基于遗传算法与支持向量机的火炮内膛疵病分类方法研究

Classifying Method for Gun Bore Flaws Based on Improved Genetic Algorithm and Support Vector Machine

支持向量机能够较好解决小样本、非线性数据特征的多分类问题,适用于火炮内膛疵病的分类,但其性能依赖于其参数的选取;对基本遗传算法进行改进,使用自适应的交叉算子和变异算子,并应用于支持向量机参数优化;通过对火炮内膛疵病分类精度的对比研究,改进的遗传算法确定支持向量机优化参数的方法取得了良好的效果,验证了该方法对疵病分类的有效性。

Support Vector Machine can solve small nonlinear multi-classification problem well,so it is applicable to classifying gun bore flaws,but its performance depends on its own parameters.Simple generic algorithm was improved by using adaptive cross operator and adaptive mutation operator.The method was applied to parameter optimization of Support Vector Machine.Then,the comparative study was carried out on classification precision of gun bore flaws and the method of improved genetic algorithm determining support vector machine parameter optimization achieved a good outcome,which proved the effectiveness of the method.