Parallel Analysis Of Swarm Intell-igence Algorithm And Co-Design On Hardware And Software

Particle Swarm Optimization algorithm as a kind of swarm intelligence algorithm, is from groups of bird and fish movement behavior. Its main feature is simple in principle, few parameters and better convergence speed. The algorithm in function optimization, neural network, combinatorial optimization, robot path planning has been widely used applications.But here introduces that the Quantum Particle Swarm Optimization(QPSO) is based on the Particle Swarm Optimization algorithm.At the same time, a new hybrid QPSO algorithm with cooperative method (CQPSO) is mentiond. They are an efficient search algorithm for global optimization, in terms relative to the PSO algorithm,which have been advantages of fast convergence and the good convergence performance. However, due to the same QPSO with PSO algorithm, they are also based on the particle as a whole to be updated, hence QPSO algorithm also has the disadvantage of limited dimensions. By combining a high-dimensional particles with complex decompose into multiple one-dimensional sub-optimization of individuals, the use of collaborative approaches CQPSO algorithm can overcome this shortcoming.Currently, FPGA technology has matured. It is no longer just for rapid ASIC prototyping, but also can be used as SoPC (System on a Programmable Chip) devices. FPGA technology is developing rapidly and widely used. BP-based online learning, CMAC estimation and control, RBF feature recognition and other neural network has been implemented and applied in the FPGA device. In the neural network learning, PSO (Particle Swarm Optimization) algorithm as a new one aroused people's attention.From the improved performance of the algorithm, it has been applied to the embedded engineering fields, such as system identification, control parameters optimization, power optimization, FPGA placement and routing optimization, and robot control design recently. This article is from the parallelism point of view , Analyzed the structure about the Quantum-behaved PSO Algorithm QPSO and hence a new hybrid QPSO algorithm with cooperative method between particles CQPSO,whose structure can be parallelism.Then combing with FPGA technology characters which can be parallel processing of information, indicated the convergence of particles performance in parallel operation mode. Programming in the Xilinx Ise10.1, when can be configured in integrated, using the main string pattern download to FPGA bo.ard.The data obtained in the software is the same as the data obtained in FPGA.