Research On Coal-rock Interface Recognition Technology Based On Ground Penetrating Radar

With the deepening reform of the energy structure,the field of coal mining is also moving towards intelligent development.Green,high efficiency and safety are the focus of development research.In the mining process,due to various factors such as the environment,the coal mining machine's determination of the coal-rock interface is mainly identified by manual observation.The ground-penetrating radar has the advantages of fast test speed and high precision,and will be used for ground penetrating radar.The identification of the coal-rock interface and integration into the shearer to achieve unmanned mining is the original intention of this paper.As a preliminary study,this paper studies the ground penetrating radar response of coal-rock interface through forward modeling and physical test methods,and uses neural network to make a simple exploration of the intelligent recognition of the interface.Through the use of ground penetrating radar,three different outdoor ground penetrating radar test methods are used for the relative dielectric constants of the coalrock media of Yuejin Coal Mine in North China and Zhaolou Coal Mine in East China(known target depth method,layered reflector method,The transmission measurement method was tested and the relationship between the water content and the dielectric constant of the coal sample medium was investigated.In this paper,the finite-time difference method is used to simulate the variables affecting the coal-rock interface(interface shape,stalk condition,water-bearing gasbearing fissure),and the response characteristics of ground penetrating radar under various variables are summarized.The effect of coal-rock media combination with different dielectric constants on the detection effect of ground penetrating radar is studied by forward modeling.The detection effects of various coal-rock media combinations are summarized.The forward modeling is established through actual geology,and the butterfly antenna is used.The simulation is carried out,and the identifiable range and treatment measures of the effective signal under the influence of random noise are given.Using the method of stacking coal and rock media,the coal and rock media corresponding to the forward model were simulated by the coal-rock media of the Yuejin Coal Mine in North China and the Zhaolou Coal Mine in East China.The coal rock in the experimental model was tested by PulseEKKO ground penetrating radar.The interface position was measured,and the coal-rock interface response under the measured effect was summarized.The mortar connection model was used to simulate the tight connection of the interface.Through the physical test,the test results of different coal-rock media in the two places were also summarized.In order to make the response information of the coal-rock interface more clear,using conventional processing methods,predictive deconvolution means and "threeinstant" feature extraction and other data processing methods,the experimentally measured data is analyzed and processed,and the response of the effective signal is enhanced.The influence of interference signals such as multiple waves is suppressed.At the end of the paper,through the experimental data,a BP neural network with three-layer structure is established.The characteristic parameters are extracted by single-channel data to form input layer data.The learning and training of multiple iterations through the network can complete the prediction of coal-rock interface.Identification;In addition,the applicability of predictive recognition in various situations under different sample training is summarized by establishing different trainings in both independent and integrated samples.