Improved ultrasonic differentiation model for structural coal types based on neural network

In order to solve the difficulty of detailed recognition of subdivisions of structural coal types, a differentiation model that combines BP neural network with an ultrasonic reflection method is proposed. Structural coal types are recognized based on a suit-able consideration of ultrasonic speed, an ultrasonic attenuation coefficient, characteristics of ultrasonic transmission and other parameters relating to structural coal types. We have focused on a computational model of ultrasonic speed, attenuation coefficient in coal and differentiation algorithm of structural coal types based on a BP neural network. Experiments demonstrate that the model can distinguish structural coal types effectively. It is important for the improved ultrasonic differentiation model to predict coal and gas outbursts.