基于支持向量机的电站锅炉煤灰熔点预测  

Prediction on Coal Ash Fusion Point in Power Station Boiler Based on Support Vector Machine

作　　者：潘文静[1] 

机构地区：[1]国核电力规划设计研究院,北京100095

出　　处：《广东电力》2015年第7期6-9,32,共5页Guangdong Electric Power

摘　　要：为解决结渣问题对电站锅炉高效稳定运行的影响,探讨了一种煤灰熔点预测模型。首先分析锅炉混煤燃烧时的煤灰结渣特性,介绍了支持向量机(support vector machine,SVM)回归方法,将煤灰中的8种氧化物成分作为输入量,以煤灰熔点作为输出量,建立煤灰熔点的SVM回归预测模型,并采用遗传算法对模型参数进行寻优。经过仿真实验,将模型的预测结果与广义回归神经网络模型的预测结果进行比较,结果表明本预测模型预测精度高、泛化能力强,有助解决电站锅炉的动力配煤技术中的结渣问题。In order to solve the problem of affect on high effective stable operation of power station boiler by slagging,a kind of prediction model for coal ash fusion point was discussed.Firstly,characteristic of coal ash slagging of the boiler at the time of blended coal combustion was analyzed,support vector machine（SVM）regression method was introduced which took eight oxide ingredients in coal ash as inputs,coal ash fusion point as output and built SVM regression prediction model for coal ash fusion point.Meanwhile,genetic algorithm（GA）was used for optimizing model parameters.By simulation experiment,comparison of prediction result of the model and that of generalized regression neural network model was made.Results indicate that prediction precision of this prediction model is high and generalization ability is good which could solve slagging problem in dynamic coal-blending technology for power station boiler.

关 键 词：电站锅炉 配煤技术 支持向量机 遗传算法 灰熔点预测 

分 类 号：TK227.1[动力工程及工程热物理—动力机械及工程]