Data Mining-based Approaches To Multilayer Process Goose Queue (PGQ) Formation Adjustments

The Process Goose Queue(PGQ)method provides a novel idea for the decomposition and optimization of industrial systems.At present,strict mathematical models of systems are used for PGQ optimizations and formation adjustments,which demand for a higher accuracy of the system model,discouraging practical engineering applications.In this context,this thesis proposes a novel method of Process Goose Queue formation adjustments combining data mining technologies.The main research contents and achievements are presented as follows.1.An in-depth study of Process Goose Queue(PGQ)method and its formation adjustment mechanism are carried out along with a study of fuzzy clustering and fuzzy association rules algorithm,laying the foundations for the further research.2.The target cluster analysis is performed on the goose position variable aiming at finding out the goose position distributions around operating points.Fuzzy clustering algorithms are used to divide the state goose and manipulated goose into fuzzy clusters,specifying the fuzzy parameter distributions under different fuzzy linguistic values.Subsequently,the divided fuzzy sets are used for mining the fuzzy association rules of PGQ.Through an example,the approach to fuzzy associations rule query table of the PGQ is given,by which,the intervene for the PGQ shift can be realized.3.A cement production process is used as the research application to establish a multilayer PGQ structure with the temperature control of the precalciner as the goal.According to the production historical data,the process production data are fuzzy clustered and divided into various fuzzy language values for each goose.By fuzzy association rules mining,the resultant fuzzy association rules query table is used to guide the goose formation adjustments.