A power battery y capacitance prediction system and a prediction method

Abstract

The application discloses a power battery Y capacitor prediction system and a prediction method, calculates the capacitor value when a power battery is tested for voltage resistance in a previous work station, analyzes data of the capacitor value and insulation resistance value, total voltage of the battery, and capacitor value tested by a later work station LCR table, finds corresponding mathematical relations, and the purpose is to calculate a relatively more accurate capacitor value by using the voltage resistance test value and combining the mathematical relations of historical test data in actual production testing, and meanwhile, the test steps of traditional modes such as the LCR table can be saved, and the effect of reducing cost and increasing benefit is achieved. The application establishes a combined prediction model by cross application of multiple regression analysis and BP neural network, the model comprehensively considers influence factors of related items in actual testing and influences of voltage resistance leakage current calculation capacitor value and direct measurement capacitor value, the BP-multiple regression prediction model has better fitting effect when predicting, greatly improves the prediction accuracy, and has better prediction effect.

Description

Technical Field

[0001] This invention relates to power battery testing technology, and in particular to a power battery Y-capacitor prediction system and prediction method. Background Technology

[0002] Y-capacitors in power batteries refer to structural capacitors existing between the positive and negative terminals of the battery and the metal casing or internal liquid cooling pipes of the battery pack. They are formed due to unreasonable internal structural design of the battery. They are mainly divided into two types: one type is a battery pack fixed with a metal tray, where the battery modules are fixed to the tray and the tray to the vehicle using bolts, forming a Y-capacitor between the battery pack electrodes and the tray; the other type is a circulating liquid-cooled battery pack, where metal pipes containing coolant are installed between the battery cells to achieve constant temperature control, forming a Y-capacitor between the battery pack electrodes and the cooling pipes. If the Y-capacitor is too large, the stored energy may cause damage to other components or people; therefore, the capacitance value of the Y-capacitor needs to be controlled.

[0003] During the production and testing of power batteries, according to the national standard (GB / T 18384-2020), capacitive coupling testing is required. On the production line, LCR meters and other methods are generally used at the testing station to test the capacitance value of the positive electrode to the casing and the negative electrode to the casing of the battery pack.

[0004] In addition to Y-capacitor testing, multiple tests are conducted during the production of power batteries, including insulation, withstand voltage, and total voltage. Among them, the leakage current obtained during AC withstand voltage testing can be used to calculate the corresponding capacitance value using a formula.

[0005] Existing technologies mainly use LCR meters or multimeters to directly measure the corresponding capacitance value of power batteries, which is a separate testing step that requires corresponding testing instruments and testing time. Summary of the Invention

[0006] The purpose of this invention is to provide a power battery Y-capacitor prediction system and method. This system calculates the capacitance value during the withstand voltage test at the front-end of the power battery process and analyzes the data with insulation resistance, total battery voltage, and capacitance values ​​measured by an LCR meter at the back-end. The aim is to find the corresponding mathematical relationship, so that in actual production testing, the mathematical relationship between the withstand voltage test value and historical test data can be used to calculate a more accurate capacitance value. This also eliminates the need for traditional testing methods such as LCR meters, achieving cost reduction and efficiency improvement.

[0007] The technical solution of this invention is: A power battery Y-capacitor prediction system includes an insulation withstand voltage tester, a voltage acquisition module, an LCR meter, a computer, and a database; wherein: The insulation withstand voltage tester measures the withstand voltage leakage current and insulation resistance values ​​of the positive and negative terminals of the power battery to the casing. The computer reads the withstand voltage leakage current value from the withstand voltage tester and calculates the corresponding capacitance value. The measured insulation resistance value, leakage current value, and calculated capacitance value are stored in a database. The voltage acquisition module is used to acquire the total voltage of the power battery, the voltage value between the positive terminal of the power battery and the casing, and the voltage value between the negative terminal of the power battery and the casing. The corresponding values ​​are read and measured by a computer and stored in a database. The LCR meter directly measures the capacitance between the positive terminal and the casing, and the negative terminal and the casing of the power battery. The computer reads the corresponding capacitance values ​​measured by the instrument and stores the capacitance values ​​in a database.

[0008] Preferably, the insulation withstand voltage test module, voltage acquisition module, and LCR meter complete their respective tests through a switching module.

[0009] Preferably, the formula for calculating the corresponding capacitance value by reading the leakage current of the withstand voltage tester is as follows: , Where C is the capacitance value, U (AC) denoted as AC voltage amplitude, f as AC voltage frequency, and I and R as withstand voltage leakage current and insulation resistance, respectively.

[0010] Preferably, in the database, a corresponding test dataset is formed from the input data. By establishing a prediction model, the cross-analysis application of BP neural network and multiple regression analysis is used to predict the measured value of Y capacitor, and the relatively more accurate capacitance value of the new battery is predicted.

[0011] A method for predicting the Y-capacitor capacity of a power battery, comprising: Step 1: Input the following data (1) to (3) to form the corresponding test datasets: (1) The withstand voltage leakage current, insulation resistance, and calculated capacitance values ​​of the positive and negative terminals of the power battery tested by the insulation withstand voltage tester; (2) The total voltage of the power battery, the voltage of the positive terminal of the power battery to the casing, and the voltage of the negative terminal of the power battery to the casing collected by the voltage acquisition module; (3) The capacitance values ​​of the positive terminal of the power battery to the casing and the negative terminal of the power battery tested by the LCR meter. Step 2: By establishing a prediction model, the cross-analysis application of BP neural network and multiple regression analysis is used to predict the measured value of Y capacitance. First, we analyze the influencing factors of Y capacitance value and the variation law and trend of capacitance value with voltage, insulation resistance and leakage current. Then, we use stepwise regression analysis to analyze the main influencing factors of Y capacitance value. Finally, we use multiple regression analysis and BP neural network to construct a BP-multiple regression prediction model to predict Y capacitance value and analyze the prediction results.

[0012] Preferably, step two specifically includes the following steps: S1: Obtain the test dataset; S2: Divide the test dataset into the training set and the test set in an 8:2 ratio; S3: Establish a stepwise regression model for screening influencing factors. The stepwise regression model is expressed as follows: Y=a0+a1X1+a2X2+a3X3+a4X4+a5X5; Where Y, X1, X2, X3, X4, and X5 represent the corresponding variables stored in the test; a0, a1, a2…a5 represent the stepwise regression coefficients; S4: Select variables from the test dataset as significant influencing factors to establish a multiple regression prediction model and calculate the predicted value yr of the Y capacitance: yr=a1Yi+a2X1i+a3X2i+a4X3i+a5X4i+a6X5i; Where Yi, X1i, X2i, X3i, X4i, and X5i represent the relevant variables; a1, a2, a3, a4, a5, and a6 represent the variable coefficients; S5: In the BP neural network prediction model, the withstand voltage leakage current value, the calculated capacitance value, and the capacitance value measured by the LCR meter are selected as the input and output for predicting the Y capacitance value. S6: Combining multiple regression and BP neural network models, the final predicted value of Y capacitance is obtained according to the following formula: y = ω1yb + ω2yr; In the formula: y represents the final Y capacitance value; yb represents the Y capacitance value predicted by the BP neural network prediction model; yr represents the Y capacitance value predicted by the multiple regression prediction model; ω1 and ω2 represent the weights of the BP neural network prediction model and the multiple regression prediction model, respectively.

[0013] Preferably, after obtaining the final Y-capacitor prediction value, the accuracy of the algorithm is tested and verified based on the test set data, and the model is optimized to finally obtain the trained Y-capacitor test prediction model; the measured data is input into the trained Y-capacitor prediction model to predict the Y-capacitor of the main positive side of the power battery and the main negative side of the battery casing.

[0014] The advantages of this invention are: 1. This invention utilizes the capacitance value calculated during the withstand voltage test of the power battery at the front-end station, and analyzes the data with the insulation resistance value, total battery voltage, and capacitance value measured by the LCR meter at the back-end station to find the corresponding mathematical relationship. The purpose is to calculate a more accurate capacitance value in actual production testing by combining the withstand voltage test value with the mathematical relationship of historical test data, while eliminating the testing steps of traditional methods such as LCR meters, thus achieving the effect of cost reduction and efficiency improvement.

[0015] 2. This invention establishes a combined prediction model through the cross-application of multiple regression analysis and BP neural network. This model comprehensively considers the influencing factors of the correlation terms in actual testing, as well as the influence of the calculated capacitance value of withstand voltage leakage current and the directly measured capacitance value. The BP-multiple regression prediction model has a better fitting effect in prediction, greatly improving the accuracy of prediction and having a better prediction effect. Attached Figure Description

[0016] The present invention will be further described below with reference to the accompanying drawings and embodiments: Figure 1 This is a diagram of the device for predicting the Y-capacitor in a power battery according to the present invention. Detailed Implementation

[0017] like Figure 1 As shown, the power battery Y-capacitor prediction system of the present invention includes an insulation withstand voltage tester, a voltage acquisition module, an LCR meter, a computer, and a database; wherein: The insulation withstand voltage tester measures the withstand voltage leakage current and insulation resistance values ​​of the positive and negative terminals of the power battery to the casing. The computer reads the withstand voltage leakage current value from the withstand voltage tester and calculates the corresponding capacitance value. The measured insulation resistance value, leakage current value, and calculated capacitance value are stored in a database. The formula for calculating the corresponding capacitance value by reading the leakage current of the withstand voltage tester is as follows: , Where C is the capacitance value, U (AC) denoted as AC voltage amplitude, f as AC voltage frequency, and I and R as withstand voltage leakage current and insulation resistance, respectively.

[0018] The voltage acquisition module is used to acquire the total voltage of the power battery, the voltage value between the positive terminal of the power battery and the casing, and the voltage value between the negative terminal of the power battery and the casing. The corresponding values ​​are read and measured by a computer and stored in a database. The LCR meter directly measures the capacitance between the positive terminal and the casing, and the negative terminal and the casing of the power battery. The computer reads the corresponding capacitance values ​​measured by the instrument and stores the capacitance values ​​in a database.

[0019] The insulation withstand voltage test module, voltage acquisition module, and LCR meter each complete their respective tests through a switching module.

[0020] In the database, a corresponding test dataset is formed from the input data. By establishing a prediction model, the cross-analysis application of BP neural network and multiple regression analysis is used to predict the measured value of Y capacitance, and the capacitance value of the new battery is predicted to be more accurate.

[0021] Specifically, the present invention provides a method for predicting the Y-capacitor capacity of a power battery, comprising: Step 1: Input the following data (1) to (3) to form the corresponding test datasets: (1) The withstand voltage leakage current, insulation resistance, and calculated capacitance values ​​of the positive and negative terminals of the power battery tested by the insulation withstand voltage tester; (2) The total voltage of the power battery, the voltage of the positive terminal of the power battery to the casing, and the voltage of the negative terminal of the power battery to the casing collected by the voltage acquisition module; (3) The capacitance values ​​of the positive terminal of the power battery to the casing and the negative terminal of the power battery tested by the LCR meter. Step 2: By establishing a prediction model, the cross-analysis application of BP neural network and multiple regression analysis is used to predict the measured value of Y capacitance. First, we analyze the influencing factors of Y capacitance value and the variation law and trend of capacitance value with voltage, insulation resistance and leakage current. Then, we use stepwise regression analysis to analyze the main influencing factors of Y capacitance value. Finally, we use multiple regression analysis and BP neural network to construct a BP-multiple regression prediction model to predict Y capacitance value and analyze the prediction results.

[0022] Step two specifically includes the following steps: S1: Obtain the test dataset; S2: Divide the test dataset into the training set and the test set in an 8:2 ratio; S3: Establish a stepwise regression model for screening influencing factors. The stepwise regression model is expressed as follows: Y=a0+a1X1+a2X2+a3X3+a4X4+a5X5; Where Y, X1, X2, X3, X4, and X5 represent the corresponding variables stored in the test; a0, a1, a2…a5 represent the stepwise regression coefficients; S4: Select variables from the test dataset as significant influencing factors to establish a multiple regression prediction model and calculate the predicted value yr of the Y capacitance: yr=a1Yi+a2X1i+a3X2i+a4X3i+a5X4i+a6X5i; Where Yi, X1i, X2i, X3i, X4i, and X5i represent the relevant variables; a1, a2, a3, a4, a5, and a6 represent the variable coefficients; S5: In the BP neural network prediction model, the withstand voltage leakage current value, the calculated capacitance value, and the capacitance value measured by the LCR meter are selected as the input and output for predicting the Y capacitance value. S6: Combining multiple regression and BP neural network models, the final predicted value of Y capacitance is obtained according to the following formula: y = ω1yb + ω2yr; In the formula: y represents the final Y capacitance value; yb represents the Y capacitance value predicted by the BP neural network prediction model; yr represents the Y capacitance value predicted by the multiple regression prediction model; ω1 and ω2 represent the weights of the BP neural network prediction model and the multiple regression prediction model, respectively. S7: After obtaining the final Y capacitance prediction value, the accuracy of the algorithm is tested and verified based on the test set data, and the model is optimized to finally obtain the trained Y capacitance test prediction model. S8: Input the measured data into the trained Y-capacitor prediction model to predict the Y-capacitor of the main positive side of the power battery and the main negative side of the battery casing.

[0023] This invention establishes a combined prediction model through the cross-application of multiple regression analysis and BP neural network. This model comprehensively considers the influence of correlation factors in actual testing as well as the influence of the calculated capacitance value of withstand voltage leakage current and the directly measured capacitance value. The BP-multiple regression prediction model has a better fitting effect in prediction, greatly improving the accuracy of prediction and having a better prediction effect.

[0024] The above embodiments are only for illustrating the technical concept and features of the present invention, and are intended to enable those skilled in the art to understand the content of the present invention and implement it accordingly. They should not be construed as limiting the scope of protection of the present invention. All modifications made according to the spirit and essence of the main technical solution of the present invention should be covered within the scope of protection of the present invention.

Claims

1. A power battery Y-capacitor prediction system, characterized in that, Includes an insulation withstand voltage tester, voltage acquisition module, LCR meter, computer, and database; among which: The insulation withstand voltage tester measures the withstand voltage leakage current and insulation resistance values ​​of the positive and negative terminals of the power battery to the casing. The computer reads the withstand voltage leakage current value from the withstand voltage tester and calculates the corresponding capacitance value. The measured insulation resistance value, leakage current value, and calculated capacitance value are stored in a database. The voltage acquisition module is used to acquire the total voltage of the power battery, the voltage value between the positive terminal of the power battery and the casing, and the voltage value between the negative terminal of the power battery and the casing. The corresponding values ​​are read and measured by a computer and stored in a database. The LCR meter directly measures the capacitance value of the positive electrode to the casing and the negative electrode to the casing of the power battery, and reads the corresponding capacitance value through a computer and stores the capacitance value in a database. Input the following data (1) to (3) to form the corresponding test datasets: (1) The withstand voltage leakage current, insulation resistance, and calculated capacitance values ​​of the positive and negative terminals of the power battery tested by the insulation withstand voltage tester; (2) The total voltage of the power battery, the voltage of the positive terminal of the power battery to the casing, and the voltage of the negative terminal of the power battery to the casing collected by the voltage acquisition module; (3) The capacitance values ​​of the positive terminal of the power battery to the casing and the negative terminal of the power battery tested by the LCR meter. By establishing a predictive model, the cross-analysis application of BP neural network and multiple regression analysis is used to predict the measured value of Y capacitance. First, we analyze the influencing factors of Y capacitance value and the variation law and trend of capacitance value with voltage, insulation resistance and leakage current. Then, we use stepwise regression analysis to analyze the main influencing factors of Y capacitance value. Finally, we use multiple regression analysis and BP neural network to construct a BP-multiple regression prediction model to predict Y capacitance value and analyze the prediction results. Obtain the test dataset; The test dataset was divided into the training and test sets in an 8:2 ratio. A stepwise regression model is established for screening influencing factors. The stepwise regression model is expressed as follows: Y=a0+a1X1+a2X2+a3X3+a4X4+a5X5; Where Y, X1, X2, X3, X4, and X5 represent the corresponding variables stored in the test; a0, a1, a2…a5 represent the stepwise regression coefficients; A multiple regression prediction model was established by selecting variables from the test dataset as significant influencing factors to calculate the predicted value yr of the Y capacitance: yr=a1Yi+a2X1i+a3X2i+a4X3i+a5X4i+a6X5i; Where Yi, X1i, X2i, X3i, X4i, and X5i represent the relevant variables; a1, a2, a3, a4, a5, and a6 represent the variable coefficients; In the BP neural network prediction model, the withstand voltage leakage current value, the calculated capacitance value, and the capacitance value measured by the LCR meter are selected as the input and output for predicting the Y capacitance value. Combining multiple regression and BP neural network models, the final predicted value of Y capacitance is obtained according to the following formula: y = ω1yb + ω2yr; In the formula: y represents the final Y capacitance value; yb represents the Y capacitance value predicted by the BP neural network prediction model; yr represents the Y capacitance value predicted by the multiple regression prediction model; ω1 and ω2 represent the weights of the BP neural network prediction model and the multiple regression prediction model, respectively.

2. The power battery Y-capacitor prediction system according to claim 1, characterized in that, The insulation withstand voltage test module, voltage acquisition module, and LCR meter each complete their respective tests through a switching module.

3. The power battery Y-capacitor prediction system according to claim 1, characterized in that, The formula for calculating the corresponding capacitance value by reading the leakage current of the withstand voltage tester is as follows: , Where C is the capacitance value, U (AC) denoted as AC voltage amplitude, f as AC voltage frequency, and I and R as withstand voltage leakage current and insulation resistance, respectively.

4. The power battery Y-capacitor prediction system according to claim 3, characterized in that, In the database, a corresponding test dataset is formed from the input data. By establishing a prediction model, the cross-analysis application of BP neural network and multiple regression analysis is used to predict the measured value of Y capacitance, and the capacitance value of the new battery is predicted to be more accurate.

5. A method for predicting the Y-capacitor capacity of a power battery, characterized in that, include: Step 1: Input the following data (1) to (3) to form the corresponding test datasets: (1) The withstand voltage leakage current, insulation resistance, and calculated capacitance values ​​of the positive and negative terminals of the power battery tested by the insulation withstand voltage tester; (2) The total voltage of the power battery, the voltage of the positive terminal of the power battery to the casing, and the voltage of the negative terminal of the power battery to the casing collected by the voltage acquisition module; (3) The capacitance values ​​of the positive terminal of the power battery to the casing and the negative terminal of the power battery tested by the LCR meter. Step 2: By establishing a prediction model, the cross-analysis application of BP neural network and multiple regression analysis is used to predict the measured value of Y capacitance. First, we analyze the influencing factors of Y capacitance value and the variation law and trend of capacitance value with voltage, insulation resistance and leakage current. Then, we use stepwise regression analysis to analyze the main influencing factors of Y capacitance value. Finally, we use multiple regression analysis and BP neural network to construct a BP-multiple regression prediction model to predict Y capacitance value and analyze the prediction results. Step two specifically includes the following steps: S1: Obtain the test dataset; S2: Divide the test dataset into the training set and the test set in an 8:2 ratio; S3: Establish a stepwise regression model for screening influencing factors. The stepwise regression model is expressed as follows: Y=a0+a1X1+a2X2+a3X3+a4X4+a5X5; Where Y, X1, X2, X3, X4, and X5 represent the corresponding variables stored in the test; a0, a1, a2…a5 represent the stepwise regression coefficients; S4: Select variables from the test dataset as significant influencing factors to establish a multiple regression prediction model and calculate the predicted value yr of the Y capacitance: yr=a1Yi+a2X1i+a3X2i+a4X3i+a5X4i+a6X5i; Where Yi, X1i, X2i, X3i, X4i, and X5i represent the relevant variables; a1, a2, a3, a4, a5, and a6 represent the variable coefficients; S5: In the BP neural network prediction model, the withstand voltage leakage current value, the calculated capacitance value, and the capacitance value measured by the LCR meter are selected as the input and output for predicting the Y capacitance value. S6: Combining multiple regression and BP neural network models, the final predicted value of Y capacitance is obtained according to the following formula: y = ω1yb + ω2yr; In the formula: y represents the final Y capacitance value; yb represents the Y capacitance value predicted by the BP neural network prediction model; yr represents the Y capacitance value predicted by the multiple regression prediction model; ω1 and ω2 represent the weights of the BP neural network prediction model and the multiple regression prediction model, respectively.

6. The method for predicting the Y-capacitor capacity of a power battery according to claim 5, characterized in that, After obtaining the final Y-capacitor prediction value, the accuracy of the algorithm is tested and verified based on the test set data, and the model is optimized to finally obtain the trained Y-capacitor test prediction model. The measured data is input into the trained Y-capacitor prediction model to predict the Y-capacitor of the main positive side of the power battery and the main negative side of the battery casing.

Images