Architecture Design Of Manufacturing Execution System Based On Microservices And Research On Microservice Clustering And Selection Methods

The development and in-depth promotion and application of industrial software is the key to accelerating the Digital transformation of manufacturing industry,and also an important foundation to support the development of intelligent manufacturing.In order to achieve large-scale application of industrial software in small and medium-sized enterprises,it is necessary to quickly and low-cost meet the personalized needs of enterprises.Manufacturing execution system(MES)is a very important industrial software,which plays a core role in enterprise production management.However,traditional MES systems have poor versatility and high development costs.This article focuses on the development requirements of modularization,componentization,and platformization in MES systems,proposes a microservice based MES system architecture,and conducts research on microservice clustering and selection methods.The main content is as follows:Analyzed the functional requirements of MES,divided its business functions into microservices,and designed an overall architecture of MES system based on microservices.Compared with traditional MES system architecture,it has the characteristics of modularization and componentization;Publish MES system functional components in the form of microservices on the internet and manage them by building a microservice library,clarifying their operational mode.Users can select different functional components according to their needs;Further propose a binding mechanism between the registry and microservice library,which can effectively respond to the dynamic changes of enterprise business processes through the flexible assembly of microservice functional components,thus achieving agile development and reconstruction of MES systems.In addition,a process for selecting the optimal microservice from the microservice library is provided to meet the requirements of MES system platform development.In order to efficiently find a set of microservices that meet functional requirements from the microservice library,a microservice clustering method is proposed.This method is based on Sentence BERT and SOM models.By converting microservices into text vectors through Sentence BERT and mapping them to the output layer nodes of SOM,it can efficiently and accurately cluster a set of microservices with similar functions from a microservice library containing a large number of microservices.The validation results indicate that compared to other clustering methods,this method can effectively improve the recall rate and F-value while ensuring accuracy.For the optimal microservice selection,non functional requirements such as availability,reliability,and response time were used as quality of service(Qo S)indicators to transform it into a multi constraint single objective optimization problem based on Qo S.An improved genetic algorithm was further proposed,redesigned in coding,crossover,mutation,selection,constraint penalty,and other aspects.The stable solution of the multi constraint single objective optimization problem was achieved using the improved genetic algorithm.The validation results indicate that the proposed microservice selection method can achieve microservice optimization and has good solving speed and quality.Finally,taking a production workshop of a certain enterprise as a scenario,based on the aforementioned MES system architecture and microservice clustering and selection methods,a MES system instance development simulation was completed.The use of microservice modules,atomic microservices,and business process microservices was described and evaluated,and the results showed that the MES system based on microservice architecture can meet user needs and achieve agile development and reconstruction.The relevant work in this article will provide reference for the development of industrial software based on microservices.