Research On Key Technologies Of Mass Customization Service By Manufacturing Cloud

It is a principal pathway to deliver mass customization services based on cloud services provided by manufacturing cloud for upgrading the service mode of discrete manufacturing enterprises.However,some new characteristics,such as the large scale and autonomy of requirements,increasing volume of candidate services,dynamic nature of service processes and complexity of service synergy have significantly challenged the effective delivery of mass customization services by manufacturing cloud.For example,(1)the large scale and autonomy of requirements put new demands into the provision of customizable requirement options and efficient design of service solutions for customized requirements;(2)the large scale of requirements and increasing volume of candidate services request an efficient service composition optimization strategy and method;(3)the dynamic and demand-driven natures of service processes make service synergy more complicated and less effective.In order to deal with the above challenges,by taking the methodology of service engineering as guidance and following a clue of the service composition lifecycle,this paper presents a series of technologies to design the service system that is used for effective and efficient delivering mass customization services by manufacturing cloud.The main research contents of this thesis include the following aspects:(1)Modeling of the mass customization service system based on manufacturing cloud.According to the basic concepts of manufacturing cloud service,life-cycle of manufacturing cloud service,manufacturing cloud and service mode,the concept of the mass customization service by manufacturing cloud is presented.Through studying its characteristics,the meta-model of the mass customization service mode is developed.Then service models specifically for the three phases of the service composition lifecycle,including the large-scale demand customization and processing,service composition optimizing,service composition validation and adaptation are presented.These service models are used to identify and decompose the main issue of each phase,as well as analyze appropriate strategies.By doing so,we finally develop a system framework for the mass customization service by cloud manufacturing.(2)Mass requirement customization and service solution clustering.By taking the service value network as a bridge between customizable requirement options and service capacities,we first present the service value network-based and Quality Function Deployment(QFD)-based methods to provide customizable requirement options and support the mass customization of requirements.Then,in order to deal with the mass customized requirements,the method used to transform requirements into service solutions and cluster them is presented.This method consists of Conditional Restricted Boltzmann Machine(CRBM)models which are proposed to effectively map a customized requirement to a feasible service solution,and an improved Affinity Propagation(AP)algorithm for incrementally clustering mass service solutions and finding exemplars.(3)Manufacturing Service composition and optimal configuration based on the found exemplar of service solutions.The Bayesian Forecasting Genetic Algorithm(BFGA)is firstly presented to effectively and efficiently implement service composition optimization with mass candidate cloud services.The service space is divided into subspaces at different levels according to employing Bayesian theorem to mine the prior knowledge of candidate services from historical service records and iterative populations,which enables BF-GA to fast locate the optimum subspace and find optimal service compositions.For effectively configuring optimized service compositions to service solutions represented by the exemplar,we further develop a genetic strategy based on Posterior Probability of Service(PPS)and improve the genetic algorithm to configure optimized service compositions with the goal of cost optimization and maximum number of satisfying requirements.(4)Verification and adaptation of manufacturing service compositions based on business processes.To automatically discover the business process of services in the context of the dynamic change of service processes,the method to discover business processes of service compositions from their event logs is presented,which includes a measure that is used to identify the change point of business processes from event logs,a weighted heuristic miner for weighting the relationship of service events among different segments of change points in event logs,and a Weighted Heuristic-based Genetic Miner(WH-GA)proposed to automatically discover newly business processes of cloud services.In order to efficiently verify the larger business process of service compositions and develop adapters,we present the maximal clique based and Minimal cut based methods to automatically split the larger business process into fragments.Pi-calculate based method is developed to concurrently and efficiently verify fragments of business process and build adapters.Finally,a services platform of life-cycle of manufacturing service for household appliances is designed based on the demands of the system for delivering mass customization services by manufacturing cloud in a home appliance company.The proposed technologies are verified according to a real-world case study.