Research on K-Medoids Clustering Integration of Multi-Source Information Data in Oil And Gas Engineering
The information data in the field of oil and gas engineering is vast and originates from multiple chan-nels.The data is widely distributed and of uneven quality.Directly integrating all data points for integration often leads to a degradation in the quality of the information matrix,making it difficult to meet practical application needs,so an oil and gas engineering multi-source information data integration algorithm based on K-medoids clustering is proposed.First,a multi-source data set is built and representative points of multi-source data are selected based on decision graph.Then,based on the nearest neighbor approximation principle,the mixed representation strategy is used to construct the sparse affinity sub-matrix and conduct the sparse processing,and the basic clustering results of the multi-source information data of oil and gas engineering are obtained by combining the nearest representative fast ap-proximation method.Finally,the Lagrange function is used to give weight to the results after basic clustering,calculate the clustering cost,and complete the integration of multi-source information data of oil and gas engineering.Experi-ments show that the average number of iterations of the proposed method for the dataset is low,the CA is always above 96%,the NMI value is above 0.94,and the curve is stable and fluctuates slightly,indicating that the clustering inte-gration is more accurate and effective.
K-medoids clusteringMulti-source information dataDecision diagramSparse affinity submatrixBasis clustering
万方数据
维普
