A Hybrid Intelligent Algorithm for Fuzzy Dynamic Inventory Problem

In this paper, a fuzzy inventory problem with multiple commodities is casted into a dynamic programming model with continuous state space and decision space. In order to solve the dynamic programming model, genetic algorithms are used to get samples of the optimal cost functions, and then neural networks are trained to approximate the optimal cost function on a randomly generated sample set, which may bypass “the curse of dimensionality”. A hybrid intelligent algorithm is thus produced to get the optimal cost functions functions that represented by neural networks. Lastly, a numerical example is given for illustrating purpose