Study On Blasting Fragmentation Forecasting Model For Rock Mass

The fragment-size distribution from blasting of ore mass and rock mass is an important index to evaluated the blasting quality quantitatively. It affects the efficiency of each follow-up producing process and the total cost in mining. In Hydraulic and Electric Engineering, the blasting fragment-size has a principal influence on the dam quality. Therefore, making a thorough study on the fragment size in rock mass blasting is of momentous practical significance. The exiting models usually describe the process by some mathematical equations based on analyzing process principles and making some assumptions and simplifications. However, most blasting processes have very complex mechanism, a lot of influencing factors and high severe nonlinearities, so simple mathematical equations is difficult to describe these processes. In these cases new modeling is adopted. As a typical kind of modeling method Artificial Neural Networks(ANN) has been applied to many chemical problems for its good performance in solving nonlinear problems. But ANN has some disadvantages such as overfitting, local minimum, etc. because its theory is based on Empirical Risk Minimization(ERM) principle. Support Vector Machines (SVM) is a new learning method based on Statistical Learning Theory(SLT).SVM based on Structural Risk Minimization(SRM) principle overcomes ANN's inherent disadvantages and greatly improves model's generalization ability. In this thesis we discuss the application of SVM in the blasting fragment-size in detail.Firstly, the exiting models at present is reviewed and the cracking mechanism of blasting rock fragment-size is analyzed. After discuss SVM's theoretical basis, computing process and optimization algorithms are demonstrated detailedly. Finally, modeling by ANN and SVM in the blasting fragment-size is introduced.This thesis mainly studies SVM's application in blasting rock fragment-size. A model to predict the fragment-size distribution from blasting of rock mass is set up byANN, training and prediction is done by improved algorithms. In addition, modeling by SVM is introduced and the model parameters' effect on the model performance is carefully studied, this model is compared with the traditional one based on ANN. The experimental results show SVM is able to model the fragment-size distribution from blasting of rock mass.