Intelligent Monitoring and Management Method of Battery SOC Based on Precise Power Algorithm

Abstract

This invention relates to a battery SOC intelligent monitoring and management method based on a precise power algorithm, belonging to the field of battery power generation management technology. It includes real-time acquisition of battery voltage, current, temperature, and historical data; calculation of the initial SOC value using the ampere-hour integration method; in the charging state and when SOC ≥ 90%, a first correction factor is generated based on voltage hysteresis characteristics and temperature coefficient, dynamically reducing the ampere-hour integration weight; in the discharging state and when SOC ≤ 20%, a second correction factor is generated by combining polarization voltage and temperature decay coefficient, increasing the Kalman filter weight; the correction factors are fused through a gradient descent optimization algorithm and closed-loop feedback is used to output the calibrated SOC value. This invention solves the problems of large estimation errors in extreme SOC ranges and poor adaptability of traditional methods by using dynamic weight allocation, a nonlinear compensation model, and a closed-loop self-optimization mechanism, significantly improving the accuracy and reliability of battery management.

Description

Technical Field

[0001] This invention relates to the field of battery power generation management technology, and in particular to a method for intelligent monitoring and management of battery SOC based on a precise power algorithm. Background Technology

[0002] In the field of battery SOC estimation, traditional methods typically combine ampere-hour integration with filtering algorithms. However, these methods face challenges such as error accumulation due to capacity decay, temperature sensitivity, and nonlinear characteristics in extreme SOC ranges (e.g., voltage hysteresis, polarization effects). Existing solutions, with their fixed-weight algorithm fusion, struggle to dynamically adapt to battery aging and operating condition fluctuations. They also lack sufficient compensation for coupling distortions in high / low SOC ranges, and their open-loop calibration mechanisms lack real-time feedback capabilities, leading to a significant decrease in estimation accuracy in extreme scenarios. Achieving dynamic weight allocation, accurate compensation for nonlinear characteristics, and closed-loop self-optimization remain critical bottlenecks that require significant technological breakthroughs. Summary of the Invention

[0003] To achieve the above objectives, the present invention provides a battery SOC intelligent monitoring and management method based on a precise power algorithm, comprising the following steps:

[0004] The battery's voltage, current, and temperature are acquired in real time, and the initial SOC value of the battery is calculated using the ampere-hour integration method in combination with the battery's historical charge and discharge data.

[0005] A dynamic weight allocation algorithm is used to adjust the calculation weights of the ampere-hour integration method and the Kalman filter algorithm respectively, thereby correcting the initial SOC value, wherein:

[0006] When the battery is in a charging state and the initial SOC value is ≥90%, a voltage hysteresis characteristic compensation model is introduced, and the product of the battery's temperature coefficient and voltage change rate is used as the first correction factor to adjust the weight of the ampere-hour integral method to the first preset value.

[0007] When the battery is in a discharged state and the initial SOC value is ≤20%, the polarization voltage compensation model is activated. The second correction factor is generated by combining the nonlinear relationship between the current change gradient and the temperature decay coefficient, and the weight of the Kalman filter algorithm is increased to the second preset value.

[0008] The first correction factor and the second correction factor are dynamically fused using a gradient descent optimization algorithm to generate a calibrated SOC value, which is then fed back to the power prediction module of the battery management system.

[0009] Furthermore, the step of acquiring the battery's voltage, current, and temperature in real time, and calculating the battery's initial SOC value using the ampere-hour integration method in conjunction with the battery's historical charge and discharge data, includes:

[0010] The system dynamically filters out outliers in real-time sampled voltage, current, and temperature data, and uses a sliding time window to perform standard deviation analysis on continuous sampled data to remove data that exceeds the preset fluctuation range and then performs interpolation processing.

[0011] Based on the number of cycles, average depth of discharge, and capacity decay trend in the historical charge and discharge data, nonlinear capacity decay data is generated, wherein the actual capacity is correlated with the square root of the number of cycles and the square of the average depth of discharge.

[0012] Based on the current temperature threshold and the direction of charging and discharging current, the corresponding ampere-hour integral calculation mode is dynamically selected: a temperature compensation coefficient is introduced to correct the nominal capacity under low temperature discharge conditions, and the nominal capacity is directly used for calculation under normal temperature charging conditions. The calculation results of the corresponding interval are then fused through a weighting coefficient.

[0013] Further, the step of generating nonlinear capacity decay data based on the number of cycles, average depth of discharge, and capacity decay trend in the historical charge-discharge data includes:

[0014] The number of cycles and the average depth of discharge of the battery are extracted as input variables. A nonlinear function is constructed based on historical capacity decay data, in which the actual capacity value is inversely proportional to the square root of the number of cycles and directly proportional to the square of the average depth of discharge.

[0015] The attenuation coefficient in the nonlinear function relationship is calibrated according to the battery type, wherein the attenuation coefficient includes a first coefficient reflecting the cycle aging rate and a second coefficient reflecting the effect of deep discharge.

[0016] Substituting the calibrated attenuation coefficient, the real-time acquired number of cycles, and the average depth of discharge into the nonlinear function relationship, the current actual capacity value is dynamically calculated and used as the capacity reference parameter for the ampere-hour integration method to obtain nonlinear capacity attenuation data.

[0017] Furthermore, when the battery is in a charging state and the initial SOC value is ≥90%, a voltage hysteresis characteristic compensation model is introduced, using the product of the battery's temperature coefficient and voltage change rate as a first correction factor to adjust the weight of the ampere-hour integral method to a first preset value. This step includes:

[0018] During charging, the voltage hysteresis curve is dynamically calibrated based on the battery's charging and discharging history data. The voltage hysteresis curve reflects the voltage difference between the charging and discharging paths within the current cycle.

[0019] When the initial SOC value is ≥90%, the charging voltage value is collected in real time and compared with the calibrated charging path hysteresis curve. If the difference between the current voltage and the calibrated curve exceeds the preset threshold, the compensation method is triggered.

[0020] The compensation method dynamically shifts and corrects the hysteresis curve based on the magnitude and direction of the deviation difference, combined with real-time temperature data. The temperature coefficient is calibrated by hysteresis voltage offset experiments of the battery at different temperatures.

[0021] The difference between the corrected hysteresis curve voltage and the real-time voltage is used as the basic compensation amount, multiplied by the temperature coefficient to generate the first correction factor, and the weight of the ampere-hour integration method is dynamically adjusted to the first preset value according to the numerical range of the first correction factor.

[0022] Furthermore, when the battery is in a discharged state and the initial SOC value is ≤20%, the polarization voltage compensation model is activated, and a second correction factor is generated by combining the nonlinear relationship between the current change gradient and the temperature decay coefficient. The step of increasing the weights of the Kalman filter algorithm to a second preset value includes:

[0023] The initial SOC value is continuously monitored during discharge. When the initial SOC value drops to 20% or below, the polarization voltage dynamic compensation mode is activated.

[0024] Real-time acquisition of discharge current data and calculation of current change gradient per unit time; at the same time, matching a pre-calibrated temperature decay coefficient based on battery temperature and historical discharge data, wherein the temperature decay coefficient reflects the non-linear growth characteristic of battery internal resistance as temperature decreases.

[0025] The current change gradient and temperature decay coefficient are input into a predefined nonlinear coupling function to generate a polarization voltage compensation amount. The nonlinear coupling function satisfies the following condition: when the current change gradient increases or the temperature decay coefficient increases, the compensation amount grows exponentially.

[0026] Based on the absolute value range of the polarization voltage compensation, the weights of the Kalman filter algorithm are progressively increased from the initial value to the second preset value in segments.

[0027] Further, the step of dynamically fusing the first correction factor and the second correction factor using a gradient descent optimization algorithm to generate a calibrated SOC value, and feeding the SOC value back to the power prediction module of the battery management system, includes:

[0028] Construct a fusion error function with the first correction factor and the second correction factor as input variables. The error function includes the absolute error between the current SOC estimate and the reference value, as well as the sliding window statistics of historical error data.

[0029] The weight allocation of the first correction factor and the second correction factor is iteratively optimized based on the gradient descent optimization algorithm. In each iteration, the weight coefficients are adjusted according to the negative gradient direction of the error function until the error converges to a preset threshold range.

[0030] The compensation amounts of the first correction factor and the second correction factor are dynamically weighted and fused according to the optimized weight coefficients to generate the calibrated SOC value, and the weight coefficients of the current iteration and the error convergence status are fed back to the parameter self-learning module of the battery management system.

[0031] Furthermore, the step of iteratively optimizing the weight allocation of the first correction factor and the second correction factor based on the gradient descent optimization algorithm includes:

[0032] Set initial weight coefficients, where the initial weight of the first correction factor is ω1, the initial weight of the second correction factor is ω2, and ω1+ω2=1;

[0033] The output value of the fusion error function is calculated based on the current weight coefficients, and the partial derivatives of the error function with respect to ω1 and ω2 are calculated to determine the negative gradient direction of the error function.

[0034] The weight coefficients are updated along the negative gradient direction according to the preset learning rate, such that ω1'=ω1-μ·∂E / ∂ω1, ω2'=ω2-μ·∂E / ∂ω2, where μ is the learning rate, and ∂E / ∂ω1 and ∂E / ∂ω2 are the partial derivatives of the error function with respect to the weight coefficients, respectively.

[0035] Determine whether the updated error function value is less than the preset convergence threshold. If it is not met, repeat the steps until the error converges.

[0036] When the error converges, the current weight coefficient is locked as the optimal weight allocation scheme, and the historical iteration data is stored in the optimization parameter database of the battery management system.

[0037] This invention also proposes a battery SOC intelligent monitoring and management device based on a precise power algorithm, comprising:

[0038] The acquisition unit is used to acquire the battery's voltage, current, and temperature in real time, and calculate the battery's initial SOC value using the ampere-hour integration method in combination with the battery's historical charge and discharge data.

[0039] The calculation unit is used to adjust the calculation weights of the ampere-hour integration method and the Kalman filter algorithm respectively using a dynamic weight allocation algorithm, thereby correcting the initial SOC value, wherein:

[0040] When the battery is in a charging state and the initial SOC value is ≥90%, a voltage hysteresis characteristic compensation model is introduced, and the product of the battery's temperature coefficient and voltage change rate is used as the first correction factor to adjust the weight of the ampere-hour integral method to the first preset value.

[0041] When the battery is in a discharged state and the initial SOC value is ≤20%, the polarization voltage compensation model is activated. The second correction factor is generated by combining the nonlinear relationship between the current change gradient and the temperature decay coefficient, and the weight of the Kalman filter algorithm is increased to the second preset value.

[0042] The feedback unit is used to dynamically fuse the first correction factor and the second correction factor through the gradient descent optimization algorithm to generate a calibrated SOC value, and feed the SOC value back to the power prediction module of the battery management system.

[0043] The present invention also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the above-described intelligent monitoring and management method for battery SOC based on a precise power algorithm.

[0044] The present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the above-described intelligent monitoring and management method for battery SOC based on a precise power algorithm.

[0045] The battery SOC intelligent monitoring and management method based on a precise power algorithm provided by this invention has the following beneficial effects:

[0046] (1) By using dynamic weight allocation and nonlinear voltage hysteresis / polarization compensation model, the estimation error in the high / low SOC range (≥90% or ≤20%) is significantly reduced, and the accuracy under extreme scenarios is improved;

[0047] (2) The algorithm weights and compensation parameters are dynamically adjusted based on the battery aging state and real-time operating conditions (temperature and current change rate) to adapt to different battery degradation levels and complex environmental fluctuations.

[0048] (3) By fusing multiple correction factors through the gradient descent algorithm and feeding them back to the self-learning module, dynamic calibration of SOC estimation parameters is achieved, reducing the error accumulation rate during long-term use;

[0049] (4) Dynamic cleaning of abnormal data and nonlinear modeling of capacity decay reduce the input error of the ampere-hour integration method and improve the accuracy of the initial SOC value benchmark. Attached Figure Description

[0050] Figure 1 This is a flowchart illustrating a method for intelligent monitoring and management of battery SOC based on a precise power algorithm in one embodiment of the present invention.

[0051] Figure 2 This is a structural block diagram of a battery SOC intelligent monitoring and management device based on a precise power algorithm according to an embodiment of the present invention;

[0052] Figure 3 This is a schematic block diagram of the structure of a computer device according to an embodiment of the present invention.

[0053] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0054] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0055] Reference Figure 1 This is a flowchart illustrating a battery SOC intelligent monitoring and management method based on a precise power algorithm proposed in this invention. The method includes the following steps:

[0056] S1, real-time acquisition of battery voltage, current and temperature, and calculation of the initial SOC value of the battery using the ampere-hour integration method in combination with the battery's historical charge and discharge data;

[0057] S2, a dynamic weight allocation algorithm is used to adjust the calculation weights of the ampere-hour integration method and the Kalman filter algorithm respectively, thereby correcting the initial SOC value, wherein:

[0058] S21, when the battery is in a charging state and the initial SOC value is ≥90%, a voltage hysteresis characteristic compensation model is introduced, and the product of the battery's temperature coefficient and voltage change rate is used as the first correction factor to adjust the weight of the ampere-hour integral method to the first preset value.

[0059] S22, when the battery is in a discharge state and the initial SOC value is ≤20%, activate the polarization voltage compensation model, combine the nonlinear relationship between the current change gradient and the temperature decay coefficient to generate a second correction factor, and increase the weight of the Kalman filter algorithm to the second preset value.

[0060] S3, dynamically fuse the first correction factor and the second correction factor through the gradient descent optimization algorithm to generate a calibrated SOC value, and feed the SOC value back to the power prediction module of the battery management system.

[0061] In one embodiment, the step of acquiring the battery's voltage, current, and temperature in real time, and calculating the battery's initial SOC value using the ampere-hour integration method in conjunction with the battery's historical charge and discharge data, includes:

[0062] The system dynamically filters out outliers in real-time sampled voltage, current, and temperature data, and uses a sliding time window to perform standard deviation analysis on continuous sampled data to remove data that exceeds the preset fluctuation range and then performs interpolation processing.

[0063] Based on the number of cycles, average depth of discharge, and capacity decay trend in the historical charge and discharge data, nonlinear capacity decay data is generated, wherein the actual capacity is correlated with the square root of the number of cycles and the square of the average depth of discharge.

[0064] Based on the current temperature threshold and the direction of charging and discharging current, the corresponding ampere-hour integral calculation mode is dynamically selected: a temperature compensation coefficient is introduced to correct the nominal capacity under low temperature discharge conditions, and the nominal capacity is directly used for calculation under normal temperature charging conditions. The calculation results of the corresponding interval are then fused through a weighting coefficient.

[0065] In the specific implementation process, the standard deviation analysis for the sliding window is as follows:

[0066] The sliding time window length is set to N=10 (corresponding to a 100ms sampling period). The window stores 10 consecutive sets of raw sampled values ​​for voltage U, current I, and temperature T. The standard deviation is calculated using the following formula: ,in, For the k-th sampled value within the window, This represents the mean of the data within the window. This represents the standard deviation of the data within the window.

[0067] The method for identifying and handling outliers in the sampled voltage, current, and temperature data is as follows:

[0068] If the current sample value satisfy If it is an outlier, it is determined to be an outlier. The formula shows that it needs to be removed. The collected data is then interpolated; this process can be understood as interpolation compensation. According to this formula, interpolation is performed using... The outlier is replaced by the average of the two normal sample values.

[0069] Case 1: The impact of different window lengths on data quality;

[0070]

[0071] The generation of nonlinear capacity decay data is essentially the construction of a nonlinear capacity decay model. Specifically, it involves extracting the battery's cycle count and average depth of discharge as input variables, constructing a nonlinear function based on historical capacity decay data, where the actual capacity value is inversely proportional to the square root of the cycle count and directly proportional to the square of the average depth of discharge. The decay coefficient in the nonlinear function is calibrated according to the battery type, and this decay coefficient includes a first coefficient reflecting the cycle aging rate and a second coefficient reflecting the effect of deep discharge. The calibrated decay coefficient, along with the real-time acquired cycle count and average depth of discharge, are substituted into the nonlinear function to dynamically calculate the current actual capacity value, which is then used as the capacity reference parameter for the ampere-hour integration method to obtain the nonlinear capacity decay data.

[0072] The model is constructed using the formula In the formula, This indicates the battery's nominal capacity at the time of manufacture (e.g., 50Ah). This indicates the cumulative number of charge-discharge cycles of the battery. This represents the historical average depth of discharge.

[0073] Case 2: Comparison of Capacity Decay Model Fitting

[0074]

[0075] In conclusion, through Reflecting the decreasing rate of cyclic aging (rapid decay in the early stages, slowing down in the later stages), through... Enhance the accelerated decay effect of deep discharge.

[0076] Based on the current temperature threshold and the direction of charging and discharging current, the corresponding ampere-hour integral calculation mode is dynamically selected. Under low-temperature discharge conditions, a temperature compensation coefficient is introduced to correct the nominal capacity; under normal-temperature charging conditions, the nominal capacity is directly used for calculation. The calculation results for corresponding intervals are then fused using weighting coefficients. The calculation formula is dynamically switched as follows:

[0077] Low temperature discharge mode in, This represents the Coulomb efficiency, which is 0.98. This represents the temperature compensation coefficient, which is 0.005 / ℃. This represents the absolute value of the ambient temperature.

[0078] room temperature charging mode

[0079] The above-mentioned weighting coefficient fusion calculation refers to the calculation using weighting coefficient ω when the temperature is in the transition range (-5℃ < T ≤ 5℃). The weighting coefficient ω is distributed linearly according to temperature. .

[0080] Case 3: Comparison of initial SOC calculation errors at different temperatures

[0081]

[0082] In one embodiment, when the battery is in a charging state and the initial SOC value is ≥90%, the step of introducing a voltage hysteresis characteristic compensation model and using the product of the battery's temperature coefficient and voltage change rate as a first correction factor to adjust the weight of the ampere-hour integral method to a first preset value includes:

[0083] During charging, the voltage hysteresis curve is dynamically calibrated based on the battery's charging and discharging history data. The voltage hysteresis curve reflects the voltage difference between the charging and discharging paths within the current cycle.

[0084] When the initial SOC value is ≥90%, the charging voltage value is collected in real time and compared with the calibrated charging path hysteresis curve. If the difference between the current voltage and the calibrated curve exceeds the preset threshold, the compensation method is triggered.

[0085] The compensation method dynamically shifts and corrects the hysteresis curve based on the magnitude and direction of the deviation difference, combined with real-time temperature data. The temperature coefficient is calibrated by hysteresis voltage offset experiments of the battery at different temperatures.

[0086] The difference between the corrected hysteresis curve voltage and the real-time voltage is used as the basic compensation amount, multiplied by the temperature coefficient to generate the first correction factor, and the weight of the ampere-hour integration method is dynamically adjusted to the first preset value according to the numerical range of the first correction factor.

[0087] In the specific implementation process, during the dynamic calibration of the voltage hysteresis curve based on the battery's historical charge and discharge data, the historical data acquisition method is as follows: extract the voltage SOC data from the most recent K=5 complete charge and discharge cycles, and record the voltage values ​​for the charging (Chg) and discharging (Dchg) paths respectively; simultaneously, the dynamic calibration formula is:

[0088]

[0089] For example, according to the formula, when SOC=90%, the average hysteresis voltage difference after 5 cycles is 0.12V.

[0090] The voltage hysteresis curve reflects the voltage difference between the charging and discharging paths within the current cycle, but this curve needs to be smoothed. An exponentially weighted moving average (EWMA) is used to update the hysteresis curve to suppress noise interference. For example, see Case 4;

[0091] Case 4: Schematic diagram of dynamic hysteresis curve calibration

[0092]

[0093] In conclusion, dynamic calibration reduces the hysteresis characteristic deviation caused by battery aging, thus mitigating the problem of error increasing with cycling.

[0094] If the difference between the current voltage and the calibration curve exceeds a preset threshold, the voltage deviation is calculated in the compensation method as follows: ,in To measure the actual charging voltage, The SOC reference voltage corresponding to the calibration curve; the condition for triggering the compensation method is that... ,when At that time, the battery polarization effect causes the voltage to be forward biased. At this time, the electrolyte concentration gradient causes a negative voltage bias, see Case 5;

[0095] Case 5: Typical Deviations in Different SOC Ranges

[0096]

[0097] The compensation method dynamically shifts and corrects the hysteresis curve based on the magnitude and direction of the deviation difference, combined with real-time temperature data. The temperature coefficient is calibrated through hysteresis voltage offset experiments on the battery at different temperatures. The dynamic shift correction method is as follows:

[0098] Curve translation formula In one example, the sample data is as follows:

[0099]

[0100] The example calculation shows that when T = -10℃, the curve shifts upward by 2.00 × (−10−25) = −70mv.

[0101] In the step of using the difference between the corrected hysteresis curve voltage and the real-time voltage as the basic compensation amount, multiplying it by a temperature coefficient to generate a first correction factor, and dynamically adjusting the weight of the ampere-hour integration method to a first preset value based on the numerical range of the first correction factor, the calculation of the first correction factor is as follows:

[0102]

[0103] In this formula, Indicates the rate of change of voltage;

[0104] The numerical range of the first correction factor, and the adjustment rule for dynamically adjusting the weights of the ampere-hour integration method to the first preset value are as follows:

[0105]

[0106] In one embodiment, when the battery is in a discharged state and the initial SOC value is ≤20%, the polarization voltage compensation model is activated, and a second correction factor is generated by combining the nonlinear relationship between the current change gradient and the temperature decay coefficient. The step of increasing the weights of the Kalman filter algorithm to a second preset value includes:

[0107] The initial SOC value is continuously monitored during discharge. When the initial SOC value drops to 20% or below, the polarization voltage dynamic compensation mode is activated.

[0108] Real-time acquisition of discharge current data and calculation of current change gradient per unit time; at the same time, matching a pre-calibrated temperature decay coefficient based on battery temperature and historical discharge data, wherein the temperature decay coefficient reflects the non-linear growth characteristic of battery internal resistance as temperature decreases.

[0109] The current change gradient and temperature decay coefficient are input into a predefined nonlinear coupling function to generate a polarization voltage compensation amount. The nonlinear coupling function satisfies the following condition: when the current change gradient increases or the temperature decay coefficient increases, the compensation amount grows exponentially.

[0110] Based on the absolute value range of the polarization voltage compensation, the weights of the Kalman filter algorithm are progressively increased from the initial value to the second preset value in segments.

[0111] In the specific implementation process, after the polarization voltage dynamic compensation mode is activated, the current change gradient is calculated. The time interval is 5 seconds; for example, when the current decreases from 10A to 5A, the gradient is -1A / s. The pre-calibrated temperature decay coefficients are shown in the table below:

[0112]

[0113] The predefined nonlinear coupling function generates the polarization voltage compensation amount using the following formula: Polarization Compensation Amount Generation Formula In the formula The polarization intensity coefficient, This is a reference temperature.

[0114] In one embodiment, the step of dynamically fusing the first correction factor and the second correction factor using a gradient descent optimization algorithm to generate a calibrated SOC value, and feeding the SOC value back to the power prediction module of the battery management system, includes:

[0115] Construct a fusion error function with the first correction factor and the second correction factor as input variables. The error function includes the absolute error between the current SOC estimate and the reference value, as well as the sliding window statistics of historical error data.

[0116] The weight allocation of the first correction factor and the second correction factor is iteratively optimized based on the gradient descent optimization algorithm. In each iteration, the weight coefficients are adjusted according to the negative gradient direction of the error function until the error converges to a preset threshold range.

[0117] The compensation amounts of the first correction factor and the second correction factor are dynamically weighted and fused according to the optimized weight coefficients to generate the calibrated SOC value, and the weight coefficients of the current iteration and the error convergence status are fed back to the parameter self-learning module of the battery management system.

[0118] In the specific implementation process, the steps of iteratively optimizing the weight allocation of the first correction factor and the second correction factor based on the gradient descent optimization algorithm include:

[0119] Set initial weight coefficients, where the initial weight of the first correction factor is ω1, the initial weight of the second correction factor is ω2, and ω1+ω2=1;

[0120] The output value of the fusion error function is calculated based on the current weight coefficients, and the partial derivatives of the error function with respect to ω1 and ω2 are calculated to determine the negative gradient direction of the error function.

[0121] The weight coefficients are updated along the negative gradient direction according to the preset learning rate, such that ω1'=ω1-μ·∂E / ∂ω1, ω2'=ω2-μ·∂E / ∂ω2, where μ is the learning rate, and ∂E / ∂ω1 and ∂E / ∂ω2 are the partial derivatives of the error function with respect to the weight coefficients, respectively.

[0122] Determine whether the updated error function value is less than the preset convergence threshold. If it is not met, repeat the steps until the error converges.

[0123] When the error converges, the current weight coefficient is locked as the optimal weight allocation scheme, and the historical iteration data is stored in the optimization parameter database of the battery management system.

[0124] Reference Appendix Figure 2 The present invention also proposes a battery SOC intelligent monitoring and management device based on a precise power algorithm, comprising:

[0125] The acquisition unit is used to acquire the battery's voltage, current, and temperature in real time, and calculate the battery's initial SOC value using the ampere-hour integration method in combination with the battery's historical charge and discharge data.

[0126] The calculation unit is used to adjust the calculation weights of the ampere-hour integration method and the Kalman filter algorithm respectively using a dynamic weight allocation algorithm, thereby correcting the initial SOC value, wherein:

[0127] When the battery is in a charging state and the initial SOC value is ≥90%, a voltage hysteresis characteristic compensation model is introduced, and the product of the battery's temperature coefficient and voltage change rate is used as the first correction factor to adjust the weight of the ampere-hour integral method to the first preset value.

[0128] When the battery is in a discharged state and the initial SOC value is ≤20%, the polarization voltage compensation model is activated. The second correction factor is generated by combining the nonlinear relationship between the current change gradient and the temperature decay coefficient, and the weight of the Kalman filter algorithm is increased to the second preset value.

[0129] The feedback unit is used to dynamically fuse the first correction factor and the second correction factor through the gradient descent optimization algorithm to generate a calibrated SOC value, and feed the SOC value back to the power prediction module of the battery management system.

[0130] Reference Figure 3 This invention also provides a computer device, which can be a server, and its internal structure can be as follows: Figure 3 As shown, the computer device includes a processor, memory, display screen, input device, network interface, and database connected via a system bus. The processor provides computing and control capabilities. The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores the operating system, computer programs, and database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The database stores the data corresponding to this embodiment. The network interface is used to communicate with external terminals via a network connection. When the computer program is executed by the processor, it implements the above-described method.

[0131] Those skilled in the art will understand that Figure 3 The structures shown are merely block diagrams of some structures related to the present invention and do not constitute a limitation on the computer devices on which the present invention is applied.

[0132] An embodiment of the present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-described method. It is understood that the computer-readable storage medium in this embodiment can be a volatile readable storage medium or a non-volatile readable storage medium.

[0133] In summary, this invention acquires real-time battery voltage, current, temperature, and historical data, and calculates the initial State of Charge (SOC) value using the ampere-hour integration method. During charging and with SOC ≥ 90%, a first correction factor is generated based on voltage hysteresis characteristics and temperature coefficient, dynamically reducing the ampere-hour integration weight. During discharging and with SOC ≤ 20%, a second correction factor is generated by combining polarization voltage and temperature decay coefficient, increasing the Kalman filter weight. The correction factors are then fused using a gradient descent optimization algorithm with closed-loop feedback to output the calibrated SOC value. This invention, through dynamic weight allocation, a nonlinear compensation model, and a closed-loop self-optimization mechanism, solves the problems of large estimation errors in extreme SOC ranges and poor adaptability of traditional methods, significantly improving the accuracy and reliability of battery management.

[0134] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the present invention and embodiments can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual-rate SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM, etc.

[0135] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, apparatus, article, or method that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, apparatus, article, or method. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, apparatus, article, or method that includes that element.

[0136] The above description is only a preferred embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent structural or procedural transformations made based on the content of the present invention specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. A battery SOC intelligent monitoring and management method based on accurate power algorithm, characterized in that, The method comprises the following steps: real-time acquisition of voltage, current and temperature of the battery, and combination of historical charging and discharging data of the battery to calculate an initial SOC value of the battery by using an ampere-hour integral method; a dynamic weight distribution algorithm is used to adjust the calculation weights of the ampere-hour integral method and the Kalman filtering algorithm, and then the initial SOC value is corrected, wherein: when the battery is in a charging state and the initial SOC value is greater than or equal to 90%, a voltage hysteresis characteristic compensation model is introduced, the product of a temperature coefficient and a voltage change rate of the battery is taken as a first correction factor, and the weight of the ampere-hour integral method is adjusted to a first preset value; when the battery is in a discharging state and the initial SOC value is less than or equal to 20%, a polarization voltage compensation model is activated, a second correction factor is generated in combination of a nonlinear relationship between a current change gradient and a temperature attenuation coefficient, and the weight of the Kalman filtering algorithm is increased to a second preset value; a gradient descent optimization algorithm is used to dynamically fuse the first correction factor and the second correction factor, generate a calibrated SOC value, and feed back the SOC value to an electric quantity prediction module of a battery management system; the step of real-time acquisition of voltage, current and temperature of the battery, and combination of historical charging and discharging data of the battery to calculate an initial SOC value of the battery by using an ampere-hour integral method comprises: abnormal values in the real-time sampled voltage, current and temperature data are dynamically screened, a sliding time window is used to perform standard deviation analysis on the continuous sampling data, data exceeding a preset fluctuation range is removed and interpolation processing is performed; nonlinear capacity attenuation data are generated according to the cycle number, average discharge depth and capacity attenuation trend in the historical charging and discharging data, wherein the actual capacity is associated with the square root of the cycle number and the square of the average discharge depth, the cycle number and the average discharge depth of the battery are extracted as input variables, a nonlinear function relationship is constructed based on the historical capacity attenuation data, wherein the actual capacity value is inversely proportional to the square root of the cycle number and is proportional to the square of the average discharge depth, and the attenuation coefficient in the nonlinear function relationship is calibrated according to the battery type, wherein the attenuation coefficient includes a first coefficient reflecting the cycle aging rate and a second coefficient reflecting the deep discharge effect; the calibrated attenuation coefficient and the real-time acquired cycle number and average discharge depth are substituted into the nonlinear function relationship, the current actual capacity value is dynamically calculated, and is taken as a capacity reference parameter of the ampere-hour integral method to obtain the nonlinear capacity attenuation data; based on a current temperature threshold and a charging and discharging current direction, a corresponding ampere-hour integral calculation mode is dynamically selected: a temperature compensation coefficient is introduced to correct the nominal capacity in a low-temperature discharging working condition, the nominal capacity is directly used for calculation in a normal-temperature charging working condition, and the calculation results in the corresponding interval are fused through a weight coefficient; in the step of introducing the voltage hysteresis characteristic compensation model when the battery is in a charging state and the initial SOC value is greater than or equal to 90%, the product of the temperature coefficient and the voltage change rate of the battery is taken as the first correction factor, and the weight of the ampere-hour integral method is adjusted to the first preset value, the step comprises: In the charging state, the voltage hysteresis curve is dynamically calibrated based on the charging and discharging history data of the battery, which reflects the voltage difference of the charging and discharging path in the current cycle; When the initial SOC value is greater than or equal to 90%, the charging voltage value is collected in real time and compared with the calibrated charging path hysteresis curve. If the difference between the current voltage and the calibrated curve exceeds the preset threshold, the compensation method is triggered; Through the compensation method, the hysteresis curve is dynamically translated and corrected according to the amplitude and direction of the deviation value, combined with real-time temperature data, and the temperature coefficient is calibrated through the hysteresis voltage offset experiment of the battery at different temperatures; The difference between the corrected hysteresis curve voltage and the real-time voltage is taken as the basic compensation amount, multiplied by the temperature coefficient to generate the first correction factor, and the weight of the ampere-hour integral method is dynamically adjusted to the first preset value according to the numerical interval of the first correction factor; The step of activating the polarization voltage compensation model when the battery is in the discharging state and the initial SOC value is less than or equal to 20%, generating a second correction factor based on the nonlinear relationship between the current change gradient and the temperature attenuation coefficient, and improving the weight of the Kalman filtering algorithm to a second preset value, comprising: Continuously monitor the initial SOC value in the discharging state, and activate the polarization voltage dynamic compensation mode when the initial SOC value is detected to drop to 20% or below; Real-time acquisition of discharge current data and calculation of current change gradient per unit time, and matching of pre-calibrated temperature attenuation coefficient based on battery temperature and historical discharge data, wherein the temperature attenuation coefficient reflects the nonlinear growth characteristics of the battery internal resistance with temperature drop; Input the current change gradient and temperature attenuation coefficient into a predefined nonlinear coupling function to generate a polarization voltage compensation amount, and the nonlinear coupling function satisfies: when the current change gradient increases or the temperature attenuation coefficient rises, the compensation amount increases exponentially; According to the absolute value range of the polarization voltage compensation amount, the weight of the Kalman filtering algorithm is segmented from the initial value to the second preset value; The step of dynamically fusing the first correction factor and the second correction factor through the gradient descent optimization algorithm to generate the calibrated SOC value, and feeding back the SOC value to the power prediction module of the battery management system, comprising: Construct a fusion error function with the first correction factor and the second correction factor as input variables, and the error function includes the absolute error of the current SOC estimation value and the reference value and the sliding window statistics of the historical error data; Iterative optimization of the weight distribution of the first correction factor and the second correction factor based on the gradient descent optimization algorithm, wherein the weight coefficient is adjusted according to the negative gradient direction of the error function in each iteration until the error converges to a preset threshold range; According to the optimized weight coefficient, the compensation amount of the first correction factor and the second correction factor is dynamically weighted and fused to generate the calibrated SOC value, and the weight coefficient and error convergence state of the current iteration are fed back to the parameter self-learning module of the battery management system; The step of iterative optimization of the weight distribution of the first correction factor and the second correction factor based on the gradient descent optimization algorithm, comprising: setting initial weight coefficients, wherein an initial weight of the first correction factor is , an initial weight of the second correction factor is , and ; calculating an output value of the fusion error function based on the current weight coefficients and calculating a partial derivative of the error function with respect to and determining a negative gradient direction of the error function; updating the weight coefficient along the negative gradient direction according to a preset learning rate, so that , wherein is a learning rate, and are partial derivatives of the error function with respect to the weight coefficient, respectively. determining whether the updated error function value is less than a preset convergence threshold, and repeating the steps until the error converges if the updated error function value is not less than the preset convergence threshold; locking the current weight coefficient as an optimal weight distribution scheme and storing historical iteration data to an optimization parameter database of the battery management system when the error converges.

2. A battery SOC intelligent monitoring and management device based on accurate electric quantity algorithm, characterized in that, The battery SOC intelligent monitoring and management device based on the precise electric quantity algorithm is used to execute the battery SOC intelligent monitoring and management method based on the precise electric quantity algorithm in claim 1, comprising: The acquisition unit is used to acquire the voltage, current and temperature of the battery in real time, and calculate the initial SOC value of the battery by using the ampere-hour integral method in combination with the historical charge and discharge data of the battery; The calculation unit is used to adjust the calculation weights of the ampere-hour integral method and the Kalman filtering algorithm respectively by using the dynamic weight distribution algorithm, and then correct the initial SOC value, wherein: when the battery is in a charging state and the initial SOC value is greater than or equal to 90%, a voltage hysteresis characteristic compensation model is introduced, the product of the temperature coefficient and the voltage change rate of the battery is taken as a first correction factor, and the weight of the ampere-hour integral method is adjusted to a first preset value; when the battery is in a discharging state and the initial SOC value is less than or equal to 20%, a polarization voltage compensation model is activated, a second correction factor is generated by combining the nonlinear relationship between the current change gradient and the temperature attenuation coefficient, and the weight of the Kalman filtering algorithm is improved to a second preset value; The feedback unit is used to dynamically fuse the first correction factor and the second correction factor by using a gradient descent optimization algorithm, generate a calibrated SOC value, and feed back the SOC value to an electric quantity prediction module of the battery management system.

3. A computer device comprising a memory and a processor, the memory having stored therein a computer program, characterized in that, The processor executes the computer program to realize the steps of the battery SOC intelligent monitoring and management method based on the precise electric quantity algorithm in claim 1.

4. A computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to realize the steps of the battery SOC intelligent monitoring and management method based on the precise electric quantity algorithm in claim 1.

Images