基于逆问题求解的汽车操纵性能分析

采用人-车-路闭环系统操纵性能的客观评价方法，通过优化评价指标得到了双移线道路输入时驾驶员最优的预瞄时间和跟随阶数。针对原始的和经过优化的两个闭环系统，利用径向基函数神经网络建立了汽车侧向位移和其他响应之间的映射关系。将待跟踪路径输入两个训练好的神经网络，反解出汽车其他的响应，可以方便地比较两个闭环系统跟踪同一条路径时的操纵性能。仿真结果表明：这种方法能够避免由于闭环系统参数变化而跟踪不同行驶路径进而对汽车操纵性能的比较产生的不利影响；基于径向基网络的逆问题求解方法是可行的，并且具有求解精度高、运算速度快及抗干扰能力强等优点。

Analysis of Lane-change Vehicle Maneuverability Based on Solution of Inverse Problems

Using closed-loop evaluation method,the paper obtained the optimal preview time and follow order of the driver through minimizing the comprehensive evaluation index in double lane change course.Based on the Radial Basis Function neural network,the mapping relationship between vehicle lateral displacement response and other responses could be found corresponding to the original and optimized closed-loop systems.One prescribed road was taken as inputs of the two trained RBF neural networks and other responses of the two systems could be identified respectively.Thus the handling performance of the two systems can be compared conveniently when they followed the same road.Simulation results show that the method can avoid the disadvantageous influence on the comparison of vehicles' handling performance when the system parameters are modified,which can cause different following roads.In addition,the solution method based on the RBF neural networks is feasible,and with high accuracy,little computation requirement and good stability.