Lithium battery internal temperature estimation method based on AWPSO thermal parameter identification and IKF

Abstract

The invention discloses a lithium battery internal temperature estimation method based on thermal parameter identification of AWPSO and IKF, and aims to solve the problems that thermal parameter identification is susceptible to noise interference and the temperature estimation precision is insufficient in the prior art, the method comprises the following steps: firstly, collecting battery voltage, current and surface temperature data, and constructing a lumped parameter thermal model; an adaptive weighted particle swarm optimization (AWPSO) algorithm is adopted to dynamically adjust the inertia weight, and thermal resistance and thermal capacity parameters are identified by taking the error minimization of the model calculation temperature and the actually measured temperature as a target; secondly, based on a thermal model of identification parameters, the internal temperature is estimated in real time through an incremental Kalman filter (IKF) algorithm, a residual feedback mechanism is introduced into the algorithm to dynamically update noise covariance, and the state prediction and measurement updating process is optimized. According to the method, the robustness of thermal parameter identification and the accuracy of internal temperature estimation are remarkably improved, and an effective technical support is provided for safety monitoring and service life management of the lithium battery.

Description

Technical Field

[0001] This invention relates to the technical field of lithium-ion batteries, and in particular to a method for estimating the internal temperature of lithium batteries based on thermal parameter identification of AWPSO and IKF. Background Technology

[0002] Lithium-ion batteries, with their fast response, high energy density, and flexible deployment, have become one of the fastest-growing and most widely used energy storage technologies. To meet the demands of high power and high voltage, large numbers of lithium-ion batteries are connected in series. Lithium batteries generate heat during charging and discharging, and elevated temperatures can affect battery performance and lifespan. Excessively high temperatures not only accelerate the degradation of internal chemical reactions and reduce cycle life, but can also lead to a decrease in battery capacity and even thermal runaway, causing safety hazards such as battery fires or explosions.

[0003] Therefore, accurate estimation of the internal temperature of lithium batteries is necessary to ensure battery safety and stability, and is also crucial for extending battery life and improving equipment performance.

[0004] Estimating the internal temperature of a lithium battery mainly involves three aspects: 1. Establishment of thermal model (1) Analysis of thermal generation in lithium-ion batteries During the charging and discharging process of a lithium-ion battery, energy storage and release are achieved through the continuous conversion between internal chemical energy and electrical energy. Some energy is lost and dissipated as heat, essentially stemming from the heat generation effect during the electrochemical reaction process. Based on the differences in heat generation mechanisms, the heat generation of lithium-ion batteries can be classified into reversible heat, irreversible heat, side reaction heat, and mixed heat. The heat generation formula for a lithium-ion battery can be expressed as: In the formula, This heat is reversible, originating from the entropy change effect accompanying the insertion / extraction of lithium ions at the positive and negative electrodes. The direction of this thermal effect is related to the charge / discharge state. This heat is directly related to the reaction entropy and the direction of the current, exhibiting significant reversibility. This is irreversible heat, composed of both ohmic heat and polarization heat. Ohmic heat originates from the charge transport impedance in the electrode materials, current collectors, and electrolyte; polarization heat arises from the electrochemical polarization effect during lithium ion migration at the solid-liquid interface. Both types of heat sources exhibit irreversible exothermic processes. This is the heat generated by side reactions, in addition to the heat generated during charging and discharging. Under healthy battery conditions, the activity of these reactions is extremely low, and the heat generated is negligible. This heat is a mixed heat generated by the local concentration gradient relaxation effect caused by the non-uniform reaction of the electrode active materials. Under normal operating conditions, the material distribution inside the battery is relatively uniform, and the heat generated can be ignored.

[0005] Based on the above analysis of battery heat generation, lithium-ion batteries undergo numerous internal reactions during use, resulting in a highly complex heat generation mechanism. Accurately quantifying all heat generation presents significant challenges in model construction and computational verification. Therefore, when calculating the total heat generated by the battery, factors with relatively minor effects, such as side reaction heat and mixing heat, can be ignored. After simplifying the heat generation process, a simplified formula for lithium-ion battery heat generation is derived based on the Bernardi heat generation model: (1) In equation (1), The heat generation rate is expressed in W; the first part on the right side of the above equation represents irreversible heat, and its sign is always positive. I represents the battery current. The second part on the right represents reversible heat, the direction of which can be positive or negative. Battery temperature (unit: K). and These are the battery open-circuit voltage and terminal voltage, respectively. This is the temperature entropy coefficient of the battery (unit: V / K).

[0006] (2) Establishment of thermal model When selecting a lumped parameter model as the thermal model for the battery, the following assumptions need to be made: ① The heat of the lithium-ion battery is generated uniformly in the core region; ② The axial temperature of the lithium-ion battery is uniformly distributed; ③ The generated heat is transferred outward only through the center in the radial direction.

[0007] Based on the above assumptions, the battery thermal model is as follows: Figure 1 As shown. Among them. , These represent the internal thermal resistance of the battery and the thermal resistance of the battery casing, respectively. , These represent the internal temperature and surface temperature of the battery, respectively. This indicates the ambient temperature of the battery. , These represent the core heat capacity and the casing heat capacity, respectively.

[0008] Based on the above analysis, heat capacity is equivalent to capacitance, thermal resistance to resistance, heat source to current source, and temperature to electromotive force. Combining circuit theory, the battery thermal model expression is as follows: (2) The parameter matrix is ​​obtained based on the lithium battery thermal model. , Output matrix The expressions are as follows: (3) 2. Thermal parameters , , , Identification By identifying the thermal parameters of lithium batteries using the least squares method, thermal model parameters, including thermal resistance and thermal capacity, can be extracted from actual measurement data. This method optimizes battery temperature estimation by establishing a thermal model of the battery, collecting experimental data, and minimizing the error between the model and actual data.

[0009] 3 Temperature Estimation Algorithm The most commonly used algorithms include electrochemical impedance spectroscopy, data-driven estimation, and model-based filtering estimation.

[0010] The main drawback is: (1) Among the algorithms for estimating the internal temperature of lithium batteries, electrochemical impedance spectroscopy is relatively cumbersome to operate, and its measurement accuracy is limited for some systems. Data-driven estimation methods usually require a large amount of historical data for training and learning, and the data quality has a significant impact on the results. Model-based filtering estimation methods may lead to unstable or inaccurate filtering estimations if the assumptions about noise are inaccurate.

[0011] (2) Although the least squares method is commonly used in thermal parameter identification, it has some limitations. It is sensitive to noise, cannot effectively handle nonlinear problems, and is prone to getting trapped in local optima. In addition, the least squares method is highly dependent on the initial value and has high computational complexity when dealing with complex high-dimensional data.

[0012] Therefore, a new method for accurately estimating the internal temperature of lithium batteries is needed. Summary of the Invention

[0013] The purpose of this invention is to solve the problems in the prior art by proposing a method for estimating the internal temperature of lithium batteries based on thermal parameter identification using the Adaptive Weighted Particle Swarm Optimization (AWPSO) algorithm and the Incremental Kalman Filter (IKF) algorithm. AWPSO can better handle nonlinearity, noise, and global optimization problems, providing a more accurate and stable parameter identification structure, which helps to improve the accuracy of temperature estimation. IKF can solve the noise uncertainty problem in the Kalman Filter (KF) algorithm, thus improving the accuracy of temperature estimation.

[0014] To achieve the above objectives, this invention proposes a method for estimating the internal temperature of a lithium battery based on thermal parameter identification using AWPSO and IKF, comprising the following steps: Step S1, lithium battery information acquisition and thermal parameter identification, specifically includes S11 battery data acquisition, S12 battery thermal model temperature calculation, and S13 adaptive weighted particle swarm optimization algorithm (AWPSO) thermal model parameter identification. Step S13 uses the AWPSO algorithm to identify the parameters of the thermal model. , , , Specifically, the parameters to be identified are used as particle positions, and the root mean square error between the model-calculated temperature and the measured temperature is used as the fitness function; the inertia weights are dynamically adjusted. ,in, and These represent the maximum and minimum values ​​of the inertia weight, respectively. It is the average fitness value of the current population. This is the optimal fitness value. It is the first The fitness value of the generation is calculated; by iteratively updating the particle position and velocity, the fitness function is minimized, and the optimal thermal parameters are output. Step S2, Lithium-ion battery internal temperature estimation based on incremental Kalman filter (IKF) algorithm: The thermal parameters obtained in step S1 are input into the lumped parameter thermal model, and the internal temperature is estimated in real time using IKF. Specifically, this includes: S20 state prediction, S21 measurement update, S22 residual calculation, and S23 R... k and Q k Dynamic update, S24 Kalman gain calculation, S25 State update, S26 Covariance update, S27 Temperature estimation update, S28 Determine if all iterations are complete, S29 Output results and evaluation; The formula for calculating the residual in step S22 is as follows: In the formula Represents residual, This is the current measurement of the battery's internal temperature. It is the measured value of the ambient temperature at the current moment. It is a predicted estimate of the battery's internal temperature; Step S23 and The dynamically updated formula is In the formula, and This represents the old system noise covariance and observation noise covariance; and Represent the new system noise covariance and observation noise covariance; The adjustment factor represents the adjustment step size; The residual reflects the difference between the predicted and actual values.

[0015] Furthermore, steps S11 and S12 are as follows: The lithium battery information collected includes temperature, voltage, and current. A lumped-parameter thermal model for a lithium battery is constructed, where thermal capacity is equivalent to capacitance, thermal resistance to resistance, heat source to current source, and temperature to electromotive force. Its state-space expression is as follows: ,in , These represent the internal thermal resistance of the battery and the thermal resistance of the battery casing, respectively. , These represent the internal temperature and surface temperature of the battery, respectively. This indicates the ambient temperature of the battery. , These represent the core heat capacity and the battery casing heat capacity, respectively. The rate of heat generation; Parameter matrix , Output matrix The expressions are as follows: ; The collected data is input into the lithium battery thermal model.

[0016] Furthermore, step S11 specifically involves collecting battery data through a lithium battery information acquisition module. The lithium battery information acquisition module includes a voltage sensor, a current sensor, and a temperature sensor. The voltage sensor is used to collect the voltage information of each battery cell, the current sensor is used to collect the current information of each battery cell, and the temperature sensor is used to collect the temperature information of each battery cell.

[0017] Furthermore, the fitness function mentioned in step S13 is: ,in, To calculate the internal temperature of the battery, To measure the internal temperature of the battery, This represents the number of measured battery temperature data points.

[0018] Furthermore, in step S2, the specific steps are as follows: Step S20: State prediction, the formula is as follows ,in It is the state prediction vector at the current moment. It is the state transition matrix. It is the state estimation vector from the previous moment. It is a control input matrix. It is the system input heat generation rate at the previous moment; Step S21 Measurement Update, formula is: , It is the measurement update value at the current moment. The measurement matrix is ​​used to transform the state space into a measurement space; Step S24: Kalman gain calculation, the formula is as follows , Indicates Kalman gain, This is a measurement matrix used to transform the state space into a measurement space; , The prior covariance matrix for prediction. Estimate the covariance matrix of the state at the previous time step; Step S25: Status update, formula is as follows In the formula: This is the updated state estimate. It is the difference between the current measured value and the predicted measured value; Step S26: Covariance update, formula is as follows In the formula: It is the updated state estimate covariance matrix; Step S27 Temperature estimation update, internal temperature estimate is ,in, It is the updated state estimation vector; Step S28 determines whether all iterations have been completed and whether all temperature estimates have been completed. If not, continue with IKF estimation; if completed, output and evaluate the results. Step S29 outputs results and evaluations, estimating the internal temperature of the battery and evaluating its accuracy.

[0019] Furthermore, the method is performed under HPPC conditions at 25°C or constant current discharge conditions at 0.7°C, and the accuracy of temperature estimation is evaluated by root mean square error (RMSE) and mean absolute error (MAE).

[0020] The present invention also proposes a lithium battery management system, characterized in that the internal temperature of the battery is estimated by applying any of the methods described above.

[0021] The beneficial effects of this invention are: 1. This invention provides an adaptive weighted particle swarm optimization algorithm for parameter identification in lithium battery thermal models, which can better solve problems involving nonlinearity, noise, and global optimization, and provides a more accurate and stable parameter identification structure, thus helping to improve the accuracy of temperature estimation.

[0022] 2. This invention addresses the noise uncertainty problem in the Kalman filter algorithm by proposing an incremental Kalman filter algorithm to improve the accuracy of temperature estimation.

[0023] The features and advantages of the present invention will be described in detail through embodiments and in conjunction with the accompanying drawings. Attached Figure Description

[0024] Figure 1 It is a lumped parameter thermal model diagram in the existing technology; Figure 2 This is an overall flowchart of a method for thermal parameter identification and internal temperature estimation of lithium batteries based on AWPSO and IKF according to the present invention. Figure 3 This is a flowchart of the particle swarm optimization algorithm for a lithium battery internal temperature estimation method based on AWPSO thermal parameter identification and IKF according to the present invention. Figure 4 This is an AWPSO fitness function graph of a lithium battery internal temperature estimation method based on AWPSO thermal parameter identification and IKF according to the present invention. Figure 5 This is a comparison chart of the estimated internal temperatures of KF and IKF under 0.7C operating conditions; Figure 6 This is a comparison chart of the estimated internal temperatures of KF and IKF under HPPC operating conditions; Figure 7 This is a diagram showing the estimated internal temperature error between KF and IKF under HPPC operating conditions. Figure 8 This is a diagram showing the estimated internal temperature error between KF and IKF under a 0.7°C operating condition. Figure 9 This is a comparison chart of the model's output temperature and the measured temperature.

[0025] Figure 4 In the diagram: the horizontal axis represents the number of iterations, and the vertical axis represents the fitness value; Figure 5 , Figure 6 , Figure 9 In the figure, the horizontal axis represents time in seconds, and the vertical axis represents temperature in degrees Celsius. Figure 7 , Figure 8 In the graph, the horizontal axis represents time in seconds (s), and the vertical axis represents error. Detailed Implementation

[0026] See Figures 1-9 The present invention includes the following steps: Step S1, lithium battery information acquisition and thermal parameter identification, specifically includes S11 battery data acquisition, S12 battery thermal model temperature calculation, and S13 adaptive weighted particle swarm optimization algorithm (AWPSO) thermal model parameter identification. Step S2, Lithium-ion battery internal temperature estimation based on incremental Kalman filter (IKF) algorithm: The thermal parameters obtained in step S1 are input into the lumped parameter thermal model, and the internal temperature is estimated in real time using IKF. Specifically, this includes: S20 state prediction, S21 measurement update, S22 residual calculation, and S23 R... k and Q k Dynamic update, S24 Kalman gain calculation, S25 State update, S26 Covariance update, S27 Temperature estimation update, S28 Determine whether all iterations have been completed, S29 Output results and evaluation.

[0027] The present invention also proposes a lithium battery management system, characterized by using the method of the present invention to estimate the internal temperature of the battery.

[0028] Working process of this invention: The present invention provides a method for thermal parameter identification of AWPSO and internal temperature estimation of lithium batteries based on IKF. The working process is described in conjunction with the accompanying drawings.

[0029] (1) Thermal parameter identification In the established battery thermal model (invention content), the parameter to be identified is: , , , A global optimization algorithm can be used to identify the parameters. By comparing the measured battery temperature with the calculated output, the objective function criterion is minimized to determine the optimal parameter combination. An adaptive weighted particle swarm optimization algorithm is used to identify the parameters.

[0030] (2) Estimation of internal temperature of lithium battery Under the hybrid pulse power characteristic (HPPC) test conditions at 25℃ and the constant current discharge condition at 0.7C, the internal temperature was estimated using the IKF algorithm and compared with the KF estimation results. To evaluate the accuracy of the algorithm, the root mean square error (RMSE) and mean absolute error (MAE) were used as evaluation metrics to verify the accuracy of the estimation algorithm.

[0031] Step one (S1) involves collecting lithium battery information (temperature, voltage, current). The collected data is input into the lithium battery thermal model, and the AWPSO algorithm is used to identify the parameters of the thermal model. , , , This includes S11 battery data acquisition, S12 battery thermal model temperature calculation, and S13 AWPSO thermal model parameter identification.

[0032] Lithium-ion battery information acquisition module (S11): The sensor and information acquisition module includes: a voltage sensor, a current sensor, and a temperature sensor. The voltage sensor is used to acquire the voltage information of each individual battery cell. The current sensor is used to acquire the current information of each individual battery cell. The temperature sensor is used to acquire the temperature information of each individual battery cell.

[0033] Battery thermal model (S12): The internal temperature of the battery can be calculated by inputting the voltage and current collected in S11.

[0034] AWPSO Thermal Parameter Identification (S13): In the Particle Swarm Optimization (PSO) algorithm, "particles" represent the set of parameters to be identified in the thermal model, while "fitness" is evaluated by the "difference" between the temperature data obtained after substituting these parameters into the thermal model and the measured battery temperature data. This "difference" constitutes the objective function in the optimization process. Particles iteratively adjust the parameters to find the optimal solution that minimizes the objective function, thus achieving the identification and optimization of the thermal model parameters. The Particle Swarm Optimization (PSO) algorithm flowchart is attached. Figure 3 As shown.

[0035] In the initial stage of the algorithm, key parameters for particle swarm optimization (such as particle number, learning factor, inertia weight, etc.) and the fitness function are first initialized. Subsequently, the algorithm iteratively solves the objective function, recording the optimal solution for each particle in its historical search process during each iteration. ) and the best solution found so far for the entire population ( Based on formula (4), each particle dynamically adjusts its position and velocity to guide the search process toward a better solution. When the algorithm detects that the fitness value has reached the preset convergence condition or meets the optimization objective, the iterative process terminates and the final result is output.

[0036] (4) In the formula, and Representing velocity and position respectively; This represents the inertial weight, which affects the particle update rate; and Indicates the acceleration constant; and It is a random number between [0,1].

[0037] For particle swarm optimization, Changing the value affects both global and local search capabilities. Increasing... Values ​​that enhance global search capabilities and reduce [the risk of data loss]. The inertia weight is usually chosen as a fixed value based on experience in Particle Swarm Optimization (PSO), which lacks systematicity and precision, and makes it difficult to balance the relationship between global search and local search. Therefore, this invention uses an AWPSO algorithm that considers the change in fitness to dynamically adjust the inertia weight. The particle swarm can flexibly perform adaptive optimization according to the current search situation, which can explore the global search and use local information to avoid getting trapped in local optima. Formula (5) is used to balance global search and local search.

[0038] (5) In the formula, and These represent the maximum and minimum values ​​of the inertia weight, respectively. It is the average fitness value of the current population; This is the optimal fitness value; It is the first The fitness value of the generation.

[0039] The objective function is the root mean square error (RMSE) between the calculated and measured temperatures. During parameter identification, RMSE measures the difference between the calculated and measured temperatures. The objective function is shown in equation (6).

[0040] (6) In the formula, The calculated internal temperature of the battery (°C); The measured internal temperature of the battery (°C) is used for actual measurement. This represents the number of measured battery temperature data points.

[0041] Under a constant temperature environment of 25℃ and a discharge condition of 0.7C, the voltage, current, and temperature data of the battery were collected as inputs for parameter identification. Combined with the heat generation model, AWPSO was used for parameter optimization. The identification results are shown in Table 1, and the fitness function graph is attached. Figure 4 As shown, the optimal function value is 0.197948.

[0042] Table 1 Parameter Identification Results Step two (S2) is the process of estimating the internal temperature of the lithium battery based on the IKF algorithm. It includes S20 state prediction, S21 measurement update, S22 residual calculation, S23 dynamic update of Rk and Qk, S24 Kalman gain calculation, S25 state update, S26 covariance update, S27 temperature estimation update, S28 determining whether all iterations have been completed, and S29 outputting results and evaluation.

[0043] S20 State Prediction In Kalman filtering, state prediction is used to estimate the state at the current moment, based on the state estimate at the previous moment and the current control input. This is one of the key steps in the Kalman filtering algorithm, as it provides a prior estimate of the system state before obtaining the current measurement data.

[0044] State prediction formula: In the discrete-time model, the formula for state prediction is: (7) in It is the state prediction vector at the current moment. It is the state transition matrix. It is the state estimation vector from the previous moment. It is a control input matrix. It is the system input heat generation rate at the previous moment.

[0045] S21 Measurement Update In Kalman filtering, measurement update is a crucial step used to integrate current measurement data into the state estimate, thereby correcting the predicted state. This process involves not only directly updating the state variables but also adjusting the covariance matrix to reflect changes in uncertainty introduced by new measurement information. The detailed steps and formulas for measurement update are as follows: (8) It is the measurement update value at the current moment, which represents the measurement estimate based on the current predicted state. The measurement matrix is ​​used to transform the state space into a measurement space, and it defines the relationship between state variables and measurement variables. It is the state prediction vector at the current moment, which contains a prior estimate of the system state.

[0046] S22 Residual Calculation Residual calculation, as a key step in the incremental Kalman filter (IKF) algorithm, focuses on accurately quantifying the deviation between the measured and predicted values, as shown in formula (9). Specifically, a predicted temperature difference model is first constructed based on the current ambient temperature compensation term and the predicted internal battery temperature estimate. This model aims to reflect the expected change in battery surface temperature under conditions without external measurement interference. Subsequently, the actual measured internal battery temperature is compared with the output of the predicted temperature difference model, and the residual signal is obtained by accurately calculating the difference between the two.

[0047] (9) In the formula Represents residual, This is the current measurement of the battery's internal temperature. It is the measured value of the ambient temperature at the current moment. It is a predicted estimate of the internal temperature of the battery.

[0048] S23 and Dynamic updates In Kalman filtering, and These represent the measurement noise covariance matrix and the process noise covariance matrix, respectively. Their dynamic updates are crucial for improving filtering accuracy and adaptability. System noise covariance and observation noise covariance Typically set empirically, these parameters affect the accuracy of the algorithm. In practical applications, noise variance exhibits uncertainty. Therefore, a method is proposed to adjust the measurement noise covariance matrix Rk and the process noise covariance matrix. This method aims to improve the accuracy and adaptability of the Kalman filter algorithm. The specific steps are as follows: 1. Residual Calculation: Calculate the residual at the current moment, which is the difference between the measured value and the predicted value.

[0049] 2. Dynamic updates and : Dynamically adjust based on the square of the residuals and the current noise covariance matrix. and The specific formula is as follows: (10) In the formula, and This represents the old system noise covariance and observation noise covariance; and Represent the new system noise covariance and observation noise covariance; The adjustment factor represents the adjustment step size; The residual reflects the difference between the predicted and actual values.

[0050] Dynamic noise adjustment is achieved through residual feedback: an increase in residuals indicates model mismatch or increased noise, requiring further adjustment. and To improve adaptability; when residuals decrease, it reflects accurate prediction, thus reducing... and This enhances the confidence in measured data. This mechanism significantly improves the filter's environmental adaptability and estimation accuracy.

[0051] S24 Kalman Gain Calculation 1. To predict the prior covariance matrix, the prior estimated covariance matrix needs to be updated before calculating the Kalman gain. The formula is as follows: (11) In the formula: The prior covariance matrix for prediction. Here is the state transition matrix. Estimate the covariance matrix of the state at the previous time step. This is the dynamic process noise covariance matrix.

[0052] 2. Calculate Kalman gain Calculate the Kalman gain based on the prior covariance matrix and the measurement noise covariance matrix: (12) In the formula: Indicates Kalman gain, This is a measurement matrix used to transform the state space into a measurement space. This is for dynamically measuring the noise covariance matrix.

[0053] S25 Status Update In Kalman filtering, state update involves correcting the previous state estimate using the Kalman gain and residuals to obtain a more accurate current state estimate. The formula is: (13) In the formula: This is the updated state estimate. It is a predicted state. It is Kalman gain. This represents the difference between the current measured value and the predicted measured value. This process integrates model predictions and actual measurement information, reducing estimation errors and improving the accuracy of state estimation.

[0054] S26 Covariance Update In Kalman filtering, covariance updates are used to correct for the uncertainty in the estimate.

[0055] (14) In the formula: It is the updated state estimate covariance matrix. It is the predicted prior covariance matrix. It is Kalman gain. This is the measurement matrix. This step adjusts the covariance matrix by combining the Kalman gain and the measurement matrix, making the estimated uncertainty more accurately reflect the current system state.

[0056] S27 Temperature Update In the Kalman filtering process, the temperature update is based on the state estimation result to obtain the estimated value of the battery's internal temperature at the current moment, and the formula is as follows: Internal temperature estimate: (15) in, This is an estimated value for the internal temperature. It is the updated state estimation vector.

[0057] S28 determines whether all iterations have been completed. If not, it continues with IKF estimation; if completed, it outputs and evaluates the results.

[0058] S29 Output Results and Evaluation: Estimates the internal temperature output of the battery and evaluates its accuracy.

[0059] Internal temperature estimation was performed under HPPC and 0.7°C conditions, and the IKF algorithm was compared with the KF algorithm. See the attached figure. RMSE and MAE data are shown in the table.

[0060] Table 2 Temperature Estimation Evaluation Indicators After identifying the thermal parameters using the AWPSO algorithm, they are input into the thermal model. The comparison between the model output temperature and the measured temperature is shown in the attached figure.

[0061] The above embodiments are illustrative of the present invention and are not intended to limit the present invention. Any simple modifications to the present invention are within the scope of protection of the present invention.

Claims

1. A method for estimating the internal temperature of a lithium battery based on thermal parameter identification using AWPSO and IKF, characterized in that: Includes the following steps: Step S1, lithium battery information acquisition and thermal parameter identification, specifically includes S11 battery data acquisition, S12 battery thermal model temperature calculation, and S13 adaptive weighted particle swarm optimization algorithm (AWPSO) thermal model parameter identification. Step S13 uses the AWPSO algorithm to identify the parameters of the thermal model. , , , Specifically, the parameters to be identified are used as particle positions, and the root mean square error between the model-calculated temperature and the measured temperature is used as the fitness function; the inertia weights are dynamically adjusted. ,in, and These represent the maximum and minimum values ​​of the inertia weight, respectively. It is the average fitness value of the current population. This is the optimal fitness value. It is the first The fitness value of the generation is calculated; by iteratively updating the particle position and velocity, the fitness function is minimized, and the optimal thermal parameters are output. Step S2, Lithium-ion battery internal temperature estimation based on incremental Kalman filter (IKF) algorithm: The thermal parameters obtained in step S1 are input into the lumped parameter thermal model, and the internal temperature is estimated in real time using IKF. Specifically, this includes: S20 state prediction, S21 measurement update, S22 residual calculation, and S23 R... k and Q k Dynamic update, S24 Kalman gain calculation, S25 State update, S26 Covariance update, S27 Temperature estimation update, S28 Determine if all iterations are complete, S29 Output results and evaluation; The formula for calculating the residual in step S22 is as follows: In the formula Represents residual, This is the current measurement of the battery's internal temperature. It is the measured value of the ambient temperature at the current moment. It is a predicted estimate of the battery's internal temperature; Step S23 and The dynamically updated formula is In the formula, and This represents the old system noise covariance and observation noise covariance; and Represent the new system noise covariance and observation noise covariance; The adjustment factor represents the adjustment step size; The residual reflects the difference between the predicted and actual values.

2. The method for thermal parameter identification and IKF-based internal temperature estimation of lithium batteries according to claim 1, characterized in that: Steps S11 and S12 are as follows: The lithium battery information collected includes temperature, voltage, and current. A lumped-parameter thermal model for a lithium battery is constructed, where thermal capacity is equivalent to capacitance, thermal resistance to resistance, heat source to current source, and temperature to electromotive force. Its state-space expression is as follows: ,in , These represent the internal thermal resistance of the battery and the thermal resistance of the battery casing, respectively. , These represent the internal temperature and surface temperature of the battery, respectively. This indicates the ambient temperature of the battery. , These represent the core heat capacity and the battery casing heat capacity, respectively. The rate of heat generation; Parameter matrix , Output matrix The expressions are as follows: ； The collected data is input into the lithium battery thermal model.

3. The method for thermal parameter identification and IKF-based internal temperature estimation of lithium batteries as described in claim 1, characterized in that: Step S11 specifically involves collecting battery data through a lithium battery information acquisition module. The lithium battery information acquisition module includes a voltage sensor, a current sensor, and a temperature sensor. The voltage sensor is used to collect the voltage information of each battery cell, the current sensor is used to collect the current information of each battery cell, and the temperature sensor is used to collect the temperature information of each battery cell.

4. The method for thermal parameter identification and IKF-based internal temperature estimation of lithium batteries according to claim 1, characterized in that: The fitness function mentioned in step S13 is ,in, To calculate the internal temperature of the battery, To measure the internal temperature of the battery, This represents the number of measured battery temperature data points.

5. The method for thermal parameter identification and IKF-based internal temperature estimation of lithium batteries as described in claim 1, characterized in that: In step S2, the specific steps are as follows: Step S20: State prediction, the formula is as follows ,in It is the state prediction vector at the current moment. It is the state transition matrix. It is the state estimation vector from the previous moment. It is a control input matrix. It is the system input heat generation rate at the previous moment; Step S21 Measurement Update, formula is: , It is the measurement update value at the current moment. The measurement matrix is ​​used to transform the state space into a measurement space; Step S24: Kalman gain calculation, the formula is as follows , Indicates Kalman gain, This is a measurement matrix used to transform the state space into a measurement space; , The prior covariance matrix for prediction. Estimate the covariance matrix of the state at the previous time step; Step S25: Status update, formula is as follows In the formula: This is the updated state estimate. It is the difference between the current measured value and the predicted measured value; Step S26: Covariance update, formula is as follows In the formula: It is the updated state estimate covariance matrix; Step S27 Temperature estimation update, internal temperature estimate is ,in, It is the updated state estimation vector; Step S28 determines whether all iterations have been completed and whether all temperature estimates have been completed. If not, continue with IKF estimation; if completed, output and evaluate the results. Step S29 outputs results and evaluations, estimating the internal temperature of the battery and evaluating its accuracy.

6. The method for thermal parameter identification and IKF-based internal temperature estimation of lithium batteries according to claim 1, characterized in that: The method is performed under HPPC conditions at 25°C or constant current discharge conditions at 0.7°C, and the accuracy of temperature estimation is evaluated by root mean square error (RMSE) and mean absolute error (MAE).

7. A lithium battery management system, characterized in that... The method described in any one of claims 1-6 is used to estimate the internal temperature of the battery.

Images