Research On Prediction Model Of Gas Concentration Of Driving Ventilation In High Gas Coal Mine Based On Neural Network

With the increase of depth and scale of coal mining, gas emission quantity from a seamis rising, mine gas concentration overrun accidents occur frequently, particularly in headingface gas concentration overrun incidents. Study found that irrational driving ventilationsystem is one of the important reasons that lead to the gas concentration overrun accident ofthe heading face. Through investigates and studies on the spot to the coal mine scene andgained information, layout and configuration parameters of ventilation equipment are the twomain reasons for irrational driving ventilation system. On the basis of comprehensive analysison the gas concentration of different driving ventilation equipment layout and configurationparameters in different excavation stages and different gas emission, using neural networktechnology, neural network prediction model was established, which is the relationshipbetween various driving ventilation key factors and gas concentration. Gas concentrationprediction model can dynamically predict gas concentration in the driving process, and alsoprovides a theoretical basis to gain the optimized program. Main research works completedare as follows:(1) Research at home and abroad on the driving ventilation gas concentrations wasstudied, and on the basis of the analysis coal mine site to obtain the actual sample data,driving ventilation key factors caused by changes in heading face gas concentrations wereanalyzed from two aspects of layout and configuration parameters of ventilation equipment.Ventilation equipment configuration parameters affected by auxiliary fan power, air-ductdiameter; the influencing factors of ventilation equipment layout air-duct length, roadwayexcavation distance; gas concentration is closely related to the environmental factors ofventilation absolute gas emission; (2) For the complex nonlinear relationships between driving ventilation key factors andthe gas concentration, neural network take the advantage of dealing with nonlinear problemsin neural network technology, BP and RBF neural network have been chosen to establishdriving ventilation gas concentration predictions model. Design of BP forecast model is aLayer3network structure. BP structure design using three-layer network architectureincluding one hidden layer, the actual sample data was trained on different hidden layerneurons numbers, to get the optimal hidden layer neurons number by comprehensivecomparison of training results. The hidden layer of RBF prediction model structure can onlybe as a layer by their own factors. As RBF prediction model creation function is selected,hidden layer neurons numbers is automatically created. Driving ventilation key factors weretaken as the input layer of two prediction models, while gas concentration as the output layer;(3) With the help of MATLAB software and the coal mining site to obtain the actualsample data, driving ventilation air concentration of BP and RBF prediction model wasestablished. The choice of the training function was selected by comparing different trainingfunctions of training performance in BP model. Select small training error, short training time,steady iterative process function as BP model of optimal training function. Spread of RBFmodel was selected by comparing the prediction error values. Select the minimum value ofprediction error as optimal spread value of RBF model. And the actual sample data was usedto test accuracy and performance of BP and RBF prediction model,and comparative analysisto get a better accuracy and performance RBF prediction model;(4) The RBF prediction model was studied on applications of driving ventilation gasconcentration. First, the analyses of gas concentration in driving ventilation equipment layoutand configuration parameters of different driving ventilation programs, and a comprehensivecomparison of both of effect on gas concentration was done. The small gas concentration andlow energy consumption ventilation program is the best option, and thus achieve a reasonableand effective optimization of driving ventilation equipment layout and configuration program,and reduce production costs.