基于LS-SVM的再热汽温趋势预测与控制优化

Forecast and Control Optimization of Reheat Steam Temperature Based on LS-SVM

提出了一种基于LS-SVM算法的再热汽温控制优化方法。在烟气挡板调节方式下,再热系统抗干扰能力差,易于波动。LS-SVM算法以执行器的动作趋势为输入样本,预测再热汽温的未来走向。建模过程利用网格法搜索最优核参数,参数评价标准根据交叉验证法确定。根据预测结果,得到未来时刻的再热汽温波动偏差。经过折算得到再热汽温控制系统的前馈控制量,从而减小外界扰动带来的再热汽温波动。经仿真结果显示,上述方法有效地优化了再热汽温系统的控制效果。

This paper presents a method of optimization of reheat steam temperature control based on LS-SVM.In the flue gas damper control mode,reheat system anti-interference ability is poor,which is easy to be fluctuant.To predict the future of reheat steam temperature,LS- SVM algorithm takes the actuator movement trend as the input samples.Grid method is used to search the optimal kernel parameters of mode ling process.Parameter evaluation criteria are determined based on the method of cross validation.According to the prediction results,we can get the deviation of reheat steam temperature fluctuations.The feedforward quantity of reheat steam temperature control system is obtained through conversion,so as to reduce the fluctuations of outside disturbance.The simulation results show that this method can effectively optimize the control effect of reheat steam temperature system.