热电厂主蒸汽温度控制策略仿真研究

Simulation on Main Steam Temperature Control Strategy in Thermal Power Plants

在火力发电锅炉温度优化控制问题的研究中,针对热电厂主蒸汽温度控制时存在实时性差的缺点,提出采用遗传算法与反向传播神经网络结合的PID主蒸汽温度控制策略,通过遗传算法优化反向传播神经网络的权值和阈值,使得神经网络能根据辨识提供的学习信号在线调整PID控制器使控制系统具有较好的自学习和自适应能力,通过在MATLAB上进行仿真证明:控制策略的三个参数在响应速度、最大误差和对变化的敏感程度上均优于传统的PID控制系统。采用GA-BP神经网络的PID控制系统能够保证良好的动、静态特性,是一种适合热电厂主蒸汽温度控制的有效的控制策略。

This paper presents a PID main steam temperature control strategy based on GA-BP neural net, and optimizes the weights and thresholds of BP neural network with genetic algorithm. It makes the neural network output for the parameters of PID controller, and it bases on the identification signal provided by the study to adjust the three parameters of PID controller on-line. Through that, the control system has good self-learning and adaptive capacity. By comparing the response speed, maximum error and reaction speed to change of the three parameters, the simula- tion shows that the control strategy is much more efficient than traditional PID control. The conclusion is that the PID control system based on GA-BP neural net can ensure good dynamic and static characteristics, and it is an effective control strategy which is suitable for main steam temperature control in thermal power plants.