Regression Analysis Based Automotive Dynamics Simulation

Auto dynamic is the most important performance of the most basic properties, their quality directly affect the quality of the car basic properties of automobile,fine performance depends largely on the auto dynamic transmission system matching design. Therefore the reasonable design, reasonably determine the parameters of the car drive transmission system is very important. This paper build regression analysis model, through the collection of more than 500 cars'dynamic system parameters and the sample training. The user can get vehicle dynamic result by only inputting all or part of dynamic system parameters of vehicle. This is the new easy to use vehicle dynamic simulation tool.This paper research the model methods of vehicle dynamic simulation, and analysis principle and adaptability of these methods in detail. Then put forward key factors influence the vehicle dynamic simulation and limitations of key factors from classic model methods. This paper do the sample training respectively, and get the regression analysis models, introduces three kinds of regression analysis about least squares, BP algorithm of neural network, genetic algorithm. Then we build more than 500 vehicle dynamic system database in WEKA, and do the sample training respectively. Finally, we presentation the result by example, to prove this method can simulate the vehicle dynamic.For simulation parameters vacancy problem, we combination the vehicle dynamic simulation model and the BP Neural network summarize the vacancy forecast model. The biggest error rate of model result to the real value is 3.5%, is suitable to use.For lack of simulation precision problem, we combination the vehicle dynamic simulation model and the BP Neural network, Genetic algorithm summarize the forecast model. The biggest error rate of model result to the real value is smaller than traditional methods.