Research On Model Servitization Technology Of Complex System Simulation Based On Cloud Computing

Cloud computing-based complex system simulation is a new simulation mode that provides modeling services for complex system simulation by taking advantage of cloud computing such as resources sharing and strong computing power.The research of simulation model servitization is an important support for realizing this simulation mode.Current simulation model servitization technologies lack the description of the semantic correlation of models;mainly provide the local invocation interface for simulation model,so it is difficult to support the sharing of model resources under the cloud architecture.At the same time,they are difficult to select model that satisfies the requirement according to the model correlation and user's QoS(Quality of Service)preference.Therefore,the research on model servitization technology of complex system simulation based on cloud computing has an important theoretical significance and a practical value for effectively describing simulation model,realizing model resources sharing under the cloud architecture,supporting efficient search and selection of model,improving the utilization rate of simulation model,and improving the execution efficiency of complex system simulation application.Aiming at solving the problems existing in current research on model servitization technology of complex system simulation based on cloud computing,this paper carries out research focusing on the key technical issues: description of simulation model,servitization encapsulation of simulation model,semantic search and selection of simulation model.The main work and innovations are as follows:Firstly,the description of simulation model provides an important support for the discovery and invocation of model resources in the cloud environment.Current simulation model description method mainly concentrates on the characteristics of model itself which is insufficient in describing the semantic correlation of models.For this reason,this paper proposes a RDF triples-based simulation model resources description method by knowledge graph.This method summarizes the static information,correlation,dynamic function and interface information that need to be described for simulation model,and transforms the description information into four RDF triples,stores them in knowledge graph database,which effectively improves the ability to describe simulation model and its semantic correlations.Secondly,servitization encapsulation of simulation model is the premise of efficient sharing of simulation model resources under the cloud architecture.Current simulation model development specification mainly provides local invocation interface for simulation model,so it is difficult to support the sharing of model resources under the cloud architecture.For this reason,the paper proposes a lightweight container-based servitization encapsulation method for component simulation model.This method uses the network data transmission interface to encapsulate the simulation model,and realizes the communication between the model service and the user terminal through the lightweight container-based architecture and proxy model,which realizes the servicized invocation of simulation model resources under the cloud architecture.Thirdly,searching and selecting the needed simulation model is the basis of constructing complex system simulation application.Current simulation model search and selection methods are difficult to select model that satisfies the requirement according to the model correlation and user's QoS preference.For this reason,the paper proposes a knowledge graph semantic search and QoS weighted –based simulation model selection method.According to the model attributes and model correlation set by users,the initial set of simulation model that meets the conditions is found based on semantic search of knowledge graph,and the simulation model that meets the user's QoS preferences is selected based on the designed QoS weighted-based optimization selection algorithm,which supports the correlation search and customized QoS-based selection for simulation models.On the basis of above research results,the cloud computing-based complex system simulation model servitization software module is designed and implemented.The module can support user in adding simulation model description information into knowledge graph and encapsulate simulation model into shareable service.Users can search for simulation model according to model attributes and correlation,and select simulation model according to user's QoS preference,and then to invocate simulation service under the cloud architecture.