Design And Implementation Of An Interactive Data Analysis Platform For Manufacturing

With the advent of the era of big data,manufacturing industry has accumulated a large amount of data in the process of production,sales and after-sales service.Many enterprises hope to improve product quality and solve the existing problems in the process of production and management by mining the value of these data,so as to convert the value of data into actual profits.However,the type of the data analysis tasks and the techniques used by different departments make a big difference,so it is difficult to realize the unified management and collaboration of data analysis tools in enterprises.According to the above problems,entrusted by the cooperative enterprise,an interactive data analysis platform for manufacturing based on B/S structure is implemented,the system supports multi-language mixed interactive interpreter(Read-Evaluation-Print-Loop,REPL).According to the type of the code written by users and the resource scheduling strategy designed,system will execute the code by calling the remote interpreter service.By using the system,users can choose the language they are most familiar with or best suited to the task for data analysis and collaboration.The main work of the thesis is as follows:(1)By analyzing the interpreters of the programming languages selected,a set of unified abstract interfaces which are used to control the life cycle of different interpreters are put forward.Then the interfaces are implemented for different interpreters according to their nature.The dependencies which are used to establish data connections are loaded dynamically by importing a local resource or downloading from MVN repository.(2)A unified JSON format is designed to package the information related to the process of coding,running,result displaying,and the JSON statement is wrote into a file.The files saved in server are synchronized to the distributed file system at the appropriate time.Due to the feature that the most core part of the file is the code written by users,JGit is used to manage the version of the files,in this way,it is convenient for users to backtrack the historical version of the code.(3)By controlling the number of JVM launched and interpreter groups created during the execution of the code written by users,several different code running strategies which can be chosen by users after considering all the factors,such as server resource occupancy,the robustness of the interpreter process,and the ease of object exchange in the code,are provided to the users.(4)Requirement analysis is completed under the guidance of the requirements provided by the partner,the functional requirements and performance requirements of the system are confirmed at this stage.Then the function modules including interpreter module,resource scheduling module,code management module,visual module.Design and implement related modules are implemented.(5)Deploy the system in the laboratory environment,each function of the system are tested according to the use case table,and the preset performance requirements are verified.Through the system test,the interactive data analysis platform designed and implemented in this thesis is concluded that the logic processing of each function of the system is correct,each interpreter can be used normally,and meets the performance requirements of the enterprise for the system.To sum up,the interactive data analysis platform for manufacturing can support users to select the most appropriate language or technology according to the task and conduct complex data analysis,it is of great significance to promote manufacturing enterprises to mine the value of data generated during enterprise operation.