Abstract: A BP neural network algorithm based hierarchical identification method， which divides the coal seam hardness into six levels， was proposed for identifying the coal seam hardness. The identification signals were taken from the stator currents of both the cutting motor and the traction motor of the mining machine， as well as the pressure signal of the height adjustment cylinder. The wavelet packet decomposition was used for extracting the characteristics of each signal， and these signals were taken as the input signals for training and testing the neural network. The experimental results show that the identification accuracy reaches 96.7% and 93.3% toward the simulation data and the real data， respectively， validating the effectiveness of the method. The method proposed provides the foundation for precisely identification of coal seam hardness.

Key words: drum shearer, cutting impedance, coal seam hardness identification, wavelet packet decomposition, BP neural network, feature vector