基于遗传算法优化支持向量机的震级预测模型研究  

Earthquake Magnitude Prediction Model Based on Support Vector Machine Optimized by Genetic Algorithm

作　　者：张小涛 张新东 王晨晖[1,2] ZHANG Xiao-tao;ZHANG Xin-dong;WANG Chen-hui(Hebei Hongshan National Observatory on Thick Sediments and Seismic Hazards,Xingtai 054000,China;Xingtai Central Seismic Station,Xingtai 054000,China;Hebei Earthquake Agency,Shijiazhuang 050021,China)

机构地区：[1]河北红山巨厚沉积与地震灾害国家野外科学观测研究站,河北邢台054000 [2]邢台地震监测中心站,河北邢台054000 [3]河北省地震局,河北石家庄050021

出　　处：《河北地质大学学报》2023年第6期41-46,共6页Journal of Hebei Geo University

基　　金：唐山震源区密集台阵观测与孕震环境研究(DZ20200827056);河北省地震科技星火计划(DZ2021110500001)。

摘　　要：基于震级与其影响指标之间复杂的非线性关系,引入支持向量机(support vector machine,SVM)实现地震震级的预测,采用主成分分析法(principal component analysis,PCA)降低影响指标集的维度,同时将新生成的主成分作为模型输入,利用遗传算法(genetic algorithm,GA)寻找SVM最优参数,建立了基于PCA-GA-SVM的地震震级预测模型,通过实际测试样本验证了模型性能的可靠性,结果表明:PCA-GA-SVM模型预测结果平均相对误差为2.13%,具有良好的预测效果。Based on the complex nonlinear relationship between magnitude and its impact indicators,support vector machine(SVM)was introduced to predict earthquake magnitude.Principal component analysis(PCA)was used to reduce the dimensions of the impact indicator set.At the same time,the newly generated principal component was used as the model input,and genetic algorithm(GA)was used to find the optimal parameters of SVM,the earthquake magnitude prediction model based on PCA-GASVM was established.The reliability of the model performance was verified by actual test samples.The results show that the average relative error of the prediction results of PCA-GA-SVM model is 2.13%,which has good prediction effect.

关 键 词：地震震级 主成分分析法 遗传算法 支持向量机 

分 类 号：TU457[建筑科学—岩土工程]