煤元素组成特性的数据挖掘

Data mining for information of coal elements constitution

随机收集整理了693个典型的中国煤的数据样本，它们来自不同的矿区，干燥无灰基挥发分变化范围Vdaf=1．71％-61．8％．分别采用BP人工神经网络和多维变量统计二次型，对样本煤的干燥无灰基元素含量的信息进行了数据挖掘，结果表明，碳、氢的含量Hdaf，Odaf可以根据其它成分来预测，挥发分Kdaf，含量也可以用Cdaf，Hdaf，Ndaf的含量表示，从而得到了适用于我国各种类型煤的元素组成通用规律．基于样本统计分析，并给出神经网络和多维变量统计二次型等模型的相对误差分布函数．本研究成果可以应用于电厂锅炉入炉煤元素成分的软测量在线监测．

693 coal samples coming from different mine, and of which, volatile matter content in dry and ash free basis vary from 1.71% to 61.8% , were settled. BP neuron network and quadratic function of multiple variables statistical model were used in data mining for information of coal elements constitution, and conclusions show that hydrogen, oxygen as well as volatile matter content can be forecasted with different combination of carbon, nitrogen and sulfur content in dry and ash free basis, and universal rules of governing elements constitution for various kinds of Chinese coal were obtained. Relative error distribution functions of BP neuron network and quadratic function of multiple variables statistical model were provided based on statistical analysis. The results of this paper are for using in real time identification of coal ultimate analysis in operation boiler in power plant.