一种变结构单神经元--Smith预估控制在再热汽温中的仿真研究

本文针对300MW机组喷燃器倾角-再热汽温被控对象,具有大迟延和参数时变,而难于控制的特点,提出了一种基于变结构的单神经元-Smith预估控制算法,该控制算法采用Smith预估控制克服被控对象的大迟延特性,采用变结构的单神经元PID控制提高对被控对象参数变化的自适应能力,提高控制系统的响应速度,并抑制被控参数严重超调.仿真试验表明该控制算法具有良好的控制品质,抗干扰能力和鲁棒性,综合性能明显优于常规的PID控制.

Simulation of a kind of variable structure single neuron-Smith predictor control in reheat steam temperature

Considering the disadvantage in the controlled object of reheat stream temperature, adjusted by changing the obliquity of fuel nozzle, with large time delay and varying parameters,the paper presents a kind of variable structure single neuron-Smith predictor in which Smith predictor is designed to minimize the large time delay .Variable structure neural network PID controller in the algorithm not only can improve the self adaptability and increase the response speed, but also may restrain overshoot. The simulation results show the controller processes excellent control quality, anti-disturbance and robustness. Its overall control performance is better than the general PID controller.