A LOCATION-ALLOCATION MODEL FOR SITING SCIENTIFIC TRAINING FACILITIES

In location-allocation problems a certain number of facilities or supply centers are needed to be located in an area of interest in order to satisfy a predefined customer demands with knownlocations, and it is not fixed or known apriori the allocation of demands to supply centers. This problem has attracted much research in the optimization literature over the last few decades since 1960s, where restricted exact algorithms as well as a lot of heuristics for searching a near optimal solution have been proposed. Among these heuristics is to consider the LA problem as identical to certain cluster analysis. In this paper we ll apply self organizing feature maps which are artificial neural networks capable of extracting the spatial structure of its input space, for finding a near optimal solution for a real life LA problem for locating a certain number of scientific training facilities to serve the set of governorates all over Egypt. In approximately all LA problems found in the literature; customer demands are predefined prior to the problem solution, but here in our problem there is no direct information available about customer demands, but instead some indicators were available about each demand center from which we could calculate each center s demand. We applied the multiple attribute value theory which is a well known multi-criteria decision making model to calculate the demand value associated with each demand center