摘要
利用火力电厂数据仓库中的128065条历史记录,对一台双进双出球磨机的煤粉磨制过程进行数据挖掘建模,所建人工神经网络模型得到设备实际运行状态的验证;采用该模型对煤粉制备过程进行模拟分析,着重研究了料位参数的优化问题。研究结果表明,磨煤机通风量不变时,随着给煤量增加,磨内存煤量增加,磨煤机电流升高,电流上升到峰值后又有所下降,开始下降的区域对应的料位是最佳运行料位,此时,钢球间隙中刚好充满了煤粉颗粒,磨煤机接近最大功率运行,粉碎效率达到最高。给出的煤粉粉碎过程神经网络模型,不仅被大量实际运行数据所验证,而且模型推断与现有研究结论相符,是一个有良好概括性的健壮模型。
128 065 records from the data warehouse in a coal-fired power plants were used for data mining modeling of coal grinding process in a double charge-discharge coal ball mill,the artificial neural network model built to be verified by the actual operation of the ball mill,was employed to simulate the coal grinding process,focusing on the particle filling optimization. The results show that the mill current and particle filling increase with the coal feed flow when ventilation rate remained the same,mill current starts to decline after rising to a peak,the beginning of the decline region corresponds to the optimal particle filling,at this point,the gap between balls are just full of coal particles,the mill is running close to maximum power,and grinding efficiency reaches the highest. The coal grinding neural network model is not only validated by large quantity of running data in a actual mill,but also the model inferences are conformed by existing research findings,and prove to be robust and of high degree of generality.
出处
《煤炭学报》
EI
CAS
CSCD
北大核心
2010年第5期850-854,共5页
Journal of China Coal Society
关键词
数据挖掘
球磨机
粉碎
人工神经网络
物料充填率
data mining
ball mill
comminution
artificial neural network
powder filling
