Research On Dynamic Alliance Oriented Manufacturing Service Composition

Based on social cyber-physical system(SCPS),big data and cloud computing technology,wisdom manufacturing model can encapsulate and visualize multi-level manufacturing resource into social manufacturing service based on cloud platform,automatically organize appropriate granularity resources according to the business requirements,and flexibly construct dynamic alliance to seize market opportunities and achieve the main objectives,such as large-scale collaboration,comprehensive perception,real time decision-making,social resources utilization and meeting the personalized consumers needs.As multi-granularity service units become the basic components of dynamic alliance,their individual function,service QoS and collaboration relationship directly determine the alliance efficiency and business execution performance in common.How to select high quality service units based on task requirements on cloud platform,and how to adjust the service execution activities adaptively are becoming the key issues that restrict the application collaboration efficiency of wisdom manufacturing.Therefore,it is an important organization model to construct dynamic alliance through multi-resource orchestration.Taking service organization of the dynamic alliance as the research topic,this dissertation integrates the theories of social network,service collaboration and situation awareness,and proposes a service selection method based on synergy capacity,a two-sided service matching method based on competition and synergy effect,and a service self-adaptive adjustment decision method based on situation awareness,in view of the new characteristics of wisdom manufacturing,such as social interconnection of manufacturing service,complex association of customized tasks and dynamic time-varying of the environment.Then,a new dynamic alliance oriented manufacturing service composition method system is perfectly constructed,to effectively deal with the novel challenges faced in the construction process of dynamic alliance under the new environment.The main contents and contributions of our research are descried as follows:(1)Manufacturing business process-oriented dynamic service alliance model In wisdom manufacturing environment,small-medium enterprises can encapsulate their equipment,production lines,plants and factories as independent functional units to participate in the collaboration manufacturing process in the manner of virtualized services on the cloud platform.Muti-granularity manufacturing service units become independent subjects involved in manufacturing value network,and profoundly affect the organization and operation mode of the manufacturing resource,resulting in the change of the structure and behavior of the previous enterprise alliance.Dynamic service alliance based on multiple manufacturing units cooperation has become a novel resource integration and collaboration model to construct complex business processes and meet individual consumers demands.Therefore,based on the analysis of manufacturing mode innovation caused by new emerging information technologies,a wisdom manufacturing cloud platform architecture that includes resource layer,perception layer,service layer,value chain layer,application layer and key support technology,is firstly proposed to provide basic information environment and technical support for multi-granularity service organization of dynamic alliance.According to the analysis of personalized requirements and muti-level system of manufacturing service units,a business process-oriented dynamic service alliance model is studied.The basic characteristics of the alliance are described from three dimensions of service relationship,business logic,and network structure.The formation and construction of task-driven dynamic alliance in wisdom manufacturing system is an iterative complex space-time optimized process.From the perspective of the full life cycle,dynamic alliance oriented social service composition process can be divided into four stages.Firstly,the task is modeled using Petri-net method,and task decomposition and synthesis are performed based on the matching of task and service characteristics;Secondly,the granular computing and semantic matching methods are used to match the corresponding manufacturing services for each sub-task in different level;Moreover,QoS,collaboration capacity and both parties' satisfaction are applied as the significant evaluation indexes to select the optimal service combination from a large number of candidate solutions.Finally,through perceiving and discovering abnormal events in service execution process,adaptive adjustment strategies are triggered for service re-selection and scheduling in dynamic environment.A synchronous and collaborative optimization mind is proposed in the whole service composition process.Then,service selection and adaptive decision-making approach become two core problems for dynamic alliance that we focus on in our research.(2)Manufacturing service selection method based on collaboration capacity In manufacturing service socialized environment,multiple service units with different functions are required to collaborate and interact with each other to complete complex manufacturing tasks together.The individual capabilities and collaborative relationships of these service units jointly determine the task completion performance.It is important to consider collaboration capacity as a critical competitive factor when business work-flow oriented dynamic alliance is orchestrated with online services.Hence,integrating social network,service computing and synergy theory,the paper attempts to proposed a service selection method based on weighted synergy network,where synergy effect is measured by aggregating five social relationships strength.Firstly,service social network is extracted and five types of relationships strengths are calculated.Then,the service synergy effect can be obtained through the weighted aggregation,and service weighted synergy network is naturally constructed.In order to maximize the overall QoS and synergy effect,a muti-objective optimization model for manufacturing service selection is put forward.Applying two-way learning,population update and group interaction based speed update mechanism,a improved GSA algorithm is built to solve the model.Through the simulation experiment of intelligent automobile manufacturing,the validity of our model and algorithm is verified,and the results show that our approach can significantly improve the actual performance of service composition.Under complex task scenario,there exist muti-dimensional materials delivery,information and knowledge interactions relations among manufacturing sub-tasks.And service units matched to sub-tasks must have consistent collaboration capacity according to the synergy requirements.The mismatch of task relations and service collaboration relations will bring out additional coordination costs,reduce the efficiency of resource allocation,and lead to non-optimal service performance.Therefore,this paper further considers the influence of the consistency matching between hybrid task network and service synergy network,and proposes a service adaptation selection method based on task relationship constraints.Based on hybrid task network built above,task importance is considered and a weighted aggregation method for service competence is given based on task weight.According to service synergy requirements constrained by task correlations,the calculation methods for horizontal and vertical collaboration levels among services are presented.Maximizing service competence and synergy effect as objectives,a mathematical optimization model for manufacturing service selection is constructed.The Pareto optimal solution set is obtained by using the improved non-dominated particle swarm optimization algorithm,and the optimal composition scheme is acquired based on the weighted TOPSIS evaluation approach.(3)Two-sided matching decision methods for manufacturing services based on competition and synergy effectsCloud platform provides a transaction and cooperation place for both task requester and service providers,where related business activities such as registration,quality assessment,negotiation and free trade can be carried out in time,and it owns the typical characteristics of the bilateral market.Hence,the individual interest and market positioning of supply-demand two sides must be integrally considered when service combinations occur on the platform.Two-sided matching decision method is applied in service selection process to maximize the satisfaction of both parties.Aiming at the collaboration and adaptive learning characteristics of intelligent services in cloud,a two-sided matching decision approach for manufacturing service considering learning and synergy effects is proposed.Due to the dynamics and repeatability of cloud service transactions,service units involved in multiple tasks can accumulate their experience and knowledge to improve its service quality.A calculation method of dynamic capability is put forward based on learning effect model,and mutual satisfaction for task and service can be aggregated by applying expected utility theory.Meanwhile,synergy network is used to describe service social relationships,and collaboration satisfaction among services can be measured through synergy effect based on social network theory.Thus,maximizing the tasks'satisfaction,services'satisfaction and inter-service collaborative satisfaction as objective,a one to one two-sided matching muti-objective model is constructed considering the influence of learning and synergy effect on task execution performance.Through automobile cloud manufacturing case study,the optimal matching scheme is obtained to verify the effectiveness and advantages of our model.Moreover,both supply-demand subjects are not independent of each other,but in a specific circle or social network.Their value judgment and satisfaction are affected by the neighbors'action strategy and decision behavior of other members with social relationships(competition,conflict,cooperation,and friendship)in the network,enhancing the complexity of cloud service composition problems.Further considering the task competition relationship fighting for market share and quality resources,a one-to-many service two-sided matching method based on competition and synergy effects is proposed,based on the impact analysis of task competition network and service collaboration network on the subject satisfaction in service matching process.Applying the expected utility theory to calculate the satisfaction of both parties,the aggregation methods of satisfaction between tasks based on competition relationship and satisfaction between services based on social network are raised respectively.To maximize the task satisfaction,service satisfaction,satisfaction among tasks and inter-service satisfaction as the objective,a two-sided matching muti-objective optimization model is built,the best solution is acquired by using the improved non-dominated PSO and weighted TOPSIS.Compared with the traditional,only considering task competition relation and only considering service synergy effect three models in the experiment,the results validate the effectiveness and advantages of our model and algorithm.(4)Adaptive adjustment decision-making method for manufacturing composite service based on situation awarenessDue to the open business requirements and dynamic operating environment faced by the composite service,such as the changes of consumer demand,service joining and leaving,manufacturing condition varying,service failure and business process anomaly may occur naturally.Hence,applying intelligent Agent technology and situation awareness theory into the whole life cycle of service composition,an adaptive adjustment decision method based on situation awareness is proposed to identify and deal with abnormal events in manufacturing process in time.First of all,from the perspective of individual smart manufacturing unit at the bottom,an adaptive adjustment system framework for manufacturing services is constructed.Aiming at the impact of manufacturing environment and context constraints on the comprehension of the objects state event,an adaptive situation recognition method based on complex event processing is proposed.By constructing the multi-level event model and the event new operator constrained by context factors,the manufacturing situation calculus process based on event aggregation and the real-time situation recognition algorithm based on the improved hybrid clustering are proposed.Finally using 4 real data sets and 1 simulation data set of manufacturing process,our experiment results have verified the validity and adaption of the proposed model and method,and discover that context factors can significantly improve the accuracy of event judgment and situation recognition in manufacturing process.After identifying the current complex manufacturing situation,the smart service unit applies the existing decision rules to take the most reasonable activities based on the subject status of its own role,resources and knowledge.The same service unit may adopt different actions in response to the same situation in various subject states.Hence,according to the changing application scenarios,two adaptive real-time decision-making mechanisms including situation awareness and Subject-Situation-Action(SSA)mechanisms are designed respectively,fully considering the impact of subject status and context factors on event understanding and decision process.Meanwhile,the decision experiment environment is built by muti-agent technology and association rule method is used to mine decision rule and knowledge.Through the experiment of automobile part manufacturing cloud service,the performance of two kinds of mechanisms in different scenes is verified,and the advantages of situation awareness and SSA mechanism under dynamic environment are proved in terms of decision accuracy,time and service QoS.