To enhance the exploitation of the DA algorithm, you can consider hybridizing it with a local search algorithm. The great deluge algorithm (GDA) is a good suggestion. GDA is known for its ability to maintain the diversity of population and avoid the trap in local minima. The distinguishing characteristic of GDA is its deterministic level-based acceptance criterion. Beside the acceptance criterion, GDA requires only one parameter (called decay rate) to be determined.
You can refer to our paper which combines the bat algorithm with GDA.
Bisan Alsalibi, Ibrahim Venkat, Mohammed Azmi Al-Betar, A membrane-inspired bat algorithm to recognize faces in unconstrained scenarios, Engineering Applications of Artificial Intelligence, Volume 64, 2017.