Abstract: In order to solve the present problems existed in the coal-rock interface recognition by labor, an automatic recognition technology of the coal-rock interfa ce was studied. According to the differences of the coal and rock dielectric characteristics, the Bruggeman effective medium theory was applied to establish the coal-roc k boundary model based on the coal content detection. Based on the coal-rock mixed medium powder as an example, a test was conducted for the verification. Witha s imulation on the real-time varied coal-rock mixed medium of the coal content, the relative dielectric constants of the coal-rock mixed medium were real timely obtained. The model was applied to the calculation on the coal content of the mixture and a comparison to the threshold value was conducted. Thus the cutting status of the coal shearer was judged. The test results showed that within the test frequency of 0.01 ~ 30 MHz, the coal content of the coal-rock mixed medium all could pass the obtaine d value calculated with the method. Within the frequency of 0.5 ~ 30 MHz, there would be a small relative error between the theoretical coal content and the actual coal content and could be controlled below 2.60% in average. The calculation results of the model all could accurately recognize the coal-rock interface and could provide a new coal-rock recognition method to realize a minerless mining in the underground mine.