Wireless Channel Modeling And Accurate Prediction In Mine Based On Support Vector Machine

The Mining insustry plays an important role in our country’s economic development, which will not be replaced in a certain time. In mining operations, communication systems provide an important guarantees for personnel safety in the underground, improving the efficiency of production and optimizing its production. However, mine roadway is a specially limited space,in which the condition is complex and the electromagnetic wave cannot propagate well, then the development of underground wireless communication system is relatively slow. This paper provide theoretical and practical significance for the research of underground wireless communication system by analysing on characteristics of underground electromagnetic wave propagation and using the support vector machine(SVM) to establish the wireless channel model on the mine laneway.On the basis of studying support vector machine(SVM) and propagation characteristics of electromagnetic wave on the mine laneway, the paper takes the rectangular channels as an object and highly analy the influence in transmission decay of the electromagnetic wave caused by the changes of the mode of antenna polarization, the carrier frequency, the electric parameters of surrounding rock, the cross-section size, the roughness of laneway walls, the inclination of tunnel walls, the distance between the sending and receiving antenna and so on.In view of the large scale we take the above affecting factors as input of support vector machine(SVM) network model, the transmission decay of the electromagnetic wave as the output of the model, finally train the support vector machine(SVM) model for the decline of large scale network. In the stage of feasibility analysis, the experimental datas are obtained by simulation experiments, the model of simulation experiments including the superposition model of electromagnetic propagation: the decay model, the decay roughness of laneway walls,the decay inclination of tunnel and the multipath Rayleigh fading. For the fading channel of small scale, this paper studies the forecasting model of multipath fading channel which is based on support vector machine algorithm, and reconstruct the changing about fading channelby using the refactoring theory of phase space, and train the fading model of support vector machine(SVM) channel with the data obtained after refactoring.In this paper, the selection about kernel parameter of support vector machine(SVM) model applis the multi-layers dynamic adaptive optimizing algorithm.Through MATLAB experimental simulation, we can find that support vector machine(SVM) can learn the change of the wireless channel in roadway and the error for its prediction is smaller compared with other traditional model. The results show that it is a feasible way applying support vector machine(SVM) technology to built the model of the electromagnetic wave propagation.