Rapid Health Testing Method for Continuous Variation Pulse Width Current Excitation of Automotive Battery Packs

Abstract

This invention discloses a rapid health detection method for automotive battery packs using continuous variable pulse width current excitation. The invention designs a pulse charging current waveform, including first to third waveforms. The first waveform is used to charge the automotive battery pack, collecting terminal voltage and current measurements. Initial battery parameters are identified offline to determine the optimal battery model, resulting in second and third waveforms. While the second and third waveforms are charging the battery pack, battery parameters are identified online at different time scales to accurately estimate the battery pack's state of charge (SOC) and state of health (SOH) and determine their convergence. This invention enables an active current excitation method to obtain ideal data conducive to battery health detection, selects a suitable battery model, and ultimately achieves accurate battery health detection through parameter identification and state estimation methods at multiple time scales.

Description

Technical Field

[0001] This invention relates to automotive battery pack testing technology, specifically to a rapid health testing method for automotive battery packs using continuous variable pulse width current excitation. Background Technology

[0002] The power battery pack accounts for a significant portion of the cost of a pure electric vehicle. Accurate testing and evaluation of the actual usable capacity and health of the battery pack will benefit vehicle residual value assessment and promote the circulation of the used car market. However, technical challenges such as the difficulty in obtaining real-time operating data, the lack of basic battery data, and the difficulty in accurately identifying model parameters and state variables make it difficult for organizations other than vehicle manufacturers to accurately assess the health status of battery packs in vehicles in use. Specific problems include:

[0003] 1. Under DC charging conditions, the real-time reported state of charge (SOC), current, and individual cell voltage and temperature values ​​of the battery pack can be obtained from the communication protocol between the off-board conductive charger and the battery management system of the electric vehicle. However, for charging safety and to ensure battery consistency during discharge, OEMs will lock the battery pack to a certain extent. That is, the battery capacity corresponding to the 0-100% range of the battery pack SOC displayed on the vehicle is not the actual usable capacity of the battery. Therefore, the calculation result obtained by using the formula: Capacity = Ampere-hour integral / Change in battery pack SOC does not match the actual battery capacity.

[0004] 2. During the DC charging process of the whole vehicle, the current is a unidirectional charging current and is relatively constant. The battery management system (BMS) usually adopts "ampere-hour integration + voltage correction". It requires the OEM to rely on the pre-stored basic data as support. Third-party testing agencies have difficulty in making accurate and effective tests and evaluations of batteries with unknown capacity degradation, actual internal resistance and open circuit voltage (OCV).

[0005] 3. Currently, the estimation of state variables such as the State of Charge (SOC) of a battery pack is achieved by combining the construction of equivalent circuits and online identification of model parameters. This includes methods such as recursive least squares, Kalman filtering, and maximum likelihood estimation. All of these algorithms require sufficient input information, specifically the voltage response under varying current excitation conditions. However, under DC charging conditions in a vehicle, the Battery Management System (BMS) pre-sets charging rules and controls the current in real-time based on collected battery voltage and temperature parameters. This results in a relatively constant and continuous charging current, making it impossible to obtain the varying current excitation that is beneficial for online identification of battery parameters and state estimation.

[0006] As the vehicle is used, parameters such as the actual usable capacity and internal resistance of the battery will change to varying degrees. It is necessary to design an active current excitation and its combination method in the early stage of charging that is conducive to accurately obtaining the state of charge (SOC) of the battery pack and identifying relevant parameters of the battery model, based on the voltage response during the initial current excitation after a certain period of rest. Combined with online parameter identification and state estimation technology, the usable capacity and health status of the vehicle battery pack can be quickly and accurately detected without relying on the battery's basic data. At the same time, based on the convergence of the real-time calculation results of the battery's usable capacity and health status, charging can be automatically stopped and the final measurement results can be output. Summary of the Invention

[0007] To address the aforementioned issues, this invention provides a rapid health detection method for automotive battery packs using continuous variable pulse width current excitation. This method employs active current excitation for overall vehicle health detection and utilizes data acquired through active current excitation to achieve rapid overall vehicle health detection.

[0008] The present invention provides a rapid health detection method for automotive battery packs using continuous variable pulse width current excitation, comprising the following steps:

[0009] 1) Design the waveform of the pulse charging current, which includes a first waveform, a second waveform, and a third waveform:

[0010] The first waveform includes the first to fifth time points t1 to t5. From time zero to time one t1, zero current is used to allow time for the charging device to start up and respond. From time one t1 to time two t2, the first zero-state response stage uses a short-time charging current, which is a constant charging current. The duration T of the short-time charging current is... i1 The current value is the maximum required current I. MAX It is used to identify the initial ohmic internal resistance and initial polarization internal resistance of the battery during short-term charging.

[0011] The first zero-input response phase, from the second time t2 to the third time t3, uses zero current and has a duration of n. t ×T i1 n t The multiple of the duration is a natural number greater than 2, used to identify the time constant of the polarization (RC) stage during short-term charging. The second zero-state response stage, from the third time t3 to the fourth time t4, uses a long-term charging current, which is a constant charging current. The duration T of this long-term charging current is... i2 The current value is the maximum required current I. MAX It is used to identify the initial ohmic internal resistance and initial polarization internal resistance of the battery during long-term charging; the second zero-input response stage from the fourth time t4 to the fifth time t5 adopts zero current and has a duration of n. t ×T i2This is used to identify the time constant of the polarization process during long-term charging.

[0012] The second waveform, spanning from time t5 to t6, employs a periodic square wave current, with the high-level current of the square wave being the maximum demand current I. MAX The low-level current is the minimum required current I. MIN The width of the square wave in the second waveform is T. i3 The duty cycle is 0.5–0.8. One cycle consists of a combination of high-level and low-level currents. The number of cycles in the second waveform is Nt, where Nt is a natural number greater than 10. The second waveform features a current step to identify the battery open-circuit voltage OCV and the battery's internal ohmic resistance R0. It also has a high current switching frequency for online identification of polarization parameters, including the polarization resistance R0. p and polarization capacitor C p The width of the square wave in the second waveform is T. i3 The duty cycle is 0.5 to 0.8 to eliminate battery polarization during long-term constant current charging, prevent the battery from entering the voltage plateau too early, and provide richer raw data for the algorithm.

[0013] The third waveform uses a continuous constant current, and the current of the continuous constant current is the maximum demand current I. MAX The duration of constant current for the third waveform is T. i4 For the concentration polarization parameters reflected in the second-order RC equivalent circuit model, the concentration polarization parameters include the concentration polarization internal resistance R2 and the concentration polarization capacitance C2. The concentration polarization response time is relatively long, and the concentration polarization parameters are low-frequency parameters. Using long-term constant current data as input, the concentration polarization parameters are identified through optimization algorithms.

[0014] 2) Connect the vehicle battery pack to the charging equipment. The charging equipment has a built-in data acquisition device and controller. The first waveform of the pulse charging current designed in step 1) is used to charge the vehicle battery pack. At the same time, the data acquisition device of the charging equipment samples the terminal voltage measurement value and current measurement value corresponding to the first waveform. After the first waveform charging is completed, the data sampling ends and the charging equipment stops charging the vehicle battery pack. The sampled terminal voltage measurement value and current measurement value corresponding to the first waveform are input to the controller.

[0015] 3) The controller determines the optimal battery model:

[0016] The battery model consists of a first-order RC equivalent circuit model and a second-order RC equivalent circuit model. The initial battery parameters of the first-order RC equivalent circuit model include the initial ohmic internal resistance R. 0ini Initial polarization internal resistance R pini and initial polarization capacitance C pini The initial battery parameters of the second-order RC equivalent circuit model include the initial ohmic internal resistance R.0ini Initial electrochemical polarization internal resistance R 1ini Initial electrochemical polarization capacitance C 1ini Initial concentration polarization internal resistance R 2ini and initial concentration polarization capacitance C 2ini The initial battery parameters are identified offline using a short-time charging current at the first and third time points t1 to t3, and a long-time charging current is used to identify the initial battery parameters offline at the third to fifth time points t3 to t5. The optimal battery model is selected by comparing the fitting accuracy of the terminal voltage of the two battery models and the consistency of the parameter identification results of the same battery model under the two current excitations. The initial battery parameters of the optimal battery model are also output.

[0017] 4) Initialization of pre-stored parameters and pulse current curve:

[0018] The initial battery parameters obtained in step 3) are used as initial values ​​for subsequent calculations, and the time constant is used as the width T of the square wave in the second waveform. i3 and the duration of constant current T of the third waveform i4 The second and third waveforms of the pulse charging current are pre-stored;

[0019] 5) The charging equipment charges the vehicle battery pack by using the second and third waveforms of the pulse charging current pre-stored in step 4). The second and third waveforms are executed in a loop. During the charging process, the charging equipment's built-in data acquisition device samples the terminal voltage and current measurement values ​​corresponding to the second and third waveforms in real time.

[0020] 6) Online identification of battery parameters at multiple time scales:

[0021] While the charging equipment is charging the vehicle battery pack, the terminal voltage and current measurements corresponding to the second and third waveforms are sampled in real time and input to the controller built into the charging equipment. Simultaneously, based on the battery model's fast and slow change characteristics, battery parameters are identified online at different time scales.

[0022] When the battery model is a first-order RC equivalent circuit model, the battery parameters of the first-order RC equivalent circuit model include: open-circuit voltage OCV, ohmic internal resistance R0, polarization internal resistance Rp, and polarization capacitance Cp. At the microscopic time scale, the recursive least squares algorithm with forgetting factor (FFRLS) is used to identify the battery open-circuit voltage OCV and ohmic internal resistance R0 online through the second waveform. At the macroscopic time scale, the adaptive extended Kalman filter (AEKF) algorithm is used to identify the polarization internal resistance Rp and polarization capacitance Cp online through the second waveform.

[0023] When the battery model is a second-order RC equivalent circuit model, the battery parameters of the second-order RC equivalent circuit model include: open-circuit voltage OCV, ohmic internal resistance R0, electrochemical polarization internal resistance R1, electrochemical polarization capacitance C1, concentration polarization internal resistance R2, and concentration polarization capacitance C2. At the microscopic time scale, a recursive least squares algorithm with a forgetting factor is used to identify the open-circuit voltage OCV and ohmic internal resistance R0 online through the second waveform. At the macroscopic time scale, an adaptive extended Kalman filter algorithm is used to identify the electrochemical polarization internal resistance R1 and electrochemical polarization capacitance C1 online through the second waveform. The electrochemical polarization internal resistance R1 and electrochemical polarization capacitance C1 are collectively referred to as electrochemical polarization parameters. A genetic algorithm (GA) is used to identify the concentration polarization internal resistance R2 and concentration polarization capacitance C2 online through the third waveform.

[0024] 7) Accurately estimate the battery pack's state of charge (SOC) and state of health (SOH) and determine its convergence:

[0025] While the charging equipment is charging the vehicle battery pack, the terminal voltage and current measurements corresponding to the second and third waveforms are sampled in real time and input to the charging equipment controller. Simultaneously, the battery pack's state of charge (SOC) and state of health (SOH) are estimated at different time scales, and their convergence is assessed.

[0026] At the microscale, the State of Charge (SOC) of the battery pack is estimated using an adaptive extended Kalman filter algorithm. Simultaneously, the difference between the predicted and observed voltages is output, and this difference is called innovation. The State of Health (SOH) is estimated at the current macroscale using this innovation update. The result of the SOH estimation at the macroscale is input for calculation at the next macroscale. The SOC and SOH estimates mutually correct each other, i.e., a joint estimation of the SOC and SOH. When the SOH estimation meets the convergence criteria, charging ends, and the current SOH is output, completing a rapid health detection.

[0027] In step 1), designing the pulse charging current refers to designing the trajectory of the pulse charging current magnitude as a function of time. The pulse charging current includes: the first waveform of the current variation trajectory before the fifth time point t5; the second waveform of the pulse charging current variation trajectory from the fifth to the sixth time point t5-t6; and the third waveform of the current variation trajectory of constant current charging after the sixth time point t6. There are two current magnitudes: zero current and the high current I required by the vehicle battery pack. MAX .

[0028] In step 2), in practical application, the charging device outputs a charging current of a corresponding magnitude to charge the vehicle battery pack according to the designed pulse charging current first waveform. During this process, the built-in data acquisition device collects, records, and analyzes the communication data between the vehicle battery pack and the charging device; the voltage and current measurement values ​​corresponding to the collected first waveform will be used in the calculation of subsequent algorithms.

[0029] In step 3), a higher order results in better simulation of battery polarization effects, but also reduces computational efficiency and increases the difficulty of offline identification of initial battery parameters. The accuracy of the battery model's terminal voltage fitting is proportional to the order of the RC circuit, but as the order increases, the complexity of the battery model increases, the difficulty of parameter identification increases dramatically, and computational efficiency is significantly reduced. Considering both terminal voltage fitting accuracy and computational efficiency, the optimal battery model is determined by comparing the consistency of the terminal voltage fitting accuracy and initial battery parameter identification results between the first-order and second-order RC equivalent circuit models. The identification results are then used as the initial values ​​for the multi-timescale joint estimation algorithm. Determining the optimal battery model requires establishing both a first-order and a second-order RC equivalent circuit model.

[0030] Determining the optimal battery model involves the following steps:

[0031] a) Offline identification of initial Ohmic resistance:

[0032] The first-order RC equivalent circuit model and the second-order RC equivalent circuit model of the battery affect the initial ohmic internal resistance R. 0ini The identification method is the same: the initial ohmic internal resistance R is obtained by the ratio of the voltage drop generated at the instant the current is introduced to the current. 0ini :

[0033]

[0034] In the formula, ΔU 充 I represents the voltage change during a current step during charging. 充 ΔU is the change in current during the instantaneous current step during charging. 静 I is the voltage change during a step current change at the moment of rest. 静 This represents the change in current during a step change at the moment of rest.

[0035] b) Offline identification of initial polarization parameters:

[0036] Identification of initial polarization parameters for a first-order RC equivalent circuit model:

[0037] i. The current is zero during the first and second zero-input response phases, and the time constant τ = R. pini C pini R piniFor the initial polarization internal resistance, C pini The initial polarization capacitor; the terminal voltage U at time k. L (k) In the zero-input response phase:

[0038]

[0039] In the formula, U OCV (k) is the open-circuit voltage at time k, U p (k-1) is the polarization voltage at time k-1, Δt is the sampling interval, and the time constant τ is obtained by fitting using the least squares method;

[0040] ii. The current during the first and second zero-state response phases is the maximum demand current I. MAX The terminal voltage U at time k L (k) In the zero-state response phase:

[0041]

[0042] In the formula, I(k) is the charging current at time k;

[0043] Substituting the time constant τ obtained in step i) above, the initial polarization resistance R is obtained by fitting using the least squares method. pini ,

[0044] The initial polarization capacitance C is further obtained. pini ;

[0045] iii. Through the initial polarization internal resistance R pini and initial polarization capacitance C pini Substituting the formula for the terminal voltage at time k in the zero-state response stage, we obtain the fitted value U of the terminal voltage at time k. 拟合 ;

[0046] Identification of initial polarization parameters for the second-order RC equivalent circuit model:

[0047] i. The current is zero during the first and second zero-input response phases, and the electrochemical time constant τ1 = R 1ini C 1ini Concentration time constant τ2=R 2ini C 2ini R 1ini R is the initial electrochemical polarization resistance. 2ini For the initial concentration polarization internal resistance, C 1ini For the initial electrochemical polarization capacitance, C 2ini The initial electrochemical polarization capacitance, and the terminal voltage U at time k. L (k) at zero input

[0048] The input response phase is as follows:

[0049]

[0050] In the formula, U1(k-1) is the electrochemical polarization voltage at time k-1, and U2(k-1) is the concentration polarization voltage at time k-1.

[0051] Pressure; the electrochemical time constant τ1 and concentration time constant τ2 were obtained by least squares fitting;

[0052] ii. The current during the first and second zero-state response phases is the maximum demand current I. MAX The terminal voltage U at time k L (k) at zero

[0053] The state response phase is as follows:

[0054]

[0055] Substituting the electrochemical time constant τ1 and the concentration time constant τ2, the initial electrochemical polarization resistance R is obtained by least squares fitting. 1ini and initial concentration polarization internal resistance R 2ini The initial electrochemical polarization capacitance C was further obtained. 1ini and initial concentration polarization capacitance C 2ini ;

[0056] c) Determine the optimal battery model:

[0057] Calculate the fitting accuracy of the terminal voltage of the next-order RC equivalent circuit model under current excitation from the first time step (t1 to t3) to the third time step (t3 to t5), respectively.

[0058] The method for calculating the fitting accuracy of the terminal voltage is as follows:

[0059]

[0060] In the formula, U 拟合 U is the fitted value of the terminal voltage. 测量 ε1 represents the measured terminal voltage value, and ε1 represents the offline identified terminal voltage error, where 0 < ε1 < 5%.

[0061] If the terminal voltage fitting accuracy of the first-order RC equivalent circuit model under current excitation from the first time to the third time t1~t3 and from the third time to the fifth time t3~t5 both satisfy Formula 2.1, then the terminal voltage fitting accuracy of the first-order RC equivalent circuit model is deemed to meet the requirements. Further, the consistency of the parameter identification results for the battery initial parameters is assessed. Battery polarization is closely related to battery charging duration. Short-duration pulse charging and long-duration pulse charging are used to determine whether the parameter identification results of the first-order RC equivalent circuit model for charging currents of different durations are consistent. If the consistency judgment condition for parameter identification results is met, it proves that the first-order RC equivalent circuit model can simulate battery polarization.

[0062] The initial polarization resistance R is obtained from the initial polarization parameters of the first-order RC equivalent circuit model in step b). pini and initial polarization capacitance C pini The offline identification results R of polarization internal resistance from the first time point to the third time point t1 to t3 are extracted respectively. pt1-t3 Offline identification results of polarization capacitor C pt1-t3 The offline identification results of polarization internal resistance from the third time point to the fifth time point t3 to t5, R pt3-t5 Offline identification results of polarization capacitor C pt3-t5 The consistency of the parameter identification results is judged as follows:

[0063]

[0064] The consistency of the parameter identification results is judged by formula 2.2, where ε2 is the polarization parameter identification error, 0 < ε2 < 10%;

[0065] If Formula 2.2 is satisfied, the consistency of the parameter identification results of the first-order RC equivalent circuit model is deemed to meet the requirements, and the optimal battery model is determined to be the first-order RC equivalent circuit model. If either the terminal voltage fitting accuracy or the consistency of the parameter identification results does not meet Formula 2.2, the optimal battery model is determined to be the second-order RC equivalent circuit model.

[0066] In step 4), the initial parameters of the battery are determined by the optimal battery model determined in step 3). The initial value of the ohmic internal resistance R0 is adopted from the initial ohmic internal resistance R under the current excitation at the first time to the third time t1~t3 and the third time to the fifth time t3~t5 in step 3). 0ini The average value of the calculation results is used, and the initial values ​​of the polarization parameters are the average values ​​of the offline identification results of the initial polarization parameters under current excitation from the first time to the third time t1~t3 and from the third time to the fifth time t3~t5.

[0067] When the battery model is a first-order RC equivalent circuit model, the width T of the square wave in the second waveform is determined by the time constant τ.i3 ,

[0068] T i3 =τ(3.1)

[0069] The duration of constant current T of the third waveform i4 The value range is 100s to 1000s;

[0070] When the battery model is a second-order RC equivalent circuit model, the width T of the square wave in the second waveform is determined by the electrochemical time constant τ1 and the concentration time constant τ2. i3 and the third waveform's continuous constant current time T i4 :

[0071]

[0072] The pulse charging current is pre-stored based on the second and third waveforms.

[0073] In step 5), the charging equipment continues to charge the vehicle battery pack. The charging current uses the pre-stored pulse charging current composed of the second and third waveforms from step 4), and the pre-stored pulse charging current is executed cyclically until the battery health convergence judgment condition in step 7) is met, at which point charging ends. During the charging process, the terminal voltage and current measurements corresponding to the pre-stored pulse charging current are collected in real time by the data acquisition device built into the charging equipment. The collected terminal voltage and current measurements are input into the controller built into the charging equipment for calculations in steps 6) and 7).

[0074] In step 6), the ohmic internal resistance and polarization internal resistance in the battery model correspond to impedances at different frequencies. Battery parameter identification needs to be performed on different time scales. Two time scales are defined: a microscopic time scale and a macroscopic time scale. The step size of the microscopic time scale is consistent with the sampling interval, denoted by k. The step size of the macroscopic time scale is an integer multiple of the step size of the microscopic time scale, denoted by m.

[0075]

[0076] In the formula, This indicates rounding up; when the battery model is a first-order RC equivalent circuit model, τ′=τ; when the battery model is a second-order RC equivalent circuit model, τ′=τ1.

[0077] The identification of polarization parameters in the first-order RC equivalent circuit model and the electrochemical polarization parameters in the second-order RC equivalent circuit model were carried out simultaneously with the estimation of SOH under the healthy state on a macroscopic time scale:

[0078] Online identification of open-circuit voltage OCV and ohmic internal resistance R0 of first-order and second-order RC equivalent circuit models at microscale time:

[0079] The parameter identification scheme for the first-order RC equivalent circuit model is implemented by a combination of recursive least squares with forgetting factor (FFRLS) and adaptive extended Kalman filter (AEKF); the parameter identification scheme for the second-order RC equivalent circuit model is implemented by a combination of recursive least squares with forgetting factor (FFRLS), adaptive extended Kalman filter (AEKF), and genetic algorithm (GA). The prerequisite for online parameter identification is obtaining the discretized difference equations of the battery model. The difference equations of the first-order RC equivalent circuit model are as follows:

[0080] U L (k)=α1U L (k-1)+α2I(k)+α3I(k-1)+α4(4.2)

[0081] In the formula, U L (k-1) is the terminal voltage at time k-1, I(k-1) is the charging current at time k-1, α1 is the coefficient of the terminal voltage at time k-1 in the first-order RC equivalent circuit model, α2 is the coefficient of the charging current at time k in the first-order RC equivalent circuit model, α3 is the coefficient of the charging current at time k-1 in the first-order RC equivalent circuit model, and α4 is the natural number related to the open-circuit voltage at time k in the first-order RC equivalent circuit model, α4=(1-α1)U OCV (k); The ohmic internal resistance R0 and open-circuit voltage U are derived from the coefficients of the differential equations in the first-order RC equivalent circuit model. OCV The analytical expression:

[0082]

[0083] Difference equations for the second-order RC equivalent circuit model:

[0084] U L (k)=α5U L (k-1)+α6U L (k-2)+α7I(k)+α8I(k-1)+α9I(k-2)+α 10 (4.3)

[0085] In the formula, U L (k-2) is the terminal voltage at time k-2, I(k-2) is the charging current at time k-2, α5 is the coefficient of the terminal voltage at time k-1 in the second-order RC equivalent circuit model, α6 is the coefficient of the terminal voltage at time k-2 in the second-order RC equivalent circuit model, α7 is the coefficient of the charging current at time k in the second-order RC equivalent circuit model, α8 is the coefficient of the charging current at time k-1 in the second-order RC equivalent circuit model, α9 is the coefficient of the charging current at time k-2 in the second-order RC equivalent circuit model, α 10 α is a natural number related to the open-circuit voltage at time k in the second-order RC equivalent circuit model.10 =(1-α5-α6)U OCV (k), U OCV (k) is the open-circuit voltage at time k;

[0086] The ohmic resistance R0 and open-circuit voltage U are derived from the coefficients of the difference equations in the second-order RC equivalent circuit model. OCV The analytical expression:

[0087]

[0088] Online parameter identification of the ohmic internal resistance R0 and open-circuit voltage OCV is performed using the recursive least squares method with forgetting factor FFRLS. After the recursive process ends, the coefficient matrix of the difference equation is obtained. The corresponding ohmic internal resistance R0 and open-circuit voltage OCV are obtained through the above analytical expression. The recursive least squares method with forgetting factor FFRLS introduces a forgetting factor λ when calculating the gain and error covariance. It is usually a constant less than 1, which represents the reduction of the weight of past data relative to the current data. The value is between 0.95 and 1.

[0089] Online identification of polarization resistance Rp and polarization capacitance Cp for a first-order RC equivalent circuit model, and electrochemical polarization resistance R1, electrochemical polarization capacitance C1, concentration polarization resistance R2, concentration polarization capacitance C2, and healthy state SOH estimation for a second-order RC equivalent circuit model on a macroscopic time scale:

[0090] In the second-order RC equivalent circuit model, concentration polarization parameters are calculated on a macroscopic time scale using a genetic algorithm. The genetic algorithm searches through a survival-of-the-fittest mechanism, generating better approximate solutions generation by generation, avoiding local optima and achieving accurate parameter identification. When the current enters the third waveform, current and voltage measurements are stored until the third waveform ends. Using the current measurement as input and the voltage measurement as observation, the concentration polarization parameters are identified using the genetic algorithm.

[0091] The optimization objective of the Genetic Algorithm (GA) is to minimize the sum of squares of the differences between the measured terminal voltage and the model-predicted terminal voltage. In the GA, battery parameters are represented as individuals in the genetic space, and the individual fitness function is U. F Let it be the reciprocal of the mean square error of the terminal voltage:

[0092]

[0093] In the formula, N represents the total number of sampling points. The terminal voltage measurement at time k;

[0094] Determine whether there is an individual among all individuals that meets the optimization termination condition. The judgment condition is set as the terminal voltage error of the optimal individual being less than ω1, 0<ω1<0.05V. If the condition is met, then this individual is taken as the identification result of the concentration polarization parameter.

[0095] In step 7), the joint estimation of the battery pack's state of charge (SOC) and state of health (SOH) across multiple time scales includes SOC estimation at the micro-time scale and capacity estimation at the macro-time scale. Simultaneously, the polarization parameters in the first-order RC equivalent circuit model and the electrochemical polarization parameters in the second-order RC equivalent circuit model are also estimated along with the capacity at the macro-time scale. For the first 10–1000 micro-steps after the joint estimation begins, the polarization parameters and capacity remain unchanged from their initial values. After 10–1000 micro-steps, the parameter estimator is activated, using the data from the first 10–1000 micro-steps to perform the first update of the polarization parameters and capacity values. The updated values ​​are input into the micro-time scale for SOC calculation. When the micro-time scale calculation meets the criteria for the next macro-time scale, the macro-time scale parameter estimator is activated, updating the polarization parameters and capacity values, and continuing to input them into the micro-time scale for SOC calculation. This process iterates continuously until the calculation is complete.

[0096] The State of Health (SOH) is defined as follows:

[0097]

[0098] In the formula, Cn is the battery capacity, C ini This is the initial capacity value;

[0099] The state of health (SOH) is determined by comparing it with the SOH calculated using a joint estimation scheme, based on the ratio of accumulated ampere-hours (AH) to the change in the battery pack's state of charge (SOC). Ah For the results of health status calculation using the ampere-hour integral method, the method for estimating the state of health (SOH) using the ampere-hour integral method at a single macroscopic time scale is as follows:

[0100]

[0101] In the formula, I m,k The charging current at time k on the m-th macroscopic time scale; SOC m,L-1 The state of charge (SOC) of the battery at time L-1 on the m-th macroscopic time scale. m,0 The state of charge of the battery at time 0 on the m-th macroscopic time scale is given.

[0102] Capacity estimation reliability assessment:

[0103]

[0104] In the formula, NSOH The number of macroscopic time scales used to determine capacity reliability ranges from 3 to 10, where n represents the nth macroscopic time scale (n>3); ε3 is the relative error of state of health (SOH), where 0<ε3<10%. When N consecutive... SOH The ampere-hour integral estimate of the state of health (SOH) at a macroscopic time scale satisfies Equation 5.2, and the joint estimation algorithm estimates the state of health (SOH) in accordance with Equation 5.3. Therefore, the capacity estimation of the state of health (SOH) is considered reliable, and charging is stopped.

[0105] Advantages of this invention:

[0106] This invention enables the acquisition of ideal data for battery health detection through active current excitation, selection of appropriate battery models, and the use of parameter identification and state estimation methods at multiple time scales to ultimately achieve accurate detection of battery health. Attached Figure Description

[0107] Figure 1 This is a schematic diagram of the waveform of the pulse charging current in an embodiment of the rapid health detection method for continuously variable pulse width current excitation of automotive battery packs according to the present invention.

[0108] Figure 2 This is a flowchart illustrating the specific steps of the rapid health detection method for a vehicle battery pack under continuous variable pulse width current excitation according to the present invention.

[0109] Figure 3 This is a flowchart summarizing the rapid health detection method for continuously variable pulse width current excitation of automotive battery packs according to the present invention. Detailed Implementation

[0110] The present invention will be further described below with reference to the accompanying drawings and specific embodiments.

[0111] This embodiment presents a rapid health detection method for automotive battery packs using continuous variable pulse width current excitation, such as... Figure 2 and 3 As shown, it includes the following steps:

[0112] 1) Design the waveform of the pulse charging current, such as... Figure 1 As shown, the pulse charging current includes a first waveform, a second waveform, and a third waveform:

[0113] The first waveform includes the first to fifth time points t1 to t5. From time zero to time one (t1 = 10s), zero current is used to allow time for the charging device to start up and respond. From time one (t1) to time two (t2 = 20s), the first zero-state response stage uses a short-time charging current, which is a constant charging current. The duration T of the short-time charging current is... i1 The current value is the maximum required current I. MAX IMAX =100A, used to identify the initial ohmic internal resistance and initial polarization internal resistance of the battery during short-term charging; the first zero-input response stage, from the second time t2 to the third time t3 = 50s, adopts zero current and has a duration of n. t ×T i1 n t =3, used to identify the time constant of the polarization (RC) element during short-term charging. From the third time t3 to the fourth time t4 = 110s, the second zero-state response stage uses a long-term charging current, which is a constant charging current. The duration T of the long-term charging current is... i2 The current value is the maximum required current I. MAX =100A, used to identify the initial ohmic internal resistance and initial polarization internal resistance of the battery during long-term charging; the second zero-input response stage, from the fourth time t4 to the fifth time t5 = 290s, adopts zero current and has a duration of n. t ×T i2 This is used to identify the time constant of the polarization process during long-term charging.

[0114] The second waveform, spanning from time t5 to t6, employs a periodic square wave current, with the high-level current of the square wave being the maximum demand current I. MAX =100A, low-level current is the minimum required current I MIN The width of the square wave in the second waveform is T. i3 =10s, duty cycle is 0.5, i.e., 10s rest. The working time of a set of high-level current plus low-level current is one cycle. The number of cycles of the second waveform is Nt, Nt=100. The second waveform has a current step, which is used to identify the battery open-circuit voltage OCV and the battery internal resistance R0. The current switching frequency is relatively high, which is used for online identification of polarization parameters, including polarization internal resistance R. p and polarization capacitor C p The width of the square wave in the second waveform is T. i3 And the settling time T i3 The model can eliminate battery polarization during long-term constant current charging, prevent the battery from entering the voltage plateau too early, and provide richer raw data for the algorithm;

[0115] The third waveform uses a continuous constant current, and the current of the continuous constant current is the maximum demand current I. MAX The duration of constant current for the third waveform is T. i4 For the concentration polarization parameters reflected in the second-order RC equivalent circuit model, the concentration polarization parameters include the concentration polarization internal resistance R2 and the concentration polarization capacitance C2. The concentration polarization response time is relatively long, and the concentration polarization parameters are low-frequency parameters. Long-term constant current data is used as input, and the concentration polarization parameters are identified through optimization algorithms.

[0116] 2) Connect the vehicle battery pack to the charging equipment. The charging equipment has a built-in data acquisition device and controller, using step 1).

[0117] The first waveform of the designed pulse charging current charges the vehicle battery pack. At the same time, the data acquisition device built into the charging equipment samples the terminal voltage and current measurement values ​​corresponding to the first waveform. After the first waveform finishes charging, the data sampling ends, the charging equipment stops charging the vehicle battery pack, and the sampled terminal voltage and current measurement values ​​corresponding to the first waveform are input to the controller.

[0118] 3) The controller determines the optimal battery model:

[0119] The battery model consists of a first-order RC equivalent circuit model and a second-order RC equivalent circuit model. The initial battery parameters of the first-order RC equivalent circuit model include the initial ohmic internal resistance R. 0ini Initial polarization internal resistance R pini and initial polarization capacitance C pini The initial battery parameters of the second-order RC equivalent circuit model include the initial ohmic internal resistance R. 0ini Initial electrochemical polarization internal resistance R 1ini Initial electrochemical polarization capacitance C 1ini Initial concentration polarization internal resistance R 2ini and initial concentration polarization capacitance C 2ini The initial battery parameters are identified offline using a short-time charging current at the first and third time points (t1-t3), and a long-time charging current at the third to fifth time points (t3-t5). The optimal battery model is selected by comparing the end-voltage fitting accuracy of the two battery models and the consistency of parameter identification results for the same battery model under the two current excitations. The initial battery parameters of the optimal battery model are also output.

[0120] a) Offline identification of initial Ohmic resistance:

[0121] The first-order RC equivalent circuit model and the second-order RC equivalent circuit model of the battery affect the initial ohmic internal resistance R. 0ini The identification method is the same: the initial ohmic internal resistance R is obtained by the ratio of the voltage drop generated at the instant the current is introduced to the current. 0ini :

[0122]

[0123] In the formula, ΔU 充 I represents the voltage change during a current step during charging. 充 ΔU is the change in current during the instantaneous current step during charging. 静 I is the voltage change during a step current change at the moment of rest. 静 This represents the change in current during a step change at the moment of rest.

[0124] b) Offline identification of initial polarization parameters:

[0125] Identification of initial polarization parameters for a first-order RC equivalent circuit model:

[0126] i. The current is zero during the first and second zero-input response phases, and the time constant τ = R. pini C pini R pini For the initial polarization internal resistance, C pini The initial polarization capacitor; the terminal voltage U at time k. L (k) In the zero-input response phase:

[0127]

[0128] In the formula, U OCV (k) is the open-circuit voltage at time k, U p (k-1) is the polarization voltage at time k-1, Δt is the sampling interval, and the time constant τ is obtained by fitting using the least squares method;

[0129] ii. The current during the first and second zero-state response phases is the maximum demand current I. MAX The terminal voltage U at time k L (k) In the zero-state response phase:

[0130]

[0131] In the formula, I(k) is the charging current at time k;

[0132] Substituting the time constant τ obtained in step i) above, the initial polarization resistance R is obtained by fitting using the least squares method. pini The initial polarization capacitance C is further obtained. pini ;

[0133] iii. Through the initial polarization internal resistance R pini and initial polarization capacitance C pini Substituting the formula for the terminal voltage at time k in the zero-state response stage, we obtain the fitted value U of the terminal voltage at time k. 拟合 ;

[0134] Identification of initial polarization parameters for the second-order RC equivalent circuit model:

[0135] iii. The current is zero during the first and second zero-input response stages, and the electrochemical time constant τ1 = R 1ini C 1ini Concentration time constant τ2=R 2ini C 2ini R 1ini R is the initial electrochemical polarization resistance.2ini For the initial concentration polarization internal resistance, C 1ini For the initial electrochemical polarization capacitance, C 2ini The initial electrochemical polarization capacitance, and the terminal voltage U at time k. L (k) In the zero-input response phase:

[0136]

[0137] In the formula, U1(k-1) is the electrochemical polarization voltage at time k-1, and U2(k-1) is the concentration polarization voltage at time k-1; the electrochemical time constant τ1 and the concentration time constant τ2 are obtained by least squares fitting.

[0138] iv. The current during the first and second zero-state response phases is the maximum demand current I. MAX The terminal voltage U at time k L (k) In the zero-state response phase:

[0139]

[0140] Substituting the electrochemical time constant τ1 and the concentration time constant τ2, the initial electrochemical polarization resistance R is obtained by least squares fitting. 1ini and initial concentration polarization internal resistance R 2ini The initial electrochemical polarization capacitance C was further obtained. 1ini and initial concentration polarization capacitance C 2ini ;

[0141] c) Determine the optimal battery model:

[0142] Calculate the fitting accuracy of the terminal voltage of the next-order RC equivalent circuit model under current excitation from the first time step (t1 to t3) to the third time step (t3 to t5), respectively.

[0143] The method for calculating the fitting accuracy of the terminal voltage is as follows:

[0144]

[0145] In the formula, U 拟合 U is the fitted value of the terminal voltage. 测量 ε1 represents the measured terminal voltage value, and ε1 is the offline identification error of the terminal voltage, ε1 = 3%.

[0146] If the terminal voltage fitting accuracy of the first-order RC equivalent circuit model under current excitation from the first time to the third time t1~t3 and from the third time to the fifth time t3~t5 both satisfy Formula 2.1, then the terminal voltage fitting accuracy of the first-order RC equivalent circuit model is deemed to meet the requirements. Further, the consistency of the parameter identification results for the battery initial parameters is assessed. Battery polarization is closely related to battery charging duration. Short-duration pulse charging and long-duration pulse charging are used to determine whether the parameter identification results of the first-order RC equivalent circuit model for charging currents of different durations are consistent. If the consistency judgment condition for parameter identification results is met, it proves that the first-order RC equivalent circuit model can simulate battery polarization.

[0147] The initial polarization resistance R is obtained from the initial polarization parameters of the first-order RC equivalent circuit model in step b). pini and initial polarization capacitance C pini The offline identification results R of polarization internal resistance from the first time point to the third time point t1 to t3 are extracted respectively. pt1-t3 Offline identification results of polarization capacitor C pt1-t3 The offline identification results of polarization internal resistance from the third time point to the fifth time point t3 to t5, R pt3-t5 Offline identification results of polarization capacitor C pt3-t5 Consistency of parameter identification results

[0148] The consistency of the parameter identification results is judged as follows:

[0149]

[0150] The consistency of the parameter identification results is judged by formula 2.2, where ε2 is the polarization parameter identification error, ε2 = 5%;

[0151] If Formula 2.2 is satisfied, the consistency of the parameter identification results of the first-order RC equivalent circuit model is deemed to meet the requirements, and the optimal battery model is determined to be the first-order RC equivalent circuit model. If either the terminal voltage fitting accuracy or the consistency of the parameter identification results does not meet Formula 2.2, the optimal battery model is determined to be the second-order RC equivalent circuit model; 4) Initialization of pre-stored parameters and pulse current curve:

[0152] The initial parameters of the battery will be determined by the optimal battery model determined in step 3). The initial value of the ohmic internal resistance R0 will be the initial ohmic internal resistance R0 under the current excitation conditions from the first time to the third time t1~t3 and from the third time to the fifth time t3~t5 in step 3). 0ini The average value of the calculation results is used, and the initial values ​​of the polarization parameters are the average values ​​of the offline identification results of the initial polarization parameters under current excitation from the first time to the third time t1~t3 and from the third time to the fifth time t3~t5.

[0153] When the battery model is a first-order RC equivalent circuit model, the width T of the square wave in the second waveform is determined by the time constant τ. i3 ,

[0154] T i3 =τ(3.1)

[0155] The duration of constant current T of the third waveform i4 The value range is 100s to 1000s;

[0156] When the battery model is a second-order RC equivalent circuit model, the width T of the square wave in the second waveform is determined by the electrochemical time constant τ1 and the concentration time constant τ2. i3 and the third waveform's continuous constant current time T i4 :

[0157]

[0158] Pre-store the pulse charging current based on the second and third waveforms;

[0159] 5) The charging equipment charges the vehicle battery pack by using the second and third waveforms of the pulse charging current pre-stored in step 4). The second and third waveforms are executed in a loop. During the charging process, the charging equipment's built-in data acquisition device samples the terminal voltage and current measurement values ​​corresponding to the second and third waveforms in real time.

[0160] 6) Online identification of battery parameters at multiple time scales:

[0161] In the battery model, ohmic internal resistance and polarization internal resistance correspond to impedances at different frequencies. Battery parameters need to be identified at different time scales. Two time scales are defined: a microscopic time scale and a macroscopic time scale. The step size of the microscopic time scale is consistent with the sampling interval, denoted by k. The step size of the macroscopic time scale is an integer multiple of the step size of the microscopic time scale, denoted by m.

[0162]

[0163] In the formula, This indicates rounding up; when the battery model is a first-order RC equivalent circuit model, τ′=τ; when the battery model is a second-order RC equivalent circuit model, τ′=τ1.

[0164] The identification of polarization parameters in the first-order RC equivalent circuit model and electrochemical polarization parameters in the second-order RC equivalent circuit model is carried out simultaneously with the estimation of SOH in the healthy state on a macroscopic time scale.

[0165] Online identification of the open-circuit voltage OCV and ohmic internal resistance R0 of first-order and second-order RC equivalent circuit models at microscale: The parameter identification scheme for the first-order RC equivalent circuit model is implemented by a combination of recursive least squares with forgetting factor (FFRLS) and adaptive extended Kalman filter (AEKF); the parameter identification scheme for the second-order RC equivalent circuit model is implemented by a combination of recursive least squares with forgetting factor (FFRLS), adaptive extended Kalman filter (AEKF), and genetic algorithm (GA). The prerequisite for online parameter identification is obtaining the discretized difference equation of the battery model. The difference equation of the first-order RC equivalent circuit model is as follows:

[0166] U L (k)=α1U L (k-1)+α2I(k)+α3I(k-1)+α4(4.2)

[0167] In the formula, U L (k-1) is the terminal voltage at time k-1, I(k-1) is the charging current at time k-1, α1 is the coefficient of the terminal voltage at time k-1 in the first-order RC equivalent circuit model, α2 is the coefficient of the charging current at time k in the first-order RC equivalent circuit model, α3 is the coefficient of the charging current at time k-1 in the first-order RC equivalent circuit model, and α4 is the natural number related to the open-circuit voltage at time k in the first-order RC equivalent circuit model, α4=(1-α1)U OCV (k); The ohmic internal resistance R0 and open-circuit voltage U are derived from the coefficients of the differential equations in the first-order RC equivalent circuit model. OCV The analytical expression:

[0168]

[0169] Difference equations for the second-order RC equivalent circuit model:

[0170] U L (k)=α5U L (k-1)+α6U L (k-2)+α7I(k)+α8I(k-1)+α9I(k-2)+α 10 (4.3)

[0171] In the formula, U L (k-2) is the terminal voltage at time k-2, I(k-2) is the charging current at time k-2, α5 is the coefficient of the terminal voltage at time k-1 in the second-order RC equivalent circuit model, α6 is the coefficient of the terminal voltage at time k-2 in the second-order RC equivalent circuit model, α7 is the coefficient of the charging current at time k in the second-order RC equivalent circuit model, α8 is the coefficient of the charging current at time k-1 in the second-order RC equivalent circuit model, α9 is the coefficient of the charging current at time k-2 in the second-order RC equivalent circuit model, α 10α is a natural number related to the open-circuit voltage at time k in the second-order RC equivalent circuit model. 10 =(1-α5-α6)U OCV (k), U OCV (k) is the open-circuit voltage at time k;

[0172] The ohmic resistance R0 and open-circuit voltage U are derived from the coefficients of the difference equations in the second-order RC equivalent circuit model. OCV The analytical expression:

[0173]

[0174] Online parameter identification of the ohmic internal resistance R0 and open-circuit voltage OCV is performed using the recursive least squares method with forgetting factor FFRLS. After the recursive process ends, the coefficient matrix of the difference equation is obtained. The corresponding ohmic internal resistance R0 and open-circuit voltage OCV are obtained through the above analytical expression. The recursive least squares method with forgetting factor FFRLS introduces a forgetting factor λ when calculating the gain and error covariance. It is usually a constant less than 1, which represents the reduction of the weight of past data relative to the current data. The value is 0.99.

[0175] Online identification of polarization resistance Rp and polarization capacitance Cp for a first-order RC equivalent circuit model, and electrochemical polarization resistance R1, electrochemical polarization capacitance C1, concentration polarization resistance R2, concentration polarization capacitance C2, and healthy state SOH estimation for a second-order RC equivalent circuit model on a macroscopic time scale:

[0176] In the second-order RC equivalent circuit model, the concentration polarization parameters are calculated using a genetic algorithm on a macroscopic time scale. The genetic algorithm searches through a survival-of-the-fittest genetic mechanism, generating better approximate solutions generation by generation, avoiding local optima and achieving accurate parameter identification. When the current enters the third waveform, the current and voltage measurements are stored until the third waveform ends. Using the current measurement as input and the voltage measurement as observation, the concentration polarization parameters are identified using the genetic algorithm.

[0177] The optimization objective of the Genetic Algorithm (GA) is to minimize the sum of squares of the differences between the measured terminal voltage and the model-predicted terminal voltage. In the GA, battery parameters are represented as individuals in the genetic space, and the individual fitness function is U. F Let it be the reciprocal of the mean square error of the terminal voltage:

[0178]

[0179] In the formula, N represents the total number of sampling points. This represents the measured terminal voltage at time k.

[0180] Determine whether there is an individual among all individuals that meets the optimization termination condition. The judgment condition is that the terminal voltage error of the optimal individual is less than ω1, where ω1 = 0.01V. If the condition is met, then this individual is taken as the identification result of the concentration polarization parameter.

[0181] 7) Accurately estimate the battery pack's state of charge (SOC) and state of health (SOH) and determine its convergence:

[0182] While the charging equipment is charging the vehicle battery pack, the terminal voltage and current measurements corresponding to the second and third waveforms are sampled in real time and input to the charging equipment controller. Simultaneously, the battery pack's state of charge (SOC) and state of health (SOH) are estimated at different time scales, and their convergence is assessed.

[0183] The joint estimation of the battery pack's State of Charge (SOC) and State of Health (SOH) across multiple time scales includes SOC estimation at the micro-time scale and capacity estimation at the macro-time scale. Simultaneously, polarization parameters from the first-order RC equivalent circuit model and electrochemical polarization parameters from the second-order RC equivalent circuit model are estimated along with capacity at the macro-time scale. For the first 10–1000 micro-steps after the joint estimation begins, the polarization parameters and capacity remain unchanged from their initial values. After 10–1000 micro-steps, the parameter estimator is activated, using data from the first 10–1000 micro-steps to update the polarization parameters and capacity values ​​for the first time. These updated values ​​are then input into the micro-time scale for SOC calculation. When the micro-time scale calculation meets the criteria for the next macro-time scale, the macro-time scale parameter estimator is activated, updating the polarization parameters and capacity values, and continuing to input data into the micro-time scale for SOC calculation. This process iterates continuously until the calculation is complete.

[0184] The State of Health (SOH) is defined as follows:

[0185]

[0186] In the formula, Cn is the battery capacity, C ini This is the initial capacity value;

[0187] The state of health (SOH) is determined by comparing it with the SOH calculated using a joint estimation scheme, based on the ratio of accumulated ampere-hours (AH) to the change in the battery pack's state of charge (SOC). Ah For the results of health status calculation using the ampere-hour integral method, the method for estimating the state of health (SOH) using the ampere-hour integral method at a single macroscopic time scale is as follows:

[0188]

[0189] In the formula, I m,kThe charging current at time k on the m-th macroscopic time scale; SOC m,L-1 The state of charge (SOC) of the battery at time L-1 on the m-th macroscopic time scale. m,0 The state of charge of the battery at time 0 on the m-th macroscopic time scale is given.

[0190] Capacity estimation reliability assessment:

[0191]

[0192] In the formula, N SOH N represents the number of macroscopic timescales used to determine capacity reliability. SOH =3, where n represents the nth macroscopic time scale, n>3; ε3 is the relative error of the state of health (SOH), ε3=5%, when N consecutive SOH The ampere-hour integral estimate of the state of health (SOH) at a macroscopic time scale satisfies Equation 5.2, and the joint estimation algorithm estimates the state of health (SOH) in accordance with Equation 5.3. Therefore, the capacity estimation of the state of health (SOH) is considered reliable, and charging is stopped.

[0193] Finally, it should be noted that the purpose of disclosing the embodiments is to help further understand the present invention. However, those skilled in the art will understand that various substitutions and modifications are possible without departing from the spirit and scope of the present invention and the appended claims. Therefore, the present invention should not be limited to the content disclosed in the embodiments, and the scope of protection of the present invention is defined by the claims.

Claims

1. A rapid health detection method for automotive battery packs using continuous variable pulse width current excitation, characterized in that, The rapid health detection method includes the following steps: 1) Design the waveform of the pulse charging current, which includes a first waveform, a second waveform, and a third waveform: The first waveform includes the first to fifth time points t1 to t5. From time zero to time one t1, zero current is used to allow time for the charging device to start up and respond. From time one t1 to time two t2, the first zero-state response stage uses a short-time charging current, which is a constant charging current. The duration T of the short-time charging current is... i1 The current value is the maximum required current I. MAX It is used to identify the initial ohmic internal resistance and initial polarization internal resistance of the battery during short-term charging; the first zero-input response stage from the second time t2 to the third time t3 adopts zero current and has a duration of n. t ×T i1 n t The multiple of the duration is a natural number greater than 2, used to identify the time constant of the polarization stage during short-term charging. The second zero-state response stage, from the third time t3 to the fourth time t4, uses a long-term charging current, which is a constant charging current. The duration T of this long-term charging current is... i2 The current value is the maximum required current I. MAX It is used to identify the initial ohmic internal resistance and initial polarization internal resistance of the battery during long-term charging; the second zero-input response stage from the fourth time t4 to the fifth time t5 adopts zero current and has a duration of n. t ×T i2 This is used to identify the time constant of the polarization process during long-term charging. The second waveform, spanning from time t5 to t6, employs a periodic square wave current, with the high-level current of the square wave being the maximum demand current I. MAX The low-level current is the minimum required current I. MIN The width of the square wave in the second waveform is T. i3 The duty cycle is 0.5–0.

8. One cycle consists of a combination of high-level and low-level currents. The number of cycles in the second waveform is Nt, where Nt is a natural number greater than 10. The second waveform features a current step to identify the battery open-circuit voltage OCV and the battery's internal ohmic resistance R0. It also has a high current switching frequency for online identification of polarization parameters, including the polarization resistance R0. p and polarization capacitor C p The width of the square wave in the second waveform is T. i3 The duty cycle is 0.5 to 0.8 to eliminate battery polarization during long-term constant current charging, prevent the battery from entering the voltage plateau too early, and provide richer raw data for the algorithm. The third waveform uses a continuous constant current, and the current of the continuous constant current is the maximum demand current I. MAX The duration of constant current for the third waveform is T. i4 For the concentration polarization parameters reflected in the second-order RC equivalent circuit model, the concentration polarization parameters include the concentration polarization internal resistance R2 and the concentration polarization capacitance C2. The concentration polarization response time is relatively long, and the concentration polarization parameters are low-frequency parameters. Using long-term constant current data as input, the concentration polarization parameters are identified through optimization algorithms. 2) Connect the vehicle battery pack to the charging equipment. The charging equipment has a built-in data acquisition device and controller. The first waveform of the pulse charging current designed in step 1) is used to charge the vehicle battery pack. At the same time, the data acquisition device of the charging equipment samples the terminal voltage measurement value and current measurement value corresponding to the first waveform. After the first waveform charging is completed, the data sampling ends and the charging equipment stops charging the vehicle battery pack. The sampled terminal voltage measurement value and current measurement value corresponding to the first waveform are input to the controller. 3) The controller determines the optimal battery model: The battery model consists of a first-order RC equivalent circuit model and a second-order RC equivalent circuit model. The initial battery parameters of the first-order RC equivalent circuit model include the initial ohmic internal resistance R. 0ini Initial polarization internal resistance R pini and initial polarization capacitance C pini The initial battery parameters of the second-order RC equivalent circuit model include the initial ohmic internal resistance R. 0ini Initial electrochemical polarization internal resistance R 1ini Initial electrochemical polarization capacitance C 1ini Initial concentration polarization internal resistance R 2ini and initial concentration polarization capacitance C 2ini The initial battery parameters are identified offline using a short-time charging current at the first and third time points t1 to t3, and a long-time charging current is used to identify the initial battery parameters offline at the third to fifth time points t3 to t5. The optimal battery model is selected by comparing the fitting accuracy of the terminal voltage of the two battery models and the consistency of the parameter identification results of the same battery model under the two current excitations. The initial battery parameters of the optimal battery model are also output. 4) Initialization of pre-stored parameters and pulse current curve: The initial battery parameters obtained in step 3) are used as initial values ​​for subsequent calculations, and the time constant is used as the width T of the square wave in the second waveform. i3 and the duration of constant current T of the third waveform i4 The second and third waveforms of the pulse charging current are pre-stored; 5) The charging equipment charges the vehicle battery pack by using the second and third waveforms of the pulse charging current pre-stored in step 4). The second and third waveforms are executed in a loop. During the charging process, the charging equipment's built-in data acquisition device samples the terminal voltage and current measurement values ​​corresponding to the second and third waveforms in real time. 6) Online identification of battery parameters at multiple time scales: While the charging equipment is charging the vehicle battery pack, the terminal voltage and current measurements corresponding to the second and third waveforms are sampled in real time and input to the controller built into the charging equipment. Simultaneously, based on the battery model's fast and slow variation characteristics, battery parameters are identified online at different time scales: When the battery model is a first-order RC equivalent circuit model, the battery parameters include: open-circuit voltage OCV, ohmic internal resistance R0, polarization internal resistance Rp, and polarization capacitance Cp; at the microscopic time scale, a recursive least squares algorithm with a forgetting factor is used to identify the battery open-circuit voltage OCV and ohmic internal resistance R0 online through the second waveform; at the macroscopic time scale, an adaptive extended Kalman filter algorithm is used to identify the polarization internal resistance Rp and polarization capacitance Cp online through the second waveform. When the battery model is a second-order RC equivalent circuit model, the battery parameters of the second-order RC equivalent circuit model include: open-circuit voltage OCV, ohmic internal resistance R0, electrochemical polarization internal resistance R1, electrochemical polarization capacitance C1, concentration polarization internal resistance R2, and concentration polarization capacitance C2. At the microscopic time scale, a recursive least squares algorithm with a forgetting factor is used to identify the open-circuit voltage OCV and ohmic internal resistance R0 online through the second waveform. At the macroscopic time scale, an adaptive extended Kalman filter algorithm is used to identify the electrochemical polarization internal resistance R1 and electrochemical polarization capacitance C1 online through the second waveform. The electrochemical polarization internal resistance R1 and electrochemical polarization capacitance C1 are collectively referred to as electrochemical polarization parameters. A genetic algorithm is used to identify the concentration polarization internal resistance R2 and concentration polarization capacitance C2 online through the third waveform. 7) Accurately estimate the battery pack's state of charge (SOC) and state of health (SOH) and determine its convergence: While the charging equipment is charging the vehicle battery pack, the terminal voltage and current measurements corresponding to the second and third waveforms are sampled in real time and input to the charging equipment controller. Estimating the State of Charge (SOC) and State of Health (SOH) of the battery pack at different time scales and determining their convergence: At the microscale, the State of Charge (SOC) of the battery pack is estimated using an adaptive extended Kalman filter algorithm. Simultaneously, the difference between the predicted and observed voltages is output, and this difference is called innovation. The State of Health (SOH) is estimated at the current macroscale using this innovation update. The result of the SOH estimation at the macroscale is input for calculation at the next macroscale. The SOC and SOH estimates mutually correct each other, i.e., a joint estimation of the SOC and SOH. When the SOH estimation meets the convergence criteria, charging ends, and the current SOH is output, completing a rapid health detection.

2. The rapid health detection method as described in claim 1, characterized in that, In step 3), determining the optimal battery model includes the following steps: a) Offline identification of initial Ohmic resistance: The first-order RC equivalent circuit model and the second-order RC equivalent circuit model of the battery affect the initial ohmic internal resistance R. 0ini The identification method is the same: the initial ohmic internal resistance R is obtained by the ratio of the voltage drop generated at the instant the current is introduced to the current. 0ini : In the formula, ΔU 充 I represents the voltage change during a current step during charging. 充 ΔU is the change in current during the instantaneous current step during charging. 静 I is the voltage change during a step current change at the moment of rest. 静 This represents the change in current during a step change at the moment of rest. b) Offline identification of initial polarization parameters: Identification of initial polarization parameters for a first-order RC equivalent circuit model: i. The current is zero during the first and second zero-input response phases, and the time constant τ = R. pini C pini R pini For the initial polarization internal resistance, C pini The initial polarization capacitor; the terminal voltage U at time k. L (k) In the zero-input response phase: In the formula, U OCV (k) is the open-circuit voltage at time k, U p (k-1) is the polarization voltage at time k-1, Δt is the sampling interval, and the time constant τ is obtained by fitting using the least squares method; ii. The current during the first and second zero-state response phases is the maximum demand current I. MAX The terminal voltage U at time k L (k) In the zero-state response phase: In the formula, I(k) is the charging current at time k; Substituting the time constant τ obtained in step i) above, the initial polarization resistance R is obtained by fitting using the least squares method. pini The initial polarization capacitance C is further obtained. pini ; iii. Through the initial polarization internal resistance R pini and initial polarization capacitance C pini Substituting the formula for the terminal voltage at time k in the zero-state response stage, we obtain the fitted value U of the terminal voltage at time k. 拟合 ; Identification of initial polarization parameters for the second-order RC equivalent circuit model: i. The current is zero during the first and second zero-input response phases, and the electrochemical time constant τ1 = R 1ini C 1ini Concentration time constant τ2=R 2ini C 2ini R 1ini R is the initial electrochemical polarization resistance. 2ini For the initial concentration polarization internal resistance, C 1ini For the initial electrochemical polarization capacitance, C 2ini The initial electrochemical polarization capacitance, and the terminal voltage U at time k. L (k) In the zero-input response phase: In the formula, U1(k-1) is the electrochemical polarization voltage at time k-1, and U2(k-1) is the concentration polarization voltage at time k-1; the electrochemical time constant τ1 and the concentration time constant τ2 are obtained by least squares fitting. ii. The current during the first and second zero-state response phases is the maximum demand current I. MAX The terminal voltage U at time k L (k) In the zero-state response phase: Substituting the electrochemical time constant τ1 and the concentration time constant τ2, the initial electrochemical polarization resistance R is obtained by least squares fitting. 1ini and initial concentration polarization internal resistance R 2ini The initial electrochemical polarization capacitance C was further obtained. 1ini and initial concentration polarization capacitance C 2ini ; c) Determine the optimal battery model: The terminal voltage fitting accuracy of the next-order RC equivalent circuit model under current excitation from the first time step to the third time step t1 to t3, and the terminal voltage fitting accuracy of the next-order RC equivalent circuit model under current excitation from the third time step to the fifth time step t3 to t5 are calculated respectively. The calculation method for the terminal voltage fitting accuracy is as follows: In the formula, U 拟合 U is the fitted value of the terminal voltage. 测量 ε1 represents the measured terminal voltage value, and ε1 represents the offline identified terminal voltage error. If the terminal voltage fitting accuracy of the first-order RC equivalent circuit model under current excitation from the first time to the third time t1~t3 and from the third time to the fifth time t3~t5 both satisfy Formula 2.1, then the terminal voltage fitting accuracy of the first-order RC equivalent circuit model is deemed to meet the requirements. Further, the consistency of the parameter identification results for the battery initial parameters is assessed. Battery polarization is closely related to battery charging duration. Short-duration pulse charging and long-duration pulse charging are used to determine whether the parameter identification results of the first-order RC equivalent circuit model for charging currents of different durations are consistent. If the consistency judgment condition for parameter identification results is met, it proves that the first-order RC equivalent circuit model can simulate battery polarization. The initial polarization resistance R is obtained from the initial polarization parameters of the first-order RC equivalent circuit model in step b). pini and initial polarization capacitance C pini The offline identification results R of polarization internal resistance from the first time point to the third time point t1 to t3 are extracted respectively. pt1-t3 Offline identification results of polarization capacitor C pt1-t3 The offline identification results of polarization internal resistance from the third time point to the fifth time point t3 to t5, R pt3-t5 Offline identification results of polarization capacitor C pt3-t5 The consistency of the parameter identification results is judged as follows: The consistency of the parameter identification results is judged by formula 2.2, where ε2 is the polarization parameter identification error; If Formula 2.2 is satisfied, then the consistency of the parameter identification results of the first-order RC equivalent circuit model is deemed to meet the requirements. The optimal battery model is determined to be a first-order RC equivalent circuit model. If either the terminal voltage fitting accuracy or the consistency of parameter identification results does not meet the conditions of Formula 2.2, then the optimal battery model is determined to be a second-order RC equivalent circuit model.

3. The rapid health detection method as described in claim 1, characterized in that, In step 4), the initial parameters of the battery are determined by the optimal battery model determined in step 3). The initial value of the ohmic internal resistance R0 is adopted from the initial ohmic internal resistance R under the current excitation at the first time to the third time t1~t3 and the third time to the fifth time t3~t5 in step 3). 0ini The average value of the calculation results is used, and the initial values ​​of the polarization parameters are the average values ​​of the offline identification results of the initial polarization parameters under current excitation from the first time to the third time t1~t3 and from the third time to the fifth time t3~t5. When the battery model is a first-order RC equivalent circuit model, the width T of the square wave in the second waveform is determined by the time constant τ. i3 , T i3 =τ(3.1) The duration of constant current T of the third waveform i4 The value range is 100s to 1000s; When the battery model is a second-order RC equivalent circuit model, the width T of the square wave in the second waveform is determined by the electrochemical time constant τ1 and the concentration time constant τ2. i3 and the third waveform's continuous constant current time T i4 : The pulse charging current is pre-stored based on the second and third waveforms.

4. The rapid health detection method as described in claim 1, characterized in that, In step 6), the ohmic internal resistance and polarization internal resistance in the battery model correspond to impedances at different frequencies. Battery parameter identification needs to be performed on different time scales. Two time scales are defined: a microscopic time scale and a macroscopic time scale. The step size of the microscopic time scale is consistent with the sampling interval, denoted by k. The step size of the macroscopic time scale is an integer multiple of the step size of the microscopic time scale, denoted by m. In the formula, This indicates rounding up; when the battery model is a first-order RC equivalent circuit model, τ′=τ; when the battery model is a second-order RC equivalent circuit model, τ′=τ1.

5. The rapid health detection method as described in claim 4, characterized in that, Online identification of open-circuit voltage OCV and ohmic internal resistance R0 of first-order and second-order RC equivalent circuit models at microscale time: The parameter identification scheme for the first-order RC equivalent circuit model is implemented by a combination of recursive least squares with forgetting factor (FFRLS) and adaptive extended Kalman filter (AEKF); the parameter identification scheme for the second-order RC equivalent circuit model is implemented by a combination of recursive least squares with forgetting factor (FFRLS), adaptive extended Kalman filter (AEKF), and genetic algorithm (GA). The prerequisite for online parameter identification is obtaining the discretized difference equations of the battery model. The difference equations of the first-order RC equivalent circuit model are as follows: U L (k)=α1U L (k-1)+α2I(k)+α3I(k-1)+α4(4.2) In the formula, U L (k-1) is the terminal voltage at time k-1, I(k-1) is the charging current at time k-1, α1 is the coefficient of the terminal voltage at time k-1 in the first-order RC equivalent circuit model, α2 is the coefficient of the charging current at time k in the first-order RC equivalent circuit model, α3 is the coefficient of the charging current at time k-1 in the first-order RC equivalent circuit model, and α4 is the natural number related to the open-circuit voltage at time k in the first-order RC equivalent circuit model, α4=(1-α1)U OCV (k); Derive the analytical expressions for the ohmic internal resistance R0 and the open-circuit voltage UOCV from the coefficients of the differential equations in the first-order RC equivalent circuit model: Difference equations for the second-order RC equivalent circuit model: U L (k)=α5U L (k-1)+α6U L (k-2)+α7I(k)+α8I(k-1)+α9I(k-2)+α 10 (4.3) In the formula, U L (k-2) is the terminal voltage at time k-2, I(k-2) is the charging current at time k-2, α5 is the coefficient of the terminal voltage at time k-1 in the second-order RC equivalent circuit model, α6 is the coefficient of the terminal voltage at time k-2 in the second-order RC equivalent circuit model, α7 is the coefficient of the charging current at time k in the second-order RC equivalent circuit model, α8 is the coefficient of the charging current at time k-1 in the second-order RC equivalent circuit model, α9 is the coefficient of the charging current at time k-2 in the second-order RC equivalent circuit model, α 10 α is a natural number related to the open-circuit voltage at time k in the second-order RC equivalent circuit model. 10 =(1-α5-α6)U OCV (k), U OCV (k) is the open-circuit voltage at time k; The analytical expressions for the ohmic internal resistance R0 and the open-circuit voltage UOCV are derived from the coefficients of the difference equations in the second-order RC equivalent circuit model: Online parameter identification of the ohmic internal resistance R0 and open-circuit voltage OCV is performed using the recursive least squares method with forgetting factor FFRLS. After the recursive process ends, the coefficient matrix of the difference equation is obtained. The corresponding ohmic internal resistance R0 and open-circuit voltage OCV are obtained through the above analytical expression. The recursive least squares method with forgetting factor FFRLS introduces the forgetting factor λ when calculating the gain and error covariance.

6. The rapid health detection method as described in claim 4, characterized in that, Online identification of polarization resistance Rp and polarization capacitance Cp for a first-order RC equivalent circuit model, and electrochemical polarization resistance R1, electrochemical polarization capacitance C1, concentration polarization resistance R2, concentration polarization capacitance C2, and healthy state SOH estimation for a second-order RC equivalent circuit model on a macroscopic time scale: In the second-order RC equivalent circuit model, the concentration polarization parameters are calculated using a genetic algorithm on a macroscopic time scale. The genetic algorithm searches through a survival-of-the-fittest genetic mechanism, generating better approximate solutions generation by generation. It will not get stuck in local optima and can achieve accurate parameter identification. When the current enters the third waveform, the current measurement value and voltage measurement value are stored until the end of the third waveform. Using the current measurement value as input and the voltage measurement value as observation, the concentration polarization parameters are identified through a genetic algorithm. The optimization objective of the genetic algorithm is to minimize the sum of squares of the differences between the measured terminal voltage and the model-predicted terminal voltage. In the genetic algorithm, battery parameters are represented as individuals in the genetic space, and the individual fitness function is U. F Let it be the reciprocal of the mean square error of the terminal voltage: In the formula, N represents the total number of sampling points. The terminal voltage measurement at time k; Determine whether there is an individual among all individuals that meets the optimization termination condition. The judgment condition is set as the terminal voltage error of the optimal individual being less than ω1. If the condition is met, this individual is taken as the identification result of the concentration polarization parameter.

7. The rapid health detection method as described in claim 1, characterized in that, In step 7), the joint estimation of the battery pack's state of charge (SOC) and state of health (SOH) at multiple time scales includes SOC estimation at the micro-time scale and capacity estimation at the macro-time scale. Simultaneously, the polarization parameters in the first-order RC equivalent circuit model and the electrochemical polarization parameters in the second-order RC equivalent circuit model are also estimated along with the capacity at the macro-time scale. For the first 10–1000 micro-steps after the joint estimation begins, the polarization parameters and capacity remain unchanged from their initial values. After 10–1000 micro-steps, the parameter estimator is activated, using the data from the first 10–1000 micro-steps to perform the first update of the polarization parameters and capacity values. The updated values ​​are input into the micro-time scale for SOC calculation. When the micro-time scale calculation meets the criteria for the next macro-time scale, the macro-time scale parameter estimator is activated, updating the polarization parameters and capacity values, and continuing to input them into the micro-time scale for SOC calculation. This process iterates continuously until the calculation is complete.

8. The rapid health detection method as described in claim 7, characterized in that, The State of Health (SOH) is defined as follows: In the formula, Cn is the battery capacity, C ini This is the initial capacity value; The state of health (SOH) is determined by comparing it with the SOH calculated using a joint estimation scheme, based on the ratio of accumulated ampere-hours (AH) to the change in the battery pack's state of charge (SOC). Ah For the results of health status calculation using the ampere-hour integral method, the method for estimating the state of health (SOH) using the ampere-hour integral method at a single macroscopic time scale is as follows: In the formula, I m,k The charging current at time k on the m-th macroscopic time scale; SOC m,L-1 The state of charge (SOC) of the battery at time L-1 on the m-th macroscopic time scale. m,0 The state of charge of the battery at time 0 on the m-th macroscopic time scale is given. Capacity estimation reliability assessment: In the formula, N SOH The number of macroscopic time scales used to determine capacity reliability is 3 to 10, where n represents the nth macroscopic time scale and is a natural number greater than 3. ε3 represents the relative error of SOH in the healthy state, when N consecutive SOH The ampere-hour integral estimate of the state of health (SOH) at a macroscopic time scale satisfies Equation 5.2, and the joint estimation algorithm estimates the state of health (SOH) in accordance with Equation 5.

3. Therefore, the capacity estimation of the state of health (SOH) is considered reliable, and charging is stopped.

Images