An Optimized Model for Blasting Parameters in Underground Mines' Deep-hole Caving Based on Rough Set and Artificial Neural Network

For better predicting and optimizing the blasting parameters in underground deep-hole mining, 16 groups of deep-hole blasting parameters are collected and collated, combining rough set and artificial neuron network theory, an optimized model for basting parameters in underground mines' long-hole caving based on rough set and artificial neural network is set up. Adopting the rough set software for data reduction, then using the reduced data and raw data as the inputs of the ANN software, the predictions have completed. The input attributes of the ANN model are 6, the RS-ANN model input attributes are 5, both training samples are 12, both forecast samples are 3, the former average prediction accuracy is 0.91 similar to 13.7%, the latter is 0.12 similar to 7.97%. This study shows that rough set is effective in data reduction while retaining key information; the predicted results of RS ANN model coincide with the actual situation, and the overall accuracy increased by more.