Demand prediction of earthquake emergency materials using CBR optimized weight distribution by GT-SAGA-AHP algorithm
Zhanzan Zhou, Yalin Chen, Youdong Lv, Jingyuan Wang, Chengcheng Wang, Yajun Li · 2024
This paper presents a case-based reasoning (CBR) optimised weight distribution by GT-SAGA-AHP algorithm for earthquake emergency materials demand forecasting, aiming to improve the accuracy of emergency resources demand prediction. The approach sets the number of disaster-affected population as the prediction target, selects seven seismic hazard indicators such as earthquake magnitude, depth of hypocenter, time, population density, number of collapsed buildings, seismic fortification level, earthquake intensity as research factors to accurately predict the disaster-affected population. Combined with the theory of safety inventory, developing an earthquake emergency materials demand forecasting model to calculate the demand for all kinds of emergency supplies after the earthquake. The experiment results show that the prediction model optimized by GT-SAGA-AHP algorithm achieves a smaller mean relative error (MRE) of the predicted values compared to the models optimized by the GA and SAGA algorithms, with reductions of 89.57% and 87.51%, respectively. This signifies that the feature weight distribution refined through the GT-SAGA-AHP is more rational, and the CBR-based prediction model exhibits greater accuracy.