基于状态修正的自学习控制系统的结构设计

The structure design of self-learning control system based on state modification

以往的自学习控制系统都是通过实际输出与理想输出之差来修正控制器参数的,但输出是系统的外部,只部分地反映了系统的本质,只有状态才能全面地反映系统的本质,为此,文中提出了一种基于状态修正的自学习控制思想,并给出了完整的系统结构,对系统中的各个环节进行了详细的描述,在已知系统理想响应的情况下,将被控对象的准状态与模拟系统的理想状态之差作为对控制系统性能评价的依据,并利用性能评价的结果,以优化策略实现对控制器参数的自动调整和学习,由于在设计中充分考虑了自学习的收敛性问题,并且在相当宽的选择范围内控制器参数能保证良好的收敛性,所以该文提出的这种控制系统结构是一种完全可行的结构,最后,还给出了仿真实验的结果.

The difference between the real output and the ideal output is always used traditionally to construct the self-learning control system, but only the states of the system can reflect the whole behavior of the system. So a serf-learning controller based on the modification of system states including its structure and design of every part are proposed in this paper. The difference between the quasi-states of the controlled object and the ideal states of the simulating system is used to evaluate the performance of the control system, and the result of the evaluation encour- ages the learning and modification of controller parameters with an optimal strategy. Because of much concentration on the convergence of self-learning, and very wide parameter selection range that can guarantee the convergence, the structure proposed here is very efficient. The simulation results are presented.