Research On Service Configuration Method For Complex Cloud Manufacturing Tasks

With the arrival of a new round of scientific and technological revolution focusing on the application of informatization,the upgrading and transformation of informatization in traditional manufacturing industry is imperative.In order to comply with it,cloud manufacturing has emerged.The cloud manufacturing tasks represented by complex manufacturing tasks will be the key issues the cloud platform needs to solve.This thesis focused on the service configuration problem of complex cloud manufacturing tasks,and three issues that the optimal decomposition of complex cloud manufacturing tasks,screening of candidate services,and comprehensive selection of composite services in the cloud platform were studied on the basis of establishing service configuration framework for complex cloud manufacturing taslks.The main research contents are summari:zed as follows:The optimal decomposition method of complex cloud manufacturing tasks based upon spectral clustering was presented.For the complex and detachable cloud manufacturing tasks,the basic principles for the preliminary separation of tasks as well as related definitions were given.Then,based on the correlation relationships among pairwise meta-tasks,an association undirected graph was established which was then segmented optimally adopting spectral clustering algorithm.This reached the goal of task optimization decomlposition.For problems that selcecting initial clustering center tand clustering mnumber of K-means from spectral clustering algoritlhm,stochastlic serlection and "elbow method" were used to solve,which achieved rapid and reasonable decomposition of complex cloud manufacturing tasks.A candidate service screening method based upon historical data mining was proposed.Based on the QoS index constraints of meta-tasks from cormplex cloud manufacturing task,the QoS index model of sub-tasks was given,and on the basis,the methods obtaining the user's ideal demand indexes under differfent conditions were presented.Then,not only was cosine similarity used to coarsely screen the candidate services,but also the variable precision rough set mining objective weights from historical data was used to help comprehensive evaluation and selection of candidates services,which guaranteed the quality and efficiency of subsequent combined services.A complex cloud manufacturing task composition service optimization model considering user fuzzy evaluation was established.Based on the analysis of services composing process of complex cloud manufacturing tasks,a complex cloud manufacturing task composition service optimization model was constructed.Then,it was solved using the genetic algorithm which was improved by adopting adaptive chaotic mutation operator.Based on the set of solutions obtained by the improved genetic algorithm,the users' fuzzy evaluation of the candidate composition service was processed through the intuitionistic fuzzy set.Finally,combined the composition services' QoS evaluation value with fuzzy evaluation value,the comprehensive selection of candidate composition services was achieved.Prototype system for complex cloud manufacturing task service configuration was designed.After analyzing the system function modules and business process,the development of complex cloud manufacturing task prototype system was completed by using the C#programming language to develop the front-end interface of the prototype system on the Visual Studio 2013 development platform,by realizing all algorithms,programmed upon MATLAB R2014a,which were called by front-end interface to run in the prototype system background,and by using the Access database to store data.With the successful running of the prototype system,the feasibility and effectiveness of the methods proposed above were verified.