SOC estimation method and system of lithium battery pack

Abstract

The invention relates to the technical field of SOC estimation, in particular to an SOC estimation method and system for a lithium battery pack. The method comprises the following steps: collecting basic parameters and voltage data of a lithium battery pack, generating steady-state voltage data by using singular value decomposition and time domain reconstruction, further performing multi-level feature mapping and performance reconstruction to obtain basic discharge data, calculating battery energy efficiency and deducing normal power consumption based on the data, and calculating the energy efficiency of the lithium battery pack. Discharging simulation is carried out to evaluate the temperature rise condition, discharging capacity attenuation is predicted in combination with working environment parameters, relative errors are calculated, finally, the charge state is corrected, electric quantity self-calibration is carried out, and intelligent state estimation of the lithium battery pack is achieved. According to the invention, a more efficient SOC estimation method is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of SOC estimation, and in particular to a SOC estimation method and system for a lithium battery pack. Background Art

[0002] Lithium battery packs play a vital role in modern new energy technologies and are widely used in electric vehicles, energy storage systems, and portable electronic devices. With the growing demand for high-performance lithium batteries, it is particularly important to accurately evaluate the state and performance of batteries. In particular, accurate estimation of the state of charge (SOC) is of great significance for optimizing battery efficiency and extending battery life. However, many traditional SOC estimation methods currently have problems with insufficient accuracy and poor generalization ability, which makes it difficult to meet performance requirements in practical applications. Existing SOC estimation methods often rely on simple algorithm models, ignoring the complex nonlinear characteristics of batteries and the diversity of working environments, resulting in large deviations in estimation results under different working conditions. In addition, the lack of effective real-time monitoring means makes it impossible to reflect changes in dynamic battery status in a timely manner, limiting the intelligence level of battery management systems. Summary of the invention

[0003] Based on this, it is necessary to provide a SOC estimation method and system for a lithium battery pack to solve at least one of the above technical problems.

[0004] To achieve the above object, a method for estimating the SOC of a lithium battery pack comprises the following steps:

[0005] Step S1: Collect basic parameters of the lithium battery pack and battery pack terminal voltage data; perform singular value decomposition on the battery pack terminal voltage data to obtain principal component voltage data; perform time domain reconstruction processing on the principal component voltage data to generate group terminal steady-state voltage data;

[0006] Step S2: Perform multi-level feature mapping on basic parameters of the lithium battery pack to obtain a parameter association matrix; perform battery pack performance reconstruction on the parameter association matrix to generate basic discharge data of the lithium battery pack;

[0007] Step S3: Based on the basic discharge data of the lithium battery pack, the steady-state voltage data of the pack end is decomposed into battery power to generate battery energy efficiency; the normal power consumption of the battery energy efficiency is deduced to obtain the normal state of charge;

[0008] Step S4: performing a discharge simulation according to a normal state of charge to obtain simulated discharge data; performing a discharge temperature rise calculation on the simulated discharge data based on basic parameters of the lithium battery pack to generate a discharge temperature rise condition of the battery pack;

[0009] Step S5: collecting working environment parameters of the lithium battery pack; predicting the discharge capacity attenuation according to the working environment parameters of the lithium battery pack and the discharge temperature rise of the battery pack to obtain the predicted attenuation; performing relative error calculation on the basic discharge data of the lithium battery pack based on the predicted attenuation to obtain the discharge relative error data;

[0010] Step S6: estimating and correcting the normal state of charge based on the discharge relative error data to generate a corrected state of charge; performing a power self-calibration process according to the corrected state of charge to execute the intelligent SOC estimation method for the lithium battery pack.

[0011] The present invention provides a reliable basis for subsequent analysis by collecting basic parameters and battery terminal voltage data through the SOC estimation method based on the lithium battery pack. The steady-state voltage data generated by singular value decomposition and time domain reconstruction processing improves the accuracy of battery state evaluation. The parameter association matrix obtained by multi-level feature mapping provides a clear view for battery pack performance reconstruction. The generation of basic discharge data lays a foundation for battery performance analysis. The energy efficiency obtained by battery power deconstruction provides an important basis for optimizing battery use. The normal power consumption derivation result provides a reference for battery state monitoring. The combination of discharge simulation and temperature rise calculation improves the accuracy and reliability of the discharge process. The collection of working environment parameters combined with discharge capacity attenuation prediction provides a scientific basis for battery life evaluation. The data generated by relative error calculation provides support for the accurate determination of discharge performance. The correction of the state of charge and the self-calibration of the power volume improve the intelligent management level of the lithium battery pack, and the overall performance monitoring, state evaluation and management efficiency of the lithium battery pack are improved.

[0012] Preferably, step S1 comprises the following steps:

[0013] Step S11: collecting basic parameters of the lithium battery pack and battery pack terminal voltage data; performing wavelet denoising on the battery pack terminal voltage data to obtain pure voltage data;

[0014] Step S12: performing matrix decomposition on the pure voltage data to generate a voltage characteristic matrix; performing singular value screening on the voltage characteristic matrix to obtain principal component voltage data;

[0015] Step S13: Perform Fourier reconstruction on the main component voltage data to obtain frequency domain feature data; perform inverse transformation mapping on the frequency domain feature data to generate a voltage time domain sequence;

[0016] Step S14: performing smoothing and filtering processing on the voltage time domain sequence to obtain group-end steady-state voltage data.

[0017] The present invention significantly improves the signal-to-noise ratio of voltage data through wavelet denoising processing, effectively extracts voltage features through matrix decomposition technology, ensures the validity and relevance of principal component voltage data through singular value screening, provides in-depth analysis of frequency domain features through Fourier reconstruction, realizes the conversion of frequency domain and time domain data through inverse transform mapping, eliminates noise and fluctuations in time domain sequences through smoothing filtering processing, and generates steady-state voltage data at the group end that provides a reliable basis for subsequent SOC estimation, thereby improving the accuracy and stability of lithium battery pack status assessment as a whole and promoting the intelligent and efficient development of battery management technology.

[0018] Preferably, step S2 comprises the following steps:

[0019] Step S21: performing hierarchical clustering on basic parameters of the lithium battery pack to obtain parameter hierarchical data; performing feature extraction on the parameter hierarchical data to generate feature vector data;

[0020] Step S22: quantifying the correlation of basic parameters of the lithium battery pack based on the characteristic vector data to obtain a parameter correlation matrix;

[0021] Step S23: Performing performance projection according to the parameter association matrix to generate battery performance characteristics; performing battery capacity mapping on the battery performance characteristics to generate discharge capacity data;

[0022] Step S24: feature fusion is performed on the discharge capacity data to obtain basic discharge data of the lithium battery pack.

[0023] The present invention realizes the systematic analysis of the basic parameters of lithium battery packs through hierarchical clustering technology. Feature extraction improves the expressiveness of data. The generated feature vector data provides a clear basis for subsequent analysis. The quantification of parameter association matrix provides a clear perspective for the relationship between different parameters. The performance projection technology effectively reveals the performance characteristics of the battery. The battery capacity mapping ensures the reliability of the discharge capacity data. The feature fusion technology integrates multi-dimensional data, improves the integrity and accuracy of the basic discharge data of the lithium battery pack, optimizes the basis of SOC estimation as a whole, and promotes the intelligent and refined development of the battery management system.

[0024] Preferably, step S3 comprises the following steps:

[0025] Step S31: dynamically filter and transform the basic discharge data of the lithium battery pack to obtain the steady-state power of the lithium battery;

[0026] Step S32: performing joint harmonic decomposition on the steady-state power of the lithium battery based on the steady-state voltage data at the group end to generate a transient power spectrum; performing inverse transformation and reconstruction on the transient power spectrum to generate battery energy efficiency;

[0027] Step S33: deriving the battery normal state from the battery energy efficiency to obtain power consumption characteristic data; identifying the power consumption pattern from the power consumption characteristic data to obtain the power consumption pattern data;

[0028] Step S34: quantifying the state of charge of the power consumption mode data based on the battery energy efficiency to generate a normal state of charge.

[0029] The present invention realizes accurate extraction of steady-state power of lithium batteries through dynamic filtering transformation, and combines harmonic decomposition to enhance the ability to analyze power fluctuations. The generation of transient power spectrum provides a comprehensive perspective of battery energy efficiency, and inverse transformation reconstruction ensures the credibility of battery energy efficiency data. The derivation of battery normal state provides a deep understanding of power consumption characteristic data, and power consumption pattern recognition technology reveals the power consumption behavior and laws of batteries. State of charge quantification provides a scientific basis for the generation of normal state of charge, which improves the accuracy and reliability of SOC estimation as a whole and promotes the intelligence and application value of lithium battery management systems.

[0030] Preferably, step S32 includes the following steps:

[0031] Perform joint Fourier expansion on the steady-state voltage data of the group end and the steady-state power data of the lithium battery to obtain a joint harmonic power sequence; perform bandpass filtering on the joint harmonic power sequence to output fundamental frequency component data;

[0032] Performing phase correction on the fundamental frequency component data to obtain phase spectrum data; performing frequency band synthesis processing on the phase spectrum data to obtain power density data;

[0033] The spectral peaks of the joint harmonic power sequence are reorganized based on the power density data to generate a transient power spectrum;

[0034] Performing spectral line compensation processing on the transient power spectrum to obtain compensated power data; performing multi-dimensional reconstruction based on the compensated power data to generate reconstructed power data;

[0035] The reconstructed power data is mapped for power loss according to the basic parameters of the lithium battery pack to generate power loss data; efficiency calibration is performed based on the power loss data to produce battery energy efficiency.

[0036] The present invention realizes the deep fusion of steady-state voltage data and steady-state power data through the joint Fourier expansion. The bandpass filtering process effectively extracts the fundamental frequency component, ensuring the accuracy and stability of the data. The phase correction improves the phase information accuracy of the power data. The frequency band synthesis process enhances the expressiveness of the power density data. The spectrum peak reconstruction technology provides clear features for the generation of transient power spectrum. The spectrum line compensation process eliminates interference in the signal and improves the reliability of the compensated power data. The multi-dimensional reconstruction realizes the comprehensive analysis of the power data. The power loss mapping provides an in-depth understanding of the battery performance. The efficiency calibration process ensures the scientificity and accuracy of the battery energy efficiency evaluation, which improves the accuracy and application value of the SOC estimation as a whole and promotes the intelligent and refined development of the lithium battery management technology.

[0037] Preferably, step S4 comprises the following steps:

[0038] Step S41: performing Poisson distribution fitting on the normal state of charge to obtain load characteristic data;

[0039] Step S42: performing time series state transfer modeling on the load characteristic data to generate a state transfer matrix; performing sequence prediction on the load characteristic data according to the state transfer matrix to obtain simulated discharge data;

[0040] Step S43: performing thermodynamic modeling on the simulated discharge data based on the basic parameters of the lithium battery pack to obtain the temperature rise characteristics of the lithium battery; dynamically evolving the temperature rise characteristics of the lithium battery to generate temperature rise evolution data;

[0041] Step S44: reconstructing the lithium battery state based on the temperature rise evolution data to generate the battery pack discharge temperature rise condition.

[0042] The present invention provides accurate load characteristic data for normal state of charge through Poisson distribution fitting, and the time series state transition modeling enhances the ability to capture load changes. The generation of state transfer matrix lays the foundation for sequence prediction. The generation of simulated discharge data realizes the dynamic simulation of battery discharge process. Thermodynamic modeling ensures the accurate evaluation of lithium battery temperature rise characteristics. The dynamic evolution of temperature rise characteristics provides a comprehensive perspective of temperature rise changes. State reconstruction technology realizes an in-depth understanding of the temperature rise condition of lithium battery discharge, which improves the accuracy and reliability of SOC estimation as a whole and promotes the intelligent and refined development of lithium battery management system.

[0043] Preferably, step S5 comprises the following steps:

[0044] Step S51: collecting working environment parameters of the lithium battery pack; constructing a temperature field for the working environment parameters of the lithium battery pack to obtain a virtual working temperature field;

[0045] Step S52: performing attenuation modeling based on the virtual operating temperature field and the battery pack discharge temperature rise condition to generate a capacity attenuation model; performing discharge capacity attenuation prediction according to the capacity attenuation model to obtain a predicted attenuation condition;

[0046] Step S53: Calculating discharge error on basic discharge data of the lithium battery pack based on the predicted attenuation condition to obtain discharge error data;

[0047] Step S54: performing relative quantization processing on the discharge error data to generate error quantization data; performing feature fusion on the error quantization data to obtain discharge relative error data.

[0048] The present invention provides basic data for the performance evaluation of lithium battery packs by collecting working environment parameters, realizes the generation of virtual working temperature field by temperature field construction, ensures the scientificity and accuracy of capacity decay model by attenuation modeling, provides a forward-looking perspective of future performance changes by discharge capacity decay prediction, effectively reveals the difference between actual discharge and theoretical value by discharge error calculation, enhances the accuracy of error analysis by relative quantitative processing of discharge error data, integrates multi-dimensional error information by feature fusion technology, generates discharge relative error data, improves the accuracy and reliability of SOC estimation as a whole, and promotes the intelligent and efficient development of lithium battery management system.

[0049] Preferably, step S52 includes the following steps:

[0050] Perform temperature surface topological mapping on the virtual working temperature field data to obtain the temperature distribution characteristic matrix;

[0051] Performing drift mapping processing on the virtual working temperature field according to the temperature distribution characteristic matrix to generate a temperature drift template;

[0052] Based on the temperature drift template and the battery pack discharge temperature rise condition, the lithium battery capacity impact analysis is performed to obtain the lithium battery capacity change data; based on the lithium battery capacity change data, the attenuation model is performed to generate the capacity attenuation model;

[0053] Perform discharge temperature rise simulation according to the capacity decay model to obtain simulated lithium battery discharge temperature rise data;

[0054] Based on the capacity decay model, the discharge capacity decay is predicted for the simulated lithium battery discharge temperature rise data to obtain the predicted decay situation.

[0055] The present invention provides a temperature distribution feature matrix through temperature surface topological mapping, which enhances the understanding of the virtual working temperature field. The temperature drift template generated by drift mapping processing ensures the accurate reflection of temperature changes. The lithium battery capacity impact analysis reveals the actual impact of temperature changes on battery capacity. The capacity change data lays the foundation for capacity decay modeling. The generation of simulated lithium battery discharge temperature rise data realizes the dynamic evaluation of the discharge process. The predicted decay situation based on the capacity decay model provides an important reference for the future performance of battery performance, which improves the accuracy and reliability of SOC estimation as a whole and promotes the intelligent and efficient development of lithium battery management systems.

[0056] Preferably, step S6 comprises the following steps:

[0057] Step S61: performing iterative distribution analysis on the discharge relative error data to obtain an error distribution spectrum;

[0058] Step S62: estimating and correcting the normal state of charge based on the error distribution spectrum to generate a corrected state of charge;

[0059] Step S63: performing state boundary detection on the corrected state of charge to obtain a state threshold range; performing feedback compensation on the corrected state of charge based on the state threshold range to obtain calibration power data;

[0060] Step S64: quantify the accuracy based on the calibration power data to obtain the SOC output result to execute the intelligent SOC estimation method of the lithium battery pack.

[0061] The present invention provides an in-depth statistical understanding of the discharge relative error data through iterative distribution analysis. The generation of the error distribution spectrum reveals the error characteristics and distribution laws. The correction of the charge state estimation improves the accuracy of the charge state. The state boundary detection ensures a comprehensive evaluation of the battery state. The determination of the state threshold range provides a clear basis for feedback compensation. The generation of calibrated power data realizes the precise adjustment of the power. The precision quantization technology improves the reliability and accuracy of the SOC output results. The intelligent SOC estimation method of the lithium battery pack is optimized as a whole, and the intelligent and efficient development of the battery management system is promoted.

[0062] The present invention also provides a SOC estimation system for a lithium battery pack, which is used to execute the SOC estimation method for a lithium battery pack as described above. The SOC estimation system for a lithium battery pack comprises:

[0063] The parameter acquisition module is used to collect basic parameters of lithium battery packs and battery pack terminal voltage data; perform singular value decomposition on the battery pack terminal voltage data to obtain principal component voltage data; perform time domain reconstruction processing on the principal component voltage data to generate group terminal steady-state voltage data;

[0064] The parameter mapping module is used to perform multi-level feature mapping on the basic parameters of the lithium battery pack to obtain a parameter association matrix; the parameter association matrix is ​​used to reconstruct the battery pack performance to generate basic discharge data of the lithium battery pack;

[0065] The power derivation module is used to deconstruct the battery power of the steady-state voltage data at the group end based on the basic discharge data of the lithium battery group to generate the battery energy efficiency; the normal power consumption of the battery energy efficiency is deduced to obtain the normal state of charge;

[0066] The discharge simulation module is used to perform discharge simulation according to the normal state of charge to obtain simulated discharge data; the discharge temperature rise calculation is performed on the simulated discharge data based on the basic parameters of the lithium battery pack to generate the discharge temperature rise status of the battery pack;

[0067] The capacity attenuation prediction module is used to collect the working environment parameters of the lithium battery pack; the discharge capacity attenuation is predicted according to the working environment parameters of the lithium battery pack and the discharge temperature rise of the battery pack to obtain the predicted attenuation situation; the relative error of the basic discharge data of the lithium battery pack is calculated based on the predicted attenuation situation to obtain the discharge relative error data;

[0068] The charge correction module is used to estimate and correct the normal state of charge based on the discharge relative error data to generate a corrected state of charge; and perform power self-calibration processing according to the corrected state of charge to execute the intelligent SOC estimation method of the lithium battery pack.

[0069] The present invention realizes efficient acquisition of basic parameters and battery pack terminal voltage data through a parameter acquisition module through a SOC estimation system based on a lithium battery pack, providing a reliable data source for subsequent analysis. The steady-state voltage data generated by singular value decomposition and time domain reconstruction processing enhances the accuracy of battery state evaluation. The multi-level feature mapping and parameter association matrix generation of the parameter mapping module provide a clear structure for battery pack performance reconstruction. The generation of basic discharge data lays a foundation for performance analysis. The energy efficiency analysis of the power derivation module provides a basis for battery optimization use. The normal power consumption derivation result provides a reference for battery state monitoring. The combination of discharge simulation and temperature rise calculation improves the reliability of the discharge process. The capacity decay prediction module realizes scientific battery life evaluation in combination with working environment parameters. The relative error calculation based on the predicted decay situation provides support for discharge performance determination. The intelligent self-calibration processing of the charge correction module improves the management level of the lithium battery pack, and improves the performance monitoring, state evaluation and management efficiency of the lithium battery pack as a whole. BRIEF DESCRIPTION OF THE DRAWINGS

[0070] Figure 1 A schematic flow chart of a method for estimating the SOC of a lithium battery pack;

[0071] Figure 2Detailed implementation flow chart of step S2;

[0072] Figure 3 is a schematic diagram of a detailed implementation process of step S3;

[0073] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0074] The technical method of the present invention is described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by technicians in this field without creative work are within the scope of protection of the present invention.

[0075] In addition, the accompanying drawings are only schematic illustrations of the present invention and are not necessarily drawn to scale. The same reference numerals in the figures represent the same or similar parts, and their repeated description will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities and do not necessarily correspond to physically or logically independent entities. The functional entities can be implemented in software form, or implemented in one or more hardware modules or integrated circuits, or implemented in different networks and / or processor methods and / or microcontroller methods.

[0076] It should be understood that, although the terms "first", "second", etc. may be used herein to describe various units, these units should not be limited by these terms. These terms are used only to distinguish one unit from another unit. For example, without departing from the scope of the exemplary embodiments, the first unit may be referred to as the second unit, and similarly the second unit may be referred to as the first unit. The term "and / or" used herein includes any and all combinations of one or more of the listed associated items.

[0077] To achieve this, please refer to Figures 1 to 3 , a method for estimating the SOC of a lithium battery pack, comprising the following steps:

[0078] Step S1: Collect basic parameters of the lithium battery pack and battery pack terminal voltage data; perform singular value decomposition on the battery pack terminal voltage data to obtain principal component voltage data; perform time domain reconstruction processing on the principal component voltage data to generate group terminal steady-state voltage data;

[0079] Step S2: Perform multi-level feature mapping on basic parameters of the lithium battery pack to obtain a parameter association matrix; perform battery pack performance reconstruction on the parameter association matrix to generate basic discharge data of the lithium battery pack;

[0080] Step S3: Based on the basic discharge data of the lithium battery pack, the steady-state voltage data of the pack end is decomposed into battery power to generate battery energy efficiency; the normal power consumption of the battery energy efficiency is deduced to obtain the normal state of charge;

[0081] Step S4: performing a discharge simulation according to a normal state of charge to obtain simulated discharge data; performing a discharge temperature rise calculation on the simulated discharge data based on basic parameters of the lithium battery pack to generate a discharge temperature rise condition of the battery pack;

[0082] Step S5: collecting working environment parameters of the lithium battery pack; predicting the discharge capacity attenuation according to the working environment parameters of the lithium battery pack and the discharge temperature rise of the battery pack to obtain the predicted attenuation; performing relative error calculation on the basic discharge data of the lithium battery pack based on the predicted attenuation to obtain the discharge relative error data;

[0083] Step S6: estimating and correcting the normal state of charge based on the discharge relative error data to generate a corrected state of charge; performing a power self-calibration process according to the corrected state of charge to execute the intelligent SOC estimation method for the lithium battery pack.

[0084] The present invention provides a reliable basis for subsequent analysis by collecting basic parameters and battery terminal voltage data through the SOC estimation method based on the lithium battery pack. The steady-state voltage data generated by singular value decomposition and time domain reconstruction processing improves the accuracy of battery state evaluation. The parameter association matrix obtained by multi-level feature mapping provides a clear view for battery pack performance reconstruction. The generation of basic discharge data lays a foundation for battery performance analysis. The energy efficiency obtained by battery power deconstruction provides an important basis for optimizing battery use. The normal power consumption derivation result provides a reference for battery state monitoring. The combination of discharge simulation and temperature rise calculation improves the accuracy and reliability of the discharge process. The collection of working environment parameters combined with discharge capacity attenuation prediction provides a scientific basis for battery life evaluation. The data generated by relative error calculation provides support for the accurate determination of discharge performance. The correction of the state of charge and the self-calibration of the power volume improve the intelligent management level of the lithium battery pack, and the overall performance monitoring, state evaluation and management efficiency of the lithium battery pack are improved.

[0085] In an embodiment of the present invention, the SOC estimation method of the lithium battery pack includes the following steps:

[0086] Step S1: Collect basic parameters of the lithium battery pack and battery pack terminal voltage data; perform singular value decomposition on the battery pack terminal voltage data to obtain principal component voltage data; perform time domain reconstruction processing on the principal component voltage data to generate group terminal steady-state voltage data;

[0087] In this embodiment, a multi-channel data acquisition device is used to record the basic parameters of the lithium battery pack. The basic parameters include the nominal capacity, charge and discharge cycle number, operating current and operating voltage range of the lithium battery pack. The data acquisition device model may be NI USB-6361 (multi-function data acquisition device). The battery pack terminal voltage data is acquired in real time by accessing the data interface of the battery management system (Battery Management System). The collected terminal voltage data is processed by the singular value decomposition (Singular Value Decomposition) toolbox in MATLAB software (matrix laboratory software). First, the voltage data is constructed in matrix form for input, the decomposition order is set to 5, the main singular values ​​are retained, the principal components after decomposition are extracted, and then the principal component voltage data is processed using the time domain reconstruction method. The reconstruction process needs to set the reconstruction window length to 10 seconds, and the reconstruction result is output as a group terminal steady-state voltage data file.

[0088] Step S2: Perform multi-level feature mapping on basic parameters of the lithium battery pack to obtain a parameter association matrix; perform battery pack performance reconstruction on the parameter association matrix to generate basic discharge data of the lithium battery pack;

[0089] In this embodiment, an analytic hierarchy process (Analytic Hierarchy Process, in Chinese, analytic hierarchy process) is used to perform multi-level feature mapping on the collected basic parameters of the lithium battery pack. The basic parameters include battery internal resistance, battery pack temperature, electrolyte concentration, etc. Each parameter is input into a feature mapping model, and the weights between the parameters are determined according to an expert scoring method (Expert Scoring Method, in Chinese, expert scoring method). The weight data and the basic parameters form a parameter association matrix. After the matrix is ​​generated, a battery performance analysis model (Battery Performance Analysis Model, in Chinese, battery performance analysis model) is used to deconstruct the parameter relationship in the association matrix, extract basic discharge characteristic data, and finally generate basic discharge data of the lithium battery pack.

[0090] Step S3: Based on the basic discharge data of the lithium battery pack, the steady-state voltage data of the pack end is decomposed into battery power to generate battery energy efficiency; the normal power consumption of the battery energy efficiency is deduced to obtain the normal state of charge;

[0091] In this embodiment, according to the basic discharge data of the lithium battery pack, the battery power deconstruction algorithm (Battery Power Deconstruction Algorithm, Chinese: Battery Power Deconstruction Algorithm) in Python is used to process the group end steady-state voltage data generated in step S1. First, the basic discharge data is matched with the group end steady-state voltage data, and a battery energy efficiency data table is generated according to the relationship between the battery's state of charge (State of Charge, Chinese: State of Charge) and the discharge curve. The table contains indicators such as energy utilization rate and remaining capacity percentage. Subsequently, the energy efficiency data is used to deduce the normal state of charge. By setting the charge and discharge rate to 0.5C, the normal state of charge of the lithium battery pack is derived.

[0092] Step S4: performing a discharge simulation according to a normal state of charge to obtain simulated discharge data; performing a discharge temperature rise calculation on the simulated discharge data based on basic parameters of the lithium battery pack to generate a discharge temperature rise condition of the battery pack;

[0093] In this embodiment, based on the normal state of charge, Simulink (simulation and modeling tool) is called to simulate the normal power consumption process of the battery. The discharge current of the lithium battery pack is set to 10A, the initial voltage is 3.7V, and the termination voltage is 2.5V. After the simulation, the simulated discharge data is exported. The simulated discharge data is processed by a temperature rise calculation script written in Python. First, the thermal conductivity coefficient, specific heat capacity and ambient temperature of the battery are used as input parameters to calculate the temperature rise of the battery during discharge. The temperature rise results are displayed in the form of a two-dimensional chart.

[0094] Step S5: collecting working environment parameters of the lithium battery pack; predicting the discharge capacity attenuation according to the working environment parameters of the lithium battery pack and the discharge temperature rise of the battery pack to obtain the predicted attenuation; performing relative error calculation on the basic discharge data of the lithium battery pack based on the predicted attenuation to obtain the discharge relative error data;

[0095] In this embodiment, the working environment parameters of the lithium battery pack, including ambient temperature, humidity, and air pressure, are collected, and an intelligent sensor module (such as Bosch BME280, which is called environmental sensor module in Chinese) is used to record the working environment data in real time. According to the discharge temperature rise condition and environmental parameters of the battery pack generated in step S4, the discharge capacity attenuation of the lithium battery pack is predicted through an LSTM (long short-term memory network) model, and the attenuation prediction result is output as a JSON format file. Based on the attenuation prediction result, the basic discharge data of step S2 is calculated for relative error. The error calculation adopts a residual analysis method (Residual Analysis Method, which is called residual analysis method in Chinese) to generate discharge relative error data.

[0096] Step S6: estimating and correcting the normal state of charge based on the discharge relative error data to generate a corrected state of charge; performing a power self-calibration process according to the corrected state of charge to execute the intelligent SOC estimation method for the lithium battery pack.

[0097] In this embodiment, the discharge relative error data is input into the correction algorithm model, and the deviation adjustment method is used to estimate and correct the normal state of charge. The corrected state of charge is processed by the power calibration model. During the calibration process, the step current method (Step Current Method, Chinese for step current method) is used as the basis to dynamically adjust the current input parameters, and finally the corrected lithium battery pack SOC (State of Charge, Chinese for charge state) estimation result is obtained. The estimation result is displayed in the form of a visual dashboard and exported in Excel table format for easy archiving and management.

[0098] Preferably, step S1 comprises the following steps:

[0099] Step S11: Collecting basic parameters of the lithium battery pack and battery pack terminal voltage data; performing wavelet denoising on the battery pack terminal voltage data to obtain pure voltage data;

[0100] Step S12: performing matrix decomposition on the pure voltage data to generate a voltage characteristic matrix; performing singular value screening on the voltage characteristic matrix to obtain principal component voltage data;

[0101] Step S13: Perform Fourier reconstruction on the main component voltage data to obtain frequency domain feature data; perform inverse transformation mapping on the frequency domain feature data to generate a voltage time domain sequence;

[0102] Step S14: performing smoothing and filtering processing on the voltage time domain sequence to obtain group-end steady-state voltage data.

[0103] In this embodiment, a multi-channel data acquisition system is used to collect basic parameters and terminal voltage data of the lithium battery pack. The data acquisition system can use the NI USB-6361 device (multi-function data acquisition device). The basic parameters of the lithium battery pack include battery internal resistance, capacity, voltage upper and lower limits, number of cycles, etc. The data interface is connected to the lithium battery management system to obtain terminal voltage data in real time. The acquisition frequency is set to 1000Hz, and the collected terminal voltage data is stored as a .csv format file. Then, the wavelet transform (Wavelet Transform (wavelet transform in Chinese) is used to denoise the terminal voltage data. Daubechies wavelet is selected as the basis function, and the number of decomposition layers is set to 5. The high-frequency noise signal is processed by the threshold denoising method, and the denoised signal is reconstructed to obtain pure voltage data. The pure voltage data is imported into the MATLAB environment, and the matrix decomposition method is used to process the data. The pure voltage data is input in the form of a matrix. The matrix dimension is determined according to the acquisition time and the number of sampling points. The decomposition algorithm uses the singular value decomposition method, and the first 5 singular values ​​are selected as the basis of the main component voltage data. By setting the threshold for singular value screening, the characteristic matrix after singular value decomposition is screened to obtain the voltage characteristic matrix, and the characteristic matrix is ​​output to a .csv file and saved. At the same time, the screened main component voltage data is normalized and its range is adjusted to between 0 and 1. The main component voltage data is imported into the Python environment and the fast Fourier transform (FastFourier Transform (Fast Fourier Transform in Chinese) is used to extract frequency domain features of the main component voltage data, and frequency domain feature data is generated after extraction. The frequency domain feature data includes frequency components and amplitude information. The frequency range is set to 0 to 500 Hz. After extracting the main frequency components, inverse transformation mapping is performed. The inverse transformation adopts the discrete inverse Fourier transform method to map the frequency domain feature data back to the time domain to generate a voltage time domain sequence. The voltage time domain sequence is imported into the filtering processing module. The filtering module uses Python's SciPy library (Scientific Python, which means scientific computing Python library in Chinese) to implement the filtering operation. The Savitzky-Golay filter is used to smooth the time domain sequence. The filtering parameters are set to a window length of 11 and an order of 3. Noise and fluctuation are eliminated by smoothing the time domain sequence point by point, and finally the steady-state voltage data of the group end is generated. The smoothed steady-state voltage data is output and saved in the .csv file format. The corresponding relationship between the voltage data and the time point is recorded in the file to support the input requirements of subsequent SOC estimation.

[0104] Preferably, step S2 comprises the following steps:

[0105] Step S21: performing hierarchical clustering on basic parameters of the lithium battery pack to obtain parameter hierarchical data; performing feature extraction on the parameter hierarchical data to generate feature vector data;

[0106] Step S22: quantifying the correlation of basic parameters of the lithium battery pack based on the characteristic vector data to obtain a parameter correlation matrix;

[0107] Step S23: Performing performance projection according to the parameter association matrix to generate battery performance characteristics; performing battery capacity mapping on the battery performance characteristics to generate discharge capacity data;

[0108] Step S24: feature fusion is performed on the discharge capacity data to obtain basic discharge data of the lithium battery pack.

[0109] In this embodiment, the basic parameters of the lithium battery pack are obtained through the data acquisition module, including the internal resistance of the battery pack, the upper and lower limits of voltage, the capacity, the number of cycles, etc. The collected basic parameters are input into the hierarchical clustering model in the form of a two-dimensional matrix. The hierarchical clustering model adopts an agglomerative algorithm based on Euclidean distance. The clustering hierarchy is divided into three levels, wherein the first-level division includes single physical parameters such as internal resistance and capacity, the second-level division combines and clusters multidimensional characteristic parameters, and the third-level division generates global hierarchical feature data. The clustered data is saved in a nested array format, and then the clustered parameter hierarchical data is analyzed using a feature extraction algorithm. The feature extraction method uses principal component analysis (PCA). The cumulative contribution rate is set to 95% to extract key eigenvectors, generate eigenvector data, and input the eigenvector data into the association quantification module. The association quantification module uses the Pearson correlation coefficient algorithm to analyze the linear correlation between eigenvectors. The calculation results are output in the form of a parameter association matrix. The rows and columns of the parameter association matrix represent different basic parameters, and each element in the matrix is ​​the correlation value between the corresponding parameters. The association matrix value range is set to -1 to 1, where a positive value indicates a positive correlation and a negative value indicates a negative correlation. The larger the absolute value, the stronger the correlation. After the calculation is completed, the association matrix is ​​stored as a .txt file, which contains the matrix values ​​and the corresponding parameter identification information. The performance projection module is imported into the matrix. The performance projection module adopts the linear projection algorithm. After the rows and columns of the association matrix are aligned, the matrix transformation is performed to generate the battery performance characteristics. The battery performance characteristics include multiple indicators such as internal resistance, discharge rate and thermal management efficiency. The performance characteristics are normalized and all performance characteristic values ​​are scaled to between 0 and 1 for subsequent calculations. The normalized data is stored in the form of a multidimensional array. Then, the capacity mapping is performed based on the battery performance characteristics. The capacity mapping model adopts a regression algorithm based on an artificial neural network. The network model input is the battery performance characteristics, and the output is the battery discharge capacity data. The network structure adopts a 3-layer fully connected structure, and the activation function is ReLU (rectified linear unit). The training The training data comes from the battery historical performance data set. The discharge capacity data is imported into the feature fusion module. The feature fusion module uses the weighted average method to fuse different capacity data according to weights. The weights are allocated according to the battery working environment parameters and the frequency of capacity use. The specific parameter weights are implemented by using the pandas library (data analysis library) in Python. The fused data generates the basic discharge data of the lithium battery pack. The basic discharge data includes the capacity mean, capacity variance and capacity attenuation rate during discharge. The basic discharge data is stored in a tabular format as an .xlsx file. The data column names include timestamps, capacity values ​​and related environmental parameters, which are used for input processing in the subsequent SOC estimation process.

[0110] Preferably, step S3 comprises the following steps:

[0111] Step S31: dynamically filter and transform the basic discharge data of the lithium battery pack to obtain the steady-state power of the lithium battery;

[0112] Step S32: performing joint harmonic decomposition on the steady-state power of the lithium battery based on the steady-state voltage data at the group end to generate a transient power spectrum; performing inverse transformation and reconstruction on the transient power spectrum to generate battery energy efficiency;

[0113] Step S33: deriving the battery normal state from the battery energy efficiency to obtain power consumption characteristic data; identifying the power consumption pattern from the power consumption characteristic data to obtain the power consumption pattern data;

[0114] Step S34: quantifying the state of charge of the power consumption mode data based on the battery energy efficiency to generate a normal state of charge.

[0115] In this embodiment, the basic discharge data of the lithium battery pack is imported into the dynamic filtering transformation module, and the basic discharge data is decomposed by wavelet decomposition technology. The Daubechies wavelet is used as the basis function to decompose the signal into four layers of frequency components, and the low-frequency components are extracted for filtering processing. The filtering adopts the adaptive threshold hard threshold method, and the noise signal is suppressed by setting the decomposition layer number and the threshold parameter. The low-frequency component after filtering is reconstructed by wavelet to generate the steady-state power data of the lithium battery. The data format is a two-dimensional array, the first column is the timestamp, and the second column is the steady-state power value at the corresponding time. The steady-state power data of the lithium battery and the steady-state voltage data of the group end are input into the joint harmonic decomposition module, and the harmonic decomposition algorithm based on fast Fourier transform (FFT, fast Fourier transform) is adopted. The steady-state power and voltage data are jointly processed, and the transient power spectrum is generated by calculating its complex form of spectral characteristics. The transient power spectrum uses frequency and power amplitude as main data features, and the data format is The transient power spectrum is in matrix form, and then the inverse fast Fourier transform (IFFT, inverse fast Fourier transform) is used to inverse transform and reconstruct the transient power spectrum to generate battery energy efficiency data. The data uses time as the horizontal axis. The battery energy efficiency data is input into the normal state derivation module. The normal state derivation adopts the support vector machine (SVM, support vector machine) algorithm for classification and derivation. The input of the model is the battery energy efficiency data, and the classification target is the normal power consumption state and the abnormal power consumption state. The derivation result is output as the power consumption feature data. The power consumption feature data includes information such as power consumption frequency, energy loss ratio and power consumption distribution mean. Then, the power consumption feature data is subjected to power consumption pattern recognition. The power consumption pattern recognition adopts the K-means clustering algorithm to cluster the data according to the power consumption characteristics. The output power consumption pattern data includes the characteristic values ​​of each category, the center point information and the power consumption pattern mark. The battery energy efficiency data and the power consumption pattern data are input into the charge state quantification module, and the Markov chain (Markov Chain, Markov chain) state transition model, quantizes and calculates the state transition probability of the data, and the quantized value of the state of charge is represented by the state transfer matrix. The input of the state of charge quantification model is energy efficiency and mode tag, and the output is normal state of charge data. The state of charge data includes the current charge ratio, future state prediction and charge efficiency change rate. The data format is in tabular form, and the fields include state tag, time interval and charge value. Finally, the generated state of charge data is stored as .xlsx file for input parameter preparation of SOC estimation process.

[0116] Preferably, step S32 includes the following steps:

[0117] Perform joint Fourier expansion on the steady-state voltage data of the group end and the steady-state power data of the lithium battery to obtain a joint harmonic power sequence; perform bandpass filtering on the joint harmonic power sequence to output fundamental frequency component data;

[0118] Performing phase correction on the fundamental frequency component data to obtain phase spectrum data; performing frequency band synthesis processing on the phase spectrum data to obtain power density data;

[0119] The spectral peaks of the joint harmonic power sequence are reorganized based on the power density data to generate a transient power spectrum;

[0120] Performing spectral line compensation processing on the transient power spectrum to obtain compensated power data; performing multi-dimensional reconstruction based on the compensated power data to generate reconstructed power data;

[0121] The reconstructed power data is mapped for power loss according to the basic parameters of the lithium battery pack to generate power loss data; efficiency calibration is performed based on the power loss data to produce battery energy efficiency.

[0122] In this embodiment, the steady-state voltage data of the group end and the steady-state power data of the lithium battery are input into the joint Fourier expansion module, and the discrete Fourier transform (DFT) algorithm is used to jointly expand the two data sets. The specific operation is to firstly perform amplitude normalization processing on the voltage and power data respectively, and limit the data range to [0,1]. Then, the two sets of data are superimposed into a joint harmonic signal by using the time domain to frequency domain conversion formula, and a joint harmonic power sequence is output. The joint harmonic power sequence is stored with frequency as the horizontal coordinate and power amplitude as the vertical coordinate.csv file, and input the joint harmonic power sequence into the bandpass filtering processing module. The filter uses a digital bandpass filter based on the second-order Butterworth filter design, and the bandpass frequency band is set to [10Hz, 100Hz]. The filter parameters include cutoff frequency, sampling frequency and filter coefficient. During processing, the high-frequency and low-frequency noise components in the sequence are eliminated, and the effective signal components within the fundamental frequency range are retained. The fundamental frequency component data output after filtering is stored in a two-dimensional matrix, in which the first column is the frequency and the second column is the amplitude. The fundamental frequency component data is input into the phase correction module, and the Hilbert transform technology is used to achieve phase compensation. The specific operation of phase correction is to calculate the instantaneous phase based on the filtered signal, compensate for the initial phase offset and recalibrate the phase value, generate phase spectrum data, and input the phase spectrum data into the frequency band synthesis processing module. The synthesis operation adopts The weighted superposition method is used to perform weighted calculation on the phases of different frequency bands. The weight parameters are determined by the energy distribution characteristics of the frequency bands. The power density data is generated by calculating the weighted sum of each frequency band. The power density data is stored in a three-column table format, including frequency, density value and corresponding frequency band information. The power density data is input into the spectrum peak reconstruction module. The peak detection algorithm is used to calibrate the main peak and secondary peak positions in the power spectrum. The main and secondary peak positions are selected to generate a new sequence to complete the spectrum peak reconstruction. The output transient power spectrum data contains frequency, amplitude and peak position parameters. The transient power spectrum data is input into the spectrum line compensation module. The compensation processing is based on the Lagrange interpolation method to complete the missing frequency bands in the power spectrum. The power gap value is calculated by the interpolation function and added to the original data. The compensated power data is input into the multidimensional reconstruction module and the singular value decomposition (SVD, Singular Value Decomposition) is used. The value decomposition method is used to perform multi-dimensional mapping of different frequency components of power data. The mapping parameters are determined by the frequency range and amplitude distribution. The reconstructed power data generated after processing is in a multi-dimensional matrix format. The rows and columns of the matrix represent the frequency and time dimensions respectively. The reconstructed power data and the basic parameters of the lithium battery pack are input into the power loss mapping module. The loss model is used to calculate the data point by point. The model is based on the distribution characteristics of heat loss and internal resistance loss. The output power loss data is in a tabular form, including time, power loss ratio and cumulative loss value. The power loss data is input into the efficiency calibration module. The power loss is normalized by using the efficiency calibration table. The generated battery energy efficiency data is stored as a two-dimensional array, in which the first column is the time point and the second column is the efficiency value. .

[0123] Preferably, step S4 comprises the following steps:

[0124] Step S41: performing Poisson distribution fitting on the normal state of charge to obtain load characteristic data;

[0125] Step S42: performing time series state transfer modeling on the load characteristic data to generate a state transfer matrix; performing sequence prediction on the load characteristic data according to the state transfer matrix to obtain simulated discharge data;

[0126] Step S43: performing thermodynamic modeling on the simulated discharge data based on the basic parameters of the lithium battery pack to obtain the temperature rise characteristics of the lithium battery; dynamically evolving the temperature rise characteristics of the lithium battery to generate temperature rise evolution data;

[0127] Step S44: reconstructing the lithium battery state based on the temperature rise evolution data to generate the battery pack discharge temperature rise condition.

[0128] In this embodiment, the normal state of charge data is input into the Poisson distribution fitting module. The Poisson distribution fitting is based on the maximum likelihood estimation method. The Poisson distribution parameter λ is determined by calculating the mean and variance of the power consumption data. The fitting function is written using a Python script, and the optimization function in the Scipy library is called for parameter estimation. The power consumption data is fitted with time as the horizontal axis and power consumption as the vertical axis. The load characteristic data is output, and the load characteristic data is input into the time series state transfer modeling module. The Markov Chain modeling method is used for time series state analysis. First, the load characteristic data is discretized into states, and the power consumption load value is divided into different intervals to generate a state set. Then, the transition frequency between each state is counted to construct a state transfer probability matrix. The size of the matrix depends on the number of state sets. The matrix is ​​normalized using the NumPy library, and the state transfer matrix is ​​output and stored in a JSON file in the form of a matrix. Then, the state transfer matrix and the load characteristic data are input into the sequence prediction module, and the Hidden Markov Model (HMM) is used. The simulated discharge data and basic parameters of the lithium battery pack are input into the thermodynamic modeling module, and the temperature rise characteristics of the battery are calculated using a thermodynamic model based on the Fourier heat conduction principle. The model involves heat source power, thermal conductivity coefficient and ambient temperature parameters. First, the simulated discharge data is converted into heat source input data in watts (W), and then the thermal characteristic parameters of the lithium battery, such as internal resistance, heat capacity and heat dissipation coefficient, are input. The thermodynamic modeling function is written through Python code, and the numerical integration function in the SciPy library is called to calculate the temperature rise. The temperature rise characteristic data of the lithium battery is output, including time, temperature increment and discharge stage identification. The temperature rise characteristic data of the lithium battery is input into the dynamic evolution module, and the temperature rise characteristic data is processed in multiple steps based on the time series dynamic evolution model. The dynamic evolution adopts the autoregressive integrated moving average model (ARIMA, Autoregressive Integrated Moving Average Average) is used to fit the temperature rise trend. First, the temperature rise characteristic data is tested for stability and differential operation is performed to remove the trend term. Then, the ARIMA function in the statsmodels library is called to fit the temperature rise sequence. The model parameters include AR (autoregressive order), I (difference order) and MA (sliding average order). After evolution, the temperature rise evolution data is output, including the predicted time point and the corresponding temperature rise value. The temperature rise evolution data is input into the lithium battery state reconstruction module, and the lithium battery state is calculated using the reconstruction algorithm based on the neural network. The specific implementation steps are as follows: first, the temperature rise evolution data is normalized, and the data range is mapped to [0,1], then the normalized data is input into the multilayer perceptron (MLP) neural network model. The network structure includes an input layer, two hidden layers and an output layer. The hidden layer activation function uses ReLU (linear rectification function), and the output layer activation function uses Sigmoid (logistic regression function). The network weights and biases are adjusted by the back propagation algorithm. The training process is based on the temperature rise evolution data and the real state label of the lithium battery. Finally, the battery pack discharge temperature rise status data is generated and stored in CSV format, including time points and discharge status values. The status values ​​reflect the discharge performance and temperature rise status of the lithium battery pack at different time points.

[0129] Preferably, step S5 comprises the following steps:

[0130] Step S51: collecting working environment parameters of the lithium battery pack; constructing a temperature field for the working environment parameters of the lithium battery pack to obtain a virtual working temperature field;

[0131] Step S52: performing attenuation modeling based on the virtual operating temperature field and the battery pack discharge temperature rise condition to generate a capacity attenuation model; performing discharge capacity attenuation prediction according to the capacity attenuation model to obtain a predicted attenuation condition;

[0132] Step S53: Calculating discharge error on basic discharge data of the lithium battery pack based on the predicted attenuation condition to obtain discharge error data;

[0133] Step S54: performing relative quantization processing on the discharge error data to generate error quantization data; performing feature fusion on the error quantization data to obtain discharge relative error data.

[0134] In this embodiment, a high-precision multi-channel data acquisition device is used to collect the working environment parameters of the lithium battery pack, including the ambient temperature, humidity, air flow velocity, and radiant heat intensity. Specifically, an industrial-grade temperature and humidity sensor and an infrared radiation intensity detector are used for real-time monitoring. The data acquisition frequency is set to once per second. The collected data is transmitted to the central processing unit through the RS485 interface for storage and processing. The central processing unit calls a dedicated physical modeling software (such as COMSOL Multiphysics) constructs the temperature field of the collected environmental parameters. During the construction process, a three-dimensional steady-state heat conduction model is selected. The boundary conditions and initial conditions are determined by inputting the collected temperature, air flow velocity, and thermal radiation intensity values. The temperature distribution is calculated using the finite element method, and the virtual working temperature field is output. The virtual working temperature field and the battery pack discharge temperature rise data are input into the attenuation modeling module. The capacity attenuation model is constructed using the life prediction model based on the Arrhenius equation. First, the virtual working temperature field and the discharge temperature rise data are interpolated and matched in the spatial dimension to ensure that the temperature data completely corresponds to the spatial position of each battery cell. Then, the attenuation rate curve is fitted based on the time characteristics of temperature change and discharge behavior. The attenuation rate model uses linear regression method for parameter estimation to generate a capacity attenuation model. The model coefficients and formula parameters are stored in JSON format. The capacity attenuation model is used to Discharge capacity attenuation prediction, prediction uses Monte Carlo simulation method, randomly generates discharge curves under different working conditions according to historical discharge curves, calculates the capacity loss of each curve in combination with the capacity attenuation model, outputs the predicted attenuation situation, inputs the predicted attenuation situation and the basic discharge data of the lithium battery pack into the error calculation module, and uses the error calculation method based on time series analysis to calculate the discharge error. First, the two sets of data are time-aligned to ensure the consistency of the time series, and then the difference between the actual discharge amount and the predicted discharge amount at each time point is calculated. The error value is recorded in absolute value form, and the operation function of the Pandas library in Python is used to implement batch calculation, output the discharge error data, and input the discharge error data into the relative quantization module. Quantization processing is performed based on the ratio of discharge error to actual discharge amount. First, the error data and actual discharge data are normalized, and the data range is mapped to [0,1] interval, the normalization method uses the minimum-maximum normalization (Min-Max Normalization), and then calculates the error ratio according to the normalized value to generate error quantization data, and records the time point and error ratio value in CSV file format. Then, the error quantization data is subjected to feature fusion processing. The feature fusion adopts the principal component analysis method (PCA, Principal Component Analysis). First, a multidimensional feature matrix containing error ratio, time series and other auxiliary features is constructed, and then the principal component features are extracted through singular value decomposition (SVD, Singular Value Decomposition), and finally the discharge relative error data is output. ,

[0135] Preferably, step S52 includes the following steps:

[0136] Perform temperature surface topological mapping on the virtual working temperature field data to obtain the temperature distribution characteristic matrix;

[0137] Performing drift mapping processing on the virtual working temperature field according to the temperature distribution characteristic matrix to generate a temperature drift template;

[0138] Based on the temperature drift template and the battery pack discharge temperature rise condition, the lithium battery capacity impact analysis is performed to obtain the lithium battery capacity change data; based on the lithium battery capacity change data, the attenuation model is performed to generate the capacity attenuation model;

[0139] Perform discharge temperature rise simulation according to the capacity decay model to obtain simulated lithium battery discharge temperature rise data;

[0140] Based on the capacity decay model, the discharge capacity decay is predicted for the simulated lithium battery discharge temperature rise data to obtain the predicted decay situation.

[0141] In this embodiment, three-dimensional temperature field data is used as input, and the temperature distribution data is converted into a two-dimensional temperature surface through the Surface Plot tool in MATLAB (matrix laboratory). The temperature value of each node is extracted from the three-dimensional space, and a contour map is drawn on the XY plane to represent the temperature change at different positions. The temperature field data will be interpolated to fill the data gaps. The interpolation method adopts bilinear interpolation to ensure the smoothness and accuracy of the temperature surface. The generated temperature distribution feature matrix is ​​stored in a matrix data format. Each row of the matrix corresponds to the temperature value of a spatial position. The temperature matrix is ​​translated using the image transformation algorithm provided in the Opencv library (open source computer vision library). First, a reference point is selected as a reference point, and the temperature matrix is ​​translated relative to the reference point. The drift amount is dynamically adjusted according to the change of the real-time working environment parameters. The translated data is used to obtain a temperature drift template through matrix operation. The generated template records the temperature change law of the battery pack under different working conditions. The temperature drift template is in the form of.The data are saved in json format. Each set of templates corresponds to the temperature drift of the battery pack under different environmental conditions. The discharge data of the battery pack is extracted from the battery management system (BMS). Combined with the temperature change information in the temperature drift template, the temperature change of the battery pack is weighted and calculated. A mathematical model between temperature and capacity change is established. The impact of temperature change on battery capacity is obtained by regression analysis (such as polynomial regression) using historical discharge data and real-time temperature data. The coefficient of temperature change on battery capacity is the fitting parameter in the regression analysis. Feature extraction is performed on the temperature change data and capacity change data to generate a feature vector. Each element in the feature vector represents a capacity change corresponding to a temperature change. The sklearn library in Python is used to train the random forest regression model. 50 trees are selected for training. Finally, a mathematical model of battery capacity attenuation is obtained. The model can predict the change of battery capacity according to real-time temperature changes. Based on the existing battery discharge curve and capacity attenuation model, MATLAB is used to simulate the discharge process of the battery. First, the initial state of the battery pack (SOC, State of Charge), and then perform discharge simulation based on the initial temperature of the battery and the discharge power of the battery pack. During the simulation, the capacity decay model is used to calculate the battery capacity that changes with time. The simulation results include the temperature change and capacity change at each time node during the battery discharge process. The simulation data is simulated multiple times according to the different battery discharge powers to obtain the discharge temperature rise data under different conditions. The simulated discharge temperature rise data is combined with the capacity decay model to predict the capacity decay of each set of discharge data. The prediction process calculates the remaining capacity of the battery at each time point by applying the trained decay model. The prediction results are calculated through time series to obtain the final battery capacity loss data, and finally the predicted decay situation is obtained. The decay prediction results are output in JSON format, including the expected capacity change and corresponding temperature information at each time node. .

[0142] Preferably, step S6 comprises the following steps:

[0143] Step S61: performing iterative distribution analysis on the discharge relative error data to obtain an error distribution spectrum;

[0144] Step S62: estimating and correcting the normal state of charge based on the error distribution spectrum to generate a corrected state of charge;

[0145] Step S63: performing state boundary detection on the corrected state of charge to obtain a state threshold range; performing feedback compensation on the corrected state of charge based on the state threshold range to obtain calibration power data;

[0146] Step S64: quantify the accuracy based on the calibration power data to obtain the SOC output result to execute the intelligent SOC estimation method of the lithium battery pack.

[0147] In this embodiment, error data during battery discharge is extracted from a lithium battery management system (BMS, Battery Management System), and the data includes a relative error value at each time node during the battery discharge process. The error data is preliminarily analyzed using the Histogram function in MATLAB, and an error distribution diagram is drawn. An iterative analysis is performed based on the distribution characteristics of the error data, and a suitable error distribution model is selected. The error distribution is fitted using the maximum likelihood estimation (MLE) method to obtain an error distribution spectrum. The error distribution spectrum describes the variation law of the error during battery discharge under different working conditions. The output of the error distribution spectrum is a data set containing different error intervals and corresponding probability densities. The state of charge is corrected using the error distribution spectrum according to the discharge data of the battery pack under normal state of charge. The correction process matches the error value in the error distribution spectrum with the battery discharge curve, and uses the weighted average method to correct the state of charge of the battery pack. The correction weight is determined by the error distribution value in the error distribution spectrum. The corrected state of charge is weightedly calculated according to the probability density of different intervals in the error distribution spectrum to obtain a corrected state of charge. The corrected state of charge reflects the actual capacity state of the battery during the current discharge process. The upper and lower limits of the corrected state of charge are determined. Boundary, boundary detection is to cluster the data of the corrected state of charge, use the K-means clustering algorithm to group the corrected state of charge, divide the corrected state of charge data into multiple state intervals, calculate the upper and lower limits of each interval as the state threshold range, and the K value of the K-means algorithm is determined by the cross-validation method. The determined state threshold range represents the variable range of the battery capacity under the current battery discharge state. The result of the state threshold range is used for subsequent feedback compensation. The corrected state of charge is compared with the state threshold range, and the adaptive control algorithm (AdaptiveControl) is used to compensate the corrected state of charge. The compensation process is adjusted by comparing the deviation between the corrected state of charge and the state threshold range. The deviation value is calculated according to the difference between the actual corrected state of charge and the target state threshold. When the deviation value is greater than a certain set value, linear feedback compensation is performed to adjust the corrected state of charge to within the target range, and finally generate the calibrated power data. The difference analysis of the calibrated power data is performed by calculating the difference between the calibrated power and the predicted power at each time node, and using the root mean square error (RMSE, Root The accuracy of the calibration power data is calculated by using the Mean Square Error method. The accuracy quantification process compares the calibration power data with the historical power data, calculates the error value, and obtains the accuracy correction coefficient of the battery pack through cumulative error analysis. The accuracy correction coefficient reflects the estimation accuracy of the battery pack. Finally, the accuracy correction coefficient is applied to the SOC estimation model to obtain an accurate SOC estimation result.The estimation result is the final output of the intelligent SOC estimation method of the lithium battery pack and is used to guide the operation of the battery management system and monitor the battery health status.

[0148] The present invention also provides a SOC estimation system for a lithium battery pack, which is used to execute the SOC estimation method for a lithium battery pack as described above. The SOC estimation system for a lithium battery pack comprises:

[0149] The parameter acquisition module is used to collect basic parameters of lithium battery packs and battery pack terminal voltage data; perform singular value decomposition on the battery pack terminal voltage data to obtain principal component voltage data; perform time domain reconstruction processing on the principal component voltage data to generate group terminal steady-state voltage data;

[0150] The parameter mapping module is used to perform multi-level feature mapping on the basic parameters of the lithium battery pack to obtain a parameter association matrix; the parameter association matrix is ​​used to reconstruct the battery pack performance to generate basic discharge data of the lithium battery pack;

[0151] The power derivation module is used to deconstruct the battery power of the steady-state voltage data at the group end based on the basic discharge data of the lithium battery group to generate the battery energy efficiency; the normal power consumption of the battery energy efficiency is deduced to obtain the normal state of charge;

[0152] The discharge simulation module is used to perform discharge simulation according to the normal state of charge to obtain simulated discharge data; the discharge temperature rise calculation is performed on the simulated discharge data based on the basic parameters of the lithium battery pack to generate the discharge temperature rise status of the battery pack;

[0153] The capacity attenuation prediction module is used to collect the working environment parameters of the lithium battery pack; the discharge capacity attenuation is predicted according to the working environment parameters of the lithium battery pack and the discharge temperature rise of the battery pack to obtain the predicted attenuation situation; the relative error of the basic discharge data of the lithium battery pack is calculated based on the predicted attenuation situation to obtain the discharge relative error data;

[0154] The charge correction module is used to estimate and correct the normal state of charge based on the discharge relative error data to generate a corrected state of charge; and perform power self-calibration processing according to the corrected state of charge to execute the intelligent SOC estimation method of the lithium battery pack.

[0155] The present invention realizes efficient acquisition of basic parameters and battery pack terminal voltage data through a parameter acquisition module through a SOC estimation system based on a lithium battery pack, providing a reliable data source for subsequent analysis. The steady-state voltage data generated by singular value decomposition and time domain reconstruction processing enhances the accuracy of battery state evaluation. The multi-level feature mapping and parameter association matrix generation of the parameter mapping module provide a clear structure for battery pack performance reconstruction. The generation of basic discharge data lays a foundation for performance analysis. The energy efficiency analysis of the power derivation module provides a basis for battery optimization use. The normal power consumption derivation result provides a reference for battery state monitoring. The combination of discharge simulation and temperature rise calculation improves the reliability of the discharge process. The capacity decay prediction module realizes scientific battery life evaluation in combination with working environment parameters. The relative error calculation based on the predicted decay situation provides support for discharge performance determination. The intelligent self-calibration processing of the charge correction module improves the management level of the lithium battery pack, and improves the performance monitoring, state evaluation and management efficiency of the lithium battery pack as a whole.

[0156] Therefore, the embodiments should be regarded as illustrative and non-restrictive from all points, and the scope of the present invention is limited by the appended claims rather than the above description, and it is therefore intended that all changes falling within the meaning and range of equivalent elements of the application documents are included in the present invention.

[0157] The above description is only a specific embodiment of the present invention, so that those skilled in the art can understand or implement the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but should conform to the widest scope consistent with the principles and novel features invented herein.

Claims

1. A method for estimating the SOC of a lithium battery pack, characterized in that: The following steps are involved: Step S1: Collect basic parameters of the lithium battery pack and battery pack terminal voltage data; perform singular value decomposition on the battery pack terminal voltage data to obtain principal component voltage data; perform time domain reconstruction processing on the principal component voltage data to generate group terminal steady-state voltage data; Step S2: Perform multi-level feature mapping on basic parameters of the lithium battery pack to obtain a parameter correlation matrix; Reconstruct the battery pack performance by using the parameter correlation matrix to generate basic discharge data for the lithium battery pack; Step S3: Based on the basic discharge data of the lithium battery pack, the battery power is decomposed on the steady-state voltage data of the group end to generate the battery energy efficiency; Derivation of normal power consumption of battery energy efficiency to obtain normal state of charge; Step S4: performing a discharge simulation according to a normal state of charge to obtain simulated discharge data; performing a discharge temperature rise calculation on the simulated discharge data based on basic parameters of the lithium battery pack to generate a discharge temperature rise condition of the battery pack; Step S5: collecting working environment parameters of the lithium battery pack; predicting the discharge capacity attenuation according to the working environment parameters of the lithium battery pack and the discharge temperature rise of the battery pack to obtain the predicted attenuation; performing relative error calculation on the basic discharge data of the lithium battery pack based on the predicted attenuation to obtain the discharge relative error data; Step S6: estimating and correcting the normal state of charge based on the discharge relative error data to generate a corrected state of charge; A power self-calibration process is performed according to the corrected state of charge to implement an intelligent SOC estimation method for a lithium battery pack.

2. The SOC estimation method of a lithium battery pack according to claim 1, characterized in that: Step S1 includes the following steps: Step S11: collecting basic parameters of the lithium battery pack and battery pack terminal voltage data; performing wavelet denoising on the battery pack terminal voltage data to obtain pure voltage data; Step S12: performing matrix decomposition on the pure voltage data to generate a voltage characteristic matrix; performing singular value screening on the voltage characteristic matrix to obtain principal component voltage data; Step S13: Perform Fourier reconstruction on the main component voltage data to obtain frequency domain feature data; perform inverse transformation mapping on the frequency domain feature data to generate a voltage time domain sequence; Step S14: performing smoothing and filtering processing on the voltage time domain sequence to obtain group-end steady-state voltage data.

3. The SOC estimation method of a lithium battery pack according to claim 1, characterized in that: Step S2 includes the following steps: Step S21: performing hierarchical clustering on basic parameters of the lithium battery pack to obtain parameter hierarchical data; performing feature extraction on the parameter hierarchical data to generate feature vector data; Step S22: quantifying the correlation of basic parameters of the lithium battery pack based on the characteristic vector data to obtain a parameter correlation matrix; Step S23: Performing performance projection according to the parameter association matrix to generate battery performance characteristics; performing battery capacity mapping on the battery performance characteristics to generate discharge capacity data; Step S24: feature fusion is performed on the discharge capacity data to obtain basic discharge data of the lithium battery pack.

4. The SOC estimation method of a lithium battery pack according to claim 1, characterized in that: Step S3 includes the following steps: Step S31: dynamically filter and transform the basic discharge data of the lithium battery pack to obtain the steady-state power of the lithium battery; Step S32: performing joint harmonic decomposition on the steady-state power of the lithium battery based on the steady-state voltage data at the group end to generate a transient power spectrum; performing inverse transformation and reconstruction on the transient power spectrum to generate battery energy efficiency; Step S33: deriving the battery normal state from the battery energy efficiency to obtain power consumption characteristic data; identifying the power consumption pattern from the power consumption characteristic data to obtain the power consumption pattern data; Step S34: quantifying the state of charge of the power consumption mode data based on the battery energy efficiency to generate a normal state of charge.

5. The SOC estimation method of a lithium battery pack according to claim 4, characterized in that: Step S32 includes the following steps: Perform joint Fourier expansion on the steady-state voltage data of the group end and the steady-state power data of the lithium battery to obtain a joint harmonic power sequence; perform bandpass filtering on the joint harmonic power sequence to output fundamental frequency component data; Performing phase correction on the fundamental frequency component data to obtain phase spectrum data; performing frequency band synthesis processing on the phase spectrum data to obtain power density data; The spectral peaks of the joint harmonic power sequence are reorganized based on the power density data to generate a transient power spectrum; Performing spectral line compensation processing on the transient power spectrum to obtain compensated power data; performing multi-dimensional reconstruction based on the compensated power data to generate reconstructed power data; The reconstructed power data is mapped for power loss according to the basic parameters of the lithium battery pack to generate power loss data; efficiency calibration is performed based on the power loss data to produce battery energy efficiency.

6. The SOC estimation method of a lithium battery pack according to claim 1, characterized in that: Step S4 includes the following steps: Step S41: performing Poisson distribution fitting on the normal state of charge to obtain load characteristic data; Step S42: performing time series state transfer modeling on the load characteristic data to generate a state transfer matrix; performing sequence prediction on the load characteristic data according to the state transfer matrix to obtain simulated discharge data; Step S43: performing thermodynamic modeling on the simulated discharge data based on the basic parameters of the lithium battery pack to obtain the temperature rise characteristics of the lithium battery; dynamically evolving the temperature rise characteristics of the lithium battery to generate temperature rise evolution data; Step S44: reconstructing the lithium battery state based on the temperature rise evolution data to generate the battery pack discharge temperature rise condition.

7. The SOC estimation method of a lithium battery pack according to claim 1, characterized in that: Step S5 includes the following steps: Step S51: collecting working environment parameters of the lithium battery pack; constructing a temperature field for the working environment parameters of the lithium battery pack to obtain a virtual working temperature field; Step S52: performing attenuation modeling based on the virtual operating temperature field and the battery pack discharge temperature rise condition to generate a capacity attenuation model; performing discharge capacity attenuation prediction according to the capacity attenuation model to obtain a predicted attenuation condition; Step S53: Calculating discharge error on basic discharge data of the lithium battery pack based on the predicted attenuation condition to obtain discharge error data; Step S54: performing relative quantization processing on the discharge error data to generate error quantization data; performing feature fusion on the error quantization data to obtain discharge relative error data.

8. The SOC estimation method of a lithium battery pack according to claim 7, characterized in that: Step S52 includes the following steps: Perform temperature surface topological mapping on the virtual working temperature field data to obtain the temperature distribution characteristic matrix; Performing drift mapping processing on the virtual working temperature field according to the temperature distribution characteristic matrix to generate a temperature drift template; Based on the temperature drift template and the battery pack discharge temperature rise condition, the lithium battery capacity impact analysis is performed to obtain the lithium battery capacity change data; based on the lithium battery capacity change data, the attenuation model is performed to generate the capacity attenuation model; Perform discharge temperature rise simulation according to the capacity decay model to obtain simulated lithium battery discharge temperature rise data; Based on the capacity decay model, the discharge capacity decay is predicted for the simulated lithium battery discharge temperature rise data to obtain the predicted decay situation.

9. The SOC estimation method of a lithium battery pack according to claim 1, characterized in that: Step S6 includes the following steps: Step S61: performing iterative distribution analysis on the discharge relative error data to obtain an error distribution spectrum; Step S62: estimating and correcting the normal state of charge based on the error distribution spectrum to generate a corrected state of charge; Step S63: performing state boundary detection on the corrected state of charge to obtain a state threshold range; performing feedback compensation on the corrected state of charge based on the state threshold range to obtain calibration power data; Step S64: quantify the accuracy based on the calibration power data to obtain the SOC output result to execute the intelligent SOC estimation method of the lithium battery pack.

10. A SOC estimation system for a lithium battery pack, characterized in that: For executing the SOC estimation method of the lithium battery pack as claimed in claim 1, the SOC estimation system of the lithium battery pack comprises: The parameter acquisition module is used to collect basic parameters of lithium battery packs and battery pack terminal voltage data; perform singular value decomposition on the battery pack terminal voltage data to obtain principal component voltage data; perform time domain reconstruction processing on the principal component voltage data to generate group terminal steady-state voltage data; The parameter mapping module is used to perform multi-level feature mapping on the basic parameters of the lithium battery pack to obtain a parameter association matrix; the parameter association matrix is ​​used to reconstruct the battery pack performance to generate basic discharge data of the lithium battery pack; The power derivation module is used to deconstruct the battery power of the steady-state voltage data at the group end based on the basic discharge data of the lithium battery group to generate the battery energy efficiency; the normal power consumption of the battery energy efficiency is deduced to obtain the normal state of charge; The discharge simulation module is used to perform discharge simulation according to the normal state of charge to obtain simulated discharge data; the discharge temperature rise calculation is performed on the simulated discharge data based on the basic parameters of the lithium battery pack to generate the discharge temperature rise status of the battery pack; The capacity attenuation prediction module is used to collect the working environment parameters of the lithium battery pack; the discharge capacity attenuation is predicted according to the working environment parameters of the lithium battery pack and the discharge temperature rise of the battery pack to obtain the predicted attenuation situation; the relative error of the basic discharge data of the lithium battery pack is calculated based on the predicted attenuation situation to obtain the discharge relative error data; The charge correction module is used to estimate and correct the normal state of charge based on the discharge relative error data to generate a corrected state of charge; and perform power self-calibration processing according to the corrected state of charge to execute the intelligent SOC estimation method of the lithium battery pack.

Images