主成分分析—BP神经网络在备件需求预测中的应用

Application of Principal Component Analysis-BP Neural Network in Spare Parts Requirement

备件需求预测在装备维修保障中占据重要的地位,针对当前主要以经验为主进行估计,与实际需求相差较大,提出基于主成分分析—BP神经网络模型的备件需求预测方法。首先利用主成分分析方法去除原始输入数据的相关性,降低数据维度,减小网络规模,选择合适的隐含层的BP神经网络。最后通过结合实例进行分析,取得较好的效果。

The requirement forecasting of spare parts has played an important part in the equipment maintenance.Now,most forecast methods are using the empirical data which can＇t satisfy the practical requirement.A method which is based on the principal component analysis is proposed for the spare parts forecasting in the model of BP neural network.First,the principal component analysis method is used to remove the relevance of the original input data,to reduce the data dimension,to reduce the network size and select an appropriate hidden layer＇s BP neural network.In the end,get a good result through a combination of case analysis.