Improvement And Application Of Particle Swarm Optimization

Particle Swarm Optimization (PSO) is a heuristic globaloptimization algorithm based on the swarm intelligence which showed inbirds foraging behavior. PSO has been widely used in various fields ofscience and engineering, and become a new field of computationalintelligence research focus due to its simple principle, easy to achieve,less adjustable parameters, fast convergence speed, etc. Due to theweakness of the theoretical basis, premature convergence and slowconvergence speed in the late iteration, it is necessary to explore andimprove the theoretical study. In this paper, we analyze the convergencetrajectories of standard PSO algorithm and propose an improved methodof PSO to improve the convergence speed and the overall convergencerate. The novel PSO is applied to the boiler drum water level controlsystem. The main work of this paper is summarized as follows.(1) Speeding up the convergence of particle swarm optimizationalgorithm and preventing fall into local optimum are the two mostimportant facts. Based on the stability analysis of the particle trajectory,we propose a new random number selection mechanism which usesGaussian distribution instead of uniform distribution in the velocity equation of standard particle swarm algorithm. The New Gaussiandistribution Particle Swarm Optimization(NGPSO) not only improve theconvergence rate, but also has the ability to jump out from local optimumThe experimental results on four widely used test functions show that theproposed strategy is effective.(2) In the modern industry, boiler is widely used in many fields.Series of control indicators have been proposed to ensure the safety ofboiler, and boiler steam drum water level is one of the important qualityindexes. Typically, we always use the cascade PID control system of thethree impulse to control the boiler steam drum water level in the industry.But there are lots of variables that affect water level, and these variableswill constantly change with the system environment and control variables,so that the traditional PID control parameter tuning becomes moredifficult. This paper proposes a way based on improved PSO to tune thePID control parameters. Making use of the characteristics ofPSO dynamic self optimization, the control system could select suitableparameters according to the change of the system, and meets the controlrequirements set before. By modeling the boiler drum water level controlsystem, this paper proposes a method of combining PSO with commonPID controller, and further completes the design which combines the PSOwith the boiler steam drum water level system. A simulation of MATLABis given to prove the feasibility of this design, and the result of the comparison to the common PID controller shows a good performance.The paper provides a new way to design the boiler drum water controlsystem.