基于PSO-SVM算法和样本数据的岩爆预测参数敏感性分析

Sensitivity Analysis of Rock Burst Prediction Parameters Based on PSO-SVM Algorithm and Sample Data

岩爆作为深部岩石工程中一种常见的动力破坏现象,多造成严重后果,开展其相关预测研究尤为必要。文章基于PSO-SVM算法和岩爆实测数据,以应力指数T_(s)、脆性指数B、弹性能量指数W_(et)作为基本元素构成的7种评判参数组合开展了岩爆预测以及评判参数的敏感性分析。通过3种参数组合(T_(s)、B、W_(et))开展了PSO-SVM算法应用于岩爆预测的可行性分析。并以3种参数组合为参照,通过逐一去除单一参数的方式和仅以单参数的方式分别开展了评判参数的敏感性分析。结果表明,PSO-SVM算法用于岩爆预测是可行的;弹性能量指数W_(et)对岩爆预测准确性的影响最大,其次为应力指数T_(s),脆性指数B的影响最小;综合应力指数T_(s)、脆性指数B、弹性能量指数W_(et)开展岩爆预测,其结果可信度较高。

Rock burst,as a common dynamic failure phenomenon in deep rock engineering,often causes serious consequences,so it is particularly necessary to carry out prediction research related to it.Based on the PSO-SVM algorithm and the measured data of rock burst,rock burst prediction and sensitivity analysis of the evaluation parameters are carried out on 7 kinds of evaluation parameter combinations with stress index T_(s),brittleness index B and elastic energy index W_(et)as the basic elements.The feasibility of applying PSO-SVM algorithm to rock burst prediction is analyzed through the combinations of three parameters(T_(s),B,W_(et)).Moreover,the combinations of three parameters are taken as a reference,and the sensitivity analysis of evaluation parameters is carried out by means of removing single parameter one by one and only using single parameter.The results show that it is feasible to use PSO-SVM algorithm for rock burst prediction;elastic energy index W_(et)has the greatest influence on the accuracy of rockburst prediction,followed by stress index T_(s),and brittleness index B has the least influence;and the comprehensive application of stress index T_(s),brittleness index B and elastic energy index W_(et)is highly reliable for rockburst prediction.