基于偏最小二乘回归的装备备件消耗预测

Spare Parts Consumption Prediction Based on Partial Least-Squares Regression

考虑了装备使用时间、行驶里程和配备时间等影响备件消耗的多种因素,依据装备备件的消耗特点,在分析偏最小二乘回归方法原理的基础上,运用方法对小样本数据条件下装备备件的消耗数量进行预测.应用示例表明,偏最小二乘回归方法比传统多元回归分析法、逐步多元回归分析法和删除多元回归分析法具有更高的预测精度.

Through making a analysis of the factors which include operation time, actual service life and the age of the equipment and the character of spare parts consumption, the partial least-squares regression is applied to solve the problem of spare parts consumption prediction when the sample is small. The example indicates that partial least-squares regression is much more accurate than traditional multiple linear regression, stepwise multiple linear regression and erasing multiple linear regression.