基于支持向量机的岩爆预测研究与应用  

Research on Rock Burst Prediction based on a Support Vector Machine

作　　者：吴菡 WU Han(Agricultural and Rural Bureau of Jingkou District,Zhenjiang 212000,China)

机构地区：[1]镇江市京口区农业农村局,江苏镇江212000

出　　处：《粉煤灰综合利用》2024年第4期92-96,133,共6页Fly Ash Comprehensive Utilization

摘　　要：随着采矿、水利、交通等领域向着深部发展,工程中岩爆问题频发,严重危害人员安全,因此建立准确有效的岩爆预测模型至关重要。结合多种优化算法得到模型最优参数,构建了3种基于支持向量机(SVR)理论的岩爆预测模型。依托157组国内外实测岩爆案例,以模型预测准确率为识别框架,综合新的模型预测结果评价指标(平均偏差)分析预测模型性能,使用数值模拟和工程应用两种方式,验证模型有效性。结果表明:随着输入参数类别的增加,模型预测准确率也随着增加,遗传算法(GA)是、粒子群算法(PSO)和麻雀搜索算法(SSA)对支持向量机有一定优化效果,麻雀搜索算法优化效果最佳,故基于优化的SVR岩爆预测模型是可靠有效的。With the development of mining,water conservancy,transportation and other fields toward deeper areas,rockburst problems occur frequently in engineering,which seriously endanger the safety of personnel,so it is crucial to establish an accurate and effective rock burst prediction model.A rock burst prediction model based on support vector machine theory is constructed,and the optimal parameters of the model are obtained by combining the sparrow search algorithm.Relying on 157 groups of domestic and foreign measured rock burst cases,the model prediction accuracy is used as the identification framework,and the new model prediction results evaluation index(mean deviation)is integrated to analyze the prediction model performance,and both numerical simulation and engineering application are used to verify the effectiveness of the SSA-SVR model.The results show that the model prediction accuracy increases with the increase of input parameter categories.Genetic algorithm(GA),particle swarm algorithm(PSO)and sparrow search algorithm(SSA)have some optimization effect on support vector machines,so the optimized SVR rock burst prediction model is reliable and effective.

关 键 词：地下工程 岩爆 遗传算法(GA) 粒子群算法(PSO) 麻雀搜索算法(SSA) 支持向量机 

分 类 号：TU45[建筑科学—岩土工程]