CIESC Journal ›› 2017, Vol. 68 ›› Issue (2): 739-745.DOI: 10.11949/j.issn.0438-1157.20161069

Previous Articles     Next Articles

Intuitionistic set theory based multiple kernel fuzzy clustering and its application of ethylene raw material properties

CUI Xinghua1, DU Wenli1, ZHAO Liang1, LI Jiangli2, CHI Liang2   

  1. 1. State Key Laboratory of Chemical Engineering, Key Laboratory of Advanced Control and Optimization for Chemical Processes, East China University of Science and Technology, Shanghai 200237, China;
    2. PetroChina Jilin Petrochemical Company, Jilin 132000, Jilin, China
  • Received:2016-07-28 Revised:2016-10-08 Online:2017-02-05 Published:2017-02-05
  • Supported by:

    supported by the Key Program of National Natural Science Foundation of China (61590923), the National Science Fund for Excellent Young Scholars, the Young Scientists Fund of the National Natural Science Foundation of China (61422303, 61403141) and the Shanghai Municipal Education Commission and Shanghai Education Development Foundation "Dawn Project".

直觉模糊多核聚类算法及其在乙烯原料属性聚类中的应用

崔兴华1, 杜文莉1, 赵亮1, 李江利2, 池亮2   

  1. 1. 化学工程联合国家重点实验室, 华东理工大学化工过程先进控制和优化技术教育部重点实验室, 上海 200237;
    2. 中国石油天然气股份有限公司吉林石化分公司, 吉林省吉林市 132000
  • 通讯作者: 杜文莉
  • 基金资助:

    国家自然科学基金重点项目(61590923);国家自然科学基金优秀青年基金项目;国家自然科学基金青年科学基金项目(61422303,61403141);上海市教育委员会和上海市教育发展基金会“曙光计划”资助项目。

Abstract:

Along with the increasing types of ethylene cracking materials and expensive feed analyzer, clustering of ethylene cracking materials which is to improve ethylene yield modeling, ethylene yield and energy consumption has very important practical significance. In order to improve the accuracy of online identification of raw materials, an intuitionistic fuzzy kernel clustering algorithm based on the theory of intuitionistic fuzzy sets is presented. In the definition of membership, membership considers uncertain information which is the hesitation degree. At the same time, intuitionistic fuzzy entropy is incorporated in the loss function of multiple kernel clustering algorithm. That is to optimize the data points in the class. Further, the cracking material attribute feature selection using random forest, based on the main attributes of contribution of ethylene yield. Finally, the actual ethylene cracking naphtha data of industry is used to verify the effectiveness and superiority of the algorithm.

Key words: algorithm, entropy, optimization, intuitionistic fuzzy, ethylene cracking

摘要:

随着乙烯裂解原料种类的日益增多,原料分析仪价格昂贵,因此根据乙烯裂解原料属性进行在线聚类,对实现乙烯收率建模,优化乙烯产率、节能减耗具有重要现实意义。为了提高原料在聚类的准确性,提出了一种基于直觉模糊集理论的核聚类算法。即在定义直觉模糊集隶属度时通过引入犹豫度来表征数据的不确定信息,同时利用直觉模糊熵对多核聚类算法的损失函数重新定义,使类簇中的数据点最优化;进一步地,使用随机森林对裂解原料属性进行特征选择,依据对乙烯产率的贡献度选取聚类的主要特征属性。最后根据实际工业裂解的石脑油数据验证了所述算法的有效性。

关键词: 算法, 熵, 优化, 直觉模糊, 乙烯裂解

CLC Number: