神经网络相平面分区控制器研究与仿真

将神经网络的在线学习、即时调节功能与相平面分区的逻辑判断特性相结合提出了一种新的智能控制器———神经网络相平面分区控制器.该控制器将系统响应的基本性能与相应工况的控制作用力的划分和作用强度在相平面上建立起了实时的相互对应关系,从而使各分区内的控制参数能随着被控对象的变化在线自动调节参数,具有自适应能力.仿真结果表明,利用该方法设计的控制器闭环性能好、鲁棒性强.该方法为非线性控制提供了一种有效的控制器设计途径.

Research and simulation of neural network phase plane divisional control

This paper combines neural network with phase plane divisional control to generate a new intelligent controller-neural network phase plane divisional controller, which establishes, in the phase plane, real time corresponding relationships between the basic performance of the system response and not only the action logic division but also acting intensity, making control parameters in each divisional area able to automatically online regulate in accordance with the changes of the controlled object. The simulation result shows that the controller designed with this method has better close-loop performance and robustness. This method provides an effective controller designing way for nonlinear control system.