论文部分内容阅读
在线性框架下研究基于输入输出观测数据对未知系统模型中两反馈控制器的设计问题.采用虚拟参考反馈校正控制直接设计控制器,无需系统建模过程.在构造代价函数时,同时考虑闭环传函和灵敏度的模型-匹配问题.针对模型匹配过程,分别设计虚拟输入和扰动;推导用于观测数据预处理的滤波器.在设计控制器前,利用概率统计法对控制器进行检验以保证闭环系统的稳定性.针对控制设计中的参数优化,利用数值优化理论分析迭代最小二乘辨识算法所生成的迭代序列与全局最优值间的逼近式,以此式作为评价收敛性和收敛速度的依据.最后用仿真算例验证方法的有效性.
Under the linear framework, the design of two feedback controllers in the unknown system model is studied based on the input and output observation data, and the virtual reference feedback correction control is used to design the controller directly, without the need of the system modeling process.When constructing the cost function, And the sensitivity of the model - matching problem.For the model matching process, respectively, the design of virtual input and disturbance; deduced for the observation data pretreatment filter.In the design of the controller, the use of probability and statistics to test the controller to ensure that the closed-loop System stability.Aiming at the parameter optimization in control design, the numerical approximation theory is used to analyze the approximation between the iterative sequence and the global optimal value generated by the iterative least-squares identification algorithm, which is used to evaluate the convergence and convergence speed Finally, the simulation example is used to verify the validity of the method.