Using Competitive Population Evaluation in a differential evolution algorithm for dynamic environments

Abstract

This paper reports three adaptations to DynDE, an approach to using Differential Evolution to solve dynamic optimization problems. The first approach, Competitive Population Evaluation (CPE), is a multi-population DE strategy aimed at locating optima faster in the dynamic environment. This approach is based on allowing populations to compete for function evaluations based on their performance. The second approach, Reinitialization Midpoint Check (RMC), is aimed at improving the technique used by DynDE to maintain populations on different peaks in the search space. A third approach, consisting of a combination of CPE and RMC is investigated. The new strategies are empirically compared to DynDE using various problem sets. The empirical results show that the new approaches constitute an improvement over DynDE and other approaches in the literature.

Self-Adaptive Differential Evolution for Dynamic Environments with Fluctuating Numbers of Optima

“…The performance of DynDE was thoroughly investigated by Du Plessis and Engelbrecht [28]. However, to illustrate how effective DynDE is on dynamic environments compared to normal DE, Figure 1.2 depicts the offline error, current error and diversity of DynDE algorithm on the MPB for 10 changes in the environment.…”

“…In these situations, one of the sub-populations will be reinitialized, leaving one of the optima unpopulated. It was shown [28] that this problem can be partially remedied by determining whether the midpoint between the best individuals in each sub-population constitutes a higher error value than the best individuals of both sub-populations. If this is the case, it implies that a trough exists between the two sub-populations and that neither should be reinitialized (see Figure 1.8, scenario A).…”

“…However, a more robust approach, for example re-evaluating the best individual in each sub-population, can be introduced relatively easily should it be required. Dynamic Population Differential Evolution (DynPopDE) [27], an extension of DynDE and CDE [28], is aimed at dynamic optimization problems where the number of optima in the search space is unknown or fluctuates over time.…”

An adaptive multi-population artificial bee colony algorithm for dynamic optimisation problems

“…-Competitive population evaluation in a differential evolution algorithm for dynamic environments (CDE) [28].…”

Solving dynamic optimisation problems by combining evolutionary algorithms with KD-tree

“…Offline errors CDE [25] 0.92 ± 0.07 CPSO [3] 1.06 ± 0.07 MSO [26] 1.51 ± 0.04 ESCA [4] 1.53 ± 0.02 Cellular DE [5] 1.64 ± 0.02 DynDE [6] 1.75 ± 0.03 MEPSO [7] (5 detectors) 4.02 ± 0.56 jDE ( [27], implemented by [25]) 5.88 ± 0.31…”