改进前车状态预测的自适应巡航控制策略

An Adaptive Cruise Control Strategy with Improved Preceding Vehicle State Prediction

摘要:

针对自适应巡航控制（ACC）系统中前车运动状态不确定性造成的模型失配和性能下降问题，提出一种改进前车状态预测的自适应巡航控制策略。首先，基于前车的历史速度和加速度信息利用时间卷积网络预测前车的未来加速度轨迹，然后将预测加速度作为扰动构建自适应巡航系统的预测控制模型，最后在Matlab—Carsim联合仿真平台进行仿真实验。实验结果显示，时间卷积网络能够对车辆加速度取得较好的预测效果，且相对于传统MPC改进方法能够减小速度跟踪误差，并提高跟驰车辆对前车速度变化的响应速度。

Abstract:

To address the issue of model mismatch and performance degradation caused by uncertainties in the motion state of a preceding vehicle within adaptive cruise control （ACC） systems， a strategy for ACC by improving the prediction of the state of the preceding vehicle was proposed. Initially， the future acceleration trajectory of the preceding vehicle is predicted using a time convolutional network （TCN）， which makes use of historical speed and acceleration information. Subsequently， the acceleration predicted is employed as a disturbance to formulate the predictive control model for the ACC system. Finally， simulation experiments are conducted on the Matlab-Carsim joint simulation platform. The experimental results demonstrate that favorable predictive results for vehicle acceleration are achieved by the TCN. Furthermore， compared to the traditional model predictive control （MPC）， the improved method leads to the reduction in velocity tracking errors and the enhancement in responsive speed of the following vehicle towards changes in the speed of the preceding vehicle.