Hardware Particle Swarm Optimization Based on the Attractive-Repulsive Scheme for Embedded Applications

Particle Swarm Optimization (PSO) algorithms have been proposed to solve engineering problems that require to find an optimal point of operation. However, the PSO algorithm suffers from \emph{premature convergence} and high elapsed time when solving multimodal and large scale engineering problems. This problem becomes an evident drawback for embedded applications in which the micro controllers often operates at low computational capacity. This paper proposes a hardware implementation of a parallel self-adaptive PSO algorithm based on an attractive-repulsive scheme and using the efficient floating-point arithmetic which performs computations with large dynamic range and high precision. The parallel capabilities of the PSO are exploited by implementing parallel particles directly in hardware in order to decrease the running time. In addition, the attractive-repulsive technique avoids the \emph{premature convergence} problem by self-adapting the swarm behavior according to its diversity. Synthesis and simulation results for benchmark test problems were performed, demonstrating the correctness of the proposed architectures. Finally, an elapsed time comparison between the hardware and software implementations shows the suitableness of the proposed architecture for embedded applications.