基于CNN-RNN组合模型的凝汽式汽轮机能耗预测研究  

Research on Energy Consumption Prediction of Condensing Steam Turbines Based on CNN-RNN Combination Model

作　　者：豆重飞 DOU Chongfei(Zhaofeng Aluminum Power Company′s Own Power Plant,Huayang Group,Yangquan,Shanxi 045200,China)

机构地区：[1]华阳集团兆丰铝电公司自备电厂,山西阳泉045200

出　　处：《自动化应用》2023年第23期118-120,123,共4页Automation Application

摘　　要：常规的凝汽式汽轮机能耗预测方法主要采用瞬态/稳态并行分析法生成汽轮机能耗预测模型,该方法易受临界负荷的影响,导致热耗预测偏差较大过高。因此,本文提出基于卷积循环神经网络(CNN-RNN)组合模型的凝汽式汽轮机能耗预测研究,利用R检验法筛选并生成凝汽式汽轮机能耗特征参数工况库,结合CNN-RNN组合模型消除凝汽式汽轮机能耗预测偏差,从而完成凝汽式汽轮机能耗预测。结果表明,设计的基于凝汽式汽轮机能耗CNN-RNN组合模型的凝汽式汽轮机能耗预测方法的热耗预测偏差较小,证明该方法的预测效果较好,具备较高的准确性和一定的应用价值,为优化汽轮机的运行方案提供了一定贡献。The conventional energy consumption prediction me thods for condensing steam turbines mainly use transient/steady-state parallel analysis to generate a turbine energy consumption prediction model.This method is susceptible to the influence of critical loads,resulting in significant and excessive deviation in heat consumption prediction.Therefore,this paper proposes a study on the energy consumption prediction of condensing steam turbines based on the convolutional recurrent neural network(CNN-RNN)combination model.The R-test method is used to screen and generate a working condition library of condensing steam turbine energy consumption characteristic parameters.The CNN-RNN combination model is combined to eliminate the energy consumption prediction bias of condensing steam turbines,thus completing the energy consumption prediction of condensing steam turbines.The results show that the designed CNN-RNN combined model based energy consumption prediction method for condensing steam turbines has a small deviation in heat consumption prediction,which proves that the prediction effect of this method is good,has high accuracy and certain application value,and provides a certain contribution to optimizing the operation plan of steam turbines.

关 键 词：CNN-RNN组合模型 凝汽式 汽轮机 能耗 预测 

分 类 号：TK269[动力工程及工程热物理—动力机械及工程]