Key Technologies Research On Energy Management Of Pure Electric Vehicle With Dual-Energy Storage System

In order to develop vehicle industry quickly, it is confronted with two important problems which are energy crsis and environmental pollution. A huge challenge for sustainable development of our country is how to develop both energy saving and environmental protection vehicles. Though pure electric vehicle has the advantages of zero emission and high efficiency and its strong points of energy saving and environmental protection can not be ignored, its driving ranges is limited in virtue of only using a energy source to provide energy for the system. At present, driving ranges has become the bottleneck for the development of pure electric vehicle. The energy storage element such as the ultra-capacitor has high specific power and is suitable to charge and discharge with high current. It is not suitable to continuous discharge because its specific energy is small. Combine battery and ultra-capacitor together and constitute dual-energy storage system. It can not only meet the instantaneous high power demand of electric vehicle but also extend the battery life and increase the driving ranges of electric vehicle. It has become the development tendency of pure electric vehicle. In this paper, control srategy and driving ranges of pure electric vehicle with dual-energy storage system are studied. They have broad application prospects and practical significance.The pure electric vehicle with dual-energy storage system is driven under complex conditions. Owing to its nonlinear model, the fuzzy control is usually used in energy distribution. Its control effect is not satisfactory in virtue of the limitations of control algorithm. In this paper, the self-adaptive fuzzy PI control is applied firstly in Pure Electric Vehicle with Dual-Energy Storage System. Its control effect is satisfactory by tuning the parameters of PI controller in real time. The driving ranges of electric vehicle usually relies on neural network for its forecast. It has strong robustness and fault tolerance. And it is good approximation ability to nonlinear complex system. However, the forecast effect was not satisfactory due to local minimization and overfitting phenomenon. The paper introduced the support vector machine (SVM) to the vehicle and its parameters were optimized by particle swarm optimization (PSO). It can complement the shortcomings of neural network. The specific content of the paper are illustrated as follows:At first, dynamic performance and the problems of energy consumption on pure electric vehicle are analyzed and several typical working conditions are described detailedly on the basis of the introduction of energy management of pure electric vehicle with dual-energy storage system.Secondly, establish the models of pure electric vehicle according to its dynamics theory in Matlab/Simulink. The experiments on the important parameters of battery and ultra-capacitor are done on the basis of analyzing the structure and characteristics of energy storage system. And the models of battery and ultra-capacitor are established according to the experimental results. The model of motor is established according to the structure and working principle of the motor. The model of DC/DC converter is established as the voltages between battery and ultra-capacitor are not equal. And it is controlled with double closed loop mode.Thirdly, analyse energy distribution strategy of energy storage system and establish the energy distribution strategy which not only can make full use of the energy storage elements but also can meet the system requirements. The fuzzy control model was established that its inputs were the battery SOC (state of charge), the ultra-capacitor SOC and required power and its output was the factor of battery power distribution. And fuzzy control strategy was lay down. Then the self-adaptive fuzzy PI control is adopted in this paper. The algorithm compares given power with actual power and then it will obtain the deviation e. The inputs of fuzzy control model are the deviation e and deviation change rate of Δe and its outputs are proportional coefficient Kp and integral coefficient Ki. The deviation values are adjusted properly by regulating the parameters of the PI controller. Compare the fuzzy control with the self-adaptive fuzzy PI control under five typical working conditions, the results of simulation experiment indicate that the self-adaptive fuzzy PI control is better control effect and economic performance than the fuzzy control.At last, the paper proposes that the battery SOC and ultra-capacitor SOC are important factors to affect the driving ranges of electric vehicle. Two algorithms which are BP neural network and SVM are presented on the basis of the theoretical analysis of the driving ranges and then the driving ranges of pure electric vehicle is forecast. Deal with the sample data and then establish BP neural network model. The driving ranges of electric vehicles is forecast after the model is verified correctly. The results show that the self-adaptive fuzzy PI control can drive father than the fuzzy control under five typical working conditions. The maximum relative error of BP neural network is2.66percent. PSO SVM is presteted in order to forecast the driving ranges of electric vehicle since the generalization ability of neural network is weak and its learning speed is slow. When the samples data are trained by SVM, the parameters of SVM which are the penalty factor C and the width coefficient of kernel function g are optimized by PSO. The aims are to obtain the accurate model of SVM and decrease the forecast errors. Compare BP neural network with PSO SVM under five typical working conditions, the results indicate that PSO SVM has smaller relative error than BP neural network. So it is more suitable to forecast the driving ranges of electric vehicles than BP neural network.