基于专家经验的神经元PID控制主汽温系统  

Single Neuron PID Control Based on Expert Rules for Main Steam Temperature Control System

作　　者：施爱平[1] 施健[1] 施允洋[1] 

机构地区：[1]江苏大学电气信息工程学院,江苏镇江212013

出　　处：《计算机仿真》2013年第11期80-83,共4页Computer Simulation

基　　金：苏州市科技科技支撑计划(工业)项目(SG201176);扬中市科技科技支撑计划(工业)项目(VE2012016)

摘　　要：研究火电厂锅炉主蒸汽的温度控制优化问题。针对主汽温对象的大惯性、大迟延,受到的扰动因素较多,以及汽温模型的不确定性,按照的典型工况整定的固定参数PID控制难以适应负荷变化,不能获得满意的控制品质,存在实时性差和稳定性低的问题。为提高性能,提出一种基于专家经验的神经元PID的主汽温控制方法。通过神经元在线整定PID参数,采用专家控制规则调节神经元的增益系数。不仅具有神经网络的自学习能力,还保持了PID串级控制的优点,同时改善了神经元的响应快速性,提高了系统的自适应性。仿真研究结果表明采用基于专家经验的神经元PID控制器的主汽温系统的控制品质优于常规的PID控制和单神经元控制,具有良好的动态性能,在不同工况下仍保持强鲁棒性,可以为火电厂锅炉主汽温温度的优化控制提供参考。The main steam temperature object in fossil-fired power station has the characteristics of large delay, large inertia, many disturbances and the uncertainty of its model. The traditional PID controller, which is tuned at typical point, can hardly satisfy the control quality requirements at different unit loads, which results in poor adaptive ability and stability. Thus, this paper proposed a control method based on single neuron PID control with expert algo- rithm. The PID parameters were tuned by the single neuron, and the proportional coefficient of single neuron was reg- ulated by the expert learning rules. The proposed control algorithm not only has self-learning ability and the advan- tage of conventional PID cascade control, but also improves the adaptive ability and rapidity of response. Simulation results show that main steam temperature control system using single neuron PID controller based on expert rules can reach better control quality than that of traditional PID and single neuron PID controller, which displays a satisfactory dynamic performance, and strong robustness in different conditions.

关 键 词：主汽温 神经元 专家规则 仿真 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]