Abstract:

Detailed local mean decomposition (Local mean decomposition LMD) non-stationary signal processing methods, LMD algorithm with a high degree of self-adaptive, able to decompose the characteristic parameters of the data itself, it is able to arbitrarily a very complexnon-stationary signals into several PF (Product function, production function), and components with certain physical meaning emission coal Face Gas collected data LMD decomposition, multiple PF component. Then use the improved neural network method were predicted, then synthesis of different prediction results superimposed reconstructed and Gas Emission Prediction value. Actual monitoring data for the gas analysis can draw this method to predict the effect of higher prediction accuracy than the conventional neural network methods, LMD method decomposition PF component has a certain manner regularity, so more substantial increase the prediction accuracy of gas emission, the case study also shows that the predicted results with actual monitoring results extremely high consistency.

Key words: Coalface, Predict, LMD, Neural network, Gas Emission