Ensembled Residual Network (ERN) for Gas-Hazard Prediction in the Underground Coal Mine

The underground coal mines exhibit many life-threatening hazards for mining workers. In contrast, gas hazards are among the most critical challenges to handle. This paper proposes a deep neural network-based model, i.e., an ensembled residual network (ERN) for the gas hazard prediction in the underground coal mine case. The model uses two balancing parameters to integrate two different one-dimensional convolutional neural network (1DCNN) architectures. Those balancing parameters are responsible for controlling the output of both architectures. The ERN model is tested with the mine gas dataset, and the performance is compared with 1DCNN and a feed-forward deep neural network. The performance analysis of ERN for underground coal mine data is presented. An accuracy of 96.5% is achieved with an error of less than 4.7%.