多因素影响下舰载机备件需求的组合预测

Combination Forecast of Spare Parts Demand for Carrier-Based Aircraft under Influence of Multiple Factors

舰载机列装时间较短,备件的样本数据较小,而且保障中受起落次数、飞行强度、海洋恶劣环境等因素影响较大。针对舰载机这一系列保障特点,选用了对多因素影响的小样本有较好预测效果的BP神经网络、GM(1,N)预测模型和SVM回归预测模型3种预测方法,建立基于IOWA算子的组合预测模型,以误差平方和为准则对数据进行分析,并利用Matlab工具箱进行优化计算,从而得出最优组合预测结果。实例分析结果验证了该组合预测模型的科学性和优越性。

Talking into account of short service time of carrier-based aircraft, small number of sample data of spare parts, with great influence of the number of taking off and landing, flight frequency, marine environment and other factors, three forecasting methods were adopted to construct a combination forecast model based on IOWA operators for small sample problem, which were BP neural network, GM（1,N） forecast model and SVM regression forecast model. Data was analyzed with the principle of sum of squares error, and the final optimized combination forecast result was attained by Matlab used to optimize and calculate. The availability and superiority of this combination forecast model was demonstrated in an exam- ple.