A Multi-Agent Cooperative Optimization Algorithm With SNS Feature

The multi-agent theory is one of the core technologies in artificial intelligence. Recently, some evolutionary algorithms based on agents are used to solve various scientific and engineeering optimization problems, such as particle swarm optimization (PSO), ant colony optimization (ACO), Immune Algorithm, differential evolution (DE), which have acquired great success in various optimization problems.A novel and less-parameters Agent-based Cooperative optimization Algorithm (ACoA) with SNS is proposed in this paper, which is a swarm optimization algorithm and completes the search by the cooperative interactions among agents. Some concepts of trust degree, neighbor area and community leader are defined to denote the mutual information of the agents and their neighbors. Algorithm ACoA consists of two parts. One is the three operations of self-enhancement, competition and cooperation to guide the evolution for stability and diversity in each generation. The other is leader identifying operation with dynamic features in SNS. Agents and their environmental information will be updated every iteration. Incorporating SNS model to multi-agent theory is the main innovation of this research. Extensive comparison among ACoA, particle swarm algorithm (PSO) and artificial bee colony algorithm (ABC) indicates that ACoA has a better or comparable performance with PSO and ABC both in convergence rate and accuracy.