Parameters optimization of undulating grey model and its application on predicting short-Term inventory demand

摘要

Undulating grey model (UGM) GM(l,l| tan(k -τ)p, sin(k -τ)p) is especially useful for predicting non-linear undulating series. In order improve the modeling precision of UGM, an optimization model is constructed to obtain theoretical minimized error and solved by Particle Swarm Optimization (PSO) algorithm in this paper. An numerical case study show the best parameters in UGM can be identified with the highest precision. The optimization method is further applied to simulating and predicting actual short-term inventory demand data from a steel enterprise.