Search

CN-121613325-B - Power battery SOC and SOH prediction and estimation method based on temperature correction

CN121613325BCN 121613325 BCN121613325 BCN 121613325BCN-121613325-B

Abstract

The invention discloses a power battery SOC and SOH prediction and estimation method based on temperature correction, and belongs to the technical field of new energy automobile battery management. The method comprises the following steps of S1, synchronously collecting multi-source original data of the power battery through a sensor under a low-temperature working condition, recording real-time capacity, sequentially executing cleaning, alignment, environment characteristic identification and baseline calibration on the original data, and outputting a standardized data set, S2, carrying out temperature correction based on the standardized data set, fusing multiple models to construct a prediction model, outputting an SOC and SOH prediction result, S3, calculating a three-dimensional health index based on the prediction result, dividing health grades, and dynamically adjusting battery operation parameters. By adopting the power battery SOC and SOH prediction and estimation method based on temperature correction, low-temperature interference can be effectively eliminated, the estimation accuracy of the SOC and the SOH can be improved, and meanwhile, the battery operation parameters can be dynamically optimized, and the battery cycle life can be prolonged.

Inventors

  • HONG JICHAO
  • Pei Jiaqi
  • LI MENG
  • WEI LINGJUN
  • YANG JINGSONG
  • MA SHIKUN

Assignees

  • 北京科技大学

Dates

Publication Date
20260512
Application Date
20251118

Claims (9)

  1. 1. A power battery SOC and SOH prediction and estimation method based on temperature correction is characterized by comprising the following steps: S1, data acquisition and preprocessing, namely synchronously acquiring multi-source original data of a power battery through a sensor under a low-temperature working condition, recording real-time capacity, and then sequentially executing cleaning, alignment, environmental characteristic identification and baseline calibration on the original data to output a standardized data set; S2, temperature correction and prediction, wherein based on a standardized data set, a relation between temperature, capacity and internal resistance is established through an Arrhenius model, and SOC calculation deviation and SOH evaluation deviation at low temperature are corrected, and the method specifically comprises the following steps: s211, calculating the temperature Capacity temperature correction factor The formula is as follows: ; S212, correcting the SOC calculation result, wherein the formula is as follows The method comprises the following steps: ; Wherein, the As the current corrected SOC value, For the value of the SOC at the previous time, Is the temperature The correction factor of the charge and discharge efficiency is set down, Is the temperature A charge-discharge current; S213, correcting SOH evaluation results by adopting a weighting method, wherein the formula is as follows The method comprises the following steps: ; Wherein, the As the current post-correction SOH value, As the weight coefficient of the capacity, As the internal resistance weight coefficient, Is the temperature The actual available capacity of the lower battery; s214, in the constant current charging stage of the battery, calculating the capacity increment and extracting the peak voltage and the peak height of a capacity increment curve as auxiliary characteristics of the health state of the battery, wherein the calculation formula of the capacity increment The method comprises the following steps: ; Wherein, the In order to increase the charge capacity of the battery, In order to increase the voltage by a voltage increment, Is used for constant-current charging current, In order to increase the charge time in increments, Is that The value of the sampled voltage at the beginning, Is that Sampling voltage value at the end; s22, based on different characteristics of short-term dynamic change of the SOC and long-term attenuation of the SOH, fusing multiple models to construct a prediction model, wherein the method specifically comprises the following steps: S221, predicting the short-term dynamic change of the SOC by adopting an LSTM model, inputting corrected time sequence data of the SOC, voltage, current and temperature into the LSTM model, and using an Adam optimizer to obtain a mean square error As a loss function, a current SOC prediction value is output, wherein, The calculation formula of (2) is as follows: ; Wherein, the In order to train the number of samples, Is the true value of the SOC and, Is the predicted value of SOC; S222, predicting SOH long-term attenuation by adopting a fusion entropy weight method of a gray GM (1, 1) model and an LSTM model, wherein the gray GM (1, 1) model is acquired at equal time intervals Construction of the original sequence from the corrected SOH data For a pair of Generating new sequence by one-time accumulation Establishing a first-order linear differential equation model, and calculating a development coefficient by a least square method And the amount of grey effect Wherein, the formula of the first-order linear differential equation is: ; The SOH long-term prediction formula is: ; Wherein, the Identified for the starting data point of the original SOH sequence, To predict the number of steps, and introducing the temperature acceleration factor through Arrhenius model Dynamically correcting SOH attenuation prediction coefficient Wherein the temperature acceleration factor Attenuation prediction coefficient The formulas of (a) are respectively as follows: ; ; Wherein, the For the aging activation energy of the battery, Is a gas constant which is a function of the gas, For the reference to the absolute temperature of the air, For the average operating absolute temperature of the battery, Obtaining the output SOH sequence of the LSTM model in the same prediction period Standard deviation of (2) Gray GM (1, 1) model output SOH sequence Standard deviation of (2) Information entropy of the LSTM model and the gray GM (1, 1) model is calculated, and the formula is as follows: ; Wherein, the Is the information entropy of the LSTM model, Entropy of information for GM (1, 1) model 、 Normalizing to obtain a fusion weight, wherein the formula is as follows: ; the formula of the weighted and fused SOH predicted value is as follows: ; s23, outputting the SOC and SOH prediction result after temperature correction; S3, health assessment, namely calculating three-dimensional health indexes based on the prediction result of the S23, wherein the three-dimensional health indexes comprise SOC estimation errors SOH attenuation Rate Temperature sensitivity Wherein the temperature sensitivity The calculation formula of (2) is as follows: ; Wherein, the Is used as a low-temperature point of the material, Is a normal temperature point, the temperature is equal to the normal temperature point, Is that The measured internal resistance is that, Is that Actually measured internal resistance; The health grade is subdivided and the battery operating parameters are dynamically adjusted.
  2. 2. The method for predicting and estimating the SOC and the SOH of the power battery based on temperature correction according to claim 1, wherein S1 specifically comprises: S11, synchronously acquiring voltage, current and temperature data of the power battery by adopting a voltage sensor, a current sensor and a temperature sensor, wherein the original sampling frequency of the sensor is set as Simultaneously recording the real-time capacity of the battery; s12, filtering high-frequency noise interference in the voltage and current signals by adopting a low-pass filter in data cleaning, and continuously deleting time length The long-time data of seconds are marked as invalid data segments, the invalid data segments do not participate in modeling calculation, the short-time data loss is filled by adopting a linear interpolation method, and the formula of the linear interpolation method is as follows: ; Wherein, the In order to fill in the value of the value, In order to miss valid data before a point, In order to be a valid data after the missing point, In order to achieve the missing dot time, the time of the missing dot, In order to miss the valid data time before a point, Removing abnormal data by statistical threshold method, calculating arithmetic mean value of data in 5 second time window And standard deviation Reject out beyond Wherein, arithmetic mean And standard deviation The calculation formula of (2) is as follows: ; ; Wherein, the Is the number of data points in the 5 second window, Is the inside of the window Data; s13, data alignment unifies time step by downsampling aggregation Outputting the aligned voltage, current and temperature data; s14, under the standard environment, the environment characteristic identification and the baseline calibration are that the battery is discharged from a full charge state to a cut-off voltage at a rate of 0.33C, and the discharge capacity is recorded as the rated discharge capacity Standard internal resistance of battery is measured by adopting mixed pulse power characteristic test method Then change the temperature Repeating the test and recording different temperatures The actual discharge capacity And measured internal resistance To establish an OCV-SOC reference correspondence table; S15, outputting a standardized data set of voltage, current and temperature.
  3. 3. The method for predicting and estimating the SOC and SOH of a power battery based on temperature correction as set forth in claim 2, wherein S3 further comprises: S31, in the three-dimensional health index, SOC estimation error The calculation formula of (2) is as follows: ; Wherein, the As an arithmetic mean value of the values, As the predicted value of the SOC, A reference true value for the measured SOC in a standard environment; SOH attenuation Rate The calculation formula of (2) is as follows: ; Wherein, the In order to evaluate the duration of the cycle, To evaluate the period The SOH value at the start-up time, To evaluate the period SOH value at end; S32, quantifying the health state of the battery based on three-dimensional health indexes, and dividing the battery into 4 health grades, wherein the grade I is excellent, the grade II is good, the grade III is attention, the grade IV is replacement, and the specific standard is as follows: Stage I: , , ; Stage II: , , ; Class III: , , ; grade IV: , , ; s33, feeding the health grade back to the power battery management system, and adjusting operation parameters based on the health grade, wherein the operation parameters comprise maximum allowable charging current, maximum allowable discharging current, an SOC interval and a temperature alarm threshold, and the adjustment standard of a control strategy mapping table corresponding to the health grade is as follows: the I level is that the maximum allowable charge current is 1C, the maximum allowable discharge current is 2C, the SOC interval is 20% -100%, and the temperature alarm threshold value is-10 ℃; stage II, namely, the maximum allowable charging current is 0.8C, the maximum allowable discharging current is 1.5C, the SOC interval is 20% -95%, and the temperature alarm threshold is-5 ℃; III, the maximum allowable charging current is 0.5C, the maximum allowable discharging current is 1C, the SOC interval is 30% -90%, and the temperature alarm threshold is 0 ℃; IV, the maximum allowable charging current is 0.2C, the maximum allowable discharging current is 0.5C, the SOC interval is 50% -80%, and the temperature alarm threshold is 10 ℃.
  4. 4. The method for predicting and estimating the SOC and the SOH of the power battery based on temperature correction according to claim 2, wherein the first-order low-pass filter in S12 is a Butterworth low-pass filter, and the cut-off frequency is fixed at 10Hz.
  5. 5. The method for predicting and estimating SOH and SOC of a power battery based on temperature correction as set forth in claim 2, wherein the step of time in S13 The range of the value is 0.1 s-10 s, Number of samples in The calculation formula of (2) is as follows: 。
  6. 6. The method for predicting and estimating the SOC and SOH of the power battery based on temperature correction according to claim 2, wherein in the process of establishing the OCV-SOC reference correspondence table in S14, the battery SOC is charged and discharged at a constant current of 0.05C, and is kept stand for 1 hour after being lowered by 10% each time, and the corresponding OCV value is recorded until the OCV-SOC reference correspondence table covering an SOC interval of 0% -100% is established.
  7. 7. The method for predicting and estimating SOC and SOH of a power battery based on temperature correction as set forth in claim 1, wherein when calculating the capacity increment in S214, the fluctuation range of the battery charging current in the constant current charging stage Pair for preventing current fluctuation Interference of curve feature extraction.
  8. 8. The method for predicting and estimating the SOC and the SOH of the power battery based on temperature correction of claim 1, wherein the LSTM model in S221 comprises an input layer, 2 hidden layers and an output layer, wherein the input layer inputs four time sequence data characteristics of the SOC, the voltage, the current and the temperature after the correction at the last moment, the number of neurons of each hidden layer is 64-128, a ReLU activation function is adopted between the input layer and the hidden layers and between the hidden layers and the output layer, and the output layer outputs the current SOC predicted value.
  9. 9. The method for predicting and estimating the SOC and the SOH of the power battery based on the temperature correction, which is based on the temperature correction, is characterized in that S1 outputs a standardized data set of voltage, current and temperature through a data acquisition and preprocessing module, noise-free and deviation-free input data are provided for S2, the real-time predicted value of the SOC and the long-term attenuation trend of the SOH after the temperature correction are output through the temperature correction and prediction module, and S3 outputs a battery health grade report and an operation parameter adjustment instruction through a health evaluation module, so that visual management and active protection of the health state of the battery are realized.

Description

Power battery SOC and SOH prediction and estimation method based on temperature correction Technical Field The invention relates to the technical field of new energy automobile battery management, in particular to a power battery SOC and SOH prediction and estimation method based on temperature correction. Background Under the low-temperature working condition of-30 ℃ to 5 ℃ in winter, the electrochemical reaction rate of the power battery is obviously reduced, the internal resistance is increased, the capacity utilization rate is reduced, and the accurate estimation of the SOC (State of Charge) and the SOH (State of Health) is directly affected. The prior art has obvious defects that the prior conventional ampere-hour integration method and the equivalent circuit model do not introduce temperature dynamic compensation, only rely on fixed parameter calculation at normal temperature to cause SOC estimation errors of 8-10% in a low-temperature environment, and the prior SOH estimation is mostly dependent on a static capacity attenuation model, only considers single factors such as cycle times and the like, and cannot adapt to internal resistance fluctuation and capacity attenuation difference caused by dynamic temperature change. The above-mentioned techniques do not solve the problem of accumulation of estimation deviation of SOC and SOH in low temperature, and long-term use of the techniques can lead to decision misalignment of BMS (Battery management system) MANAGEMENT SYSTEM, which further causes risks such as insufficient charging, cruising deficiency or excessive loss, so that an accurate prediction and estimation scheme integrating temperature correction mechanisms is needed. Disclosure of Invention The invention aims to provide a power battery SOC and SOH prediction and estimation method based on temperature correction, which solves the technical problems. In order to achieve the above purpose, the invention provides a power battery SOC and SOH prediction and estimation method based on temperature correction, which comprises the following steps: S1, data acquisition and preprocessing, namely synchronously acquiring multi-source original data of a power battery through a sensor under a low-temperature working condition, recording real-time capacity, and then sequentially executing cleaning, alignment, environmental characteristic identification and baseline calibration on the original data to output a standardized data set; S2, temperature correction and prediction, wherein based on a standardized data set, a relation between temperature, capacity and internal resistance is established through an Arrhenius model, and SOC calculation deviation and SOH evaluation deviation at low temperature are corrected, and the method specifically comprises the following steps: s211, calculating the temperature Capacity temperature correction factorThe formula is as follows: ; S212, correcting the SOC calculation result, wherein the formula is as follows The method comprises the following steps: ; Wherein, the As the current corrected SOC value,For the value of the SOC at the previous time,Is the temperatureThe correction factor of the charge and discharge efficiency is set down,Is the temperatureA charge-discharge current; S213, correcting SOH evaluation results by adopting a weighting method, wherein the formula is as follows The method comprises the following steps: ; Wherein, the As the current post-correction SOH value,As the weight coefficient of the capacity,As the internal resistance weight coefficient,Is the temperatureThe actual available capacity of the lower battery; s214, in the constant current charging stage of the battery, calculating the capacity increment and extracting the peak voltage and the peak height of a capacity increment curve as auxiliary characteristics of the health state of the battery, wherein the calculation formula of the capacity increment The method comprises the following steps: ; Wherein, the In order to increase the charge capacity of the battery,In order to increase the voltage by a voltage increment,Is used for constant-current charging current,In order to increase the charge time in increments,Is thatThe value of the sampled voltage at the beginning,Is thatSampling voltage value at the end; s22, based on different characteristics of short-term dynamic change of the SOC and long-term attenuation of the SOH, fusing multiple models to construct a prediction model, wherein the method specifically comprises the following steps: S221, predicting the short-term dynamic change of the SOC by adopting an LSTM model, inputting corrected time sequence data of the SOC, voltage, current and temperature into the LSTM model, and using an Adam optimizer to obtain a mean square error As a loss function, a current SOC prediction value is output, wherein,The calculation formula of (2) is as follows: ; Wherein, the In order to train the number of samples,Is the true value of the SOC and,Is the