支持向量机灰熔点预测模型研究  

Study of a Support Vector Machine-based Model for Predicting Melting Points of Ash

作　　者：赵显桥[1] 吴胜杰 何国亮 王春林[3] 

机构地区：[1]山东电力研究院,山东济南250002 [2]山东沾化热电有限公司,山东沾化256800 [3]杭州电子科技大学,浙江杭州310018

出　　处：《热能动力工程》2011年第4期436-439,494-495,共4页Journal of Engineering for Thermal Energy and Power

基　　金：国家自然科学基金资助项目(60904058)

摘　　要：根据电厂对混煤灰熔点计算的需求,采用支持向量机算法和BP神经网络算法对灰熔点进行了建模和对比研究,灰熔点模型采用灰成分作为输入量,灰熔点(ST)作为输出量,用所建模型对单煤和混煤灰熔点进行预测,然后将预测结果与实验结果进行比较。支持向量机模型对单煤和混煤的预测误差分别为0.57%和1.94%,BP神经网络模型对单煤和混煤的预测误差分别为1.925%和10.43%,结果表明,支持向量机模型对单煤和混煤灰熔点的预测更精确。To the demand of ash melting point calculation of blended coal in power plants,a model for melting points of ash was established and a contrast study was conducted by using the support vector machine algorithm and BP（back propagation） neural network algorithm.The model in question used the ash composition as a input and the melting point of ash as an output.It was employed to predict the ash melting points of a single coal and blended one.Then,the prediction results were compared with the test ones.The errors of the model based on the support vector machine were 0.57% and 1.94 % respectively in predicting the single coal and blended one while those of the model based on the BP neural network were 1.925% and 10.43% respectively in predicting the above-mentioned two types of coal.The research results show that the model based on the support vector machine is more precise when predicting the ash melting points of a single coal and blended one.

关 键 词：灰熔点 支持向量机 BP神经网络 预测 

分 类 号：TQ533[化学工程—煤化学工程] TP273[自动化与计算机技术—检测技术与自动化装置]