基于遗传算法优化参数SVM的备件需求预测研究

Research on Spare Demand Prediction Based on Support Vector Machine by Genetic Algorithm Optimization Parameter

针对传统备件预测理论在小样本下预测精度不高的实际问题,将支持向量机(SVM)回归理论引入备件需求预测领域,提出基于支持向量机备件需求预测方法,并给出了具体步骤以及需求预测结果准确率的评价指标;以实际数据为例,得到遗传算法优化参数的支持向量机方法的计算结果,通过与指数平滑法、网格搜索法优化参数的支持向量机和遗传算法优化参数的支持向量机进行对比,验证了该方法精度高的优点,表明将支持向量机理论应用到备件保障领域具有重要的实用价值。

Focusing on the practical problem of low precision of the conventional prediction method,the actual data on aerial support vector machine classification of spare models are applied to verify the superiority of their classification. The exponential smoothing method,grid search method optimization parameters of support vector machines and genetic algorithm to optimize parameters of support vector machines is respectively used to forecast key aerial spare demand. The result shows that the genetic algorithm optimization of support vector machine forecasting performance is the best. Results prove that the support vector machine theory is applied to the field of aerail spare security has important practical significance.