Research On The Information Standardization And Data Mining For Coal Quality And Preparation In Mine Area

With the rapidly development of information technology and the forthcoming of information society, the coal corporations have been paying more and more attention on the information construction which can positively promote work efficiency and management level while fruitful achievements have been attained. However, some shortages and problems that could not make expectant aim and effect of informatization projects achieved have been found in the informatization process. The problems mainly display in two aspects: one is that each information system can't or hardly share data with others, which engenders a great many of'data isolations'; the other is that large amounts of business data have been stored in every information system database but little information is indeed useful to manager, which means that the utilization ratio and level is low.According to the first aspect problem, summarizing the experience in the process of mining area coal quality and coal preparation information system development and generalization, and using information standardization research achievement of other fields for reference, coal quality and preparation information standardization system has been built to make the system configuration, concrete content, adduction content about other related standards and deployment and application method of the standard system clear, which provides reference and basis for related information construction. It's also has been put forward that the standard system should include six standardization branch systems of general standard, application standard, information resources standard, network basic establishment standard, information security standard and management standard. Taking the development of mining area coal quality and coal preparation informatization system for an example, it has been introduced how to carry out the requirements of information standardization system sufficiently.According to the second aspect problem existing in the coal corporation informatization construction, in this dissertation data mining technology has been used to improve data utilization ratio and level of information system, so the technicians and managers could achieve more valuable information and knowledge. Support vector machine mining algorithm has been used to judge the problem that coal preparation production is abnormal.According to separation theory and basing on data mining thought and method, the knowledge mining of the dense-medium separation process parameters on-line prediction has been carried through, while the real-time prediction model of raw coal density composition has been constructed. Using on-line ash to predict the density composition of feed raw coal which have relatively stable origin and quality, theory yield and separation density of clean coal can be predicted by fitting washability curves. Real-time prediction model of production index has been constructed. According to the feed raw coal density composition data predicted, the practical separation density could be predicted. A whole arithmetic about dense-medium separation process parameters on-line prediction has been researched on and data mining method in point has been concluded.Rapid evaluation of dense-medium separation effect has been researched on using data mining thought. Rapid prediction model of raw coal and production density composition should be constructed firstly. Using quick float-and-sink separation yield data to correct the recent monthly comprehensive float-and-sink separation yield data of raw coal, float-and-sink separation yield after calibration approximately represents density composition of feed-cleaning raw coal. By predicting density composition of raw coal and drawing washability curves, theory yield and quantity efficiency of clean coal can be calculated. Constructing dense-medium separation effect rapid evaluating model, using density composition data of raw coal and production to calculate distribution ratio, the amount of misplaced material and possible deviation or imperfect level can be calculated finally, so the dense-medium separation effect can be evaluated.It's brought forward that using PSO to optimize fitting parameters of washability curves and distribution curves, in which way there's no need to input fitting parameters and only parameter values bound needs to be input, and it's quite convenient and easy to realize. Because fitting models in existence have bad fitting effect in some special condition, five new functions which could be used for washability curves and distribution curves have been found from S-shape growth function after amount of data fitting tests. Using the same distribution ratio data to validate the 5 fitting models newly advanced and optimizing parameters of curves fitting by PSO, it's shown that fitting models newly advanced have the better fitting effect than models in existence in some special situation.