Abstract: Manufacturing Execution System (MES) is a key component of integrated automation systems in process industries. Data integration is fundamental to MES. Achieving data integration in heterogeneous environments such as heterogeneous networks, operating systems, and databases is a challenge in system integration. This paper proposes an integration model based on a real-time data platform and applies it to practical system integration, achieving satisfactory results. Keywords: Manufacturing Execution System, Real-time Data Platform, Data Integration Currently, integrated automation systems in process industries consist of three supporting systems: Enterprise Resource Planning (ERP), Manufacturing Execution System (MES), and Process Control System (PCS). ERP is responsible for business decisions and production planning; MES is responsible for production scheduling and system process optimization; and PCS is responsible for production process control. Most process industries have underlying control systems, and many companies have implemented ERP systems. However, effective communication between the management and control layers remains lacking. MES is a key component of integrated automation systems in process industries, playing a crucial role in connecting the upper and lower layers. It serves as an information integration bridge between enterprise production and management activities, making it essential to plan data for the entire production execution process from a holistic enterprise perspective. Data integration is fundamental to the functional subsystems of MES. Based on the characteristics and importance of the data integration module, this paper proposes a data integration method based on a data platform. I. Importance and Challenges of Data Integration in MES 1. Importance of MES Data Integration MES collects production operation data, integrates raw material and product storage data, and integrates equipment status information. It then performs comprehensive processing such as merging, summarizing, standardizing, comparing, and analyzing this information. On the one hand, it provides a basis for production planning and scheduling; on the other hand, it provides timely, reliable, and accurate production and operation decision-making reference information for ERP. Data integration is the foundation of MES implementation. It summarizes and processes data such as production operation, product quality, raw material and product transportation, and energy consumption at the PCS layer, integrating real-time information from the lower-level production process with various information from upper-level enterprise resource management at the MES layer. Through information integration, it forms scheduling or instructions such as optimized control, optimized scheduling, and optimized decision-making. Simultaneously, the data integration module is also responsible for transmitting some data (such as optimized values ​​and setpoints) from the upper-level system to the PCS. 2. Challenges of MES Data Integration Enterprise production processes are complex, with diverse data sources, varied data acquisition and storage methods, and closed underlying control systems. The networks, systems, and databases used also differ significantly. Achieving comprehensive integration of data from heterogeneous networks, systems, and databases is the biggest challenge in MES data integration. II. Data Platform Characteristics and Functions 1. Introduction of the Data Platform Traditional computer application systems are developed and run directly on top of operating systems, networks, and database systems. Because these application systems are highly dependent on the underlying support environment, they lack good openness and portability with the integration environment. Furthermore, it is difficult to solve the integration problems of heterogeneous information and environments, affecting the overall efficiency of the application system. Therefore, integration based on a data platform was proposed. 2. Data Platform Structure The Real-Time Data Platform (RTDP) realizes functions such as real-time data acquisition, management, historical archiving, maintenance, writing, alarm generation, event logging, and time synchronization. Simultaneously, as a runtime platform, it provides real-time/historical data services to various application software running on it. Its overall structure is shown in Figure I. [IMG=Research and Application of Data Integration in MES of Process Industry]/uploadpic/THESIS/2008/1/2008012409535797502C.jpg[/IMG] 3 Main functions of the data platform (1) Data communication function: mainly communicates with monitoring software, application programs and various databases through OPC, DDE, ODBC interfaces to read data from the lower layer into the real-time data platform; and sends command information from the upper layer back to the corresponding system through these interfaces. (2) Data input and output processing: processes the data entering the real-time data platform. Such as data format conversion, range conversion, alarm setting, data statistics, and historical data archiving. (3) Network monitoring and reconnection: to ensure normal data transmission, the real-time data platform should have the functions of easy network monitoring and automatic reconnection. When the network fails, it can promptly prompt or alarm. When the network recovers, it can detect and automatically reconnect. (4) Online maintenance, configuration and query: To ensure the continuity of data, the real-time data platform should provide online maintenance functions; it can be configured according to different user requirements; and it provides query functions. (5) Data security guarantee: Because the data platform is connected to the lower-level control system and is related to the enterprise's production information, and may also be connected to the Internet, a security mechanism must be established to absolutely prevent unauthorized operations and ensure the security of the entire information system. III. Structural Model of Data Integration Based on Real-Time Data Platform Considering the characteristics of actual data integration, a data integration model as shown in Figure 2 is proposed. The right side shows the three-level division of the integration model, and the left side shows the correspondence between the three-level model of enterprise information integration. [IMG=Data Integration Model]/uploadpic/THESIS/2008/1/2008012409540482758H.jpg[/IMG] The entire structure is mainly divided into three levels: (1) The bottom layer is the environment layer, which refers to the environment based on various fieldbuses, DCS, various dedicated control networks and industrial Ethernet networks, various operating systems, and various database systems. It can provide real-time and non-real-time data services to the upper layer, and provide various data interfaces (such as OPC, ODBC, DDE, CORBA, etc.), and is the intermediary for information interaction between the data platform and the lowest-level equipment. (2) The middle layer is the data platform layer. On the one hand, it communicates with the lower layer using the corresponding methods according to the interface provided by the lower layer, and completes data integration and management, data services and network communication and other service functions. When there are scheduling, optimization and other instructions from the upper layer, it converts them into corresponding data and sends them to the corresponding system; on the other hand, it provides some common integration support services for the upper layer applications, and forms the data from the lower layer into a unified format, providing real-time and non-real-time information for other modules of the MES layer (such as real-time data monitoring, process simulation, production scheduling, data analysis, equipment management, inventory tracking, quality control, optimization control, etc.) and the upper-level ERP. (3) The upper layer is the application layer, which includes other functions in the MES except for the data acquisition function, as well as some functions related to ERP. The characteristics of data integration based on the data platform are as follows: (1) Data sharing and application integration between different application systems can be realized through the data platform. The data platform provides a unified integration environment for other modules in MES and the upper-level ERP, facilitating application development and integration. (2) Openness. Standard interfaces such as DDE, ODBC, Web, and OPC enable MES to connect with other applications and the Internet/Intranet, providing unified system resources and shared resources for application development. (3) Transparency. Based on platform data integration, development work is simplified. Developers can directly develop for the platform without having to consider the structure or communication mode of the lower-level data. All of these tasks are completed by the data platform. (4) The data platform provides functions such as shared data management, data services, and network communication, and supports multiple application services at the same time. This can shorten the development cycle of information systems, improve development efficiency, and more effectively realize the integration of the three-tier structure of the enterprise. IV. Integration Case Analysis The following analysis will start from the current situation of a coking gasification plant and analyze how to use a real-time data platform to achieve the integration goal. 1. Current Status of the System The plant has eight workshop-level production monitoring systems (hereinafter referred to as subsystems), including the fully automatic coal preparation system, coking production monitoring system, gas pressure conveying station production monitoring system, boiler and steam turbine generator production monitoring system, gas source peak shaving station production monitoring system, biochemical station production monitoring system, gas blower production monitoring system, and power supply and distribution integrated automation system; three weighing systems, including truck scale, rail scale and coal tower scale; and two non-continuous systems, namely coke oven number identification and coke oven temperature measurement. In order to make full use of existing resources and achieve resource sharing, it is necessary to integrate the information of the 13 subsystems, establish a production management network covering all monitoring systems in the plant, complete the construction and development of the production management scheduling system (hereinafter referred to as the scheduling system), and realize the centralized monitoring and management of the plant's production data by the scheduling center. Since the entire plant has developed over several decades, there are significant differences between the old and new systems, mainly reflected in the following aspects. (1) Hardware Platform. There are several types of computers, including workstations, personal computers, industrial control computers, and servers; in terms of communication networks, there are serial ports, fieldbuses, industrial control networks, and local area networks. (2) Operating systems. There are four types: Windows 98, Windows NT, Windows 2000, and Windows XP. (3) Databases. There are text databases (Axt), Access, SQL Server 7.0, SQL Server 2000, Paradox 5.0, and Excel spreadsheets. (4) Development tools: assembly language, Turbo C, VC, VB, Delphi, etc. Therefore, the difficulty in this system integration is how to unify the information based on various heterogeneous environments (heterogeneous networks, heterogeneous operating systems, heterogeneous databases, etc.) to meet the integration requirements. 2 System integration method According to the characteristics of the system to be integrated, we adopted the integration method based on the above integration model. The network structure adopts a local area network centered on the switch. The specific information exchange between each part is as follows: (1) The eight monitoring systems, namely the fully automatic coal preparation system, coking production monitoring system, gas pressure conveying station production monitoring system, boiler and steam turbine generator production monitoring system, gas source peak shaving station production monitoring system, biochemical station production monitoring system, gas blower production monitoring system, and power supply and distribution integrated automation system, all provide OPC servers and have standard OPC interfaces, and communicate with the real-time data platform through the OPC interface. (2) The three weighing systems, namely the truck scale, rail scale and coal tower scale, as well as the two systems, namely the coke oven number identification and coke oven temperature measurement, do not have continuous data, and all provide databases, and the corresponding data can be sent to the data platform through the ODBC interface. (3) Some data can only be entered manually because it has not yet entered the system. These data can be directly entered into the data platform through the application interface. (4) In this integration, we adopted the integrated automation system Was developed by the Integrated Automation Center of the Institute of Automation, Chinese Academy of Sciences, which is based on multiple fieldbus and Internet. Its core is the real-time data platform RTDB. This system provides multiple communication interfaces and can perform operations such as data preprocessing, alarming, publishing, archiving, querying, simulation, and analysis. On this basis, data integration is realized. (5) Based on the above solution, we adopt the integration structure shown in Figure 3. [IMG=Research and Application of Data Integration in MES of Process Industry]/uploadpic/THESIS/2008/1/2008012409541237704A.jpg[/IMG] · The network adopts a structure centered on a 100Mbps Ethernet switch. Each subsystem is connected to the real-time data platform through the switch. · The real-time data platform collects data from the lower layer and, after processing, is called by applications at the upper layer (including querying, analysis, simulation, etc.). On the other hand, it sends information from the upper layer to the corresponding system at the lower layer after appropriate transformation. · Considering the load, historical data is managed by a dedicated historical data server, and the real-time data platform is only responsible for saving the data to the server. V. Conclusion The real-time data platform technology provides great convenience for enterprise system integration in heterogeneous environments. The integration model proposed in this paper is based on the data platform technology. It utilizes various interfaces provided by the real-time data platform itself to interact with heterogeneous data sources at the lower level, while providing a unified data model for various applications at the upper level, thus facilitating information integration across the entire enterprise. Since its deployment, the system has achieved excellent results and brought significant benefits to the company.