Intelligent method and system for adjusting spare parts demand based on dynamic weight self-adaption

Abstract

The application discloses a kind of intelligent prediction method and system of aviation material demand based on dynamic weight self-adaptive adjustment, belong to the field of civil aviation material intelligent scheduling.The method includes: constructing the multi-source feature matrix of fusion historical maintenance data features, model parameter features and environmental factor features;Weighted fusion is carried out to multi-source features based on dynamic weight function containing seasonal adjustment component and gradient feedback component;The feature after weighted fusion is input into the hybrid prediction model including time series feature extraction layer and long-term correlation modeling layer to predict aviation material demand;The hybrid prediction model is trained based on the weighted loss function embedded with dynamic weight vector.The application realizes the self-adaptive adjustment of multi-source feature weight through dynamic weight function, can accurately capture the seasonal fluctuation of aviation material demand and quickly respond to sudden changes, effectively improve the prediction accuracy and inventory turnover efficiency.

Images