基于LS-SVM的新机备件需求预测

为解决因新机备件历史消耗数据相对较少而给备件预测工作带来的困难,提出应用最小二乘支持向量机(least squares support vector machine,LS-SVM)回归算法来实现新机备件需求的预测.阐述了最小二乘支持向量机的基本原理,建立了新机备件需求的预测模型,选取核函数,采用LS-SVM对训练样本进行学习,对其网格结构参数进行训练,通过十字交叉验证(cross-validation)和网格搜索(grid-search)确定最优参数,利用训练后的LS-SVM对新机备件需求进行预测,并进行算例仿真.结果表明,LS-SVM在新机备件需求预测上表现优秀.

Demand Prediction of New Aircraft Spare Parts Based on LS-SVM

Aiming at the difficulties caused by little history consumption data of new aircraft spare parts on spare parts predication, put forward the LS-SVM regression algorithm to realize the new aircraft spare parts demand prediction. Introduce the LS-SVM basic principle, establish the new aircraft demand predication model, select kernel function, use LS-SVM to learn the training sample, and train its network structure parameter. Ascertain the optimal parameter by cross-validation and grid-search. Use trained LS-SVM to predict the new aircraft spare part demand, and carry out example simulation. The results proved the excellent effectiveness of LS-SVM in new aircraft spare parts demand predication.