基于ANFIS的工况跟踪用纵向驾驶员模型开发

Development of Longitudinal Driver Model for Driving Cycle Tracking Based on ANFIS

针对基于PI控制纵向驾驶员模型的工况跟踪效果不佳等问题,采集了某PHEV试验样车的纵向驾驶行为数据,采用ANFIS建立了二输入参数的纵向驾驶员模型和考虑未来预期车速影响的预瞄纵向驾驶员模型。仿真结果表明,基于ANFIS的预瞄纵向驾驶员模型具有最好的工况跟踪效果,均方根误差为0.993 0 km/h,且其决策出的加速踏板行程、挡位和需求转矩与实车试验结果最为接近。

In view of the poor results of driving cycle tracking with PI control-based longitudinal driver model, the longitudinal driving behavior data of a tested PHEV are collected, the longitudinal driver models with two input parameters and preview longitudinal driver models with consideration of the effects of future expected vehicle speed are established by using adaptive network-based fuzzy inference system(ANFIS). The results of simulation show that the ANFIS-based preview longitudinal driver model has the best tracking effect of operation conditions with a root mean square error of 0.993 0 km/h, and the accelerator pedal travel, gear position and desired torque worked out are closest to the results of real vehicle test.