基于RBF神经网络辨识的过热蒸汽温度控制

CONTROL OF SUPERHEATED STEAM TEMPERATURE BASED ON RBF NEURAL NETWORK IDENTIFICATION

提出一种基于RBF神经网络辨识的PID串级主蒸汽温度控制策略,即将RBF神经网络与常规PID串级控制相结合构成RBF-PID控制器。该控制器不仅具有常规PID控制器的特性,而且还具有智能控制器的自学习能力,增强了系统对不确定因素的适应性。仿真研究结果表明,RBF-PID控制系统动态调节品质显著优于常规PID串级控制,能适应对象参数的变化,具有较强的鲁棒性和自适应能力。

A PID cascade main steam temperature control strategy based on radial basis function （RBF）neural network identification has been put forward,and a RBF- PID controller being constructed by combining RBF neural network with conventional PID cascade control. The said controller not only has the behavior of conventional PID cascade control, but also boasts the self- study ability of an intellegent controller, strengthening the system＇s adaptability to some uncertain factors. Results of simulation study show that the dynamic performance of RBF- PID control system is substantially superior to that of the convintional PID cascade control, can readily accommodate itself to the variation of object＇s parameters, having stronger robustness and self- adaptability.