| Particle swarm optimization(PSO)algorithm has the advantages of simple parameter settings and faster convergence speed,now it has been widely used in many fields.However,with the emergence of big data,many real problems have become more and more complex,such as the particle routing problem of distributed filter centralized resampling.Such problems often consist of multiple conflicting objective functions and a large number of constraints.In this case,the traditional PSO algorithms are difficult to achieve good results.How to improve the capability of PSO and effectively apply it to real complex optimization problems have become a current research hotspot.In view of this,this thesis studies an improved PSO algorithm with the theory of molecular force,and its applications the particle routing problem of distributed filter centralized resampling.(1)Effectively balancing the convergence and diversity of swarm has always been a difficult problem for PSO.To overcome the above problem,firstly,an improved social learning PSO algorithm(MISL-PSO)based on molecular interaction is proposed.Introducing the mechanism of molecular attraction and repulsion into PSO,an optimal imitation learning strategy based on attraction force and a variable-scale particle initialization strategy based on repulsion force are presented to adaptively balance the convergence and diversity of swarm.Furthermore,the convergence and time complexity analysis of the proposed algorithm are analyzed.By applying the proposed algorithm into 49 benchmark functions including the CEC2013 test functions and comparing it with 6 typical algorithms,the experimental results demonstrate its effectiveness.(2)Applying the proposed MISL-PSO to the distributed filter particle routing problem,a single-objective distributed filter particle routing method is proposed.Firstly,a single-objective constrained optimization model for he distributed filter particle routing problem is established with the goal of minimizing the communication cost between processing units(PU).Then,an improved variable-length coding strategy is introduced to discretize the particle position update formula,the penalty function method is used to deal with constraints,and a single-objective distributed filter particle routing method with MISL-PSO is proposed.Applying the proposed algorithm to the data assimilation system of large-scale wildfire spreading,and comparing it with other routing algorithms,the experimental results demonstrate its effectiveness.(3)Considering the two objectives,i.e.,the communication cost between PUs and the computing load capacity of each PU,a multi-objective distributed filter particle routing method based on improved PSO is proposed.Firstly,a multi-objective constrained optimization model for the distributed filter particle routing problem is established.Then,an improved hybrid discrete multi-objective PSO is proposed to solve the model.Based on the characteristics of the problem,a local search strategy based on molecular force and a constraint processing mechanism based on greedy search are proposed to improve the performance of the algorithm.By comparing with three commonly used methods for the classical particle filter problem,the experimental results show that the proposed algorithm is a very competitive method,which can provide decision-makers with a variety of high-quality Pareto optimal solutions. |
