Automatic clustering and multi-directional scheduling schemes in multi-mode scheduling projects

In this study, we tried to provide a new approach to solve multi-mode resource constrained project scheduling problem (MMRCPSP), in which each activity has multiple execution modes needing both of renewable and non-renewable resources, which are categorized as NP-hard. Three ideas are studied and evaluated in the present paper. Firstly, the single-objective MMRCPSP is converted into a MMRCPSP with three objective functions. By doing so, the non-renewable resource constraint is removed from the main constraints of the problem. Moreover, a new executive cost is defined. By eliminating these constraints and defining the new cost constraints for each activity, one can be encountered to a new larger solution space and the evolutionary operators can be simplified in this space. At the end, we defined a fitness function based on clustering methods on the new solution space. Numerical results showed that according to multi-directional scheduling schemes (multi-dss), the fitness functions based on automatic clustering methods increased the performance of the evolutionary algorithms, because this approach prevents to trap into the local optimums by increasing the diversity of obtained solutions.

کلیدواژه ها:

project scheduling ، multi-directional scheduling scheme ، clustering ، Automatic clustering ، Genetic algorithm