摘要： 针对支持向量机(SVM)参数大多凭经验选择的费时问题，提出基于遗传算法(GA)的SVM参数选取方法和基于组件对象模型(COM)技术实现Visual C#与Matlab 的混合编程方法。以质量预测系统中GA-SVM预测模型建模和程序实现为例给出2 种方法的具体实现。结果表明，使用GA优化SVM参数能充分发挥GA算法特性，降低参数选择的时间；使用COM技术的混合编程能提高程序开发和运行的 效率。

关键词: 遗传算法, 支持向量机, 组件对象模型, 混合编程

Abstract: For the choices of the Support Vector Machine(SVM) parameters mostly depending on experience is time-consuming, this paper proposes a method of selection based on GA, and a method of hybrid programming, based on the Component Object Model(COM) using Visual C # and Matlab. An example is introduced for the GA-SVM prediction modeling and procedures for implementation. Results show that optimizing the SVM parameters by the GA makes best use of it, meanwhile reducing the time for parameters selection greatly. The use of COM in integrated programming improves the development and operating efficiency.

Key words: genetic algorithm, Support Vector Machine(SVM), Component Object Model(COM), integrated programming

中图分类号: 

• TP311.5