基于PCA-GA-SVM的火成岩分类方法研究

在地质科学中,正确的岩石分类有助于研究岩石的成因、形成条件、演化过程和工程设计等.由于地质条件的多样性、变异性及复杂性,人们很难对岩石样本进行准确的分类.通过主成分分析法(PCA)从影响火成岩分类的众多氧化物评价指标中提取出主成分,用遗传算法(GA)优化支持向量机参数,并采用支持向量机方法(SVM)对实际火成岩公开数据进行训练,建立了火成岩岩石分类的PCA-GA-SVM模型,同时结合火成岩实际数据将预测结果和基于Levenberg-Marquardt算法改进的BP神经网络模型(LM-BP)的预测结果做了比较.结果表明:基于PCA-GA-SVM模型得到的火成岩分类预测结果精度较LM-BP神经网络有很大的提高,与实际分类相符,有广泛的应用前景.

Study on the Classification Method of Igneous Rocks Based on PCA-GA-SVM

In the Geological Sciences,the correct classification of rocks is helpful to the study of the origin,formation conditions,evolution process,and engineering design and so on.Due to the diversity,variability and complexity of the geological conditions,it is difficult to classify the rock samples accurately.In this paper,by means of principal component analysis method (PCA) the principal components were extracted from the evaluation indexes of various oxides affecting the igneous rock classification firstly;then the genetic algorithm (GA) was used to optimize the parameters of support vector machine (SVM) and the actual publicly available data of igneous rock were trained by the support vector machine method;and finally,PCA-GA-SVM model of the igneous rock classification was established and compared with the BP neural network optimized by Levenberg-Marquardt algorithm(LM-BP).The results show that the classification method of igneous rocks based on PCA-GA-SVM model has a higher accuracy then LM-BP model,which is consistent with the actual classification,and has a broad application prospect.