摘要
针对一类未知非线性时变系统,本文提出一种不同次迭代运行过程中期望轨迹可变的迭代学习控制算法。该算法利用高斯径向基网络逼近系统逆的未知参数,并采用迭代学习的方式修正网络逼近的系数,然后结合变结构技术设计控制律。收敛性分析表明,随着迭代次数的增加,逼近系数与最佳系数的差异逐渐减小。最后,在机械臂上的仿真验证了算法的有效性。
In this paper, for a class of unknown nonlinear time-varying systems, an iterative learning control method based on Gaussian networks is presented to track the varying trajectories in different iterative running processes. The method employs Gaussian networks to compensate the nonlinearities of inverse plant,the weight coefficients are adjusted by learning and the control law is synthesized according to variable structure control strategy. It is shown that the differences between actual and ideal weight coefficients are monotonically decreasing with the iteration for non-identical trajectories. A simulation example on the manipulator shows the algorithm's effect.
出处
《科技通报》
北大核心
2010年第1期120-124,共5页
Bulletin of Science and Technology
基金
国家自然科学基金资助项目(60574020)
浙江省自然科学基金资助项目(Y1090212)
关键词
迭代学习控制
逆系统
高斯径向基
可变轨迹
非线性时变系统
iterative learning control, inverse system, Gaussian radial basis, non-identical trajectories, nonlinear uncertain system