QoS Driven Cloud Manufacturing Service Evaluation And Composition Optimization In Fuzzy Environment

Cloud manufacturing is a new service-oriented networked manufacturing mode.Based on cloud computing,Internet of things,big data,artificial intelligence and other technologies,the manufacturing resources are encapsulated as cloud services,forming a cloud manufacturing service pool for unified management,allocating appropriate services according to the needs of users,and realizing the efficient use of manufacturing resources.With the increasing complexity of manufacturing tasks,cloud manufacturing platform needs to decompose the complex tasks into several sub-tasks,and match each sub-task with a cloud manufacturing service to form a service composition for users to call.A large number of cloud services with similar functional attributes but different non-functional attributes are also concentrated on the cloud manufacturing platform.Cloud manufacturing services need to be evaluated first to narrow the search scope of composition optimization.How to evaluate a large number of cloud services with similar functional attributes and select services from them to form the optimal cloud service composition is an important task for cloud manufacturing platform.And it's also the key of this thesis.At the same time,in reality,the evaluation of cloud manufacturing service is often fuzziness and uncertainty.Therefore,this thesis studies two stages include cloud manufacturing service evaluation and cloud manufacturing service composition optimization based on QoS in fuzzy environment.In the cloud manufacturing service evaluation model,based on the characteristics of cloud manufacturing,this thesis constructs a more suitable QoS evaluation criteria system for cloud manufacturing service evaluation.In view of the uncertainty of criteria in the evaluation,this thesis uses language description to evaluate,and uses interval-valued Pythagorean fuzzy set to describe the language evaluation information.Based on the interval-valued Pythagorean fuzzy set,the GRA-TOPSIS comprehensive evaluation model is constructed to evaluate the cloud manufacturing service from the two aspects of distance and similarity.Experiments show that the model can effectively recommend Top-n high-quality cloud services for each sub-task.In the cloud manufacturing service composition optimization model,based on the QoS evaluation criteria system,this thesis analyzes the calculation method of each criterion in different structures of cloud manufacturing service composition,constructs a high-dimensional multi-objective optimization model,and uses the multi-objective optimization algorithm to solve.Aiming at the practical problems in this thesis,the multi-objective optimization algorithm NSGA-? is improved.Firstly,aiming at the dimension difference of high-dimensional multi-objective optimization model,this thesis proposed the NSGA-II algorithm based on the improved ?-dominance,so that it can deal with the high-dimensional multi-objective optimization problem more efficiently;secondly,based on the coding method of the research problem,using the uniform two point crossover and adjacent mutation replace the original algorithm,so that it is more suitable for this study.Finally,based on the obtained non-dominated solution set,this thesis uses the AHP-ELECTRE ? algorithm to evaluate and select the best composition in the non-dominated solution set,which can fully consider the user's preference for the criteria and prevent the mutual compensation between the criteria,and provide the optimal cloud manufacturing service composition for users.In the end,a case of complex manufacturing task is given to illustrate the implementation process of the two-stage model,which proves the feasibility and effectiveness of the model.Through sensitivity analysis,the stability and comprehensiveness of the fuzzy evaluation model are verified.Through the comparative experiments,it is verified that the average level of solution set obtained by the optimization model proposed in this thesis is superior to that obtained by the standard NSGA-II algorithm and without primary selection.