神经网络技术在主蒸汽温度控制系统前馈中的应用

为了解决电厂主蒸汽温度系统的大迟延问题，提出了一种带神经网络前馈控制器的主蒸汽温度串级控制系统.通过对电厂主蒸汽温度的大迟延特性导致的控制难题的详细分析，提出合理设计的前馈控制作用能弥补反馈控制作用的不足,可以有效地解决大迟延系统的控制问题.文中根据电厂熟练运行操作员的实际操作经验和数据，在常规串级PID控制系统的基础上，增加设计了基于神经网络技术的前馈控制器.仿真结果表明，在机组发生负荷扰动时，主蒸汽温度的最大动态偏差降低了13.8％，调节时间缩短了17％，其调节品质明显改善.

Application of neural network technology on feed-forward control of main-steam temperature

For the long time-delayed characteristic of main-steam temperature object in power plant,a mainsteam temperature cascade control system with a feed-forward controller was proposed. By detailedly analyzing the control problem caused by the time-delayed characteristic of main-steam temperature, the fact that a well-designed feed-forward effect can supply the want of the feedback effect and solve the control problem of long time-delayed system was put forward. According to the experiences and data of the skilled operators, a feed-forward controller which is based on the neural network technology was added on the conventional cascade PID control system. The simulating result showed that when the unit load changes, the maximum dynamic difference of main-steam temperature drops 13.8%,the regulating time shortens 17%, and the control character is improved well.