Modelling the Performance of an Implicit Memory-Based Technique in Solving Dynamic Scheduling Problems Through Response Surface Methodology

Response Surface Methodology (RSM) is a popular class of methods used to build empirical models through statistical means. Rarely in the literature is the RSM applied to model characteristics of Evolutionary Algorithm (EA) based techniques used to solve dynamic optimization problems. This paper presents RSM as a modelling tool of some characteristics of a memory and EA-based approach referred to as McBAR – the Mapping of Task IDs for Centroid-Based Adaptation with Random Immigrants. McBAR is applied to solve a class of multi-objective dynamic resource-constrained project scheduling problems where the total number of schedule-comprising tasks varies in time. In this paper we: (a) Determine EA-related parameters of which the EA-based techniques under consideration (including McBAR) can produce high-quality solutions to the above scheduling problems. This parametric search applies an innovated Split-plot approach that applies the RSM. (b) Create models through a novel method of applying the RSM in the EA domain. These models characterize certain capabilities of the techniques equipped with the determined EA-related parameters. The capabilities are in searching for solutions to the scheduling problems. (c) Legitimize the algorithmic components of McBAR through the models. (d) Illustrate the solution-searching abilities of the techniques under the dynamics of the scheduling problems. (e) Determine the limitations of the techniques under these dynamics.