Application of Hilbert-Huang Transform and SVM to Coal Gangue Interface Detection

In order to detect coal gangue interface on fully mechanized mining face, a new method of vibration signal analysis of coal and gangue based on Hilbert-Huang transform is presented in this paper. At first Empirical mode decomposition algorithm was used to decompose the original vibration signal of coal and gangue into intrinsic modes for further extract meaningful information contained in response signals under complicated environment. By analyzing local Hilbert marginal spectrum and local energy spectrum of the first four intrinsic mode function components, we found the difference of coal and gangue at specific frequency interval that the amplitude and energy mainly distributed at frequency interval between 100Hz and 600Hz when coal fell down, while the amplitude and energy were more concentrated at 1000Hz or so when gangue fell down. Furthermore, the further analysis result from marginal spectrum of each intrinsic mode function component agreed well with the conclusion above. Combined with time-domain parameters, we defined the energy function based on the above feature as inputs of support vector machine for simulation experiment. The results show that the extracted features with the proposed approach can be served as coal gangue interface recognition. Index Terms fully mechanized mining face , vibration signal , coal gangue Interface detection, Hilbert-Huang transform, empirical mode decomposition, support vector machine