Aiming at the problems of complicated parameter tuning and poor self-adaptability of PID cascade control system for main steam temperature in power plant,an improved neural network PID cascade control method is proposed.In order to re-duce the uncertainty of the main steam temperature control system,the main neural network PID controller in the cascade con-trol is trained based on the minimum error entropy(MEE)criterion,and the entropy of the tracking error is estimated recur-sively by using the rolling time domain window method to improve the operation efficiency of the algorithm.The main steam temperature error sequence and some measurable disturbances are sent to the input layer of the neural network PID controller to achieve the integration of feedback control and feedforward control and improve the anti-interference ability of the control sys-tem.Compared with the neural network PID controller using the minimum sum of square error(MSE)criterion,the neural PID controller using MEE can reduce the fluctuation of superheated steam temperature and reduce the randomness of the con-trol system.
关键词
主蒸汽温度/神经网络PID/最小误差熵准则/最小误差平方和准则/不确定性/抗干扰
Key words
main steam temperature/neural network PID/minimum error entropy criterion/minimum sum of square error cri-terion/uncertainty/anti-interference
维普
万方数据