A multi-tiered spare parts inventory forecasting system

To achieve intelligent nuclear power management, we propose a multilevel nuclear power spare parts prediction method. We combine qualitative prediction methods with various cutting-edge quantitative prediction methods to forecast the overall inventory of nuclear power spare parts and individual categories, enabling enterprises to minimise costs and prevent stockouts. Specifically, we classified the spare parts according to their usage characteristics and developed a hybrid model that integrates the CNN+BiLSTM and DLinear models (Zeng et al., 2022), taking into account expert opinions for each category. Experimental results show a significant improvement in accuracy compared to traditional methods.