Study On Calculations For Structural Parameters Of Coal And The Indexes Of Power Blended Coal

This thesis consists of two parts mainly. The first, the calculation of the structuralparameters for coal and its applicability for coal classification, and the second, the calculationmethod of quality indexes of power blended coal.Multiple indexes are included in Chinese coal classification standard <Chineseclassification of coals>, which makes coal classification more clear, but the amount of indexesalso make it not direct and difficult to guide the practical application of coal. Based on themethod of the structural-chemical classification from relevant literature, in this thesis, thestructural parameters have been calculated for the applicability of Chinese coals classification.According to the contents of five kinds of elements: carbon, hydrogen, oxygen, nitrogen andsulfur in coal, we calculated some structural parameters of coal, then used them to classifyChinese coals.In this thesis, five structural parameters were researched, such as the degree ofunsaturation of the structure δ, the extent of reduction of the carbon backbone of the organicstructure B, aromatization–carbon ratio f_a, modified structural parameter η’, atomic ratio ofhydrogen and oxygen H/O.The results showed: for data of Chinese coal, three subclasses of anthracite were welldivided by δ and B; f_a and η’ were also working well on the division of conventional steamcoals; f_a and H/O could divide varieties of conventional coking coals to some extent.At present, the application of power coal blending technology has become popular, at thesame time, there are more and more studies on indexes of blended coal.In this thesis, the relationship between indexes of single coal and ones of blended coalwere investigated with BP neural network and Elman neural network, and a model wasestablished, in which three single coals’ proportion, moisture, volatile component, calorificvalue, ash content were input, and blended coal’s volatile component, calorific value, ashcontent were output, together and separately. The results showed: no matter BP neuralnetwork or Elman neural network, the optimal result of the prediction of separated index of blended coal was better than that of all indexes of blended coal.Making normalized processing to data is good for the results from BP neural network,but not for ones from Elman neural network; in Elman neural network there are more hiddenlayer nodes than in BP neural network when optimal value of R were got; in all calculations,the best optimal prediction results were calculated from normalized BP neural network, thatthe values of the average relative error R of volatile component, calorific value, ash contentwere8.98%,4.21%,8.01%, respectively, when the number of the model’s hidden layer nodeswere9,5,6, respectively.