Recognition of coal and gangue based on dielectric characteristics and geometric constraints under multi factors

Water and coal slime are interference factors in actual production, which adversely affect the feature extraction, feature credibility, and recognition accuracy of coal and coal gangue. In this paper, from two ways of simulation and experiment, combined with statistical methods, it is concluded that the influence of water content and slime cover on the dielectric and geometric characteristics of coal and gangue can be ignored. The Variational Mode Decomposition (VMD) denoising method is improved. After the response signal is processed by VMD- Kalman denoising method, the signal-to-noise ratio is as high as 26.2232, and the root means the square error is as low as 0.0019587. The dielectric and geometric features of coal and coal gangue under the action of water and slime are extracted to form feature vectors, and the reliability of feature vectors is tested by using classification models such as back propagation neural network, Bayesian, proximity algorithm, and support vector machine, all of which achieve recognition accuracy of over 92.86%. The research shows that the identification method of coal and gangue based on dielectric characteristics and geometric constraints can adapt to the working conditions with multiple factors.