基于自适应反馈的MPC车辆轨迹跟踪控制算法  

MPC vehicle trajectory tracking and control algorithm based on adaptive feedback

作　　者：丁炳超 王立勇[1] 苏清华 张政 DING Bingchao;WANG Liyong;SU Qinghua;ZHANG Zheng(Key Laboratory of Modern Measurement and Control Technology,Ministry of Education,Beijing Information Science and Technology University,Beijing 100192,China)

机构地区：[1]北京信息科技大学现代测控技术教育部重点实验室,北京100192

出　　处：《传感器与微系统》2024年第12期150-154,共5页Transducer and Microsystem Technologies

基　　金：国家“173”计划项目(2021JCJQJJ0022,MKF20210009);国家自然科学基金资助项目(52175074)。

摘　　要：为解决模型预测控制(MPC)设计轨迹跟踪控制器在过于简化的车辆运动学模型下高速运动时稳态误差大的问题,借助神经网络(NN)PID控制器,引入具有前瞻性的误差反馈机制,提出一种基于NNPID反馈的模型预测轨迹跟踪控制方法。通过求取预测T时域内的平均误差,利用NNPID的自适应特性,融入模型预测控制算法中,以提高对不同轨迹曲线跟踪控制的自适应能力并降低横向稳态误差。实验结果表明,改进后的算法能够有效使得稳态误差更接近零,平均误差与最大误差均降低30%以上;使用大曲率真实轨迹进行测试,最大误差降低52.99%,平均误差降低35.78%。In order to solve the problem of large steady-state error when trajectory tracking controller is designed by model predictive control during high-speed motion under oversimplified vehicle kinematic model,a prospective error feedback mechanism is introduced by using neural network PID controller,and a model predictive trajectory tracking and control method based on neural network PID feedback is proposed.By calculating the average error of T time domain of prediction,using the adaptive characteristics of neural network PID,integrating into the model predictive control algorithm,in order to improve the adaptive ability of different trajectory curve tracking and control and reduce the lateral steady-state error.The experimental results show that the improved algorithm can effectively make the steady-state error closer to zero,and both the average error and the maximum error are reduced by more than 30%.The maximum error is reduced by 52.99% and the average error is reduced by 35.78% when using the real trajectory with large curvature to test.

关 键 词：自动驾驶车辆 轨迹跟踪 模型预测控制 神经网络PID 

分 类 号：TP249[自动化与计算机技术—检测技术与自动化装置]