Research And Application Of Multi-sensor Data Fusion In The Coal Mine Safety Early-warning

The dynamic changeable underground mining environment, diverse and complex the influence factors of natural disasters lead to difficulties in downhole safety early warning. The problem can be resolved by introducing multi-sensor data fusion technology into underground mining environment used for the mine environment security early warning methods.According to different data fusion approach, multi-sensor data fusion model are classified and evaluated at present. On the basis of the evaluation results, general designing principles of the coal downhole natural disasters safety early warning model is put forward.Based on principles above, an early safety warning model is designed. The safety warning model employs the two-tier hierarchical processing fusion structure, respectively in the feature layer and decision making layer to realize data fusion. In the feature layer, the algorithm based on principal component analysis(PCA) and neural network is proposed, which uses principal component analysis to realize feature extraction of a single hazard. Choose the corresponding weight coefficient of different types indicators and use neural network to train sample data. In the decision making layer, hazards safety warning decision making is obtained by building new BPAsO function to improve D-S evidence theory algorithm.The paper uses MATLAB to simulate the early waring model with instance of Liuzhuang coal mine, and compares the early warning results with measured data. Simulation results show that after fusion of the gas single hazard, the safety assessment of the various states has good consistency compared to the measured data, the results conform to the actual situation. The research of early waring model has a certain guiding significance to the actual downhole safety warning.The thesis is supported by the National Natural Science Foundation Project(No:612610001), named "Research on key technology of data aggregation and transmission in WSNs and its application in facility agriculture in semiarid areas".