On Improving the electromagnetism algorithm for solving the resource allocation problem in probabilistic activity networks

The focus of this study is to investigate the optimal resource allocation in probability activity networks (PAN) where the duration of the activity is not only a random variable, but also a function of the resource allocated. Assume that the project manager concerns the total costs incurred in the project that includes two cost terms, namely, the resource usage cost and the tardiness cost. In order to optimally allocate the resources in PAN, we propose an improved electromagnetism algorithm (EMA). Furthermore, in order to improve the convergence of our EMA, we device two heuristics based on the so-called "critical paths cluster" in PAN, that is a subset of paths in PAN more critical in determining the project completion time than others. Our first heuristic could efficiently obtain good initial solutions, and the second heuristic conducts effective local search to secure better solutions. Our numerical experiments show that the proposed EMA significantly out-performs other approaches.