Forecast Of Floor Water Bursting Based On Information Fusion

Based on reading a wide range of relevant references at home and abroad,summed up the techniques and theories of mine floor brusting, researching actualities of information fusion and significance in floor water bursting forecast that use information Fusion, review the commonly used method of principle, the scope of application and the actual value and prospects of information fusion. Based on the theory of BP neural networks and mine pressure control,through summing up forecasting methods and theoretical of depth of destroyed floor, find out the main factors to impact depth of destroyed floor, built forecast model by BP neural networks that use Matlab software. Forecast model is used in 8812 and 9604 workface and got results .Find out got results more closer to reality than calculate results through comparing them. Through introduced basic concepts of D-S evidence theory, intergration and basic decision-making process of information fusion, analysed the main factors which is quantified to forecast floor water brusting. Combined with appilication example in Feicheng caolbed, established identification framework of system and evidence using BP neural networks, ELMAN neural networks and RBF neural networks which simulate in Matlab. Got evidence by three neural networks and give by experts. Take 8812 workface in Caozhuang mine for example, information can effectively increase the credibility of water burst probability and reducing the probability of water burst uncertainty after several integration. Based on neural, expert ananlysis and D-S evidence theory,development integration software of floor water brust using VB and Matlab platform. Got the probability of floor water brusting through preparation of applications.