%0 Journal Article %T 基于PNNPID的车辆加速度控制器设计<br>Design of Vehicle Acceleration Controller Based on Parallel Neural Network PID %A 梁军 %A 赵彤阳 %A 熊晓夏 %A 张婉婉 %A 陈龙 %A 朱宁< %A br> %A LIANG Jun %A ZHAO Tongyang %A XIONG Xiaoxia %A ZHANG Wanwan %A CHEN Long %A ZHU Ning %J 西南交通大学学报 %D 2017 %R 10.3969/j.issn.0258-2724.2017.03.026 %X 针对传统的自适应巡航(adaptive cruise control,ACC)加速度控制算法对系统变化辨识速度慢、动态性能差、调整缓慢的问题,提出了一种基于PNNPID(parallel neural network proportion integration differentiation)的加速度控制算法.通过分析传统SNNPID(serial neural network proportion integration differentiation)对误差直接反馈的不足,应用神经网络自学习功能,开发了基于并行控制原理的车辆加速度控制器;考虑车辆平顺性,设计了符合驾驶行为的滤波限幅方案.实验结果表明,相比基于SNNPID的加速度控制器,基于PNNPID的加速度控制器最大偏差控制在±0.25 m/s2以内,并具有平均误差小、调整时间短和瞬态特性良好的特点,滤波限幅方案有效提高了驾驶的舒适性.<br>: An acceleration control algorithm based on the parallel neural network PID (PNNPID) was proposed to conquer the problems of slow system change identification, poor dynamic performance, and slow adjustment in traditional acceleration control algorithms for adaptive cruise control (ACC). Through analysis of the shortcomings of the traditional serial neural network PID (SNNPID) for direct feedback error, an acceleration controller for vehicles that is based on the parallel control theory is developed using the self-learning function of neural network. In addition, taking into account the vehicle ride comfort, an amplitude limiting scheme of filter is designed to fit driving behavior. Experimental results show that the acceleration controller based on SNNPID can function within a maximum deviation of ±0.25 m/s, and is more accurate than the controller based on PNNPID. This new acceleration controller is characterized by small average error, short adjustment time, and good transient property; and the proposed filter amplitude limiting scheme can greatly improve the ride comfort of vehicles %K 自适应巡航 %K 加速度控制器 %K 动态性能 %K 神经网络 %K PID %K < %K br> %K adaptive cruise %K acceleration controller %K dynamic performance %K neural network %K PID %U http://manu19.magtech.com.cn/Jweb_xnjd/CN/abstract/abstract12448.shtml