Coupled Neural Networks and Genetic Algorithms Application in the Field of Mine Fire Extinguishing

Abstract

As one of the important causes of mine fires, predicting the distribution of gas content in mines is of great value. In order to predict the gas content in mines, this study proposes a combination of backpropagation neural networks and genetic algorithms. Firstly, the topology structure and network weights of the backpropagation neural network are determined through genetic algorithms. Then, preprocessed data is input to the updated network structure, and data errors are calculated. Finally, the prediction is completed by outputting the results. Results showed that the combined algorithm can achieve stable performance after training 8000 times, with a prediction error of around 0.5%. Applying it to the system, it can be predicted that the high gas content range in the experimental mine was concentrated between 21.02 m³·t^-1 and 28.46 m³·t^-1. Therefore, the system proposed in the study can stably predict the gas content in mines.