运用GA-SVM模型的砂石骨料分类方法

提出一种能够直接从破碎工序上识别骨料成分的算法模型.在花岗岩、石灰石、灰绿岩三种骨料剪切破碎实验的基础上,获取其相关物理特征量;然后,采用遗传算法(GA)与支持向量机(SVM)相结合的算法,对破碎、筛分后的人工砂石骨料物理特征量进行训练,建立人工砂石骨料分类的GA-SVM模型.研究结果表明:所提GA-SVM模型具有较强的分类效果,能够较好地识别人工砂石骨料成分;与传统的BP神经网络、偏最小二乘法相比,其分类准确率高.

Research on Classification Method of Sand and Stone Aggregate Based on GA-SVM

An algorithm model was proposed to directly identify the composition of aggregates in the process of crushing. According to the shear crushing experiments of three aggregates of granite, limestone, and gray greenstone, their relevant physical characteristics were obtained first; then, a genetic algorithm(GA)combined with support vector machine(SVM)algorithm was adopted to train the physical characteristics of artificial sand and gravel aggregate after crushing and sieving, and a GA-SVM model was establish for artificial sand and gravel classification. The research results show that the proposed GA-SVM model has a strong classification effect and can better recognize the components of artificial sandstone aggregates. Compared with the traditional BP neural network and partial least squares method, its classification accuracy is high.