Battery discharge rate and temperature rise prediction method

Abstract

The invention discloses a battery discharge rate and temperature rise prediction method, and belongs to the technical field of battery prediction. Comprising the following steps: acquiring polarization resistance Rx, polarization capacitance Cx, open-circuit voltage VOC and internal resistance Rs of a battery, building an equivalent circuit model, optimizing the equivalent circuit model by using a relation curve L to obtain an optimized model, and calculating the polarization resistance Rx, the polarization capacitance Cx, the open-circuit voltage VOC and the internal resistance Rs of the battery. Solving and iterating the open-circuit voltage VOC corresponding to the discrete temperature by using a square root of a state covariance matrix of a square root volume Kalman filtering algorithm, introducing a compensation model, optimizing an optimization model by using the compensation model to obtain a pre-estimation model, calculating a theoretical temperature rise value of the pre-estimated discharge depth in combination with the pre-estimation model, and calculating the temperature rise value of the pre-estimated discharge depth in combination with a current cell temperature rise deviation. The current environment temperature is used for predicting the temperature of the discharge tail end. According to the invention, the parameters of the prediction model are dynamically adjusted according to the temperature change of the actual operation environment of the battery, and the prediction accuracy of the discharge end temperature of the battery is improved.

Description

Technical Field

[0001] The invention belongs to the technical field of battery prediction, and in particular relates to a method for predicting battery discharge rate and temperature rise. Background Art

[0002] During battery use, factors such as discharge current and ambient temperature have a significant impact on battery temperature changes. Accurately estimating the temperature rise at the end of battery discharge is of great significance for the safe use, performance and life extension of the battery.

[0003] The existing prediction methods for batteries include physical models, machine learning, and other prediction methods. Among them, by establishing a physical model of the battery, taking into account the chemical reaction of the battery, the heat conduction and heat radiation inside the battery, and other factors, the temperature rise of the battery is modeled and predicted using mathematical equations. Using machine learning algorithms, a prediction model is established by training the temperature rise data and related features of the battery. The supervised learning method can be used to train the model with known battery conditions and temperature rise data, and then predict unknown conditions. However, whether it is based on physical models or machine learning algorithms, the requirements for hardware computing power are relatively high, and it is difficult to apply on embedded systems. In addition, the complex actual environment and the changing operating conditions lead to large errors in the prediction results, resulting in a shortened battery life. Summary of the invention

[0004] Purpose of the invention: To provide a method for predicting battery discharge rate and temperature rise, which solves the above problems existing in the prior art.

[0005] Technical solution: A method for predicting battery discharge rate and temperature rise, comprising the following steps:

[0006] Step 1: Get the polarization resistance R of the battery x , polarization capacitance C x , open circuit voltage V OC and the battery internal resistance R s , build an equivalent circuit model;

[0007] Step 2: Measure the open circuit voltage V corresponding to the battery at a predetermined temperature. OC , obtain the temperature and open circuit voltage V at the same current rate OC The relationship curve L 曲 ;

[0008] Step 3: Use the nonlinear curve fitting function of Fit to obtain the open circuit voltage V at different temperatures OC The functional relationship of the state of charge is obtained by fitting the temperature and the open circuit voltage V OC The relationship curve L 合 Then, using the relationship curve L 合 Optimizing the equivalent circuit model to obtain an optimized model;

[0009] Step 4: Use the square root of the state covariance matrix of the square root cubature Kalman filter algorithm to calculate the open circuit voltage V corresponding to the discrete temperature OC Solve the iterative solution, introduce the compensation model, and use the compensation model to optimize the optimization model to obtain the estimation model;

[0010] Step 5: Measure the real-time ambient temperature and discharge current of the battery online, calculate the discharge rate of the current discharge current of the battery, and update the temperature rise of the battery at the end of discharge in real time in combination with the estimation model. Calculate the theoretical temperature rise value of the estimated discharge depth, add the current battery cell temperature rise deviation, as the estimated temperature rise at the end of discharge, and add the current ambient temperature to predict the temperature at the end of discharge.

[0011] Preferably, the construction process of the equivalent circuit model is as follows:

[0012] Step 11: Obtain the polarization resistance R of the battery x , polarization capacitance C x and the voltage V across the battery circuit x , open circuit voltage V OC and the battery internal resistance R s As sample data;

[0013] Step 12: Calculate the battery charge and discharge current I by using formula (1) batt

[0014]

[0015] Where: t represents time, d represents voltage V x The derivative with respect to time t, x represents the order of the battery circuit;

[0016] Step 13: Calculate the battery terminal voltage V using formula (2): batt

[0017] V batt =V oc -R s I batt -V1-…-V x (2)

[0018] Step 14: Obtain the battery charge and discharge current and the battery terminal voltage and build an equivalent circuit model using formula (3). Formula (3) is as follows:

[0019]

[0020] Where: D is the penalty term of the fitting, D is an integer greater than 1; m e is the number of sample data; i represents the iteration variable, represents the rate of exponential decay over time t.

[0021] Preferably, τ in formula (3) x Through the battery's polarization resistance R x and polarization capacitance C x The formula (4) is used to calculate the following:

[0022] τ x =R x Cx (4).

[0023] Preferably, the relationship curve L between temperature and open circuit voltage at the same current rate is 曲 The acquisition process is as follows:

[0024] Step 21, under the same preset current magnification, adjust the preset temperature from low to high;

[0025] Step 22: During the preset temperature adjustment process, discharge the battery. When the battery is discharged and left to stand, measure the current open circuit voltage V OC ;

[0026] Step 23: Obtain the open circuit voltage V corresponding to the changing temperature OC , establish the temperature and open circuit voltage V OC The relationship curve L 曲 .

[0027] Preferably, the nonlinear curve fitting function of Fit in step 3 is used to obtain the open circuit voltage V at different temperatures. OC The fitting process of the functional relationship of the state of charge is as follows:

[0028] Step 31: Get the octave polynomial to build V OC -SOC octave polynomial model, where the octave polynomial formula is shown in formula (5):

[0029]

[0030] Where: C 初 Indicates the capacitance of the battery when it starts to discharge, Z i represents the fitting parameter, a is an integer 8;

[0031] Step 32: When the open circuit voltage V OC After fitting the functional relationship of the state of charge, the root mean square error, correlation coefficient and correction determination coefficient of the fitting results at the predetermined temperature are tested to determine the fitting relationship curve L of the octave polynomial model. 合 The precise state of

[0032] When the root mean square error of the fitting result at the predetermined temperature is less than 0.01V, and the correlation coefficient and the correction determination coefficient are greater than or equal to 0.99, it means that the eighth-order polynomial model fits well and the fitting relationship curve L 合 The accuracy of the fitting relationship curve L 合 of rough.

[0033] Preferably, the square root of the state covariance matrix of the square root cubature Kalman filter algorithm is used to calculate the open circuit voltage V corresponding to the discrete temperature. OC The process of solving the iteration is as follows:

[0034] Step 41, calculating the sampling volume point of discrete temperature and the weight of the sampling volume point;

[0035] Step 42: After completing the calculation of the sampling volume point and the weight of the sampling volume point, the predicted volume point is obtained by propagating the state equation;

[0036] Step 43: Obtain the sampling volume point, the weight of the sampling volume point and the prediction volume point, calculate the dynamic prediction value at the same time, and the square root of the prediction error covariance matrix, and complete the corresponding open circuit voltage V by calculating the square root of the dynamic prediction value and the prediction error covariance matrix. OC Solve the iterations.

[0037] Preferably, the calculation formula of the sampling volume points of the discrete temperature is as follows:

[0038]

[0039] Where: k (t) represents the kth state vector at time point t, ξ k represents the kth sampling volume point, (t) represents the predicted value at time point t;

[0040] The calculation formula of the weight of the discrete temperature sampling volume point is as follows:

[0041]

[0042] Where: M represents the dimension of the state vector, j represents the discrete temperature iteration variable, and w ij represents the i-th j The weight of each volume point.

[0043] Preferably, the calculation formula for the predicted volume point obtained by propagating the state equation is as follows:

[0044] X b (t+1 / t)=f(X b (t+1 / t),t) (8);

[0045] Wherein: t+1|t represents the prediction made at time point t+1 based on the information at time point t, and b represents the bth predicted volume point.

[0046] Preferably, the calculation formula for calculating the dynamic prediction value at the same time is as follows:

[0047]

[0048] The square root of the forecast error covariance matrix is ​​calculated as follows:

[0049] S(t+1 / t)=Tria(λ N (t+1 / t,S Q (t))) (10);

[0050]

[0051] Where: S Q represents the RMS value of the state noise covariance matrix Q, λ N (t+1 / t) represents the weighted center matrix.

[0052] Preferably, the compensation model building process is as follows:

[0053] Select a third-order polynomial to convert V OC -The fitting parameter Z in the SOC octave polynomial model is fitted with the temperature at which the fitting parameter Z is located, and the fitting formula is as follows:

[0054] Z i =Y ji T j +k (j-1)i T j-1 +…+Y (j-j)i (12);

[0055] Where: k ji represents the fitting coefficient of the compensation model, j = 0, 1, 2, 3; T represents the initial discharge capacity of the battery;

[0056] At the predetermined temperature V OC -After the fitting parameter Z in the SOC octave polynomial model is fitted with the temperature at which the fitting parameter Z is located, it is combined with the octave polynomial formula to complete the construction of the compensation model.

[0057] Beneficial effect: The present invention relates to a method for predicting battery discharge rate and temperature rise, using the polarization resistance R of the battery x , polarization capacitance C x , open circuit voltage V OC and the battery internal resistance R s, build an equivalent circuit model, and optimize the equivalent circuit to obtain an optimization model, and introduce a compensation model. Use the compensation model to optimize the optimization model to obtain a prediction model. Combine the prediction model with the real-time measured ambient temperature and discharge current, and update the battery discharge end temperature in real time to predict the battery discharge end temperature. By dynamically adjusting the parameters of the prediction model according to the temperature changes of the actual operating environment of the battery, the prediction accuracy of the battery discharge end temperature is improved. According to the real-time temperature prediction results, the battery charging power and discharge depth can be adjusted to avoid overcharging or over-discharging, thereby extending the battery life. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Figure 1 is a system block diagram of the present invention;

[0059] Figure 2 It is a flow chart of the discharge depth of the present invention;

[0060] Figure 3 It is a temperature rise process flow chart of the present invention;

[0061] Figure 4 It is a flowchart of the real-time temperature rise deviation estimation process of the present invention;

[0062] Figure 5 This is a flowchart of an update process for estimating discharge depth of the present invention. DETAILED DESCRIPTION

[0063] like Figures 1 to 5 As shown, the present invention provides a technical solution: a method for predicting battery discharge rate and temperature rise, comprising the following steps:

[0064] Step 1: Get the polarization resistance R of the battery x , polarization capacitance C x , open circuit voltage V OC and the battery internal resistance R s , build an equivalent circuit model, wherein the building process of the equivalent circuit model is as follows:

[0065] Step 11: Obtain the polarization resistance R of the battery x , polarization capacitance C x and the voltage V across the battery circuit x , open circuit voltage V OC and the battery internal resistance R s As sample data;

[0066] Step 12: Calculate the battery charge and discharge current I by using formula (1) batt

[0067]

[0068] Where: t represents time, d represents voltage V x The derivative with respect to time t, x represents the order of the battery circuit;

[0069] Step 13: Calculate the battery terminal voltage V using formula (2): batt

[0070] V batt =V oc -R s I batt -V1-…-V x (2)

[0071] Step 14: Obtain the battery charge and discharge current and the battery terminal voltage and build an equivalent circuit model using formula (3). Formula (3) is as follows:

[0072]

[0073] Where: D is the penalty term of the fitting, D is an integer greater than 1; m e is the number of sample data; i represents the iteration variable, represents the rate of exponential decay over time t.

[0074] Preferably, τ in formula (3) x Through the battery's polarization resistance R x and polarization capacitance C x The formula (4) is used to calculate the following:

[0075] τ x =R x C x (4).

[0076] Step 2: Measure the open circuit voltage V corresponding to the battery at a predetermined temperature. OC , obtain the temperature and open circuit voltage V at the same current rate OC The relationship curve L 曲 , the relationship curve between temperature and open circuit voltage at the same current rate L 曲 The acquisition process is as follows:

[0077] Step 21, under the same preset current magnification, adjust the preset temperature from low to high;

[0078] Step 22: During the preset temperature adjustment process, discharge the battery. When the battery is discharged and left to stand, measure the current open circuit voltage V OC ;

[0079] Step 23: Obtain the open circuit voltage V corresponding to the changing temperature OC , establish the temperature and open circuit voltage V OCThe relationship curve L 曲 .

[0080] Step 3: Use the nonlinear curve fitting function of Fit to obtain the open circuit voltage V at different temperatures OC The functional relationship of the state of charge is as follows:

[0081] Step 31: Get the octave polynomial to build V OC -SOC octave polynomial model, where the octave polynomial formula is shown in formula (5):

[0082]

[0083] Where: C 初 Indicates the capacitance of the battery when it starts to discharge, Z i represents the fitting parameter, a is an integer 8;

[0084] Step 32: When the open circuit voltage V OC After fitting the functional relationship of the state of charge, the root mean square error, correlation coefficient and correction determination coefficient of the fitting results at the predetermined temperature are tested to determine the fitting relationship curve L of the octave polynomial model. 合 The precise state of

[0085] When the root mean square error of the fitting result at the predetermined temperature is less than 0.01V, and the correlation coefficient and the correction determination coefficient are greater than or equal to 0.99, it means that the eighth-order polynomial model fits well and the fitting relationship curve L 合 The accuracy of the fitting relationship curve L 合 The rough result of the fitting is the temperature and open circuit voltage V OC The relationship curve L 合 Then, using the relationship curve L 合 The equivalent circuit model is optimized to obtain an optimized model. The parameters of the prediction model are dynamically adjusted according to the temperature changes in the actual operating environment of the battery to improve the prediction accuracy of the battery discharge terminal temperature.

[0086] Step 4: Use the square root of the state covariance matrix of the square root cubature Kalman filter algorithm to calculate the open circuit voltage V corresponding to the discrete temperature OC Solve the iteration, where the process of solving the iteration is as follows:

[0087] Step 41, calculating the sampling volume points of discrete temperatures and the weights of the sampling volume points, the calculation formula of the sampling volume points of discrete temperatures is as follows:

[0088]

[0089] Where: k(t) represents the kth state vector at time point t, ξ k represents the kth sampling volume point, (t) represents the predicted value at time point t;

[0090] The calculation formula of the weight of the discrete temperature sampling volume point is as follows:

[0091]

[0092] Where: M represents the dimension of the state vector, j represents the discrete temperature iteration variable, and w ij represents the i-th j The weight of each volume point.

[0093] Step 42: After the sampling volume point and the weight calculation of the sampling volume point are completed, the predicted volume point is obtained by using the state equation propagation. The calculation formula of the predicted volume point obtained by the state equation propagation is as follows:

[0094] X b (t+1 / t)=f(X b (t+1 / t),t) (8);

[0095] Wherein: t+1|t represents the prediction made at time point t+1 based on the information at time point t, and b represents the bth predicted volume point.

[0096] Step 43, obtain the sampling volume point, the weight of the sampling volume point and the prediction volume point, calculate the dynamic prediction value at the same time, and the square root of the prediction error covariance matrix. The calculation formula for calculating the dynamic prediction value at the same time is as follows:

[0097]

[0098] The square root of the forecast error covariance matrix is ​​calculated as follows:

[0099] S(t+1 / t)=Tria(λ N (t+1 / t,S Q (t))) (10);

[0100]

[0101] Where: S Q represents the RMS value of the state noise covariance matrix Q, λ N (t+1 / t) represents the weighted center matrix. After the square root of the dynamic prediction value and the prediction error covariance matrix is ​​calculated, the corresponding open circuit voltage V is completed by calculating the square root of the dynamic prediction value and the prediction error covariance matrix. OC Solve the iteration and complete the open circuit voltage V OCAfter solving the iteration, the compensation model is introduced, and the compensation model construction process is as follows:

[0102] Select a third-order polynomial to convert V OC -The fitting parameter Z in the SOC octave polynomial model is fitted with the temperature at which the fitting parameter Z is located, and the fitting formula is as follows:

[0103] Z i =Y ji T j +k (j-1)i T j-1 +…+Y (j-j)i (12);

[0104] Where: k ji represents the fitting coefficient of the compensation model, j = 0, 1, 2, 3; T represents the initial discharge capacity of the battery;

[0105] At the predetermined temperature V OC -After fitting the fitting parameter Z in the SOC octahedral polynomial model and the temperature at which the fitting parameter Z is located, the compensation model is constructed in conjunction with the octahedral polynomial formula, and the compensation model is used to optimize the optimization model to obtain the estimation model;

[0106] Step 5. Measure the real-time ambient temperature and discharge current of the battery online, calculate the discharge rate of the current discharge current of the battery, and update the temperature rise of the battery discharge end in real time in combination with the estimation model. Calculate the theoretical temperature rise value of the estimated discharge depth, add the current battery cell temperature rise deviation, as the estimated temperature rise of the discharge end, plus the current ambient temperature, to predict the temperature of the discharge end. By changing the temperature of the actual operating environment of the battery, dynamically adjust the parameters of the prediction model to improve the prediction accuracy of the battery discharge end temperature. According to the real-time temperature prediction results, the charging power and discharge depth of the battery can be adjusted to avoid overcharging or over-discharging, thereby extending the battery life.

[0107] The preferred embodiments of the present invention are described in detail above; however, the present invention is not limited to the specific details in the above embodiments. Within the technical concept of the present invention, various equivalent transformations can be made to the technical solutions of the present invention, and these equivalent transformations all belong to the protection scope of the present invention.

Claims

1. A method for predicting battery discharge rate and temperature rise, characterized in that: The following steps are involved: Step 1: Get the polarization resistance R of the battery x , polarization capacitance C x , open circuit voltage V OC and the battery internal resistance R s , build an equivalent circuit model; Step 2: Measure the open circuit voltage V corresponding to the battery at a predetermined temperature. OC , obtain the temperature and open circuit voltage V at the same current rate OC The relationship curve L 曲 ; Step 3: Use the nonlinear curve fitting function of Fit to obtain the open circuit voltage V at different temperatures OC The functional relationship of the state of charge is obtained by fitting the temperature and the open circuit voltage V OC The relationship curve L 合 Then, using the relationship curve L 合 Optimizing the equivalent circuit model to obtain an optimized model; Step 4: Use the square root of the state covariance matrix of the square root cubature Kalman filter algorithm to calculate the open circuit voltage V corresponding to the discrete temperature OC Solve the iterative solution, introduce the compensation model, and use the compensation model to optimize the optimization model to obtain the estimation model; Step 5: Measure the real-time ambient temperature and discharge current of the battery online, calculate the discharge rate of the current discharge current of the battery, and update the temperature rise of the battery at the end of discharge in real time in combination with the estimation model. Calculate the theoretical temperature rise value of the estimated discharge depth, add the current battery cell temperature rise deviation, as the estimated temperature rise at the end of discharge, and add the current ambient temperature to predict the temperature at the end of discharge.

2. A battery discharge rate and temperature rise prediction method according to claim 1, characterized in that: The construction process of the equivalent circuit model is as follows: Step 11: Obtain the polarization resistance R of the battery x , polarization capacitance C x and the voltage V across the battery circuit x , open circuit voltage V OC and the battery internal resistance R s As sample data; Step 12: Calculate the battery charge and discharge current I by using formula (1) batt Where: t represents time, d represents voltage V x The derivative with respect to time t, x represents the order of the battery circuit; Step 13: Calculate the battery terminal voltage V using formula (2): batt V batt =V oc -R s I batt -V1-…-V x (2) Step 14: Obtain the battery charge and discharge current and the battery terminal voltage and build an equivalent circuit model using formula (3). Formula (3) is as follows: Where: D is the penalty term of the fitting, D is an integer greater than 1; m e is the number of sample data; i represents the iteration variable, represents the rate of exponential decay over time t.

3. A battery discharge rate and temperature rise prediction method according to claim 2, characterized in that: τ in formula (3) x Through the battery's polarization resistance R x and polarization capacitance C x The formula (4) is used to calculate the following: t x =R x C x (4)。 4. A battery discharge rate and temperature rise prediction method according to claim 1, characterized in that: The relationship curve L between temperature and open circuit voltage at the same current rate 曲 The acquisition process is as follows: Step 21, under the same preset current magnification, adjust the preset temperature from low to high; Step 22: During the preset temperature adjustment process, discharge the battery. When the battery is discharged and left to stand, measure the current open circuit voltage V OC ; Step 23: Obtain the open circuit voltage V corresponding to the changing temperature OC , establish the temperature and open circuit voltage V OC The relationship curve L 曲 .

5. A battery discharge rate and temperature rise prediction method according to claim 1, characterized in that: The nonlinear curve fitting function of Fit in step 3 is used to obtain the open circuit voltage V at different temperatures OC The fitting process of the functional relationship of the state of charge is as follows: Step 31: Get the octave polynomial to build V OC -SOC octave polynomial model, where the octave polynomial formula is shown in formula (5): Where: C 初 Indicates the capacitance of the battery when it starts to discharge, Z i represents the fitting parameter, a is an integer 8; Step 32: When the open circuit voltage V OC After fitting the functional relationship of the state of charge, the root mean square error, correlation coefficient and correction determination coefficient of the fitting results at the predetermined temperature are tested to determine the fitting relationship curve L of the octave polynomial model. 合 The precise state of When the root mean square error of the fitting result at the predetermined temperature is less than 0.01V, and the correlation coefficient and the correction determination coefficient are greater than or equal to 0.99, it means that the eighth-order polynomial model fits well and the fitting relationship curve L 合 The accuracy of the fitting relationship curve L 合 of rough.

6. A battery discharge rate and temperature rise prediction method according to claim 5, characterized in that: The square root of the state covariance matrix of the square root cubature Kalman filter algorithm is used to calculate the open circuit voltage V corresponding to the discrete temperature. OC The process of solving the iteration is as follows: Step 41, calculating the sampling volume point of discrete temperature and the weight of the sampling volume point; Step 42: After completing the calculation of the sampling volume point and the weight of the sampling volume point, the predicted volume point is obtained by propagating the state equation; Step 43: Obtain the sampling volume point, the weight of the sampling volume point and the prediction volume point, calculate the dynamic prediction value at the same time, and the square root of the prediction error covariance matrix, and complete the corresponding open circuit voltage V by calculating the square root of the dynamic prediction value and the prediction error covariance matrix. OC Solve the iterations.

7. A battery discharge rate and temperature rise prediction method according to claim 6, characterized in that: The calculation formula of the sampling volume point of the discrete temperature is as follows: Where: k (t) represents the kth state vector at time point t, ξ k represents the kth sampling volume point, (t) represents the predicted value at time point t; The calculation formula of the weight of the discrete temperature sampling volume point is as follows: Where: M represents the dimension of the state vector, j represents the discrete temperature iteration variable, and w ij represents the i-th j The weight of each volume point.

8. A battery discharge rate and temperature rise prediction method according to claim 6, characterized in that: The calculation formula for the predicted volume point obtained by propagating the state equation is as follows: X b (t+1 / t)=f(X b (t+1 / t),t) (8); Wherein: t+1|t represents the prediction made at time point t+1 based on the information at time point t, and b represents the bth predicted volume point.

9. A battery discharge rate and temperature rise prediction method according to claim 8, characterized in that: The calculation formula for calculating the dynamic prediction value at the same time is as follows: The square root of the forecast error covariance matrix is ​​calculated as follows: S(t+1 / t)=Tria(λ N (t+1 / t,S Q (t))) (10); Where: S Q represents the root mean square value of the state noise covariance matrix Q, and λ N (t + 1|t) represents the weighted center matrix.

10. A battery discharge rate and temperature rise prediction method according to claim 6, characterized in that: The compensation model building process is as follows: Select a third-order polynomial to convert V OC -The fitting parameter Z in the SOC octave polynomial model is fitted with the temperature at which the fitting parameter Z is located, and the fitting formula is as follows: Z i =Y ji T j +k (j-1)i T j-1 +…+Y (j-j)i (12); Where: k ji represents the fitting coefficient of the compensation model, j = 0, 1, 2, 3; T represents the initial discharge capacity of the battery; At the predetermined temperature V OC -After the fitting parameter Z in the SOC octave polynomial model is fitted with the temperature at which the fitting parameter Z is located, it is combined with the octave polynomial formula to complete the construction of the compensation model.

Images