Research Of Clustering Algorithm Based On Membrane Computing

Membrane computing is a young branch of natural computing that studies how to abstract computing methods and models based on the characteristics of biological cell membranes.It is composed of membrane computing model,membrane algorithm and membrane evolutionary algorithm.Membrane computing model(also known as P system)is a parallel computing model which is distributed based on cell membrane structure and communication mechanism when the computing power of traditional computing model is limited.Due to the natural parallelism and uncertainty of membrane computing model,it has been applied in more and more fields in recent years.Membrane evolutionary algorithm is an evolutionary algorithm inspired by the life cycle characteristics of biological cells.It has the ability to find the global optimal solution and has been applied in solving NP hard problems in resent years.In the era of big data,clustering analysis is an important part of data mining,which is proposed for digging the inner structure of data set and the relation between data.Hierarchical clustering and partition clustering are the methods of clustering analysis.Among them,hierarchical clustering can form the data set into a tree-like hierarchical structure,but its computation is very large and the degree of parallelism is not high enough compared with other clustering algorithms.At the same time,partition clustering method is relatively efficient,but it needs to specify the clustering number of data sets in advance,and most partition clustering algorithms cannot find the global optimal solution affected by the initial clustering center.So in this thesis,we designed a P system?to achieve the hierarchical clustering algorithm based on the characteristics of parallelism and uncertainty of membrane computing model(P system).It improved the ability of parallelism compared with the traditional hierarchical clustering.Secondly,a heuristic clustering algorithm MEAMC is proposed based on the global search capability of the membrane evolution algorithm.It does not require a predetermined number of clusters in MEAMC,and the experimental results show that it has good performance.The main work in this thesis is shown as follows:(1)The hierarchical clustering P system?is designed based on the cell-like membrane model,including its membrane structure and evolutionary rules.This thesis also analyzes the computational performance,parallel characteristics,correctness and practicability of?through an instance.(2)MEAMC,a metaheuristic clustering algorithm is proposed based on membrane evolution algorithm,the membrane structure and evolutionary operators are also designed to solve the problems.Finally,the effectiveness of MEAMC is verified and the optimal clustering number can be found by comparing other clustering algorithms based on evolutionary algorithm.In this thesis,two fields of membrane computing and clustering analysis are combined to carry out relevant research.The research results not only extend the application field of membrane computing model and membrane evolution algorithm,but also provide a new way to solve the existing problems in clustering analysis,which has certain reference significance for the research in the field of membrane computing and clustering analysis.