The Application Of Gas Emission Prediction Based On Improved Ant Colony Algorithm

Increasingly occurrence of coal mine accident greatly threat to the safe production of coal mine, where the gas disaster was the primary problem faced by coal production. Shortage of accurate way to predict gas emission presents a great challenge to gas accident prevention. And the reasonable prediction of mine ventilation design is closely related to the gas emission, which also become an important consideration in technical and economic indicators of mine. The factors of gas emission usually express dynamic, fuzzy nonlinear changes, which causes that the modeling accuracy in traditional linear method is difficult to meet the actual production needs.After summarizing the past forecasting methods of gas emission, the paper improves the basic ant colony algorithm and tries to use it to apply to the prediction of gas emission, the paper uses neural network as a medium and then build a system predictive model by the improved ant colony algorithm to optimize the wavelet neural network. Firstly, the paper analyzes the traditional wavelet neural network has low convergence precision, easy to fall into local minima of defects, so the improved ant colony algorithm is introduced. Then the paper analyzes the basic principle and all the important parameters in ant colony algorithm, according to the problem which the algorithm is easy to fall into local issues, ant colony algorithm optimizes the maximum and minimum ranges, smoothes pheromone track and introduces penalty factor and disturbance factors. The algorithm achieves good results to solve the traveling salesman problem. Finally, using the improved ant colony algorithm and wavelet neural network to build model and simulate in historical data. The result shows that comparing with the traditional algorithm; the paper's model has faster convergence, higher convergence precision and better robustness.With the predictive tool of wavelet neural network, improved ant colony algorithm can build predictive models, which reflects its superiority in prediction. That proves in the future, improved ant colony algorithm can play a key role in gas emission prediction and its applied research also has better prospects.