基于PID神经网络的主汽温系统设计与仿真

Development and Simulation of Main Steam Temperature System Based on PID Neural Networks

PID神经网络是将PID控制规律融入神经网络的一种智能控制技术,它既具有常规PID控制结构简单、参数物理意义明确的优点,又具有神经网络非线性映射能力强、自学习、自适应的功能。针对主汽温控制系统大滞后、非线性、时变等特性,综合考虑控制系统的动态、静态性能,利用PID神经网络对控制系统进行全工况的学习与优化设计,并进行了主汽温全程控制仿真研究。仿真结果表明该方法具有较强的工况适应能力与抗干扰能力。

The PID neural networks （PIDNN） is a intellective-control technology by combining PID control rule with neural network. It has the advantages of simple structure, clear physical meanings of parameters, as well as the functions of nonlinear reflection ability and self-leaming and self-adaptive of the neural networks. Focused on the dynamic and static performances of a nonlinear time-variant system with large delay-time, the PIDNN controller was used to make full load range learning and optimizing for main steam temperature control. The simulation results demonstrate that the strategy has strong load adaptability and disturbance rejection ability.