Research On The Gas Prediction Model Based On RBF Neural Networks And Ant Colony Algorithm

Gas disaster prevention is always the important point of coal mine safety. More than 70% serious disasters and accidents of coal mine in our country are gas disasters. With the increase of mining depth and growing competition in coal industry, the number of gas disasters is showing a tendency of rising. And the result can be worse.There are many advantages of RBF neural network, such as simple structure, quick study, fit accuracy, better generalization capabilities, not easy to fall into local minima and so on. In this paper, the RBF neural network prediction mode of coal mine gas emission rate is attempted to establish, which is to do the approximation of function between coal mine gas emission rate and complicated non-linear relation of many factors. The ant colony optimization theory is used to optimize the neural network and overcome the shortcomings of the neural network. Besides that, weights and threshold value are optimized by the ant colony optimization algorithm.Enough representative dates of gas detector are used as examples in this paper. On the basis of wireless sensor gas detection system, the prediction model of coal mine gas emission rate is established by the RBF neural network algorithm, which is optimized by the ant colony algorithm. MATLAB is also used to do simulation research. The results show good forecast results.