Research On The SOC Estimatioii Of Power Li-ion Battery Based On ANFIS Model

In order to solve the energy crisis and environmental pollution, the automobilemanufacturers all over the world have been committed to research and production ofhybrid vehicles actively. The development of battery technology is of great significanceto the development of hybrid vehicles. The SOC of battery describes the number ofremaining power, is one of the most important parameters during the usage of battery.Estimate the SOC exactly can extend battery life effectively, and can predict the drivingrange.Affected by the charge and discharge rate, temperature, self-discharge, aging andother factors, SOC showed a high degree of nonlinearity in the battery using process,this brought great difficulties to estimate the SOC accurately. The battery of Hybridvehicle operating has its own characteristics: the temperature spacing, which can reach55℃; frequent and large changes in charge and discharge. Existing SOC estimationalgorithm either did not consider the temperature factor, or deal with the temperature asconstant; In addition, the relationship between the SOC of the battery and the batteryterminal voltage is nonlinear, but many algorithms handle this relationship as a lineardirectly. This has led to a great error for the SOC estimation of hybrid car battery pack,even led to the unavailable of fine management for hybrid electric vehicle energy, whichdirectly affect the economy of hybrid vehicles and its promotion process.As Lithium-ion battery for the study, this paper introduced the working principleand performance of the lithium-ion battery, analyzed the charge-discharge performanceof lithium-ion battery, made the charge and discharge experiments for100Ah powerlithium battery, and collected voltage, current, temperature, and the remaining capacityof the experimental data.On basis of traditional battery performance model, paper analyzed the advantage anddisadvantage of SOC estimation methods, introduced Adaptive Neuro-Fuzzy Inference Systemswhich integrated artificial neural network and fuzzy logic have predicted SOC of battery. Thisapproach take full advantage of the fuzzy logic is simple, strong robustness and accuracy of theneural network system, and the neural network system for the three-input single-output system, thehidden nodes is greatly reduced, and easy to implement.And so to research scientifically the factors of which affect the SOC estimation oflithium-ion battery, to establish the ANFIS model which estimation the SOC of powerLi-ion battery. even to use the collected charge and discharge data to make trainingsimulation for model. Experiment which is based on the MATLAB platform is of idealprecision of simulation. The error between prediction value and measured value of theSOC is less than6%. This result shows that ANFIS has stronger adaptive ability andgeneralization ability, and can reflect the characteristics of the battery better, so it can beused to estimate the SOC effectively.