Research On Data-drive State Estimation And SOP For Lithium-ion Batteries

Lithium-ion batteries have been widely used in the field of electric vehicles because of their high energy density,long service life and low environmental hazard.In practical operation,an efficient battery management system and accurate battery state estimation play an important role in ensuring the safe and reliable operation of batteries and extending their service life.This article focuses on the accurate estimation of the state of charge,state of health,and state of power of lithium-ion batteries from a data-driven perspective,in response to issues such as significant errors in battery state estimation and insufficient adaptability in various operating conditions.The following work has been mainly carried out:(1)Construction of the power battery experimental platform and research on the battery model: With the C-300A-05V-8CH battery comprehensive performance testing equipment as the battery charge and discharge control equipment,the experimental database of the Ningde era ATL-V0D5N0 lithium iron phosphate battery was established,providing data support for battery state estimation.A second-order RC equivalent circuit model of the battery was established,and the parameters were identified offline/online using the least squares method with forgetting factor.Finally,the accuracy of the parameter identification method was verified through the terminal voltage curves obtained from HPPC testing,DST,and FUDS operating conditions testing,serving as the estimation basis for battery SOC and SOP.(2)Research on battery SOC estimation based on AOA-UKF algorithm: From the perspective of battery operating current and terminal voltage data driven,a SOC estimation algorithm based on unscented Kalman filtering is proposed through spatial state equations.At the same time,arithmetic optimization algorithms were adopted to optimize the parameters of the system noise covariance matrix and the measurement noise covariance matrix,solving the problem of difficulty in determining the values of the noise matrix during the filtering process.Finally,the AOA-UKF algorithm proposed in this paper was compared and analyzed with the PSO-EKF algorithm and PSO-UKF algorithm through the UDDS and US06 operating conditions,verifying the effectiveness and accuracy of the proposed improvement scheme.This solved the problem of low estimation accuracy and high initial SOC requirements of traditional Kalman filtering algorithms,and served as the basis for battery SOP estimation under SOC constraints.(3)Research on battery SOH estimation based on improved BiLSTM algorithm: A BiLSTM neural network model for SOH estimation was established through data-driven approach.Analyzed the mechanism of battery aging and extracted battery health features from the IC curve during the constant current charging stage as input variables for the network model.By assigning different weights to input variables of different importance through the self attention mechanism layer,the impact of important information on model estimation is increased.Then Golden Eagle optimization algorithm is used to optimize the best combination of hyperparameter of BiLSTM model,which improves the accuracy of SOH estimation.Finally,through the ATL battery aging data tested on the experimental platform and the aging data of batteries B05 and B06 in NASA’s publicly available datasets,the improved BiLSTM algorithm proposed in this paper was compared with BiLSTM,LSTM,and GRU algorithms,verifying that the improved method can effectively improve the accuracy and fitting degree of SOH estimation,laying the foundation for battery SOP estimation under SOH constraints.(4)Research on SOP estimation of lithium-ion batteries based on multi-state constraints: To address the problem of low accuracy in estimating SOP under single conditions,a multi-state conditional constrained SOP estimation method is proposed by combining accurate estimation of terminal voltage,SOC,and SOH.A battery constant power discharge experiment was conducted to obtain the true peak power values of the battery at different discharge durations(30s,2min,5min)using curve fitting method.Then,the peak power curves constrained by terminal voltage alone,SOC alone,and multi-state conditions were compared to verify the accuracy of the method adopted in this paper.Finally,the estimation results of battery peak power under different operating conditions,durations,and degrees of aging validate the adaptability and rationality of multi state condition constraints on SOP estimation in different situations.