论文部分内容阅读
针对传统PID神经网络不能实时有效地控制非线性多变量系统的问题,设计了一种新型多变量自适应PID神经网络控制器。该控制器的隐含层带有输出反馈和激活反馈,实现了比例、微分和积分功能。利用一种基于解空间划分的改进粒子群算法对控制器参数进行优化,消除了初始值对控制器准确性的影响,并将控制器应用于并联机构控制中。由仿真结果可知:控制器控制精度高,鲁棒性和自适应性较强。这一研究为并联机构的精准控制和优化设计提供了理论基础。
Aiming at the problem that the traditional PID neural network can not control the nonlinear multivariable system in real time and effectively, a new multivariable adaptive PID neural network controller is designed. The controller’s hidden layer with output feedback and activation feedback, to achieve the proportion, differential and integral functions. The controller parameters are optimized by using an improved particle swarm optimization algorithm based on solution space partitioning, the influence of the initial value on the controller accuracy is eliminated, and the controller is applied to the parallel mechanism control. The simulation results show that the controller has high control accuracy, robustness and self-adaptability. This research provides the theoretical basis for the accurate control and optimization design of the parallel mechanism.