Research On Clustering Algorithm Based On Swarm Intelligence Optimization

With the increase in computer computing capabilities and storage capacity and the development of the mobile Internet,e-commerce,social media,and information technology will generate large amounts of data all the time.How to effectively mine useful hidden information from complex and sparse mass data is a big problem for social development in the context of artificial intelligence.As a classic unsupervised learning method in the field of data mining,cluster analysis can find potential relationships between data in the absence of training samples and has unsupervised in the context of the lack of marking of existing actual data.The clustering analysis technology of learning ability has become one of the research hotspots in the field of artificial intelligence.Due to the unsupervised learning nature,almost all clustering algorithms need to manually set key parameters at this stage.In the face of high-dimensional and complex data,it is difficult to find suitable parameters manually,so finding the suitable parameter in a large data environment has become the key problem to improving the performance of the clustering algorithm.This paper starts from this type of problem and uses the optimization features of the swarm intelligence algorithms.The fruit fly optimization algorithm and the cuckoo search algorithm are improved to enhance the algorithm's search ability.A new cluster analysis model is applied to solve the problem of parameter optimization of the affinity propagation algorithm.The main content and innovation of this article are as follows:(1)To solve the problem that the traditional fruit fly optimization algorithm lacks global searching ability and is easy to be premature,a knowledge-memory-based fruit fly optimization algorithm(KM-FOA)is proposed.The algorithm proposed individual knowledge memory mechanism.The concept of direction vector is introduced into the knowledge memory mechanism and used as knowledge to optimize the search path of the fruit fly,so as to achieve the goal of quickly locating the optimal solution area in the early stage of the algorithm.The results of experiment represent that KM-FOA has better perdormance in solving multidimensional optimization problems.(2)A new cuckoo search algorithm based on boundary recognition is proposed.Aiming at the problem of low search efficiency and slow convergence rate of the original cuckoo algorithm,a boundary recognition method is proposed.By quickly delineating community search boundaries,individual search efficiency is improved.And this model is applied to solve the design problem of PID controller.Experimental results show that the proposed model can find reasonable parameters of the less absolute error(ITAE)index.(3)Novel fruit fly optimization algorithm with trend search and co-evolution algorithm is proposed.A trend search strategy was proposed to achieve the goal of accelerating the convergence rate of the algorithm.On this basis,a co-evolutionary mechanism was adopted to avoid premature convergence and improve the global search capability.The improved algorithm is applied to optimize the bias parameters in affinity propagation clustering.The simulation results show that the proposed improved AP clustering model has good research potential and nice application value.(4)Based on the research of the semi-supervised affinity propagation clustering which introduces a priori paired constraint metrics,a fuzzy-density based fruit fly optimized semi-supervised affinity propagation clustering is proposed.Using the improved global optimization ability of fruit fly optimization algorithm,the two parameter space of preference parameter and damping factor were optimized,and the optimal clustering was determined according to the Silhouette evaluation index.To further demonstrate the research potential of the model,we use improved models to analyze seismic data.The clustering results show that the model shows a good application value.