Multiple Strategies Of Particle Swarm Optimization And Its Application In Vehicle Routing Problem

In scientific research fields such as engineering practice,computers,intelligent transportation,and artificial life,the optimization is a common problem in our lives.As the economic growth model has undergone a tremendous transformation,the forms and processes of industrial production have become more and more complex,and the growth of the dimensions of optimization problems are also very dramatic too.For example,the Vehicle Routing Problem with Capacity(CVRP)is a typical and complex optimization problem,the solution of specific path planning problems such as inspection route of robots in substation inspection robots and route of robots in garbage dumps have been well solved under the intelligence optimization algorithm.For the solution of endless optimization problems,the intelligent optimization algorithm has obvious advantages in terms of computational complexity,application scope,and the scale of problem solving have a wider adaptability and higher efficiency.Particle swarm optimization algorithm(PSO)is a new type of swarm intelligence based stochastic optimization algorithm.Compared with other evolutionary algorithms,PSO is easy to implement and has fewer parameters,and it has been widely used in science and engineering.However,the standard PSO still has some problems such as the premature convergence,local convergence,lower diversity in the later period,and the accuracy,the theoretical analysis and application research of PSO need to be further developed and improved.The disadvantages of the above algorithms block the promotion and application of particle swarm optimization.Therefore,it is very great significant to improve the insufficiency of particle swarm optimization and expand its application.The main research of this paper are as follows:(1)An adaptive filtering mechanism is introduced into the search strategy of the improved PSO,and a particle swarm optimization algorithm with adaptive filter based on health degree is proposed.Firstly,through the dynamic detection of particle health,the particle state is distinguished,the abnormal particles are processed and marked,the position of the lazy particle is adaptively filtered,and the algorithm is prevented from falling into a local optimum;secondly,the global worst particle value is updated using a leader factor,and the abnormal particle is filtered.Meanwhile,the leader factor also can avoid the invalid search and speed up convergence.Finally,compared with the standard particle swarm and other improved algorithms by the 11 standard functions are simulated to show the advantages of the proposed algorithm.The results show that the adaptive filtering particle swarm algorithm based on health has higher optimization accuracy and faster convergence.(2)Facing the problem that the particle swarm optimization algorithm can easily fall into a local optimum when dealing with a multi-dimensional and multi-valued point function,a dispersal particle swarm optimization algorithm based on random whip mechanism is proposed.Secondly,the particles of active movement are processed by the individual historical optimal positions to improve the convergence of the algorithm.The analysis of the experimental results shows that the dispersal particle swarm optimization algorithm based on random whip mechanism has higher optimization accuracy and faster convergence,and also verifies the superiority of the improved algorithm.(3)Finally,the above two improved particle swarm algorithms and original PSO algorithm are applied to the capacity-constrained vehicle path optimization problem(CVRP).CVRP is a nonlinear programming with complex constraints and belongs to the NP problem.The experimental results also verify that the two improved algorithms proposed in this paper have good advantages,so they have high application value.