Particle Swarm Optimization Algorithm And Its Application In Fuzzy Clustering Algorithm

In the era of rapid information development,the data generated by various industries contains a large amount of information,which has the characteristics of high redundancy and poor coupling.How to effectively mine massive data is a hot issue that is currently being explored in various industries.Fuzzy clustering(FCM)is an effective method to classify the data with unclear classification boundaries in the real world.It overcomes the absolute diagnosis results and makes the clustering results more suitable for the actual needs.However,the result of fuzzy clustering is greatly affected by the initial clustering center,which can easily lead to misjudgment or misjudgment.Therefore,it has become an urgent problem to optimize the initial FCM clustering center and improve the clustering accuracy.As a swarm intelligence algorithm,particle swarm optimization algorithm not only maintains the cooperative working mode among particles,but also maintains the independent searching ability,which makes the algorithm have the advantages of fast searching speed and high convergence precision.It has been proved that the particle swarm algorithm has certain advantages in FCM initial clustering center processing by many scholars who conduct a lot of experiments in Recent years.However,the particle swarm algorithm has the shortcomings of being trapped in local extreme points and oscillating in the later stage of particle search.This paper proposes a new FW-SAPSO algorithm for particle swarm optimization,and uses it to optimize the quality of the initial clustering center of fuzzy clustering.Finally,the clustering effect of FW-SAPSO algorithm is evaluated by UCI data set.The main work and results of the paper are as follows:(1)Simulated annealing particle swarm optimization algorithm for dynamically adjusting flight time and inertia weight(FW-SAPSO)is proposed.It avoids falling into local optimum and oscillations.The algorithm is improved in the following four aspects: 1)The Metropolis criterion of the simulated annealing algorithm is introduced into the FW-SAPSO algorithm,so that the improved algorithm has a certain "inclusiveness" for particles that currently appear to be not optimal,and accepts non-high-quality solutions as high-quality solutions with a small probability to prevent particles from concentrating on the current optimal solution too quickly and falling into a local extreme point;2)The mutation operation is introduced to improve the algorithm to enrich the diversity of the population.The improved algorithm can effectively avoid the reduction of population diversity in the later stage of particle search,which will affect the quality of the solution and cause the problem of premature population;3)Introduce the time-of-flight strategy into the FW-SAPSO algorithm to improve the situation of oscillations in the later stage of the particle search and missing the optimal solution;4)Inertial weight has a significant impact on the optimization performance of particle swarm optimization.This paper proposes a classification adaptive inertia weight to replace the linearly decreasing inertia weight in the standard particle swarm algorithm to achieve the purpose of timely adjustment according to the movement state of the particles,and used the benchmark function to simulate the FW-SAPSO algorithm,which verifies the superiority of the FW-SAPSO algorithm in accuracy and convergence speed.(2)The poor quality of FCM random clustering center has a great impact on the subsequent clustering process.In this paper,the FW-SAPSO algorithm is used to improve the quality of FCM initial clustering centers,and obtain a set of better quality clustering centers,then perform FCM clustering operations,and finally use UCI data set to test and analyze the optimization effect.