论文部分内容阅读
备件预测在产品物流保障中占有极其重要的地位,针对现有各种航空备件预测方法精度较低,无法满足实际需求的现状,文章提出了基于主成分分析—RBF神经网络模型的备件预测方法。首先利用主成分分析方法去除原始输入层数据的相关性,以解决RBF神经网络模拟预测备件需求时输入变量过多,网络规模过大导致效率下降的问题,最后选择合适的径向基函数密度训练神经网络。通过结合实例进行分析,取得了较好的效果。
Spare parts prediction plays an extremely important role in product logistics support. Aiming at the current situation that all kinds of aviation spare parts prediction methods have low accuracy and can not meet the actual demand, a spare part prediction method based on Principal Component Analysis - RBF neural network model is proposed. First, the principal component analysis method is used to remove the correlation of the original input layer data to solve the problem of RBF neural network simulation forecasting spare parts demand too much input variables, the network size is too large lead to efficiency decline, and finally select the appropriate radial basis function density training Neural Networks. By combining the examples for analysis, good results have been achieved.