Multi-objective Variable Weight Combination Forecasting Model Based on pccsAMOPSO

Abstract

In order to accurately predict the macroscopic material flow, aiming at the limitations of the existing medium and long-term macro material flow forecasting models, we propose a multi-objective variable weight combination prediction mode (MOVWCP) based on the parallel cell coordinates system Adaptive Multi-Objective Particle Swarm optimizer algorithm (pccsAMOPSO) to analyze and predict macro material flows. In order to improve the stability of MOVWCP, the concept of error entropy is proposed, at the same time, MOVWCP uses mean absolute percentage error (MAPE) and error entropy to build the objective functions. An intelligent heuristic algorithm based on pccsAMOPSO is designed to solve the Pareto front of variable weights during the fitting period and the variable weight Pareto solution was selected by using the sensitivity difference. A series of numerical experimental results verify the superiority of MOVWCP and its algorithm.