Estimation Of Battery SOC Based On EKF Algorithm And Neural Network

The energy fell sharply and serious environmental pollution are the mostly important influence factors in the development of both at domestic and the global economy,both issues are prominent.So researching new renewable energy is the most important issues to domestic and abroad Government.With the rapidly development of science and technology and economy of contemporary Chinese society,the new energy automobile industry is developing rapidly.Statistics show that,the proportion of new energy vehicles sales have greatly improved in recent months China's car sales.But BMS technology still has a lot of upside and requires a lot of manpower and material resources.One of the most important is the assessment of the remaining capacity of the battery.The assessment of SOC is significant.Accurate assessment is not only significant to protect the batteries but also can protect the safety of drivers.The lithium-ion battery as the object of this thesis,and made the following aspects.First of all,this article describes estimating new energy automobile battery 's SOC development situation of domestic and foreign is given,the background of battery and battery performance and main parameters.Through referring to a large number of documents and Comparative analysis of individual battery model,this thesis establishes a lithium-ion battery equivalent circuit model to performance characteristics of internal mechanism.And then a series of experiments testing the battery are carried out,getting initial value of the parameters by index fitting in least squares algorithm and obtaining different open circuit voltage under different SOC.At this time,considering the effect of discharge-rate on the battery capacity,the number relationships of the effect of discharge-rate on battery capacity is calculated.Secondly,MATLAB/Simulink module is needed to validate the parameters which are of equivalent circuit model.The test results in Simulink simulation module showed that this equivalent circuit model can simulate the characteristics of batteries.And then,the relationship between parameters in equivalent circuit model and SOC is fitted by the neural network algorithm.Finally,State-space equation is established based on the equivalent circuit model.Kalman filter algorithm is mainly used in linear systems,but this model is non-linear systems.So EKF algorithm is used.The results can be seen,the EKF with neural network assessment of battery SOC in this thesis has the following advantages:strong anti-interference,strong convergence and high precision.