Open AccessClassification of GSM-R Network Operation Quality Evaluation Indicators Based on WDFCE Algorithm
Author(s) -
Hang Zhang,
Chaiyang Peng,
Liangde Ma,
Yunru Zhao,
Qing Liu
Publication year - 2020
Publication title -
iop conference series. earth and environmental science
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.179
H-Index - 26
eISSN - 1755-1307
pISSN - 1755-1315
DOI - 10.1088/1755-1315/428/1/012069
Subject(s) - cluster analysis , weighting , k medians clustering , algorithm , boundary (topology) , data mining , mathematics , fuzzy clustering , fuzzy logic , computer science , reliability (semiconductor) , decision matrix , statistics , artificial intelligence , canopy clustering algorithm , medicine , mathematical analysis , power (physics) , physics , quantum mechanics , radiology , operations research
This paper proposed a fuzzy clustering ensemble algorithm based on membership weighting and distance decision (WDFCE) and defined a method for determining the boundary threshold of indicators’ grade intervals. Base clusters and corresponding membership matrices were obtained by multiple fuzzy C-means (FCM) algorithm. Considering the reliability of each base cluster, the decision weighting method was introduced to weight the membership matrix and construct the cumulative distance matrix. The Density Peaks (DP) algorithm was used for clustering ensemble to obtain the final result. By defining the boundary zone, the boundary zone class density and the threshold possibility, the clustering result was combined to determine the grade interval of the indicators. In order to judge the advantages and disadvantages of the improved algorithm, the S_Dbw index with the best evaluation effect was selected from the various cluster validity evaluation indicators studied by domestic and foreign scholars. The results of the indicator S_Dbw for evaluating cluster validity indicated that the improved method we proposed is superior to the general clustering ensemble algorithm and it solved the problem of determining the boundary threshold of the indicator grade interval in the GSM-R evaluation system.