Research On Impact Characteristics And Identification Method Of Coal Gangue In Fully Mechanized Caving Face

Accurate identification of coal gangue is an important foundation for building a green,safe and efficient fully mechanized caving face,and it is also an important way to realize intelligent top coal caving.Simulating the coal caving experience of coal caving workers "listening and seeing",monitoring the mixed state of coal gangue during the top coal caving process is the most effective way to achieve accurate identification and control the opening and closing of the coal caving window.The difference in the physical properties of coal and gangue will lead to different impact characteristics of coal gangue mixtures with different mixing degrees.Based on the impact characteristics of coal gangue,this thesis takes the coal caving process of the fully mechanized caving face as the research object,and the coal caving process identification is the research direction.Numerical simulation and verification experiments of single unit,simulating the impact of mixed gangue,built a mixed gangue drop experiment platform,collected the mixed gangue shock signal through the data acquisition system,and used two types of machine learning methods to conduct multi-class classification research on the mixed gangue shock signal.,and finally achieve the research goal of coal caving process identification based on the impact characteristics of coal gangue.The main work and achievements are as follows:(1)On the basis of the previous investigation of Tashan 8222 fully mechanized caving face and the understanding of coal caving process,the relationship between coal gangue impact characteristics and coal caving process identification was analyzed and explained,and a coal caving process identification method based on coal gangue impact characteristics was proposed.Considering the collapsed top coal and hydraulic support as a complete impact system,the process of top coal caving in fully mechanized caving face is that the tail beam of the hydraulic support is subjected to external "excitation" to produce an impact response,and when the response threshold is reached,it is "notified".The bracket controller closes the coal draw window.First,a threedimensional simulation model of the material single body impacting the tail beam was established by the numerical simulation software ABAQUS,and the acceleration response law of different material single body falling from different heights was obtained.Secondly,an experimental platform for different material monomers to be dropped from different heights was built.The data acquisition system was used to collect the impact acceleration signals of the material monomers.The results of the numerical simulation were further verified through analysis and research.The acceleration response generated by the material monomer impact Its variation law is consistent with the simulation results.(2)On the basis of the acceleration response and its variation law generated by the impact of the single material,and considering the conditions of the field experiment,the laboratory has built a mixed gangue impact experiment platform.The top coal caving process of the fully mechanized caving face is modeled by parametric model,and the mixed gangue ratio parameter λ is introduced to represent the mixing degree of coal and gangue in different coal caving stages.The laboratory selects the accelerometer and data acquisition instrument suitable for the scene,and uses this data acquisition system to collect the mixed gangue impact signals of different λ values.At the same time,for the collected original mixed gangue shock signal,the trend term of the original signal was eliminated by the least square method,and the signal-to-noise ratio of the original shock signal was improved by the method of wavelet threshold noise reduction.Through effective preprocessing of the original mixed gangue shock signals with different λ values,a more accurate and reliable shock signal data source is obtained,which provides data conditions for the identification of coal caving process based on shock characteristics.(3)On the basis of the preprocessing of the original mixed gangue shock signal,the mixed gangue shock signal was sampled with 12800 sampling points as the length,and the signal samples were divided into training sample set and test sample set at 7:3.A classification model of mixed gangue impact signal based on convolutional neural network and a mixed gangue impact signal classification model based on multi-class support vector machine are proposed.The latter is more suitable for a small number of signal samples,showing fast convergence speed and high classification accuracy,and the two types of models have their own applicability.The sampled signal sample set is used to train the model,and by means of model parameter optimization and other methods,after the test signal sample set test,the classification accuracy of the two types of models for mixed gangue impact signals with different λ values has reached 90.63%and 99.17%,respectively,the identification of different coal caving stages of the fully mechanized caving face is completed,and then the decision to open and close the coal caving opening can be made to realize the automatic coal caving in the true sense.