Particle Swarm Optimization And Its Application In Optimal Sensor Placement

Particle Swarm Optimization (PSO) is an optimization algorithm based on swarm intelligence, it issimple, has fast convergence ability and strong global search ability and only need to adjust a fewparameters, and has wide applicability in different optimization problems.So it was concerned widely bymany scholars since it was proposed. At present, the algorithm has been successfully applied in functionoptimization, combinatorial optimization, neural networks, pattern recognition, fuzzy system control andmany other fields. However, the particle swarm optimization algorithm originated from the simulation ofnature biomes behavior, it does not have a strict theoretical basis and have the prevalence of prematureconvergence and other defects.In this paper, Aiming at the defects of the particle swarm algorithm, some improved strategies wereproposed and the improved algorithm was used to optimal sensor placement of historic architecture, andachieved good results finally. The main research work is as follows:1. This paper analyzed influence of inertia weight on particle swarm algorithm, and showed theperformance analysis on the different inertia weight under the same number of iterations. However, thePSO algorithm is difficult to achieve dynamic balance of global exploration and local search. This paperproposed a particle swarm optimization algorithm based on cosine adaptive adjusting inertiaweight(CW-PSO).And the algorithm selects different inertia weight adaptively in the iterative process toadjust the direction and speed of particles. The test results show that the algorithm improved the ability ofglobal optimization algorithm to some extent, and are not easily fall into local optimum, and can controlthe balance of global optimization and local optimization better.2. The diversity of the population declined rapidly in an iterative process and the algorithm is easy to result in local convergence.This paper proposed a particle swarm optimization based on double populationstrategy (DP-PSO). The original population was divided into two groups with the elite group and thecommon group, and different evolutionary strategies were executed in each group respectively. At the sametime, in order to maintain the diversity of particles, this paper set a updating period T,and the twopopulations cross, share information, co-evolve every T. The results show that the algorithm iseffective to avoid the decrease of the diversity of particles, while the appropriate updating period canguarantee the algorithm achieve better convergence accuracy.3. Under the engineering application background of 《the Baoguo Temple Basilica science andtechnology protection information system》, the CW-PSO algorithm is applied in the optimal sensorplacement of historic architecture, and designed a reasonable fitness function. The output of the algorithmresults and advice by experts are basically consistent and the PSO algorithm was further validated.