Research On Adaptive Estimation Of State Of Charge Of Electric Vehicle Power Battery Under Noise Interference

As one of the key emerging strategic industries,the electric vehicle industry is an important measure for our country to respond to energy and environmental challenges,and to achieve overtaking in automotive technology.The trend of battery vehicles replacing traditional fuel vehicles has been irreversible.Vehicle power batteries and related technologies are becoming a research hotspot nowadays,and SOC estimation is the core research directions in this field.The ternary lithium-ion battery has high energy density and good consistency,and currently still has an absolute advantage in terms of installed capacity.This article is based on the study of ternary lithium-ion batteries,with the goal of improving the accuracy of SOC estimation under noise interference conditions.First,the standard available capacity and the functional relationship between open circuit voltage and SOC are calibrated through battery test experiments.In order to accurately describe the output characteristics of the battery,the structure of the dualpolarization equivalent circuit model is built.In order to improve the accuracy of the input signal and observation signal of the DP model,the wavelet threshold algorithm is used to denoise the current and voltage signals respectively,and the SNR is used to measure the denoising effect.After the structure of the DP model is determined,the value of the RC parameters need to be obtained.First,the least squares fitting method is used to identify the model parameters offline under HPPC conditions.On this basis,the adaptive recombination genetic algorithm is used to identify the parameters of DP model.To further improve the precision of DP model,FRLS and the LKF algorithm are respectively applied in the online parameter identification,and parameters of DP model are updated online according to the continuously updated current and voltage data.Finally,the idea of deviation compensation is introduced on the basis of the RLS algorithm,which is used to improve the parameter identification accuracy of the online identification algorithm in the noise interference environment.Based on the DP model,first combine the parameters of offline identification to build the SOC estimation models of the extended Kalman filter algorithm and the square root volume Kalman filter algorithm.on this basis,respectively introduce the forgetting factor adaptive algorithm and moving window adaptive algorithm to predict the statistical characteristics of noise.BCFRLS and ASRCKF are combined to realize the collaborative online estimation of the SOC value and the model parameter value to improve the estimation accuracy of the SOC.Finally,the estimation effect of the above estimation algorithm is verified under a variety of experimental conditions.Aiming at the problem that the accuracy of SOC estimation decreases sharply when the current signal is interfered by colored noise signals,an online DP model based on state expansion and a three-layer combined estimation algorithm of ASRCKF are proposed to simultaneously realize current sampling signal correction and battery model Online update of parameters and high-precision estimation of SOC.Experimental and simulation results show that the three-level estimation algorithm can realise highprecision estimation of SOC value when the current signal has a sampling deviation.