摘要
为了预测混煤的灰熔点,采用支持向量机建立煤灰软化温度模型,模型将煤的灰成分作为输入量,煤的软化温度作为输出量,利用网格搜索寻优方法对支持向量机(SVM)模型的参数进行了优化,在设定的不同精度下分别获得模型的最优参数,利用优化后的模型对单煤和混煤的灰熔点进行了预测,并将不同精度下的预测结果与实验结果进行对比.煤灰软化温度模型设定精度为0.01时,单煤样本预测相对误差最小,其最大相对误差和平均相对误差分别为3.00%和0.48%;运用此模型对混煤预测的最大相对误差和平均相对误差分别为1.74%和0.62%.预测结果表明,经网格搜索优化后的支持向量机模型对煤灰熔点预测较精确.
Support vector machine （SVM） was used to model the mixed coals＇ ash softening temperature. The coal ash compositions were employed as the inputs of SVM, and the measured ash softening temperatures were used as the outputs. Parameters of the SVM model were optimized by grid search and the best parameters were obtained for different setting precisions. Single and mixed coals~ ash fusion temperatures were predicted by the optimized SVM and the predicted values were compared with the actual measurement results. The relative error of testing single coal was the smallest when the setting precision was 0.01. The results showed that the maximum predicting error was 3.00%, and the average predicting error was 0.48%. Using the 0.01 precision for the mixed coals, the maximum relative error was 1.74%, and the average relative error was 0.62%. So the SVM model optimized by grid search has good predicting performance of ash fusion temperature.
出处
《浙江大学学报（工学版）》
EI
CAS
CSCD
北大核心
2011年第12期2181-2187,共7页
Journal of Zhejiang University：Engineering Science
基金
国家重点基础研究发展计划资助项目(2009CB219802)
浙江省自然科学基金资助项目(R107532)
新世纪优秀人才支持计划资助项目(NCET-07-0761)
全国优秀博士学位论文作者专项资金资助项目(200747)
浙江大学曹光彪高科技发展基金资助项目(2008RC001)
中央高校基本科研业务费专项资金资助项目
关键词
网格搜索
支持向量机
灰熔点
混煤
grid search
SVM
ash fusion temperature
mixed coals
