Research On Improvements And Applications Of Particle Swarm Optimization

Particle swarm optimization(PSO)is a typical intelligent optimization algorithm.It has the advantages of fast convergence,few required parameters,and easy implementation.Therefore,it is widely used in many practical problems.PSO originated from the simulation of bird flock foraging behavior.Its theoretical basis is not perfect at present,and the algorithm itself has shortcomings such as premature convergence and dimensionality disaster,which needs further research and improvement.This paper improves PSO and uses it to solve the problems of B-spline curve fitting and robot path planning.The main work includes the following three parts.First,in order to improve the quality of the initial population of PSO,this paper introduces the homotopy analysis method(HAM)to learn prior knowledge.The connection between the solved problem and the problem to be solved is established,and an improved PSO framework HAM-PSO is proposed.The framework is divided into four main steps,including obtaining prior knowledge,constructing homotopy functions,generating homotopy solution populations and solving problems to be solved.This framework is not only suitable for basic PSO,but also for other improved PSO algorithms.The feasibility and effectiveness of the framework are verified through basic PSO and three other classic improved PSO algorithms.The experimental results show that the framework can improve the convergence accuracy of the algorithm.Secondly,an improved PSO based on genetic-beetle strategy(GBPSO)was proposed to solve the problem that PSO was easy to fall into the local extreme value when solving B-spline curve fitting problem,which resulted in the poor curve fitting effect.In GBPSO,on the one hand,the beetle antenna search strategy is introduced to enhance the local search ability of the algorithm.On the other hand,the crossover and mutation operations in the genetic algorithm are introduced to enhance the global search ability of the algorithm.Use the GBPSO to optimize the node vector,and gradually reduce the curve fitting error through algorithm iteration.The simulation results of four examples show that,compared with the other three algorithms,the error of the fitting curve obtained by GBPSO optimization is smaller.Finally,this paper proposes a hybrid improved particle swarm optimization(HIPSO)for the shortcoming that PSO is easy to fall into the sub-optimal path when solving the robot path planning problem.Three improvement strategies are included in HIPSO.The first is to use the beetle antenna search strategy to enhance the local search ability of the algorithm.The second is the global exploration and local development capabilities of the random inertia weight balancing algorithm.The third is to introduce the Levi flight strategy to prevent the algorithm from falling into search stagnation in the later stage.The three improvement mechanisms cooperate with each other to enhance the algorithm’s optimization ability.At the same time,the cubic spline interpolation method is combined to make the robot moving path smoother.Compared with the other two algorithms,the experimental results show that under the same obstacle environment,the robot movement path planned by HIPSO is shorter.