摘要
针对基体位置及姿态均不受控的自由漂浮柔性空间机器人轨迹跟踪问题,提出了一种前馈多层感知器(MLP)神经网络控制策略.建立了末端柔性的自由漂浮基机器人的耦合动力学模型,再利用MLP神经网络良好的逼近能力来自适应补偿非线性柔性臂的逆动力学模型,其误差代价函数由PID控制器提供,权重及阀值的调整采用改进的BP反传算法.最后通过仿真比较详细分析了所提方案的工作机理及对非线性强耦合系统控制的有效性.
The problem of trajectory tracking for free-floating space robot with flexible manipulators are considered.An improved multilayer perceptual neural network（MLPNN） inverse-model control algorithm based on the BP algorithm is proposed in this paper.High order liberation modal is ignored based on the assumed modal method,Lagrange principle and momentum conservation.The MLPNN based controller is used to adaptively learn and compensate the inverse-dynamics model of robot,neural network parameters can be adaptively adjusted on line,and the improved BP algorithm is adopted to learn rules.Error cost function is provided by PID controller.The controller improves the control accuracy and the asymptotic convergence of tracking error.The simulation results show that the presented controller has important value.
出处
《空间控制技术与应用》
2011年第1期59-62,共4页
Aerospace Control and Application
基金
中国航天科技联合创新基金(CAST-HIT09C01)资助项目
关键词
多层感知器神经网络
逆模控制
PID控制
BP算法
multilayer perceptual neural network
inverse-modal control
PID control
BP algorithm
