Multi-objective particle swarm (PSO) analysis in collaborative working environments

摘要

In a collaborative organization, the partner selection problem consists of selecting the best combination of partners capable to accomplish all the project tasks, considering both quantitative and qualitative critical factors. The problem here considers a project that can be divided into a number of sub-projects (business processes or tasks) and for each task, several candidates are possible. Due to the existing relationships between tasks, this can be seen as a network structure containing a set of potential partners connected with each other. The process is modeled as a multi-criteria optimization problem that requires trade-offs between contradictory criteria: internal and collaboration costs, processing time and risk of failure. Due to the non-linearity and large scale of the problem (considerable number of alternatives and different criteria), global powerful optimization techniques are needed. In this study, we have implemented a multi-objective particle swarm optimization algorithm, augmented with an additional fuzzy controller for tuning algorithm parameters in order to determine an effective approximation of the Pareto front. Computational results have shown that the algorithm is able to produce high-quality solutions for all tested instances. Moreover, the algorithm is very flexible and able to obtain non-dominated alternative solutions for different scenarios and alternatives in the design of virtual enterprises.