Application Of Information Fusion Technology In Prediction Of Mine Floor Water Inrush

Coal mine water disasters frequently happen, it hidden trouble to the safety of coalmine workers at the same time it brings great threat to the economic benefit of coal minedue to water disasters. Among them, mine floor water inrush caused the most seriousthreat. Because of our country coal mine special hydrological and geological condition,the influence factors of floor water inrush are multiple and their relationship is complex.Many traditional mine floor water inrush forecast method had not been able to achievean ideal prediction effect in practice.According to the prediction technology was not perfect; a kind of multi-sourceinformation fusion technology applied to mine floor water inrush state prediction wasput forward in this paper. First of all, the research status of the floor water inrushforecast and the information fusion technology was introduced; secondly themulti-source information fusion principle, structure and fusion algorithm were discussed;Then information fusion method based on RBF neural network had been mainly studied,it collected the data affecting the floor water irruption from multiple sensors，thenprocessed as input of RBF neural network, through the analysis of choice of the quantumparticle swarm algorithm to optimize the network parameters, feature level informationfusion of floor water irruption prediction method based on RBF neural network wasestablished. Single feature level information fusion resulted in certain instability, inorder to improve the reliability, the D-S evidence theory decision fusion was introduced,every output result of subset the neural network was normalized as the basic probabilityassignment function of state of evidence theory, then D-S evidence theory informationfused each evidence, the final decision result was obtained.In this paper, the information fusion technology was introduced into the floor waterinrush prediction, the double information fusion algorithm was used and a generalframework was established. By the analysis of experiment, the results calculated bydouble information fusion method had high accuracy and low uncertainty, this methodwill have very good application prospect in the mine floor forecast area.