A reactive-proactive approach for solving dynamic scheduling with time-varying number of Tasks

Abstract

Any system (whether in the area of finance, manufacturing, administration, etc.) 
that operates in a dynamic environment needs to be adaptive to changes; it should 
also anticipate possible adverse events to remain competitive. In our previous research 
in this area we experimented with one particular approach: Mapping of Task ID for 
Centroid-Based Adaptation with Random Immigrants (McBAR) to address problems 
of environmental changes for Resource-Constrained Project Scheduling (RCPS) problem, 
especially when the latter involves changes in task numbers. However, at that time, McBAR 
was applied as reactive tool only. In this paper we extend McBAR approach to the RCPS 
problem in a proactive-reactive way. The system handles also three competing objectives: 
cost, makespan, and the risk of failure.

We have not found any papers that deal with risk on the RCPS problem and utilize the 
attributes of plans from the past environmental changes. This particular aspect is incorporated 
in McBAR - experimental results indicate the efficiency of such approach in finding optimal 
solutions for a current change. In this paper we also analyze, under the effects of environmental 
dynamics, the variation of risk computed via McBAR and of parameters related to optimization. 
Further, we compare McBAR to other Evolutionary Algorithm approach in the same problem.