主成分分析法和遗传算法优化的支持向量机模型在地震伤亡人数预测中的应用

Application of Support Vector Machine Model Optimized by Principal Component Analysis and Genetic Algorithm in the Prediction of Earthquake Casualties

为有效解决地震伤亡人数预测所需影响因子多、运算量大、模型训练烦琐等问题,构建了主成分分析法(PCA)和遗传算法(GA)优化的支持向量机(SVM)模型,采用PCA对地震伤亡人数影响因子进行降维以去除贡献率较低的主成分,将贡献率较大的主成分作为支持向量机的输入变量,以地震伤亡人数作为输出变量,利用GA对SVM模型性能参数进行优化,建立基于PCA-GA-SVM的地震伤亡人数预测模型,并对测试样本进行预测,结果表明:与SVM模型、GA-SVM模型和PCA-GA-BP模型相比,PCA-GA-SVM模型的预测准确率和运行效率分别提高4.73%、1.14%、9.99%和47.05%、36.76%、44.55%。结果显示,PCA-GA-SVM模型预测精度高,泛化能力强,能够科学合理地对地震伤亡人数作出预测。

The prediction of earthquake casualties involves many influencing factors,such as heavy computation,complicated model training,etc.In order to solve these problems,support vector machine(SVM)model optimized by genetic algorithm(GA)based on principle component analysis(PCA)was proposed.PCA was used to reduce the number of earthquake casualties influencing factors,and abandon those principal components with low contribution.The principal components with high contribution were used as input variables of SVM,and earthquake casualties were taken as output variable.Then GA was used to optimize the SVM parameters.Finally the prediction model for earthquake casualties based on PCA-GA-SVM was established,and used to predict the test samples.The result shows that the average prediction accuracy and operation efficiency of PCA-GA-SVM model respectively increased by 4.73%,1.14%,9.99%and 47.05%,36.76%,44.55%compared with prediction results of SVM model,GA-SVM model and PCA-GA-BP model.Therefore,the PCA-GA-SVM model has high prediction accuracy and strong generalization ability,which can predict earthquake casualties scientifically and reasonably.