摘要
在分析磁流变阻尼器车辆悬架非线性特性的基础上 ,设计了一类神经网络间接自适应控制器 ,并根据系统的低频特性和作动器的快响应 ,实现了悬架振动的神经网络实时控制。计算机仿真和悬架实验的结果均表明 ,神经模拟器能够逼近非线性系统 ,神经控制器能在时域和频域内以较高的精度控制悬架系统的振动。
Due to the multi degree of freedom and time varied characteristics of vehicle suspension, it is not easy, even impossible, to achieve a precise mathematical model. Thus the desired control input cannot be calculated according to the appointed response. Based on the analysis of the direct adaptive control of neural networks and the magnetorheological damper of partⅠof this work, this paper presents an indirect adaptive control of neural networks. The neural network simulator not only tracks the plant but also back propagates the errors between the desired response and the output of the plant, which can gain a more adjacent input error for the neural network controller than the direct. Though the indirect laws increase the computational, but the results of simulation show that it does not affect the convergent rate and accuracy. In experiments the low frequency property of the vehicle suspension and the rapid response of magnetorheological fluid provide basis of the realization of realtime time control using neural network. The results of computer simulation and experiments in time domain and frequency domain demonstrate the advantages of the indirect adaptive control of neural networks over the passive and the direct control strategy.
出处
《振动工程学报》
EI
CSCD
北大核心
2002年第3期285-289,共5页
Journal of Vibration Engineering
基金
国家杰出青年科学基金资助项目 (编号 :5 96 2 5 5 11)
福特 -中国研究与发展基金资助项目 (编号 :9715 5 0 8)
关键词
磁流变阻尼器
车辆悬架
主动控制
自适应控制
实验
振动控制
adaptive control
neural networks
magnetorheological damper
vehicle chassis
vibration control
