爆破器材2024，Vol.53Issue(6)：58-64,7.DOI:10.3969/j.issn.1001-8352.2024.06.009
基于粒子群优化相关向量机的爆破飞石距离预测模型
Prediction Model of Blasting Flying Rock Distances Based on Particle Swarm Optimization Relevance Vector Machine
摘要
Abstract
In order to quickly and accurately obtain the distance of blasting flying rocks and timely control blasting hazards,a prediction model of blasting flying rock distance based on particle swarm optimization(PSO)and relevance vec-tor machine(RVM)was proposed.PSO was used to optimize the core width parameter of RVM model,the optimal parame-ters could be adaptively obtained.The optimized RVM was used to establish the nonlinear mapping relationship between distances of blasting fly rock and six main influencing factors including borehole aperture,borehole length,ratio of mini-mum resistance line to borehole distance,borehole filling length,maximum section charge,and explosive consumption.Multiple indicators such as absolute relative error δ,root mean square error ERMS,mean square error EMS,mean absolute error EMA,and correlation coefficient R2 were used to evaluate the performance of the model.The model was applied to predict the distances of blasting flying rock in a mine in Johor,Malaysia,and compared with the optimal results of three models:quadratic rational Gaussian process regression model,medium Gaussian kernel support vector regression model,and double-layer neural network model using the same samples.R2 of PSO-RVM model increases by 7.1％,and ERMS decreases by 14.56％.EMS and EMA decrease by 26.99％ and 15.96％,respectively.PSO-RVM model has better reliabi-lity and fit of prediction results,and higher accuracy.关键词
粒子群优化算法/相关向量机/爆破/飞石距离/预测模型Key words
particle swarm optimization algorithm/relevance vector machine/blasting/distance of flying rock/pre-diction model分类
矿山工程引用本文复制引用
刘小明,唐北昌,李荣华,陈德斌,梁培钊..基于粒子群优化相关向量机的爆破飞石距离预测模型[J].爆破器材,2024,53(6):58-64,7.基金项目
国家自然科学基金(52068016) （52068016）
广西重点研发计划(桂科AB21196041) （桂科AB21196041）
