Lithium iron phosphate battery soc estimation method considering charge-discharge alternating hysteresis error

Abstract

The application discloses a lithium iron phosphate battery SOC estimation method considering charge-discharge alternating hysteresis error. Firstly, the open-circuit voltage curve including the open-circuit voltage boundary curve, the charge-to-discharge transition curve and the discharge-to-charge transition curve is calibrated, and the d value under different SOC in the charge-to-discharge process and the discharge-to-charge process is obtained. Then, a transition curve generation model is constructed to generate the transition curve under any SOC. Finally, the OCV-SOC curve is updated in real time according to the SOC accumulation during the battery working process. When the SOC accumulation is 0 or reaches a set threshold, the OCV-SOC curve is updated to the boundary curve of the battery working at the current time. When the SOC accumulation is between 0 and the set threshold, the OCV-SOC curve remains the original transition curve unchanged. Then, based on the OCV-SOC curve and the battery equivalent circuit model, the SOC of the lithium iron phosphate battery is estimated by using an adaptive extended Kalman filtering algorithm. The method can obtain the OCV-SOC curve suitable for the change of the battery working state, and has higher SOC estimation precision when the lithium iron phosphate battery is charged and discharged in small amounts alternately.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of power battery SOC estimation, and particularly relates to a lithium iron phosphate battery SOC estimation method considering charging-discharging alternation hysteresis error. BACKGROUND

[0002] Accurate estimation of battery state (SOC) is a crucial task for new energy vehicle battery management system (BMS), and a commonly used SOC estimation method is the equivalent circuit method. The SOC estimation based on the equivalent circuit combines the open circuit voltage method and the ampere-hour integral method, realizes closed-loop correction of SOC on the basis of state estimation, and regards the mapping relationship between SOC and open circuit voltage (OCV) as the basis for SOC correction, which greatly affects the accuracy of the estimation result.

[0003] Due to the material characteristics of the lithium iron phosphate battery, the open circuit voltage curve thereof shows a long and flat platform period, which leads to more obvious open circuit voltage hysteresis characteristics than other lithium batteries, so that the SOC-OCV mapping relationship of the lithium iron phosphate battery tends to be complex, and it is difficult to obtain an accurate and reliable open circuit voltage curve, thereby increasing the SOC estimation error. In order to reduce the SOC estimation error caused by the open circuit voltage hysteresis, researchers have established various hysteresis models. The simplest method is to take the midline of the charging boundary and discharging boundary curve as the open circuit voltage curve, which will lead to a very obvious SOC estimation error. In order to facilitate the representation of the hysteresis of the open circuit voltage, some researchers introduce a hysteresis factor based on the boundary value to represent the position of the current state open circuit voltage point from the two boundary curves. For the definition of the hysteresis factor, some are calculated by the normalized integral of the charge throughput, and some are defined based on the SOC accumulation of the OCV difference. Some researchers establish hysteresis models from mathematical models, such as single-state hysteresis model, Preisach model, Plett hysteresis model and the like. However, the existing hysteresis models mainly focus on the OCV-SOC curve of the lithium iron phosphate battery in the single working state of charging or discharging, that is, only the state of the lithium iron phosphate battery working on the boundary curve is considered, and the problem of increased SOC estimation error caused by the deviation of the open circuit voltage curve from the main hysteresis curve in the small amount of SOC charging and discharging alternation process is ignored.

[0004] Therefore, the application focuses on the open circuit voltage change of the lithium iron phosphate battery in the small amount of SOC charging and discharging alternation in the braking or inertial sliding scene of the electric vehicle in the actual use process, and proposes a lithium iron phosphate battery SOC estimation method considering charging-discharging alternation hysteresis error, which can effectively improve the SOC estimation accuracy of the lithium iron phosphate battery in the small amount of SOC charging and discharging alternation. SUMMARY

[0005] In view of the deficiencies of the prior art, the technical problem to be solved by the present application is to provide a lithium iron phosphate battery SOC estimation method considering charging and discharging alternating hysteresis error.

[0006] The present application solves the technical problem by adopting the following technical solution:

[0007] A lithium iron phosphate battery SOC estimation method considering charging and discharging alternating hysteresis error, characterized in that the method comprises the following steps:

[0008] First step: calibrate the open circuit voltage curve including the open circuit voltage boundary curve, the charging to discharging transition curve and the discharging to charging transition curve, and obtain the d value at different SOC in the charging to discharging process and the discharging to charging process; d represents the distance from the current state of the battery in the open circuit voltage curve to the original boundary curve before the transition starts, accounting for the percentage of the distance between the two boundary curves at the current SOC;

[0009] Second step: build a transition curve generation model for generating transition curves at any SOC; the transition curve generation model uses GAN to generate a plurality of simulated d values according to the d value at the SOC closest to the predicted target SOC, obtains a plurality of open circuit voltages in the transition process according to these simulated d values, the predicted target SOC and the open circuit voltage boundary curve, and performs curve fitting on these open circuit voltages to obtain the transition curve;

[0010] Third step: update the OCV-SOC curve in real time according to the SOC accumulation during the operation of the battery, when the SOC accumulation is 0 or reaches a set threshold, the OCV-SOC curve is updated to the boundary curve of the battery operation at the current time; when the SOC accumulation is between 0 and the set threshold, the OCV-SOC curve remains the original transition curve unchanged; then, based on the OCV-SOC curve and the equivalent circuit model of the battery, the SOC of the lithium iron phosphate battery is estimated by using the adaptive extended Kalman filtering algorithm.

[0011] Compared with the prior art, the present application has the following advantages:

[0012] The present application studies the OCV-SOC curve of the lithium iron phosphate battery, focuses on the transition of the OCV-SOC curve between the two boundary curves when the battery charging / discharging working state alternates, quantifies the transition degree by using the d value, and obtains the d value at different SOC through the open circuit voltage curve calibration. The transition curve generation model can generate the transition curve at different SOC, which can obtain the OCV-SOC curve that can adapt to the change of the battery working state with less calculation amount, that is, the OCV-SOC curve is updated in real time according to the battery working state, which improves the accuracy of the OCV-SOC curve. This means that when the battery small amount SOC (such as 6% SOC) changes caused by the vehicle braking energy recovery system (such as energy recovery in the scene of braking or inertial sliding), the battery management system can obtain the OCV-SOC curve that is more in line with the actual operation of the battery. For the lithium iron phosphate battery SOC estimation with obvious open circuit voltage hysteresis characteristics, the OCV-SOC curve that is more accurate and in line with the battery working state can significantly reduce the SOC estimation error. Therefore, the method has higher SOC estimation accuracy when the lithium iron phosphate battery is charged and discharged in small amount alternately. BRIEF DESCRIPTION OF DRAWINGS

[0013] Figure 1 is a schematic diagram of the generated transition curve;

[0014] Figure 2 is the SOC estimation result when considering the transition curve at 50% SOC when small amount charging and discharging alternately;

[0015] Figure 3 is the SOC estimation result when not considering the transition curve at 50% SOC when small amount charging and discharging alternately;

[0016] Figure 4 is the SOC estimation result when considering the transition curve at 74% SOC when small amount charging and discharging alternately;

[0017] Figure 5 is the SOC estimation result considering the transition curve under the DST working condition. DETAILED DESCRIPTION

[0018] Specific embodiments will be described below with reference to the accompanying drawings, which are only used to further illustrate the technical solutions of the present application and do not limit the protection scope of the present application.

[0019] The present application provides a lithium iron phosphate battery SOC estimation method considering charging and discharging alternation hysteresis error (referred to as method, see Figures 1 to 5 ), comprising the following steps:

[0020] The first step is to calibrate the open circuit voltage curve, including the open circuit voltage boundary curve, the charging to discharging transition curve and the discharging to charging transition curve.

[0021] The open circuit voltage boundary curve calibration step is as follows: 1) discharge the battery to the cut-off voltage, and stand for 2 hours, at this time the battery SOC is considered to be 0; 2) charge the battery at a constant current and constant voltage of 10% SOC, stand for 1 hour, and record the open circuit voltage; 3) repeat step 2), charge 10% SOC each time until the battery is fully charged, obtain the open circuit voltage at different SOC and use the least square method to fit the curve, and obtain the charging open circuit voltage boundary curve; 4) discharge the battery at a constant current and constant voltage of 10% SOC, stand for 1 hour, and record the open circuit voltage; 5) repeat step 4), discharge 10% SOC each time until the battery is discharged, obtain the open circuit voltage at different SOC and use the least square method to fit the curve, and obtain the discharging open circuit voltage boundary curve.

[0022] The charging to discharging transition curve calibration step is as follows: 1) discharge the battery to the cut-off voltage, and stand for 2 hours, at this time the battery SOC is considered to be 0; 2) charge the battery at a constant current of 1 / 3C to the initial SOC, stand for 1 hour, and record the open circuit voltage; 3) discharge the battery at a constant current of 1 / 3C, the discharge capacity is 2% SOC, stand for 1 hour, and record the open circuit voltage; 4) repeat step 3), discharge 2% SOC each time until the total discharge capacity reaches 10% SOC, obtain the open circuit voltage at different SOC and use the least square method to fit the curve, and obtain the charging to discharging transition curve at the initial SOC; set the initial SOC to be 0.3, 0.4, 0.5, 0.6, 0.7 and 0.8 respectively, and obtain the charging to discharging transition curve at different SOC.

[0023] The discharging to charging transition curve calibration step is as follows: 1) fully charge the battery by using the constant current and constant voltage method, stand for 2 hours, at this time the battery SOC is considered to be 100%; 2) discharge the battery at a constant current of 1 / 3C to the initial SOC, stand for 1 hour, and record the open circuit voltage; 3) charge the battery at a constant current of 1 / 3C, the charge capacity is 2% SOC, stand for 1 hour, and record the open circuit voltage; 4) repeat step 3), charge 2% SOC each time until the total charge capacity reaches 10% SOC, obtain the open circuit voltage at different SOC and use the least square method to fit the curve, and obtain the discharging to charging transition curve at the initial SOC; set the initial SOC to be 0.7, 0.6, 0.5, 0.4, 0.3 and 0.2 respectively, and obtain the discharging to charging transition curve at different SOC.

[0024] A quantitative index d is proposed to measure the extent to which the current state of the battery deviates from the original boundary curve before the transition starts; the specific definition of d is that in the open-circuit voltage curve, the distance from the current state of the battery to the original boundary curve before the transition starts accounts for a percentage of the distance between the two boundary curves at the current SOC. Through open-circuit voltage curve calibration, the d values in the charging-to-discharging process and the discharging-to-charging process at different SOCs are obtained.

[0025] Second step: build a transition curve generation model to generate a transition curve at any SOC according to the transition curve generation model;

[0026] The transition curve generation model includes d value generation and transition curve fitting. The d value generation adopts a generative adversarial network (GAN), which includes a generator and a discriminator. The generator generates simulated d values according to real d values, and the discriminator is responsible for judging the authenticity of the generated data. The real d values measured by the calibration experiment are used to train the GAN, so that the generator can generate simulated values close to the real values. The trained generator generates multiple simulated d values according to the d values at the SOC closest to the predicted target SOC; according to the simulated d values, the predicted target SOC, and the open-circuit voltage boundary curve, multiple open-circuit voltages in the transition process are obtained, and these open-circuit voltages are fitted by the least squares method to obtain the transition curve. Therefore, according to the transition curve generation model, a transition curve starting from the boundary curve and meeting the characteristics of the lithium iron phosphate battery at any SOC can be generated.

[0027] Generating simulated d values through GAN not only effectively expands the data set, but also captures and simulates the subtle changes of the transition curve, improving the model's prediction ability of the transition path at different starting SOCs.

[0028] Third step: update the OCV-SOC curve in real time according to the battery operating state; based on the OCV-SOC curve and the battery equivalent circuit model, the adaptive extended Kalman filter (AEKF) algorithm is used to estimate the SOC of the lithium iron phosphate battery;

[0029] When the battery works on the boundary curve at the current time and the next time, the OCV-SOC curve is the boundary curve on which the battery works; when the battery works on the boundary curve at the current time and changes the charging / discharging direction to work on the transition curve at the next time, the OCV-SOC curve adopts the transition curve at the current SOC; when the battery works on the transition curve at the current time and changes the charging / discharging direction at the next time, the OCV-SOC curve remains the original transition curve unchanged. In short, the OCV-SOC curve can be updated according to the SOC accumulation, when the SOC accumulation is 0 or reaches 10%, the OCV-SOC curve is updated to the boundary curve on which the battery works at the current time; when the SOC accumulation is between 0 and 10%, the OCV-SOC curve remains the original transition curve unchanged.

[0030] Figure 2 , 3 are SOC estimation results of the test battery at 50% SOC when considering transition curve and not considering transition curve respectively. Figure 2 The MAE of SOC estimation result in the middle is 0.0149, and the RMSE is 0.0152. Figure 3 The MAE of SOC estimation result in the middle is 0.0359, and the RMSE is 0.0359. Figure 4 is the SOC estimation result when considering transition curve at 74% SOC starting from uninterrupted small amount of charge and discharge alternation, and the MAE is 0.0087, and the RMSE is 0.0087. Figure 5 is the SOC estimation result considering transition curve under DST working condition, and the MAE is 0.0112, and the RMSE is 0.0113. It can be concluded that considering transition curve can effectively improve the accuracy of the open circuit voltage curve, thereby improving the accuracy of SOC estimation. Especially when the lithium iron phosphate battery is in the state of charge and discharge alternation, the open circuit voltage curve generated with the change of the battery working state is closer to the true curve, and the SOC estimation based on this is also more accurate, and the average error is within 1.5%, so the method has higher precision for SOC estimation of lithium iron phosphate battery in small amount of charge and discharge alternation state.

[0031] The unmentioned part of the application is applicable to the prior art.

Claims

1. A lithium iron phosphate battery SOC estimation method considering charge-discharge alternating hysteresis error, characterized by, The method comprises the following steps: First step: calibrate the open circuit voltage curve including open circuit voltage boundary curve, charge to discharge transition curve and discharge to charge transition curve, and obtain the d value at different SOC in the charge to discharge process and the discharge to charge process; d represents the distance from the current state of the battery in the open circuit voltage curve to the original boundary curve before the transition starts, accounting for the percentage of the distance between the two boundary curves at the current SOC; The open circuit voltage boundary curve calibration comprises: 1) discharge the battery to the cut-off voltage, and stand for a period of time until the battery SOC is 0; 2) charge the battery to 10% SOC at constant current and constant voltage, and record the open circuit voltage after standing for 1h; 3) repeat step 2), charge 10% SOC each time until the battery is fully charged, obtain the open circuit voltage at different SOC and perform curve fitting to obtain the charge open circuit voltage boundary curve; 4) discharge the battery to 10% SOC at constant current and constant voltage, and record the open circuit voltage after standing for 1h; 5) repeat step 4), discharge 10% SOC each time until the battery is fully discharged, obtain the open circuit voltage at different SOC and perform curve fitting to obtain the discharge open circuit voltage boundary curve; The charge to discharge transition curve calibration comprises: 1) discharge the battery to the cut-off voltage, and stand for a period of time until the battery SOC is 0; 2) charge the battery to the starting SOC at a current of 1 / 3C, and record the open circuit voltage after standing for 1h; 3) discharge the battery at a current of 1 / 3C, and record the open circuit voltage after standing for 1h; 4) repeat step 3), discharge 2% SOC each time until the total discharge capacity reaches 10% SOC, obtain the open circuit voltage at different SOC and perform curve fitting to obtain the charge to discharge transition curve; The discharge to charge transition curve calibration comprises: 1) fully charge the battery by adopting the mode of constant current first and then constant voltage, and stand for a period of time until the battery SOC is 100%; 2) discharge the battery to the starting SOC at a current of 1 / 3C, and record the open circuit voltage after standing for 1h; 3) charge the battery at a current of 1 / 3C, and record the open circuit voltage after standing for 1h; 4) repeat step 3), charge 2% SOC each time until the total charge capacity reaches 10% SOC, obtain the open circuit voltage at different SOC and perform curve fitting to obtain the discharge to charge transition curve; Second step: build a transition curve generation model for generating the transition curve at any SOC; the transition curve generation model generates multiple simulated d values according to the d value at the SOC closest to the predicted target SOC by using GAN, obtains multiple open circuit voltages in the transition process according to the simulated d values, the predicted target SOC and the open circuit voltage boundary curve, and performs curve fitting on the open circuit voltages to obtain the transition curve; Third step: real-time update the OCV-SOC curve according to the SOC accumulation during the battery working process, when the SOC accumulation is 0 or reaches the set threshold, the OCV-SOC curve is updated to the boundary curve of the battery working at the current time; when the SOC accumulation is between 0 and the set threshold, the OCV-SOC curve remains the original transition curve unchanged; then, based on the OCV-SOC curve and the battery equivalent circuit model, the SOC of the lithium iron phosphate battery is estimated by using the adaptive extended Kalman filtering algorithm. 2.The lithium iron phosphate battery SOC estimation method considering charge-discharge alternation hysteresis error according to claim 1, wherein, The set threshold of the SOC accumulation is 10%.

Images