In this paper we set the first steps towards the development of a commercially viable tool that uses evolutionary computation to address the Product to Shelf Allocation Problem (P2SAP). The problem is described as that of finding the numbers and locations of modules to allocate to particular products in a shop, fulfilling at the same time a number of constraints. We first justify the use of evolutionary algorithms in this problem in the bad scalability properties shown by exact methods. Then we proceed, from simpler to more complex versions of the problem, to describe different encodings, fitness functions and evolutionary operators that are suited to the problem. The variations described are tested on five different problem configurations: three with one shelf, one with two shelves and one with eight shelves. In all cases acceptable results can be obtained in a very short timescale, although there is much work to be done on the subject.
展开▼
