为提高分布式驱动电动汽车转向稳定性，解决传统神经网络控制算法收敛速度 慢、易陷入局部最优解的问题，提出一种利用粒子群算法优化神经网络的比例－积分－ 微分（PID）转向稳定控制器，利用横摆力矩和滑移率调整力矩实现横摆角速度和各轮 滑移率的控制。在此基础上研究了一种针对转向工况的最优力矩分配算法，通过模糊控 制算法对驱动力矩进行修正得到驱动修正力矩，将其与横摆力矩和滑移率调整力矩一起 作为二次规划问题进行最优分配，得到各轮最佳驱动力矩。基于联合仿真平台进行了双 移线和蛇形等典型转向工况下的性能对比测试。结果表明：文中提出的算法能在保持车 辆良好动力性同时维持稳定性，稳定控制器能将蛇形工况打滑现象降低36.4％ ，最优 力矩分配算法能将双移线工况的稳定性提高31.2％ 。

A neural network PID method based steering stability controller was proposed by using particle swarm optimization. It aims to improve the steering stability of distributed drive electric vehicle and solve the problems of slow convergence and easy to fall into local optimal solution based on neural network control algorithm. This method can realize the yaw rate control and slip rate control of each wheel by yaw moment and slip rate adjustment torque. Based on this, an optimal torque distribution algorithm under steering condition was studied. The modified torque was obtained by correcting the driving torque with the fuzzy control algorithm. Then the modified torque was optimally distributed together with the yaw moment and slip rate adjustment torque of the stability controller as a quadratic programming problem to obtain the optimal driving torque of each wheel. Finally, the performance comparison tests under typical working conditions such as double lane shifting condition and serpentine condition were carried out based on the joint simulation platform. The results show that this algorithm can maintain the stability of the vehicle while maintaining good vehicle dynamics. The stability controller can reduce the slip phenomenon by 36． 4% under slalom condition, and the optimal torque distribution algorithm can improve the stability by 31． 2% under the double lane change condition.