuracy, poor real-time performance and high computing load caused by the use of fixed frequency sampling and fixed parameter denoising throughout the process in the existing technology are solved.

[0022] like Figure 1 As shown in the embodiments of this application, the real-time estimation method for lithium battery internal resistance based on adaptive wavelet denoising includes: S201. Obtain the charging current signal and determine whether the charging is in the current ramp-up stage based on the charging current signal.

[0023] In some implementations, determining whether charging is in the current ramp-up phase based on the charging current signal includes: Based on charging current signal The rate of change of current was calculated. ;in, t For time; rate of change of current and charging current signal Compared with the preset rate of change threshold and target charging current Compare; when > and < If so, it is determined that the charging is in the current ramp-up phase.

[0024] It should be noted that when and ≤ At that time, the climb phase is considered complete; The charging rate is adjusted in real time to adapt to different scenarios.

[0025] For example, acquiring the charging current signal It calculates the first-order difference as the rate of change of current in real time; and sets a basic rate of change threshold. It is 0.1, and based on the real-time charging rate. (C) e Dynamically adjust (for the battery's rated capacity): When When =2C, the threshold is increased to 0.2. When C = 0.5, the threshold is lowered to 0.05; when both conditions are met... Greater than the adjusted threshold and Less than the target charging current When the system is in the current ramp-up phase, it is determined that the current has entered the ramp-up phase. Then, high-frequency synchronous acquisition is started once per second and continues until the current stabilizes and the rate of change falls below the threshold.

[0026] S202. If the charging is in the current ramp-up stage, the voltage signal and current signal are collected based on the preset sampling frequency.

[0027] For example, once the system determines that it has entered the charging current ramp-up phase, it immediately initiates synchronous voltage and current acquisition at a sampling frequency of once per second, continuously covering the entire ramp-up process for approximately 15 to 30 seconds; a schematic diagram of high-frequency acquisition of the internal resistance voltage signals of each cell in the battery pack is shown below. Figure 2 As shown; the real-time charging rate is calculated synchronously during the data acquisition process. Its value is the current charging current. With battery rated capacity C e The system records the current cycle number to provide an aging state reference. By implementing this local high-frequency acquisition mechanism, the system performs intensive sampling only within the time window when the signal changes most significantly. While fully capturing the transient response of the ohmic internal resistance during the current change process, it significantly reduces the amount of redundant data brought by full-cycle sampling, effectively reducing the real-time computing load and storage pressure of the embedded battery management system.

[0028] In this embodiment, high-frequency acquisition is initiated only during the significant rise phase of transient response, avoiding the waste of computing power caused by full-cycle high-frequency sampling, balancing signal acquisition accuracy and computing efficiency, and improving the real-time performance and energy efficiency of the system in embedded battery management scenarios.

[0029] S203. Perform adaptive wavelet denoising on the voltage and current signals to obtain the denoised signal.

[0030] In some implementations, the adaptive wavelet denoising of the voltage and current signals includes: Wavelet transform is used to decompose voltage and current signals to obtain low-frequency approximate components and high-frequency detail components. High-frequency detail components are threshold-filtered based on a threshold strategy to obtain a high-frequency filtered signal; The denoised signal is obtained by performing inverse wavelet transform on the high-frequency filtered signal and the low-frequency approximate component.

[0031] For example, the signal is first decomposed using a dynamically selected optimal wavelet basis function to obtain low-frequency approximate components and high-frequency detail components at each level. Then, the noise level is estimated using the noise standard deviation formula, and the noise type is distinguished based on the peak factor. High-frequency spike noise is filtered out using a hard threshold calculated based on the noise standard deviation and coefficient length, while low-frequency drift noise is processed using a corresponding soft threshold. The threshold is dynamically corrected within a range of ±10% based on real-time temperature. The processed high-frequency coefficients at each level and the retained low-frequency approximate components are reconstructed using inverse wavelet transform to finally obtain the denoised signal. A comparison diagram of the original voltage signal and the denoised voltage signal is shown below. Figure 3 As shown.

[0032] In this embodiment, threshold classification driven by signal features enables precise removal of sensor interference, polarization residues, and environmental noise, effectively preserving the key ohmic response characteristics during the current ramp-up phase and providing a high-fidelity data foundation for subsequent internal resistance calculation.

[0033] In some implementations, the wavelet decomposition process includes adaptive selection of wavelet basis functions and dynamic determination of the number of decomposition levels; The adaptive selection of the wavelet basis function involves constructing a wavelet basis evaluation index; wherein the expression for the wavelet basis evaluation index is: In the formula, The signal-to-noise ratio after signal reconstruction. The mean absolute error at the signal abrupt change point. To balance the weighting coefficients; calculate the evaluation scores of multiple wavelet basis functions in the preset wavelet basis library based on the wavelet basis evaluation index; select the wavelet basis function corresponding to the maximum evaluation score; The dynamic determination of the number of decomposition layers: the rate of change of current. The number of decomposition levels is obtained by comparing the result with a preset multi-level decomposition threshold.

[0034] It should be noted that, Used to measure the purity of the effective components in the denoised signal compared to the original signal; Used to quantize the deviation between the denoised signal and the original signal at abrupt changes (such as the start / end of a current ramp-up); balancing weighting coefficients Adjustments are made based on the real-time collected charging rate and charging temperature T to regulate... and The relative importance in the comprehensive evaluation, with balanced weighting coefficients calibrated experimentally. The standard value is 0.6. When the charging rate is high, the balance weight coefficient is reduced. When the charging temperature T is low, increase the balance weight coefficient. .

[0035] For example, wavelet base libraries include db4, db5, sym5, and coif3; the rate of change of current is compared with a preset multi-level decomposition threshold, and different decomposition levels are set for three scenarios; a) strong abrupt change scenario ( >1): Employing a 4-layer decomposition method to accurately capture high-frequency detail signals; b) Moderate abrupt change scenarios (1≥ >0.5): A 3-layer decomposition is used to balance noise suppression and detail preservation; c) Smooth scenes (0.5≥) ): A two-level decomposition is adopted to avoid redundant calculations and the introduction of noise.

[0036] In the embodiments of this application, the dynamic selection mechanism ensures the accurate matching between the wavelet decomposition structure and the real-time signal features, thereby effectively improving the denoising process's ability to preserve the fidelity of transient useful signals and its targeted suppression of complex noise.

[0037] In some implementations, the threshold filtering of high-frequency detail components based on a threshold strategy includes: Through formula Peak factor was calculated ;in, These are the wavelet decomposition coefficients corresponding to the high-frequency detail components. N is the length of the wavelet decomposition coefficients; Peak factor With the preset peak threshold Compare; when > At that time, through the formula Calculate the hard threshold The hard threshold formula is obtained; the wavelet decomposition coefficients are filtered using the hard threshold formula to obtain the high-frequency decomposition coefficients; where, The standard deviation of noise. The expression for the hard threshold formula is: ; when ≤ At that time, through the formula Calculate soft threshold The soft thresholding formula is obtained; the wavelet decomposition coefficients are filtered using the soft thresholding formula to obtain the high-frequency decomposition coefficients; wherein, the expression of the soft thresholding formula is: ; The high-frequency filtered signal is obtained by filtering the high-frequency detail components based on the high-frequency decomposition coefficients.

[0038] It should be noted that the noise standard deviation The components include sensor acquisition noise, electrolyte polarization effect, and environmental electromagnetic interference; the noise type is identified by calculating the peak factor. > The noise spikes are high-frequency peaks, mainly caused by sensor interference. Hard thresholding is used to quickly remove these peaks. ≤ The noise is low-frequency drift, mainly caused by electrolyte polarization residue. Soft thresholding is used to balance noise suppression and signal smoothness.

[0039] For example, firstly, the peak factor of each layer's coefficients is calculated by dividing the maximum absolute value of the coefficient by its mean; then, a peak threshold is set. When the peak factor is 5, When the peak factor is greater than 5, it is determined to be high-frequency spike noise caused by the sensor and hard thresholding is applied; when the peak factor is greater than 5, it is considered to be high-frequency spike noise caused by the sensor and hard thresholding is applied. If the value is less than or equal to 5, it is determined to be low-frequency drift noise caused by electrolyte polarization residue, and soft threshold processing is adopted.

[0040] In the embodiments of this application, the high-frequency filtered signal obtained after this classification threshold processing can significantly distinguish and eliminate different types of noise, while ensuring that the key transient characteristics characterizing the ohmic internal resistance during the current ramp-up stage are fully preserved.

[0041] In some implementations, the threshold strategy is also dynamically adjusted; wherein, the dynamic adjustment of the threshold strategy includes: Obtain the charging temperature T; A temperature correction coefficient is constructed based on the charging temperature T. ; Based on temperature correction factor The soft and hard thresholds are adjusted to obtain the adjusted threshold strategy; wherein, the soft and hard thresholds include the soft threshold. and hard threshold .

[0042] For example, the corrected soft and hard thresholds are: , Different correction scenarios are defined based on different charging temperatures T, a) Low temperature scenario (T<15℃): <0.9, threshold lowered by 10% to avoid filtering out weak ohmic signals; b) High temperature scenario (T>35℃): >1.1, threshold increased by 10% to enhance noise suppression capability; c) Normal temperature scenario (35℃≥T≥15℃): It is approximately 1, keeping the threshold unchanged.

[0043] In this embodiment, the temperature-related threshold correction mechanism enables the denoising process to adapt to the actual working environment of the battery, significantly improving the stability and accuracy of the system's internal resistance estimation across the entire temperature range while ensuring the integrity of the effective signal components.

[0044] S204. Perform linear fitting on the denoised signal to obtain the initial value of the internal resistance.

[0045] In some implementations, the formula for calculating the initial value of the internal resistance is: ; in, This marks the start of the current ramp-up phase. This marks the end of the current ramp-up phase. and They are respectively and The charging voltage at any given time. and They are respectively and The charging current at any given moment.

[0046] For example, the clean voltage signal obtained after adaptive wavelet denoising. With current signal Select the starting time of the climb phase Until the end time The complete data segment, with current signal as independent variable and voltage signal as dependent variable, is linearly fitted using the least squares method; among which, the initial current... With end current Corresponding to and The slope of the fitted straight line at time t is the initial value of the ohmic internal resistance. Its physical meaning is ΔU / ΔI, which reflects the average sensitivity of the terminal voltage to the change of current within this time period.

[0047] In this embodiment, the internal resistance parameter is directly extracted from the denoised transient signal using a linear fitting method based on Ohm's law. This method is not only simple and efficient in calculation, making it suitable for real-time execution in embedded systems, but also ensures that the initial value obtained has high accuracy and reliability because the signal on which the fitting is based has been effectively stripped of noise interference. This lays a precise iterative starting point for the subsequent dynamic tracking using the recursive least squares method.

[0048] S205. Input the initial value of the internal resistance into the first-order RC equivalent circuit model, and perform dynamic estimation by recursive least squares method to obtain the estimated value of the battery internal resistance.

[0049] In some implementations, the method for obtaining the estimated battery internal resistance includes: Based on a first-order RC equivalent circuit model, a state vector and a covariance matrix are constructed; wherein, the expression for the state vector is: In the formula, k For discrete time series indexes, K is an integer. The internal resistance of the battery to be estimated is... The open-circuit voltage to be estimated; The state vector is initialized based on the initial value of the internal resistance; The estimated internal resistance of the battery is obtained by iteratively updating the state vector and covariance matrix using the recursive least squares method.

[0050] It should be noted that the expression for the first-order RC equivalent circuit model is as follows: In the formula, The signal noise during the observation process follows a mean of 0 and a variance of . 2 The Gaussian distribution is used; the initialized state vector is: In the formula, The initial value of internal resistance calculated during the charging ramp-up phase. The initial value of the covariance matrix is ​​the average of the last 5 high-frequency sampled voltages during the ramp-up phase (to reduce the impact of initial voltage fluctuations); the initial value of the covariance matrix is... ; I It is a 2×2 identity matrix. Take 10 -3 (Based on BMS hardware precision calibration, ensure that the initial weights are reasonable and avoid iterative oscillations.)

[0051] An exemplary schematic diagram of a first-order RC equivalent circuit model is shown below. Figure 4As shown, based on a first-order RC equivalent circuit model, a state vector is first constructed, containing two parameters: the internal resistance and open-circuit voltage to be estimated in real time. This vector is initialized with the initial internal resistance value obtained in previous steps and the average value of the last five voltage sampling points before the end of the ramp-up phase. Simultaneously, the covariance matrix is ​​set as a second-order identity matrix with diagonal elements all equal to 0.001 to reasonably characterize the uncertainty of the initial estimate. In the real-time estimation phase, the system collects battery terminal voltage and charging / discharging current once per second and iteratively updates them using the recursive least squares method: each iteration first calculates the Kalman gain using a gain formula with a forgetting factor of 0.98, then corrects the state vector based on the current observation residuals, and simultaneously updates the covariance matrix, ultimately calculating the estimated battery internal resistance.

[0052] In some implementations, the iterative update of the state vector and covariance matrix using the recursive least squares method includes: Get k Battery terminal voltage at time and charging / discharging current And construct the input vector ; Based on input vector and state vector The residuals were calculated. ; Based on input vector Covariance Matrix Calculate the gain ;in, This is a forgetting factor used to dynamically balance the weights of old and new data. Based on residuals and gain For the state vector Covariance Matrix The battery internal resistance is estimated by iterative correction.

[0053] It should be noted that the charging and discharging current The value is positive when charging and negative when discharging; input vector Wherein, constant term 1 corresponds to the linear coefficient of the open-circuit voltage; based on the input vector and state vector Battery terminal voltage It can be rewritten in linear regression form: Forgetting factor The settings are determined based on the time-varying characteristics of the battery system, the noise environment, and the sampling frequency: when parameters change slowly, noise interference is high, or the sampling frequency is high, the forgetting factor needs to be increased to enhance smoothness; when operating conditions change abruptly, fast tracking is required, or signal quality is good, the factor needs to be decreased to improve tracking speed, thereby achieving the best balance between estimation accuracy and dynamic response capability. The closer to 1, the greater the weight of the old data, and the smoother the estimation of internal resistance.

[0054] For example, the system collects data at a frequency of once per second. k Battery terminal voltage at time With charging and discharging current The system constructs an input vector based on this vector; then, it predicts the current voltage based on this input vector and the state vector from the previous time step, and compares it with the actual acquired value. Comparison yields residuals Subsequently, using a preset forgetting factor of 0.98 and the covariance matrix of the previous time step... And the current input vector, calculate the Kalman gain according to the gain formula of the recursive least squares method. Finally, the system uses the gain and residual to correct the state vector and update the covariance matrix simultaneously, thereby outputting the estimated internal resistance at the current moment.

[0055] In this embodiment, based on the standard recursive least squares theory, the weight of historical data is dynamically adjusted by introducing a forgetting factor mechanism, which enables the internal resistance estimation to quickly track the dynamic changes of the battery while maintaining the numerical stability of the estimation process. This significantly improves the convergence speed of online estimation and the tracking ability under varying operating conditions.

[0056] In some implementations, the state vector Covariance Matrix The correction formula for iterative correction is: ; in, I Let Q be the identity matrix of order Q, where Q is a positive integer.

[0057] In the embodiments of this application, I It is a second-order identity matrix, obtained through the gain matrix. and residual Adjusting the correction magnitude of the state vector enables asymptotic optimization of the parameters; covariance matrix It reflects the uncertainty of parameter estimation, gradually converges during the iteration process, and ensures that the estimation accuracy continues to improve.

[0058] For example, the overall method flow diagram of this application is as follows: Figure 5As shown, under the 25℃ / 1C scenario, the internal resistance of the power ternary lithium battery is estimated using the method provided in this application. A schematic diagram of the estimated internal resistance of each cell in the battery pack is shown below. Figure 6 As shown.

[0059] Based on the above technical solutions, the real-time lithium battery internal resistance estimation method based on adaptive wavelet denoising provided in this application effectively solves the technical bottleneck of traditional internal resistance estimation methods by designing an adaptive wavelet denoising framework and fusing recursive least squares (RLS). In the signal preprocessing stage, adaptive wavelet denoising accurately matches the signal characteristics during the charging current ramp-up phase by dynamically selecting wavelet basis functions, adjusting the number of decomposition layers, and employing layer threshold strategies. This achieves efficient noise removal while ensuring lossless preservation of useful transient responses, improving the estimation accuracy of the initial internal resistance value from the source. Combined with a local high-frequency acquisition strategy, data acquisition is initiated only during the window period of significant transient response, significantly reducing computational power consumption. In the dynamic estimation stage, the RLS algorithm is based on a first-order RC equivalent circuit model, using high-precision initial values ​​for iterative optimization. A forgetting factor mechanism balances the weights of new and old data, enabling real-time tracking and adaptive updating of internal resistance parameters under all operating conditions, effectively addressing complex scenarios such as temperature fluctuations and charge / discharge switching. The overall approach balances estimation accuracy and computational efficiency. Its modular design enhances its engineering practicality and can be directly integrated into the BMS embedded platform, providing reliable parameter support for battery health status assessment. At the same time, it verifies the flexible adaptability value of wavelet analysis theory in nonlinear transient signal processing.

[0060] In one possible implementation, this application embodiment also provides a real-time lithium battery internal resistance estimation system 100 based on adaptive wavelet denoising, such as... Figure 7 As shown, it includes a signal acquisition and judgment module 10, a local high-frequency acquisition module 20, an adaptive wavelet denoising module 30, an internal resistance initial value calculation module 40, and a dynamic internal resistance estimation module 50. Among them, the signal acquisition and judgment module 10 is used to acquire the charging current signal and determine whether the charging is in the current ramp-up stage based on the charging current signal. The local high-frequency acquisition module 20 is used to acquire voltage and current signals based on a preset sampling frequency when the charging is in the current ramp-up stage. The adaptive wavelet denoising module 30 is used to perform adaptive wavelet denoising on voltage signals and current signals to obtain denoised signals. The internal resistance initial value calculation module 40 is used to perform linear fitting on the denoised signal to obtain the internal resistance initial value. The dynamic internal resistance estimation module 50 is used to input the initial value of the internal resistance into the first-order RC equivalent circuit model, and perform dynamic estimation by recursive least squares method to obtain the estimated value of the battery internal resistance.

[0061] For example, the signal acquisition and judgment module 10 calculates the charging current change rate in real time and compares it with a dynamic threshold. When the change rate exceeds 0.1 and the current does not reach the target value, it determines that it has entered the current ramp-up stage and triggers subsequent modules. The local high-frequency acquisition module 20 then starts synchronous acquisition of voltage and current at a frequency of once per second, with a duration covering a ramp-up window of about 15 to 30 seconds. The adaptive wavelet denoising module 30 performs dynamic wavelet decomposition and threshold processing on the acquired raw signal. It adaptively selects the wavelet basis function and determines the number of decomposition layers according to the real-time operating conditions, and uses a temperature-corrected threshold strategy to filter out sensor noise and polarization interference. The internal resistance initial value calculation module 40 then performs least squares linear fitting on the purified voltage-current data to extract a high-precision ohmic internal resistance initial value. Finally, the dynamic internal resistance estimation module 50 substitutes the initial value into the first-order RC equivalent circuit model and continuously iterates and updates it using a recursive least squares method with a forgetting factor of 0.98 to achieve dynamic and smooth tracking of internal resistance under all operating conditions.

[0062] Based on the above technical solutions, adaptive wavelet denoising effectively solves the adaptability problem of traditional wavelet denoising in the charging current ramp-up stage by adaptively selecting wavelet bases, dynamically adjusting the number of decomposition layers, and designing hierarchical and categorized threshold strategies. Compared with traditional methods, the average relative error of internal resistance estimation in this application is effectively reduced, with less computational power consumption, while balancing accuracy and real-time performance. This application verifies the application value of wavelet analysis in nonlinear transient electrical signal processing, provides a typical case for the engineering implementation of wavelet analysis theory, and also provides high-precision core parameters for the health status assessment of battery management systems.

[0063] In implementation, each step of the method provided in this embodiment can be completed by integrated logic circuits in the processor hardware or by instructions in software. The steps of the method disclosed in the embodiments of this application can be directly manifested as being executed by a hardware processor, or being executed by a combination of hardware and software modules in the processor.

[0064] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented using software programs, implementation can be, in whole or in part, in the form of a computer program product. This computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium accessible to a computer or a data storage device containing one or more servers, data centers, etc., that can be integrated with the medium. The available media can be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., DVDs), or semiconductor media (e.g., solid-state disks (SSDs)).

[0065] Although this application has been described herein in conjunction with various embodiments, those skilled in the art, by reviewing the accompanying drawings, disclosure, and appended claims, will understand and implement other variations of the disclosed embodiments in carrying out the claimed application. In the claims, the word "comprising" does not exclude other components or steps, and "a" or "an" does not exclude multiple instances. A single processor or other unit can implement several functions listed in the claims. While different dependent claims may recite certain measures, this does not mean that these measures cannot be combined to produce good results.

[0066] Although this application has been described in conjunction with specific features and embodiments, it is obvious that various modifications and combinations can be made thereto without departing from the spirit and scope of this application. Accordingly, this specification and drawings are merely exemplary illustrations of this application as defined by the appended claims, and are considered to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from the spirit and scope of this application. Thus, if such modifications and modifications of this application fall within the scope of the claims of this application and their equivalents, this application is also intended to include such modifications and modifications.

Claims

1. A real-time estimation method for the internal resistance of a lithium battery based on adaptive wavelet denoising, characterized in that, include: Acquire the charging current signal and determine whether the charging is in the current ramp-up phase based on the charging current signal; If the charging process is in the current ramp-up phase, voltage and current signals are collected based on a preset sampling frequency. Adaptive wavelet denoising is performed on the voltage and current signals to obtain the denoised signal. Linear fitting is performed on the denoised signal to obtain the initial value of the internal resistance; The initial value of the internal resistance is input into the first-order RC equivalent circuit model, and the internal resistance of the battery is estimated by dynamic estimation through recursive least squares method.

2. The real-time estimation method for lithium battery internal resistance based on adaptive wavelet denoising according to claim 1, characterized in that, The method of determining whether charging is in the current ramp-up phase based on the charging current signal includes: Based on charging current signal The rate of change of current was calculated. ;in, t For time; rate of change of current and charging current signal Compared with the preset rate of change threshold and target charging current Compare; when > and < If so, it is determined that the charging is in the current ramp-up phase.

3. The real-time estimation method for lithium battery internal resistance based on adaptive wavelet denoising according to claim 2, characterized in that, The adaptive wavelet denoising of voltage and current signals includes: Wavelet transform is used to decompose voltage and current signals to obtain low-frequency approximate components and high-frequency detail components. High-frequency detail components are threshold-filtered based on a threshold strategy to obtain a high-frequency filtered signal; The denoised signal is obtained by performing inverse wavelet transform on the high-frequency filtered signal and the low-frequency approximate component.

4. The real-time estimation method for lithium battery internal resistance based on adaptive wavelet denoising according to claim 3, characterized in that, The wavelet decomposition process includes adaptive selection of wavelet basis functions and dynamic determination of the number of decomposition levels. The adaptive selection of the wavelet basis function involves constructing a wavelet basis evaluation index; wherein the expression for the wavelet basis evaluation index is: In the formula, The signal-to-noise ratio after signal reconstruction. The mean absolute error at the signal abrupt change point. To balance the weighting coefficients; calculate the evaluation scores of multiple wavelet basis functions in the preset wavelet basis library based on the wavelet basis evaluation index; select the wavelet basis function corresponding to the maximum evaluation score; The dynamic determination of the number of decomposition layers: the rate of change of current. The number of decomposition levels is obtained by comparing the result with a preset multi-level decomposition threshold.

5. The real-time estimation method for lithium battery internal resistance based on adaptive wavelet denoising according to claim 3, characterized in that, The threshold-based strategy for threshold filtering of high-frequency detail components includes: Through formula Peak factor was calculated ;in, These are the wavelet decomposition coefficients corresponding to the high-frequency detail components. N is the length of the wavelet decomposition coefficients; Peak factor With the preset peak threshold Compare; when > At that time, through the formula Calculate the hard threshold The hard threshold formula is obtained; the wavelet decomposition coefficients are filtered using the hard threshold formula to obtain the high-frequency decomposition coefficients; where, The standard deviation of noise. The expression for the hard threshold formula is: ; when ≤ At that time, through the formula Calculate soft threshold The soft thresholding formula is obtained; the wavelet decomposition coefficients are filtered using the soft thresholding formula to obtain the high-frequency decomposition coefficients; wherein, the expression of the soft thresholding formula is: ; The high-frequency filtered signal is obtained by filtering the high-frequency detail components based on the high-frequency decomposition coefficients.

6. The real-time estimation method for lithium battery internal resistance based on adaptive wavelet denoising according to claim 5, characterized in that, It also includes dynamically adjusting the threshold strategy; wherein, the dynamic adjustment of the threshold strategy includes: Obtain the charging temperature T; A temperature correction coefficient is constructed based on the charging temperature T. ; Based on temperature correction factor The soft and hard thresholds are adjusted to obtain the adjusted threshold strategy; wherein, the soft and hard thresholds include the soft threshold. and hard threshold .

7. The real-time estimation method for lithium battery internal resistance based on adaptive wavelet denoising according to claim 1, characterized in that, The methods for obtaining the battery internal resistance estimate include: Based on a first-order RC equivalent circuit model, a state vector and a covariance matrix are constructed; wherein, the expression for the state vector is: In the formula, k For discrete time series indexes, K is an integer. The internal resistance of the battery to be estimated is... The open-circuit voltage to be estimated; The state vector is initialized based on the initial value of the internal resistance; The estimated internal resistance of the battery is obtained by iteratively updating the state vector and covariance matrix using the recursive least squares method.

8. The real-time estimation method for lithium battery internal resistance based on adaptive wavelet denoising according to claim 7, characterized in that, The iterative update of the state vector and covariance matrix using the recursive least squares method includes: Get k Battery terminal voltage at time and charging / discharging current And construct the input vector ; Based on input vector and state vector The residuals were calculated. ; Based on input vector Covariance Matrix Calculate the gain ;in, This is a forgetting factor used to dynamically balance the weights of old and new data. Based on residuals and gain For the state vector Covariance Matrix The battery internal resistance is estimated by iterative correction.

9. The real-time estimation method for lithium battery internal resistance based on adaptive wavelet denoising according to claim 8, characterized in that, The state vector Covariance Matrix The correction formula for iterative correction is: ； in, I Let Q be the identity matrix of order Q, where Q is a positive integer.

10. A real-time lithium battery internal resistance estimation system based on adaptive wavelet denoising, applied to the real-time lithium battery internal resistance estimation method based on adaptive wavelet denoising as described in any one of claims 1-9, characterized in that, It includes a signal acquisition and judgment module, a local high-frequency acquisition module, an adaptive wavelet denoising module, an internal resistance initial value calculation module, and a dynamic internal resistance estimation module; The signal acquisition and determination module is used to acquire the charging current signal and determine whether the charging is in the current ramp-up stage based on the charging current signal. The local high-frequency acquisition module is used to acquire voltage and current signals based on a preset sampling frequency when the charging is in the current ramp-up phase. The adaptive wavelet denoising module is used to perform adaptive wavelet denoising on voltage and current signals to obtain denoised signals. The internal resistance initial value calculation module is used to perform linear fitting on the denoised signal to obtain the internal resistance initial value. The dynamic internal resistance estimation module is used to input the initial value of the internal resistance into the first-order RC equivalent circuit model, and perform dynamic estimation by recursive least squares method to obtain the estimated value of the battery internal resistance.

Images