KR20120132878A - An Architecture of Real-time, Historical Database System for Industrial Process Control and Monitoring - Google Patents

Abstract

PURPOSE: A structure of a process control database system including a characteristic of process control transaction is provided to increase efficiency of storage by giving a second chance when determining which data is to be stored subsequently in SDT(Swing Door Trending). CONSTITUTION: A programming API(Application Program Interface) layer is comprised of a light weight API and query based API. A data model layer is comprised of RBTDM(Relational Based Tag Data Model). A data compression layer is composed of SDT-SC(Swing Door Trending with Second Chance). A storage layer is composed of MLSM(Multilayered Storage Model), MRMRF(Multi Replication with Multi Resolution Filter) and BWA(Bulk Write with Asynchronous). The RBTDM in the data model layer is expanded to process data in a relational data model such as ORDB(Object Relational Database).

Description

An Architecture of Real-time, Historical Database System for Industrial Process Control and Monitoring

The present invention relates to the structure of a process database, and to derive the characteristics of a process control transaction and analyze the existing process database to derive the requirements that the process database should have, and to a structure of a new process database that can reflect this well. will be.

1 illustrates a typical process control system architecture. Regarding the structure of the process control system, the process control system acquires data using a sensor unit which measures data such as temperature, pressure, voltage, etc., a programmable logic controller (PLC), a distributed communication system (DCS), a remote terminal unit (RTU), and the like. It consists of an input unit, a process database unit for storing the obtained information, and a HMI (Human Machine Interface) unit for the driver to detect and respond to abnormal conditions. The process database must efficiently manage the current, recent changes, and the entire historical data of the process data.

In the process control transaction, the characteristics of the process data compared to general data are shown in Table 1. The biggest feature is that the process data has a time continuity within 1 second. If one point is 10bytes and is stored in 1 second period, more than 31M should be stored.

Process Data Features ITEM Conventional Data Process Data Unit Record Tag, Point Complex Primitive, Complex Primitive Volume Various Massive Type Discrete Continuous Priority Even Depend on time

The characteristics of the process control transaction for processing such process data are shown in Table 2.

The Features of Process Control Transaction ITEM Conventional Data Process Data Predictable Non Predictable Predictable Real time Non Real time Real time Operation Write many, Read many Write One, Read Many ACID Tight ACID Loose ACID

Process database processes data in various ways such as self-developed, historian, memory DBMS, general purpose disk DBMS, etc. according to process data processing requirements of process control system. Historian is widely used for high performance and large data storage, and Memory DB which supports relational model is also used for real time data processing.

Regarding the system structure of the historian and the memory DB, FIG. 2 shows the system structure of the hyper historian, which is one of the historians. The main component, Real-time Data Logger, collects process data, provides information to MS-SQL server, and provides interface with various solutions of HMI.

3 shows McObject's ExtremeDB system architecture supporting a relational data model. Its main feature is to provide high performance by providing various programming APIs according to application and processing data in memory with DB memory pool.

   As mentioned above, the process database is mostly relational DB using Historian and memory technology. However, each historian and relational DB has a trade-off between performance and flexibility. However, with the development of H / W and SW technology, a process database with high performance needs and flexibility is required. To this end, according to the present invention, it is an object of the present invention to efficiently manage a data model having flexibility, an improvement of a compression scheme, and an importance of stored data.

According to a feature of the present invention for achieving the above object, the structure of the process database,

Programming API Layer consisting of Light weight API and Query based API,

Data Model Layer composed of RBTDM,

Data Compression Layer composed of SDT-SC,

The configuration features include storage layers consisting of MLSM, MRMRF, and BWA.

In the Data Model Layer, RBTDM is configured to be extended to process process data in a relational data model such as ORDB (Object Relational Database).

In the Data Compression Layer, SDT-SC improves the storage space by giving another chance when the next data to be stored is determined in the swing door trending (SDT) [7], which is widely used as a data compression technique. .

In the storage layer, the MLSM divides the data to be stored into current, trend, and total, and applies a data compression method suitable for each, and improves the writing and reading of the data.

The MRMRF adjusts the amount of data to be copied according to each data request cycle and device of the HMI. The BWA is characterized by splitting, sending in bulk, and asynchronously recording the tasks of the target to be shared among clients.

  According to the process database structure of the present invention, the process database structure includes a programming API layer, a data model layer, a data compression layer, and a storage layer. In the Data Model Layer, RBTDM can be extended to process process data in relational data models such as ORDB (Object Relational Database) .In the Data Compression Layer, SDT-SC is a swing door trending method that is widely used as a data compression technique. ), When the next data to be stored is decided, the storage space can be more efficient by giving another chance, and in the Storage Layer, MLSM divides the data to be stored into current, trend, and overall data compression methods. It can improve the writing and reading of data.

1 shows a typical process control system structure.
2 shows a system structure of Hyper Historian, which is one of Historian.
3 shows McObject's ExtremeDB system architecture supporting a relational data model.
4 shows two important measurement elements of the Process Database.
5 shows the overall architecture of the new process database.

Hereinafter, with reference to the accompanying drawings, it will be described in detail the characteristics of the structure of the process control database system of the present invention.

The main requirements to be supported in the new process database are:

Regarding flexible data model support, not only tag-based data models but also relational data models for general data processing are needed to efficiently perform real-time information analysis.

In addition, with regard to improving storage efficiency through improvements in compression techniques, a compression scheme suitable for both analog and digital data should be considered.

In addition, with regard to storage management according to the importance of data, the closer to the current time, the more important and real time is when the importance of the data is compared with time from present to past. Storage management considering this is necessary.

Regarding the improvement in the write performance in consideration of the multi-core or the multi-processor, the HW equipped with the multi-core and the multi-processor has become common. To improve the write performance of process data, we need to consider the characteristics of recent HW.

The overall architecture of the new process database is shown in FIG. 5 and consists of a programming API layer, a data model layer, a data compression layer, and a storage layer. In the Data Model Layer, RBTDM is an extended data model for processing process data in relational data models such as ORDB (Object Relational Database). In the Data Compression Layer, SDT-SC improves the storage space by giving another chance when the next data to be stored is decided in the swing door trending [7] [7], which is widely used as a data compression technique. In the storage layer, MLSM classifies data to be stored in current, trend, and total, and applies data compression method suitable for each, and improves data writing and reading. MRMRF adjusts the amount of data to be replicated according to each data request cycle and device of HMI. BWA splits, transfers in bulk, and records asynchronously so that the client's work can be shared by the client.

Claims (5)

Programming API Layer consisting of Light weight API and Query based API,
Data Model Layer composed of RBTDM,
Data Compression Layer composed of SDT-SC,
Process control database system whose configuration features include storage layers consisting of MLSM, MRMRF, and BWA In the Data Model Layer, RBTDM is a process control database system, characterized in that configured to be extended to process the process data in a relational data model, such as ORDB (Object Relational Database). In the Data Compression Layer, SDT-SC improves the storage space by giving a second chance when the next data to be stored is determined in the swing door trending (SDT) [7], which is widely used as a data compression technique. Process control database system.  In the storage layer, the MLSM divides the data to be stored into current, trending, and total data, applies a data compression method suitable for each, and improves the writing and reading of the data. The MRMRF adjusts the amount of data to be copied according to each data request cycle and device of the HMI. BWA is a process control database system that divides, transfers in bulk, and records Asynchronous the tasks of the objects to be recorded so that they can be shared among clients.

Images