基于GA-SVM的质量预测系统设计和实现

Design and Implementation of Quality Prediction System Based on GA-SVM

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。

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.