Incremental capacity analysis method of lithium iron phosphate battery health state

Abstract

The invention discloses a lithium iron phosphate battery health state increment capacity analysis method. The method comprises the following steps: acquiring voltage and capacity QV data of a battery in different health states; converting the obtained voltage and capacity data into incremental capacity IC data, and preprocessing the incremental capacity IC data; extracting, from the processed incremental capacity IC data, a voltage value V1 corresponding to an IC peak value, a voltage value V2 after a battery is charged by a fixed capacity Q from V1, and a peak height H of the IC peak; respectively performing curve fitting according to the extracted data V1, V2, H and S0H to obtain formulas H-SOH, V1-SOH and V2-SOH; s0H is predicted based on real vehicle data. On the basis of full-period offline incremental capacity analysis, the SOH of the battery can be monitored in real time, and the precision is greatly improved; a corresponding formula is obtained through an off-line incremental capacity analysis method, the SOH of the battery can be analyzed in real time through the formula, and the burden of the operational power of a BMS controller is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of battery state of health (SOH) analysis, and more particularly, to an incremental capacity analysis method for the state of health of a lithium iron phosphate battery. Background Art

[0002] With the wide application of lithium iron phosphate batteries in new energy vehicles and energy storage, the related algorithms and strategies of lithium iron phosphate batteries have attracted more and more attention. As an important part of the battery state, the monitoring and prediction of the battery state of health (SOH) have become increasingly important. Due to the special electrochemical characteristics of lithium iron phosphate, during the plateau period, as the state of charge (SOC) of the battery increases, its voltage changes very smoothly, and the constraints brought by the plateau period can be eliminated through the ICA algorithm. However, the general ICA algorithm not only requires a large amount of data but also requires real-time analysis using a neural network algorithm online, which brings a great burden to the BMS controller. In addition, using the traditional SOH linear look-up table (cumulative used Ah - SOH) ignores the influence of the complex working environment, temperature, DOD, and charge-discharge rate during the use of the battery, resulting in a large deviation in the estimation of the battery SOH. Summary of the Invention

[0003] The technical problem to be solved by the present invention is to provide an incremental capacity analysis method for the state of health of a lithium iron phosphate battery, which improves the prediction accuracy of the battery state of health and reduces the computing burden of the BMS controller.

[0004] To solve the above technical problems, the technical solution adopted by the present invention is: An incremental capacity analysis method for the state of health of a lithium iron phosphate battery, comprising: Obtaining voltage and capacity QV data of batteries in different health states; Converting the obtained voltage and capacity data into incremental capacity IC data, and preprocessing the incremental capacity IC data; Extracting the voltage value V1 corresponding to the IC peak, the voltage value V2 after the battery is charged by a fixed capacity Q starting from V1, and the peak height H of the IC peak from the processed incremental capacity IC data; Performing curve fitting on the extracted data V1, V2, H and S0H respectively to obtain the formulas H - SOH, V1 - SOH and V2 - SOH; Predicting S0H based on real vehicle data.

[0005] Further, by performing relative differentiation of the battery capacity Q with respect to the battery voltage V, the Q, V data is converted into IC data.

[0006] Further, the preprocessing of the incremental capacity IC data includes filtering processing and smoothing processing.

[0007] Furthermore, for each charging cycle, the extracted H, V1, V2, and SOH are respectively used to obtain the H-SOH, V1-SOH, and V2-SOH formulas through curve fitting and polynomial fitting in Matlab.

[0008] Furthermore, when charging the actual vehicle, the corresponding H, V1, and V2 of each charging cycle are obtained according to the algorithm. Based on the H-SOH, V1-SOH, and V2-SOH, SOH1, SOH2, and SOH3 are respectively obtained, and correlation analysis and calculation are carried out to obtain the final SOH.

[0009] Furthermore, the correlation analysis and calculation include: S1: Obtain the SOH, SOH1, SOH2, and SOH3 of the last five charging cycles to obtain four sequence arrays respectively: SOH = [SOH(1), SOH(2), SOH(3), SOH(4), SOH(5)]; SOH1 = [SOH1(1), SOH1(2), SOH1(3), SOH1(4), SOH1(5)]; SOH2 = [SOH2(1), SOH2(2), SOH2(3), SOH2(4), SOH2(5)]; SOH3 = [SOH3(1), SOH3(2), SOH3(3), SOH3(4), SOH3(5)]; S2: Calculate the average value of each sequence array and equalize it: mean(SOH) = (SOH(1) + SOH(2) + SOH(3) + SOH(4) + SOH(5)) / 5; mean(SOH1) = (SOH1(1) + SOH1(2) + SOH1(3) + SOH1(4) + SOH1(5)) / 5; mean(SOH2) = (SOH2(1) + SOH2(2) + SOH2(3) + SOH2(4) + SOH2(5)) / 5; mean(SOH3) = (SOH3(1) + SOH3(2) + SOH3(3) + SOH3(4) + SOH3(5)) / 5; ave(SOH) = SOH. / mean(SOH); ave(SOH1) = SOH1. / mean(SOH1); ave(SOH2) = SOH2. / mean(SOH2); ave(SOH3) = SOH3. / mean(SOH3); S3: Obtain the global maximum and minimum values: Concatenate SOH1, SOH2, and SOH3 into a 3×5 array A: [ SOH1(1),SOH1(2),SOH1(3),SOH1(4),SOH1(5); SOH2(1),SOH2(2),SOH2(3),SOH2(4),SOH2(5); SOH3(1),SOH3(2),SOH3(3),SOH3(4),SOH3(5)] S4: Concatenate SOH into a 3×5 array B: [ SOH(1),SOH(2),SOH(3),SOH(4),SOH(5); SOH(1),SOH(2),SOH(3),SOH(4),SOH(5); SOH(1),SOH(2),SOH(3),SOH(4),SOH(5)] min = min(min(A - B)); max = max(max(A - B)); S5: Obtain the correlation coefficient: z1 = (min + 0.5×max). / ((|SOH – SOH1|) + 0.5×max); z2 = (min + 0.5×max). / ((|SOH – SOH2|) + 0.5×max); z3 = (min + 0.5×max). / ((|SOH – SOH3|) + 0.5×max); S6: Calculate the predicted SOH value based on the correlation coefficient and SOH1, SOH2, SOH3 obtained in the current cycle: Pre(SOH) = z1 / (z1 + z2 + z3) × SOH1 + z2 / (z1 + z2 + z3) × SOH2 +z3 / (z1 + z2 + z3) × SOH3.

[0010] Based on the full-cycle offline incremental capacity analysis, the present invention can monitor the SOH of the battery in real time, with a significant improvement in accuracy; the corresponding formula is obtained by the offline incremental capacity analysis method, and the SOH of the battery can be analyzed in real time through the formula, reducing the computational burden on the BMS controller. Brief Description of the Drawings

[0011] The drawings herein are used to provide further illustration of the present invention, and constitute a part of this application. The schematic embodiments of the present invention and their descriptions are used to explain the present invention, and do not constitute an improper limitation to the present invention.

[0012] Figure 1 is a flowchart of the analysis method of the present invention; Figure 2 is a flowchart of voltage and capacity QV data acquisition; Figure 3 is a current and voltage curve; Figure 4 is a QV data curve at 0 cycle at 25°C; Figure 5 is an IC curve at different cycles at 25°C; Figure 6 is a schematic diagram of three characteristic factors obtained by the present invention. Detailed Description of the Embodiments

[0013] In order to enable those skilled in the art to better understand the present invention, the following further clearly and completely describes the present invention with reference to the accompanying drawings and in combination with embodiments. It should be noted that, without conflict, the embodiments in this application and the features in the embodiments can be combined with each other.

[0014] An incremental capacity analysis method for the state of health of a lithium iron phosphate battery provided by a typical embodiment of the present invention, as Figure 1 shown, includes: Obtain voltage and capacity QV data of batteries in different states of health; Convert the obtained voltage and capacity data into incremental capacity IC data, and preprocess the incremental capacity IC data; Extract the voltage value V1 corresponding to the IC peak, the voltage value V2 after the battery is charged with a fixed capacity Q starting from V1, and the peak height H of the IC peak from the processed incremental capacity IC data; Perform curve fitting on the extracted data V1, V2, H and SOH respectively to obtain formula ① H - SOH, formula ② V1 - SOH, and formula ③ V2 - SOH.

[0015] Predict SOH based on in-vehicle data.

[0016] In some embodiments, to obtain voltage and capacity QV data of batteries in different states of health, tests are carried out under working conditions of -20°C, 0°C, 25°C, and 50°C respectively, and the test steps are as Figure 2 shown, including: 1. Let the battery stand still for 5 minutes; 2. Charge the battery with a constant current and constant voltage (CCCV) at a charging rate of 0.3C, as Figure 3 shown; 3. Stand still for 5 min; 4. Discharge the battery at a constant current of 1C until the lower cut-off voltage; 5. Determine whether the number of charge-discharge cycles is greater than 100. If it reaches 100, perform a capacity test to obtain the current maximum available capacity C of the battery. If it is less than 100, repeat the operations in 1, 2, 3, and 4.

[0017] 6. Determine whether the SOH of the battery is less than 80%. If it is less than 80%, end the test. If not, repeat the operations in 1, 2, 3, 4, and 5.

[0018] In some embodiments, when obtaining the QV data curves at different life cycles and different temperatures, first extract the QV data of the current charging cycle at different life cycles and perform smoothing and filtering processing on it. One is filtering processing, including outlier processing and algorithm processing, and the other is smoothing processing, mainly manual processing and algorithm processing.

[0019] Among them, at 25°C, the QV data curve of the 0 cycle is as Figure 4 shown. Since the lithium iron phosphate battery has a large plateau period, during the large available capacity change process, the voltage change range is small, which brings great difficulties to the estimation and correction of SOH. By converting the QV curve into an IC curve, the smooth curve can be converted into a feature curve that is easy to extract.

[0020] In some embodiments, the Q and V data are converted into IC data by taking the relative differential of the battery capacity Q with respect to the battery voltage V.

[0021] Convert the QV data of each charging cycle, dQ / dV = (Q k+1 -Q k ) / (V k+1 -V k ), where k is the sampling record period in each charging cycle. At the same time, perform smoothing and filtering processing on the extracted IC data. One is filtering processing, including outlier processing and algorithm processing, and the other is smoothing processing, mainly manual processing and algorithm processing. The obtained IC curves at different cycle periods at 25°C are as Figure 5 shown.

[0022] Then, extract the characteristic factors from the IC data, as Figure 6 shown. The three characteristic factors proposed in this embodiment are: the peak height H of the Ⅲ peak, the voltage V1 corresponding to the Ⅲ peak, and the voltage V2 after the voltage V1 corresponding to the Ⅲ peak starts charging a fixed capacity Q. Q is the data obtained by calibration.

[0023] Peak I: Refers to an obvious peak in the incremental capacity (IC) curve of a lithium iron phosphate battery, located in the lower voltage region. It is mainly related to the electrochemical reactions at the initial stage of battery charging and reflects the lithium intercalation characteristics of the battery at this stage.

[0024] Peak II: A peak in the medium voltage region after Peak I and before Peak III. The corresponding electrochemical reaction is the lithium intercalation behavior in the middle stage of the battery during charging. In a lithium iron phosphate battery, it is relatively flat compared to Peak I and Peak III.

[0025] Peak III: A peak in the relatively high voltage region, which is a characteristic peak of the lithium iron phosphate battery when it is close to full charge. It is mainly related to the electrochemical reactions of the battery at high voltages and reflects the characteristics of the battery at this stage. The peak height and position are sensitive to the health state of the battery.

[0026] Obtaining method: After obtaining the QV data curves at different life cycles and temperatures, extract the QV data of the current charging cycle at different life cycles and perform smoothing filtering processing on it, including outlier processing and algorithm smoothing processing, etc.

[0027] Convert the QV data to IC data by taking the relative differential of the battery capacity Q with respect to the battery voltage V, that is, calculate dQ / dV = (Qk+1 - Qk) / (Vk+1 - Vk), where k is the sampling record period in each charging cycle. Perform smoothing filtering processing on the extracted IC data again.

[0028] On the processed IC curve, determine Peak I, Peak II, and Peak III according to the characteristics such as the shape, size, and position of the peaks. Generally, identify these three peaks by observing the turning points, slope changes, and the protrusion degree of the peaks on the curve.

[0029] Calibration of battery capacity Q: Experimental test method: Select multiple lithium iron phosphate batteries of the same model, same batch, and in good health state as samples. Conduct multiple complete charge and discharge cycle tests on the sample batteries under standard test conditions (such as an ambient temperature of 25°C, a charging rate of 0.3C, etc.), and record the charging capacity Q values at different cycle periods. Take the average or median of multiple measurements as the calibration value of Q to reduce the influence of random errors.

[0030] Data fitting method: Collect the actual charging data of lithium iron phosphate batteries with a large number of known states of health (SOH) under different working conditions, including data on capacity Q and voltage V. Use data fitting algorithms, such as polynomial fitting, non-linear fitting, etc., to establish a mathematical model between Q and the actual working state of the battery (such as temperature, charge and discharge rate, number of cycles, etc.). By continuously adjusting the model parameters to make the fitting result closest to the actual data, finally determine the calibration formula or parameter value of Q. In practical applications, according to the specific working conditions of the battery, substitute the corresponding parameters to obtain the calibrated Q value.

[0031] Comparison calibration method: Select a lithium iron phosphate battery that has been precisely tested and has a known state of health as a reference battery. Conduct a comparative charging test with other batteries of the same type to be calibrated under the same test conditions, and record the charging capacity difference between the two. Based on the known Q value of the reference battery and the capacity difference between the two, calculate the Q value of the battery to be calibrated, thereby completing the calibration of Q.

[0032] In some embodiments, the offline curve fitting between the characteristic factor and SOH includes: For the H, V1, V2 and SOH extracted for each SOH cycle, curve fitting and polynomial fitting are performed through polyfit in matlab. Preferably, a third-degree polynomial is used to obtain the H-SOH, V1-SOH, and V2-SOH formulas respectively.

[0033] Polynomial fitting and curve fitting can be obtained offline, reducing the real-time monitoring and analysis part in the algorithm, and achieving the purpose of reducing the computing load of the controller by streamlining the algorithm.

[0034] Then, predict the SOH based on the actual vehicle data.

[0035] When the actual vehicle is charging, obtain data such as H, V1, V2, etc. for the corresponding charging cycle according to the algorithm. Based on H-SOH, V1-SOH, and V2-SOH, obtain SOH1, SOH2, and SOH3 respectively. Conduct correlation analysis (algorithm analysis) and calculation based on these three SOHs, and obtain the final SOH.

[0036] Relatively specifically, the above-mentioned correlation analysis and calculation include: S1: Obtain the SOH for the last five charging cycles. SOH, SOH1, SOH2, and SOH3 respectively obtain four sequence arrays: SOH = [SOH(1), SOH(2), SOH(3), SOH(4), SOH(5)]; SOH1 = [SOH1(1), SOH1(2), SOH1(3), SOH1(4), SOH1(5)]; SOH2 = [SOH2(1), SOH2(2), SOH2(3), SOH2(4), SOH2(5)]; SOH3 = [SOH3(1), SOH3(2), SOH3(3), SOH3(4), SOH3(5)]; S2: Obtain the average value of each sequence array and equalize it: mean(SOH) = (SOH(1) + SOH(2) + SOH(3) + SOH(4) + SOH(5)) / 5; mean(SOH1) = (SOH1(1) + SOH1(2) + SOH1(3) + SOH1(4) + SOH1(5)) / 5; mean(SOH2) = (SOH2(1) + SOH2(2) + SOH2(3) + SOH2(4) + SOH2(5)) / 5; mean(SOH3) = (SOH3(1) + SOH3(2) + SOH3(3) + SOH3(4) + SOH3(5)) / 5; ave(SOH) = SOH. / mean(SOH); ave(SOH1) = SOH1. / mean(SOH1); ave(SOH2) = SOH2. / mean(SOH2); ave(SOH3) = SOH3. / mean(SOH3); S3: Obtain the global maximum and minimum values: Concatenate SOH1, SOH2, and SOH3 into a 3×5 array A: [SOH1(1), SOH1(2), SOH1(3), SOH1(4), SOH1(5); SOH2(1), SOH2(2), SOH2(3), SOH2(4), SOH2(5); SOH3(1), SOH3(2), SOH3(3), SOH3(4), SOH3(5)] S4: Concatenate SOH into a 3×5 array B: [SOH(1), SOH(2), SOH(3), SOH(4), SOH(5); SOH(1), SOH(2), SOH(3), SOH(4), SOH(5); SOH(1), SOH(2), SOH(3), SOH(4), SOH(5)] min = min(min(A - B)); max = max(max(A - B)); S4: Obtain the correlation coefficient: z1 = (min + 0.5 * max). / ((abs(SOH - SOH1)) + 0.5 * max); z2 = (min + 0.5 * max). / ((abs(SOH - SOH2)) + 0.5 * max); z3 = (min + 0.5 * max). / ((abs(SOH - SOH3)) + 0.5 * max); S5: Calculate the predicted SOH value based on the correlation coefficient and SOH1, SOH2, SOH3 obtained in the current cycle: Pre(SOH) = z1 / (z1 + z2 + z3) * SOH1 + z2 / (z1 + z2 + z3) * SOH2 + z3 / (z1 + z2 + z3) * SOH3.

[0037] The scope of protection claimed by the present invention is not limited to the above specific embodiments. For those skilled in the art, the present invention can have various deformations and modifications. Any modification, improvement, and equivalent replacement made within the concept and principle of the present invention should be included within the scope of protection of the present invention.

Claims

1. An incremental capacity analysis method for the state of health of a lithium iron phosphate battery, characterized in that, Including: Obtain the voltage and capacity QV data of the battery under different health states; Convert the obtained voltage and capacity data into incremental capacity IC data, and preprocess the incremental capacity IC data; Extract the voltage value V1 corresponding to the IC peak, the voltage value V2 after the battery is charged by a fixed capacity Q starting from V1, and the peak height H of the IC peak from the processed incremental capacity IC data; Perform curve fitting on the extracted data V1, V2, H and S0H respectively to obtain the formulas H-SOH, V1-SOH and V2-SOH; Predict S0H based on real vehicle data.

2. The method according to claim 1, characterized in that: Convert Q, V data into IC data by taking the relative differential of the battery capacity Q with respect to the battery voltage V.

3. The method according to claim 1 or 2, characterized in that: Preprocessing the incremental capacity IC data includes filtering and smoothing.

4. The method according to claim 3, characterized in that: For each charging cycle, the obtained H, V1, V2 and SOH are curve-fitted and polynomially fitted through polyfit in matlab to obtain the above-mentioned H-SOH, V1-SOH, V2-SOH formulas respectively.

5. The method according to claim 4, characterized in that: When charging the vehicle in real time, according to the algorithm, obtain H, V1, V2 of the corresponding charging cycle, and obtain SOH1, SOH2, SOH3 respectively based on H-SOH, V1-SOH, V2-SOH, and perform correlation analysis and calculation to obtain the final SOH.

6. The method according to claim 5, wherein: Correlation analysis and calculation include: S1: Obtain the SOH of the last five charging cycles, and SOH, SOH1, SOH2, SOH3 respectively obtain four sequence arrays: SOH = [SOH(1), SOH(2), SOH(3), SOH(4), SOH(5)]; SOH1 = [SOH1(1), SOH1(2), SOH1(3), SOH1(4), SOH1(5)]; SOH2 = [SOH2(1), SOH2(2), SOH2(3), SOH2(4), SOH2(5)]; SOH3 = [SOH3(1), SOH3(2), SOH3(3), SOH3(4), SOH3(5)]; S2: Calculate the average value of each sequence array and equalize it: mean(SOH) = (SOH(1)+SOH(2)+SOH(3)+SOH(4)+SOH(5)) / 5; mean(SOH1) = (SOH1(1)+SOH1(2)+SOH1(3)+SOH1(4)+SOH1(5)) / 5; mean(SOH2) = (SOH2(1)+SOH2(2)+SOH2(3)+SOH2(4)+SOH2(5)) / 5; mean(SOH3) = (SOH3(1)+SOH3(2)+SOH3(3)+SOH3(4)+SOH3(5)) / 5; ave(SOH) = SOH. / mean(SOH); ave(SOH1) = SOH1. / mean(SOH1); ave(SOH2) = SOH2. / mean(SOH2); ave(SOH3) = SOH3. / mean(SOH3); S3: Obtain the global maximum and minimum values: Concatenate SOH1, SOH2, and SOH3 into a 3×5 array A: [ SOH1(1),SOH1(2),SOH1(3),SOH1(4),SOH1(5); SOH2(1),SOH2(2),SOH2(3),SOH2(4),SOH2(5); SOH3(1),SOH3(2),SOH3(3),SOH3(4),SOH3(5)] S4: Concatenate SOH into a 3×5 array B: [ SOH(1),SOH(2),SOH(3),SOH(4),SOH(5); SOH(1),SOH(2),SOH(3),SOH(4),SOH(5); SOH(1),SOH(2),SOH(3),SOH(4),SOH(5)] min = min(min(A - B)); max = max(max(A - B)); S5: Obtain the correlation coefficient: z1 = (min + 0.5×max). / ((|SOH – SOH1|) + 0.5×max); z2 = (min + 0.5×max). / ((|SOH – SOH2|) + 0.5×max); z3 = (min + 0.5×max). / ((|SOH – SOH3|) + 0.5×max); S6: Calculate the predicted SOH value based on the correlation coefficient and SOH1, SOH2, SOH3 obtained in the current period: Pre(SOH) = z1 / (z1 + z2 + z3) × SOH1 + z2 / (z1 + z2 + z3) × SOH2 + z3 / (z1+ z2 + z3) × SOH3。

Images