基于支持向量机的低温受热面积灰监测研究  

Monitoring of ash deposition on low temperature heating surface based on support vector machine

作　　者：赵勇纲 白杨 陈裕辉 肖海平[2] 康志忠[2] 孙保民[2] 茹宇 ZHAO Yonggang;BAI Yang;CHEN Yuhui;XIAO Haiping;KANG Zhizhong;SUN Baomin;RU Yu(Shenhua Guoshen Group Technology Research Institute,Xi'an 710065,China;;School of Energy,Power and Mechanical Engineering,North China Electric Power University,Beijing 102206,China)

机构地区：[1]神华国神集团技术研究院,陕西西安710065 [2]华北电力大学能源动力与机械工程学院,北京102206

出　　处：《热力发电》2018年第11期77-81,共5页Thermal Power Generation

基　　金：国家科技支撑计划项目(2015BAA04B02)~~

摘　　要：针对燃用准东煤易造成锅炉结渣以及尾部受热面沾污严重的特性,通过支持向量机与遗传算法相结合建立一种预测低温过热器受热面清洁换热量的模型,进一步计算出清洁因子来判断受热面积灰程度,从而指导运行人员进行吹灰操作,相比于传统的按时吹灰更具有可靠性和经济性。本文以某330 MW机组为例,通过DCS数据对低温过热器受热面的清洁换热量进行训练预测,测试集的平均相对误差仅为1.35%,并根据历史数据得出了清洁曲线图,与实际情况较吻合,这表明该预测模型在一定程度上能较好地反映低温受热面的污染情况,可为实现大型电站锅炉积灰在线监测提供有力依据。Aiming at solving the problem of severe slagging and fouling during Zhundong coal burning,a model was established to predict the clean heat-absorption capacity of low temperature superheater,by using support vector machine(SVM)and genetic algorithm(GA).Moreover,the value of clean coefficient was calculated to judge the ash fouling degree for heating surface,thus to instruct the operators to carry out ash operation.This is more reliable and economical than the conventional blow by time.Taking a 330 MW unit as an example,the model uses the data of DCS system to practice and predict the clean absorption heat of low temperature superheater,and the average relative error of the test set is only 1.35%.Furthermore,according to the historical data,the clean curve is obtained,which is consistent with the actual situation.The results show that the prediction model can reflect the pollution of low temperature heating surface well to some extent,thus it provides a reference for developing online monitoring systems of fouling and slagging in large-scale utility boilers.

关 键 词：锅炉 准东煤 低温受热面 积灰 支持向量机 遗传算法 清洁因子 预测模型 

分 类 号：TK313[动力工程及工程热物理—热能工程]