Battery Remaining Capacity Forecasting Method For Electric Vehicle

Electric Vehicles (EV) are a kind of environmentally friendly cars powered by batteries. The performance of batteries plays an important role on the whole performance of EV, and directly influences its performance, such as, the range of driving, acceleration ability, climbing capacity and so on. This thesis focuses on the prediction of the battery remaining capacity of batteries Hope that the work done can make a meaningful attempt for the battery testing industry.At first introduced the development history and status quo of power battery, and several forecasting methods of the remaining capacity. And then analyzing each of the existing forecasting methods we got the conclusion that the current commonly used methods have various deficiencies, due to the characteristics of the battery is influenced by lots of constraints, difficult to determine the input-output relations system, because Neural network have the capacity of arbitrary approaching nonlinear, uncertain system, therefore the effect of the method used to predict the battery remaining capacity is significantly. In this paper LiFePO4 batteries is researched, by careful analysis of LiFePO4 battery's work principles, collecting the actual data battery's operating voltage and current under different temperatures, on base of analyzing the battery voltage, current and temperature characteristics, we introduced the forecasting method of BP neural networks, and established a prediction model, to overcome the shortcomings of BP neural network easily be trapped in local minimum, we further raised the forecasting method of BP neural networks and intelligent forecasting method of genetic algorithms GA, a combining GA-BP algorithm was proposed, Aspects from the structural parameters and training parameters of the network we improved the prediction model. Simulation results shows that the improved prediction model not only ensure the accuracy of the prediction but also improves efficiency.According to the work done established the testing system, using this system to predict the data which not actually measured, comparison the actual measurement results with the predicted results, we got the conclusion that the range of error is allowed of the project, verify the accuracy and reliability of the system.