Clustering Algorithm Based On Dynamic Membrane Computing System

The clustering analysis algorithm is an unsupervised learning method, and it groups the data byfinding the hidden patterns naturally and intelligently. This algorithm is based on the analysis of theclustering principle, and in essence it is a method of optimization problems. Also it is an important researchdirection in data mining. P system is a new branch of evolutionary computation, and in terms ofoptimization it has its own incomparable advantages. P system has the features of parallelism anddistributed nature, making it has a great potential to exceed the Turing machine. On the basis of the analysisof cluster algorithm and P system, this paper proposes the new idea of putting P system into use ofclustering analysis.(1)The research of clustering according to different classification standards. Studying the method ofclustering based on different types of sample data. Contrasting the advantages and disadvantages ofdifferent methods to get a more satisfactory clustering result, so that it can be timely adjust its method tosolve practical problems in the application process. Finally, the effect of the algorithm is verified bysimulation.(2)Conducting research according to the structure and objects and basic elements of principles of Psystem. It is similar to cluster analysis that different objects will form different types of algorithms. Therules of P system are more special, and its principles are based on its own custom settings, so we can makedynamic rules according to the dynamic needs of the system, and leading to a more appropriate method ofP system to deal with dynamic data objects.(3)To introduce the advantages of P system into cluster analysis, so as to getting a better clusteringalgorithm, and making the P system and cluster analysis more and more widely used. This article learns theMCMO algorithm, and making use of the dynamic structure object data of the P system to cluster analysis.It obtains the clustering results by the use of rules to the data object again and again. In the simulationexperiment, it tests the performance of P system represented by MCMO algorithm and the clusteringanalysis algorithm represented by K-means algorithm. This paper uses three test functions compared withPareto optimal boundary, and to illustrate the advantages of clustering analysis based on P system by comparing the results.