基于神经网络的超临界机组过热汽温设定值优化补偿  

Optimized Compensation for Set Values of Superheated Steam Temperature in Supercritical Unit Based on Neural Network

作　　者：马良玉[1] 陈婷婷 刘文杰 李倩倩 MA Liangyu;CHEN Tingting;LIU Wenjie;LI Qianqian(School of Control and Computer Enginee ring,North China Electric Power University,Baoding 071003,Hebei Province,China)

机构地区：[1]华北电力大学控制与计算机工程学院,河北保定071003

出　　处：《动力工程学报》2021年第1期36-42,共7页Journal of Chinese Society of Power Engineering

基　　金：国家自然科学基金资助项目(61174111)。

摘　　要：基于历史运行数据对过热器喷水减温系统特性进行神经网络建模,在不改变机组原过热汽温控制逻辑和PID参数的前提下,采用基于预测模型的前馈补偿和反馈补偿相结合的策略,在控制回路的顶层对过热汽温设定值进行实时优化补偿,以改善过热汽温控制效果,并借助600 MW超临界机组仿真机进行仿真试验研究。结果表明:采用设定值优化补偿方案可明显提升过热汽温的控制品质,验证了优化方案的有效性。Neural network models were developed for the characteristics of a superheated steam temperature(SST) system based on historical data. On the premise of not changing the original SST control logic and PID parameters, a real-time optimization strategy was proposed for the set values on the top level of control loop to improve the SST control effect, which combines feedback compensation with feedforward compensation based on ANN prediction model. Simulation experiments were carried out with the help of a full-scope simulator for a 600 MW supercritical power unit. Results show that the SST control quality can be obviously improved by using the optimized compensation scheme for set values, which is proved to be effective.

关 键 词：超临界机组 过热汽温 神经网络建模 设定值优化补偿 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]