Optimise a ventilation system in underground mines using artificial neural networks

This study aimed to optimise ventilation systems in underground mines using artificial neural networks (ANNs) to reduce temperature and relative humidity while improving airflow. A dataset of 66 samples was used to train the ANN model, which included 4 input parameters, 128 and 64 hidden neurons, and 3 output parameters. The model was tested with 70% of data in training and 30% in validation. The results demonstrated that the ANN showed strong predictive performance for temperature and humidity, achieving R2 values of 0.88 and 0.96, respectively, in the validation data. Additionally, the model achieved an R2 of 0.81 for airflow data, indicating reliable predictions. The ANN also successfully optimised the ventilation parameters, resulting in a temperature reduction of 6.13°C, a 5.61% decrease in humidity, and an increase of 110.80 CFM in airflow. In conclusion, the study confirms that artificial neural networks can effectively optimise a ventilation system in mines.