Most common approaches for solving DOPs using metaheuristics:
Search Restart
This straight-forward forward method consists on re-initializing the search in order to react to changes in the environment. This kind of approach was mainly explored in the 90s, while nowadays it is preferred to use techniques that transfer information from the past into the new state of the problem.
Tuning Evolutionary Operators
This category refers to methods like hypermutation (raising the mutation rate during a certain period of time) and adaptive mutation operators, hyperselection, etc.
Memory Schemes
Approaches in this category share a common characteristic: they all keep some kind of memory which intends to store good solutions that can be reused when the environment changes.
Multi-Populations
The motivation behind using several populations to enhance performance is trying to keep diversity. Self-Organizing Scouts, Shifting Balance GA and Multinational GA fall in this category.
Future Prediction
This kind of methods try to predict future changes, based on the fact that changes in a real problem can follow a certain pattern and could be learned. This approach, under the usage of Kalman filters, has attracted much attention recently.
University of Málaga
Department of Languages and Computation Sciences
NEO: Networking and Emerging Optimization