A method for predicting blasting grade distribution in block stone mining

Abstract

A method for predicting blasting gradation in block stone mine includes such steps as: 1, obtaining influence factors of blasting effect as input parameters, taking a distribution index of blasting gradation as an output parameter, and establishing sample database, obtaining blasting effect influence factor as input parameter, taking blasting gradation distribution index as output parameter, and setting up blasting gradation index as output parameter. 2, establishing a BP neural network; 3, optimizing a BP neural network through a genetic algorithm to form a GA-BP model; 4: through the sampledatabase of Step 1, performing GA-BP model training; 5, adopting the model obtained in the step 4 to predict the blasting grade distribution of the block stone material mining. The invention integrates the advantages of the two algorithms to recombine the blasting gradation prediction model, and establishes the GA-BP model for the blasting gradation prediction, thereby improving the Prediction Model of Blasting Grading and its Performance and Precision. The algorithm itself has the function of self-learning and self-adjusting, and has strong adaptability to the model, which can meet the engineering application requirements of gradation prediction in block stone blasting mining.

Description

technical field [0001] The invention relates to a blasting gradation prediction method, in particular to a blasting gradation prediction method for block stone mining. Background technique [0002] With the accelerated development of China's hydropower industry, domestic large-scale hydropower projects are gradually concentrated in the southwest region; restricted by topography and geology, building materials, construction conditions, etc., earth-rock dams have become an important optional dam type; earth-rock dams are in the process of construction Among them, the mining of dam materials usually adopts blasting construction, which is affected by rock mass structural surfaces such as joints and fissures. After blasting, the gradation of stone materials is difficult to meet the design requirements, thus affecting the project cost and the compaction quality of dam filling; jointed rock mass The research on the control and prediction of blasting gradation has become the main te...

AI Technical Summary

Problems solved by technology

[0005] At present, many scholars in China have carried out prediction research on the blasting gradation distribution of large-sized dam materials such as rockfill materials and transition materials based on neural networks, and achieved some research results; Optimization research [D]. Quanzhou: Huaqiao University, 2005.) Combined with the practice of bench blasting in Shaxian quarry, comprehensively applied photography, image recognition and fractal theory to establish a fractal test method for calculating the gradation composition of blast piles, and passed BP Neural network to predict the gradation composition of rock blocks; Gao Hong (Gao Hong. Research on block size distribution prediction model and engineering application of open-pit bench blasting [D]. Anhui University of Technology, 2013.) Combining with the on-site blasting production operations of Yichun Tantalum Niobium Mine, Analyzed the main factors affecting the blasting effect, introduced the neural network theory on this basis, took the blasting parameters and the block degree distribution after blasting as the input and output layers of the network respectively, and established a three-layer neural network prediction model; Zhu Wenhua et al. (Zhu Wenhua, Zhu Ruigeng, Xia Yuanyou. Research on Neural Network Method for Prediction of Blasting Blockage [J]. Journal of Wuhan University of Technology, 2001, 23(1):60-62.) Combined with engineering practice, a blasting mining level for rockfill materials was established. The BP neural network model for distribution prediction is compared with the traditional R-R, G-G-S and other empirical function distribution models to analyze the prediction effect of the gradation prediction model; however, although the neural network has a highly nonlinear mapping function and has a strong self- Learning ability and self-adjustment ability, but the algorithm itself still has problems such as slow convergence speed and easy to fall into local minimum points, so there is a lot of room for improvement in the performance and prediction accuracy of the gradation prediction model; therefore, how to improve the neural network model , to overcome the shortcomings and deficiencies of the algorithm, and to improve the performance of the gradation prediction model has become a key technical difficulty in engineering research; in summary, the existing neural network-based blasting gradation prediction methods have "slow convergence" and "easy to Converging to a local minimum" and other issues

Images