Research And Application On Distributed Real-time Database In Process Industry

With the optimization and integration of industrial structures,the production scale of process industry enterprise is continuously enlarged.Due to the insufficient data throughput,high costs,low fault-tolerant ability,low scalability and other problems,traditional centralized real-time database cannot meet the practical requirements.To solve the above problems,this dissertation combines the design philosophy of large-scale distributed systems with real-time and reliability techniques,establishing a service model of industrial distributed real-time database.Based on this model,dynamic index mechanism for massive data,modelling and parallel processing of transactions,real-time network communications with high availability,and other related technologies are studied.Then the researches are verified in practical projects.The theoretical researches and technical applications are detailed as follows:(1)On the basis of the in-depth analysis of requirements.from application scenarios in process industry,the service model of distributed real-time database is proposed as the fundamental architecture of the whole system.Meanwhile,distributed self-organizing theory of real-time database is proposed based on centralized structure and master-slave mechanism.The above theory includes master-slave nodes election strategy based on Raft protocol,service discovery model,and service deploy strategy,finally implementing the self-organization of real-time database on both service level and system level.(2)A dynamic global index mechanism based on plant logical model,as well as a local index mechanism based on two-level paging and multi-version management,is proposed according to the storage characteristics in process industry.To meet the requirements of reliability and scalability of the storage system,data partition and sharding strategy,solutions for node insertion and deletion,multiple copy mechanism,and archive file compression are given,solving the problem of data fitting error in abnormal situations and high disk space occupation by historical data.(3)A transaction model of distributed real-time database is built by analyzing the characteristics of database requests.Based on this model,the distributed parallel query model and distributed parallel computing service are researched and implemented to solve the transaction decomposition,parallel scheduling,resource management and other problems in distributed transaction processes.In addition,distributed real-time mixed concurrency control algorithm is proposed,which utilizes the non-blocking characteristic of multi-version concurrency control and optimistic concurrency control,thus optimizing the real-time performance of distributed transaction processes.(4)The hierarchical structure of real-time network communication service with high availability is proposed,in which the redundant connection session,data-forwarding channel,and network service self-organization are utilized to improve the availability of network in process industry.Meanwhile,time synchronization mechanism for industrial network is designed to solve the problem of network delay jitter and single point failure problems.According to the requirements of network information security,data encryption solutions for network transmission and user identity authentication mechanism based on digital certificate are proposed.The research results have efficiently solved the fundamental architecture,large capacity,real-time performance,availability,concurrent access,and other technical difficulties.Through system tests and practical application cases,the correctness and performance of the distributed real-time database are verified,showing that the research results can meet the demands of large-scale process industry for data services.