The Coal Mine Flood Prediction Research Based on Neural Network and D-S Theory of Evidence

To resolve the problem of underground security status misjudgment and the low accuracy of coal mine flood forecast, this paper presents a coal mine flood forecasting method. Which is based on neural network preliminary judgment and D-S (Dempster-Shafer) theory of evidence decision-making judgment? By associating information obtained from various sensing sources, this method effectively reflects the security status of the coal mine. The results normalized to output from each sensor serve as the basic probability distribution function of evidence theory. The final conclusion is drawn by applying D-S theory of evidence and fusing evidence information. The simulation results from Matlab 7.0 show that the method improves the goal mine state recognition rate significantly, reduces uncertainty and improves the accuracy of judgment of mine safety.