关键词: 相似度, 高斯混合模型, JS散度, 流程工厂, 设备集

Abstract: To measure the similarity between process plant equipment collections,a similarity calculation method based on Gaussian mixture models was proposed.This method calculated from two aspects which were the amount of equipment in collection and the constitution of equipment collection,then the two factors were integrated to achieve an integrate similar indicator.For the amount of equipment,it was calculated by Jaccard coefficient thought;For the constitution of collection,firstly the collection was mapped to be points set in two dimensional coordinate space with classification tree,then Gaussian mixture models were used to describe the regularities of points distribution,finally the similarity was calculated with the JS divergence of Gaussian mixture models.According to the grading scores and the analysis of equipment process function from process plant professionals,the rationality and effectivity of the calculations were verified.

Key words: similarity, Gaussian mixture model, JS divergence, process plant, equipment collections