Research On A New Particle Swarm Optimization Algorithm And Its Application

Particle swarm optimization(PSO)algorithm is a kind of new group intelligent Optimization algorithm inspired by social behavior ofbirdflocking, which is proposed by Kennedy and Eberhart. It is a new branch of evolutionary computation. PSO is simple in concept, few inparameters, faster convergence and easy in implementation. Therefore, when it was proposed, it attracts wide attention and gradually become a new hotspot. Currently, particle swarm optimization algorithm is applied to neural network training, function optimization, multi-objective optimization and so on. It has achieved good results and has broad application prospects. branch. Up to now, it mainly consists of two aspects for PSO research:One is the algorithm research, the other is the application study for PSO.In this paper, the following two aspects are studied. First, a new particle swarm optimization algorithm-Locust-Behaved Particle Swarm Optimization(LBPSO) technique is proposed, which is applied to Benchmark test functions. Based on the mechanism of the locusts’collective behavior, LBPSO adopts an adaptive evolutionary mechanism. The number of swarms is self-adaptively adjusted by the acquired outstanding particles coming from behind the previous global best solution. As a new multi-swarm optimization technique, LBPSO automatically generates and ends the swarms determined by the optimization results. With the self-adaptation, LBPSO prevents the blindness of initial setting of swarms. This kind of multiple species not only maintains the diversity of population but also enhances the global search ability of particles. The results show that LBPSO significantly improves the outcome performance, convergence rate and adaptation of different problems.The second part is application of LBPSO algorithm on wireless sensor network placement optimization. In this part, the maximize of the wireless sensor network coverage and energy consumption model with adjustable sensing radius is studied. Firstly, the energy consumption is formulated as functions of node sensing ranges; Then, the dymic network coverage model is proposed, namely adjustable sensing radius is energy consumption model and the maximize of network coverage is the goal. Simulation shows that LBPSO performs well on both0-1model and Statistical model.