A Novel Prediction of Blasting Vibration Parameter

Abstract: Rough set theory (RS) can find useful information from a large number of data and generate decision rules without prior knowledge. Support vector machines (SVM) have good classification performances and good capabilities of fault-tolerance and generalization. To effectively control blasting vibration effect, prediction model of blasting vibration parameters based on rough sets and support vector machines are established. Particle swarm algorithm (PS) is used to dynamically discretize attributes. Then, according to the optimal particle calculated by the PS algorithm, the decision table is established. After eliminating less importance attributes and data, then SVM is trained and established for prediction. The model can deal with both quantitative and qualitative factors such as aperture and cast direction, and analyze their importance to peak vibration acceleration value. The improvement of prediction accuracy is verified by testing data.