考虑全过程优化的支持向量机航材消耗预测方法

Aviation Material Consumption Prediction Method Based on Support Vector Machine Considering Whole Process Optimization

针对航材消耗影响因素复杂,传统支持向量机预测精度较低的问题,提出了一种考虑全过程优化的支持向量机(SVM)航材消耗预测方法。采用LASSO算法实现主要影响因素选择,通过K-means聚类算法将样本分为相关性较强的子样本集,根据不同类别分别选择合适的核函数和优化参数建立SVM预测模型。结合航材消耗数据实例分析,最后通过均方根误差与传统支持向量机模型和神经网络模型比较,结果表明全过程优化的预测模型对提高航材保障效率有积极意义。

Aiming at the problems that influence factors of the aviation material consumption are complicated and the prediction accuracy of the traditional support vector machine is low,a support vector machine(SVM)forecasting method for aviation material consumption considering the whole process optimization is proposed.Lasso algorithm is used to select the main influencing factors.Then,Kmeans clustering algorithm is used to divide the samples into sub sample sets with strong correlation,and the appropriate kernel function and optimization parameters are selected to establish SVM predication model according to different categories.The comparison with traditional support vector model and neural network model is done by root means quene,combined with data example analysis of aviation material.The results show that the prediction model of the whole process is positive to improve the efficiency of the aviation material support.