A New Cockroach Swarm Optimization Algorithm Using Mixed-Valued Spaces

Swarm intelligence algorithms gain knowledge from the collective behaviour of simple agents that interact with one another and with the immediate environment. Cockroach swarm optimization algorithm (CSO) was inspired by the emergent social behaviour of cockroaches which include chase-swarming, dispersing, hunger and ruthlessness etc. The continuous spaces version of CSO was originally introduced and later improved using digitization and discretization techniques to create binary space and discrete multivalued spaces algorithms respectively to cater for binary-valued and discrete multi-valued optimization problems. The research project reported here combines the previous designs of continuous, binary, and discrete multi-valued spaces to present a model that can solve optimization problems of any valued space. A Mixed-valued cockroach swarm optimization (MCSO) algorithm is proposed in this paper where different variable types are contained in a single cockroach agent, and different position update procedures are executed based on the type of variable e.g. continuous, binary or discrete. The performance of the proposed algorithm was evaluated via simulation studies using well-known benchmark problems, and the performance comparison of continuous, binary, and discrete version of cockroach swarm optimization was carried out. We also compared the performance of a binary version of cockroach swarm optimization with that of the existing binary particle swarm optimization. Our results show the binary cockroach swarm optimization performed better than binary particle swarm optimization.