Technology Research On The SOC Predication Of Power Battery Used In EV

Has been galloping more than hundreds of years in human’ history, in the21stcentury，automobile not only brings convenience to people, but also brings serious harm tohuman survival environment. Research shows that70%of the urban environment pollutioncome from car tail gas, car exhaust pollution has became a city big invisible killer. Atpresent the ultimate outlet to solve this problem lies in the development of low pollutionand even pollution-free transport, under such background, the new energy vehicles arestructured in time of use.At present, the new energy vehicle plays a strategic position in thecountry’s car industry development, the pure electric vehicle of them become hot spots ofthe research and manufacturing with its no pollution, the small noise, simple structure,convenient maintenance. And as the core of the new energy vehicles, power battery is oneof the bottleneck for restricting its development. This requires each auto enterprise andrelated industry personnel constantly innovate new energy vehicle power battery productand improve the detection technology level according to new energy vehicle technologydevelopment requirements,so that improve the comprehensive performance of automobileproducts, adapt to the needs of the development of automobile industry of purpose.This paper with the power of the electric car-battery key technology of SOC forecasttechniques are studied, the subject from chongqing general information engineering andindustrial Co., LTD. New energy vehicles with power battery detection technologyinnovation center. Mainly for simulation of the actual power battery operating conditions.Study the electric car power battery simulation and simulation experiment technology andmethods.This paper has conducted the research to the electric automobile power batterykey technologies——SOC forecasting technology，the topic derive from ChongqingYongtong information technology industrial corporation limited power battery with newenergy automobile detection technology innovation center. Research is mainly directedagainst simulate power battery actually operation condition. Study simulation、simulationexperiment technology and methods. Conveniently complete electric car’s power battery’seach kind of experiment through the hardware establishment and the softwareprogramming, provide the effective method to scientifically appraise its performance，make earlier period essential fundamental research for later period’s battery and theautomobile match. This article first detailed analysis the basic working principle and dischargecharacteristics of lithium iron phosphate battery charge，as well as affect the batteryremaining capacity (SOC) factors, select the main influence factors, namely, current,voltage, temperature. According to the above factors as input neural network mathematicalmodel, the output of the mathematical model is SOC. Facing the shortcomings that thetraditional BP neural network learning convergence is slow, prone to long flat areas,cannot be shrunk to a global minimum and other shortcomings, improve the neural network,and use simulation to prove such practicality of the improve method.In order to coordinate the BP network training, as well as the following convenienceoperators implement the online monitor, this paper compile a special BP network trainerusing MATLAB, realize the function of BP network to seal one kind, base on this kind,constructed the BP study software. Then, build the verifying bench, carry on theexamination experiment, gather vehicle travel signals including the electric current, thevoltage, the temperature, use the re-development software to carry on the electricautomobile SOC forecast.Finally, analytical the error between the predicted value derived from the model and theactual remaining capacity，prove the high accuracy with such a model for SOC estimation.In addition, throng contrast the battery theory charge-discharge curves and the actualdischarge curves，We can realize the judgment of the failure for the battery pack.This article describes neural network can detection online predict the remainingcapacity of the pure electric vehicle battery pack，through the improved BP algorithm andtraining software, set up the hardware platform to achieve the SOC accurate onlineprediction system.