Prediction Research Of Coal Mine Gas Concentration In The Upper Corner

In the production of coal mine, as a hazardous gas which comes from the rock in coal mine, the coal mine gas is a threat to the safety of coal mine production. Accurate measurement and real-time monitoring control are the important measures to prevent coal gas explosion of which concentration is one of the important indexes in coal mine monitoring. The upper corner of coal face near to the coal wall and the god side has the special position and structure of low airflow speed and eddy current state in local, which makes the gas accumulation easy to happen. This requires a reliable prediction of the upper corner gas concentration to avoid the harm caused by the gas accumulation, predict the gas concentration change tendency in advance and then take effective measures to lower the probability of the occurrence of the upper corner gas accumulation. As a latest forecast science and forecast technology, the prediction of the gas concentration of upper corner in coal face is based on the application of modern wireless monitoring network and project.According to the relevant information and establish basis of the research of the coal gas concentration prediction and combining with the complexity of gas concentration prediction, this topic determines the gas concentration prediction project. The simple model of gray forecast theory is established on the basis of the times series of the gas concentration. Then taking various factors which influence the gas concentration data into account and combining with the phase space reconstruction which is widely used, the structure of the neural network model is variously analyzed to determine the each layer weights and threshold of the neural network model and establish a multidimensional forecast model of neural network. Finally, by learning to know the support vector machine (SVM) model theory and modeling knowledge and analyzing the feasibility and advantage of SVM in coal mine upper corner, the selection problem of two important parameters called kernel function and punish factor selection is fully considered in the modeling process, after which the effective SVM prediction model is built.The engineering realization of gas concentration prediction is solved on the use of the wireless sensor network in this study. The prediction of gas concentration needs a large number of data to research calculation, therefore, prediction has high requirements in data stability, fast and accurate of network transmission. Based on Zigbee wireless sensor network communication protocol, it can improve the former network by improve the structure of networks and transmission algorithm to reach the network support needed in gas concentration prediction.As the main research content forecast method of the application in coal mine gas concentration prediction, this subject sets up a simple wireless monitoring network for the the experiment of forecast scheme to obtain three prediction methods, each of which has the results of the upper corner of gas concentration prediction. Compared with the measured concentration, it can be seen that the prediction strategies demonstrated in this research is reliable and the wireless sensor network designed is feasible, on basis of which we achieve the expected result.