Data conversion system using on-premises generative ai and heterogeneous PLC and edge platform-linked cloud system using same

Abstract

The present invention relates to a data conversion system using an on-premises generative AI and a heterogeneous PLC and edge platform-linked cloud system using same, and, more specifically, to a data conversion system for diagnosing the state of a production facility. The data conversion system comprises: an edge platform; a server data cloud that is data-linked with the edge platform; and an AI server for generating diagnostic data for the state of the production facility by using a learning model, wherein the server data cloud includes an on-premises generative AI for receiving a user command input and processing to output data so as to be applicable to the learning model to be trained by a user.

Description

A data conversion system using on-premises generative AI and a cloud system that links heterogeneous PLCs and edge platforms.

[0001] The present invention relates to a data conversion system using on-premise generative AI and a cloud system that links heterogeneous PLCs and edge platforms using the same.

[0002] In today's society, where social and economic situations are rapidly changing, the smart factory and smart manufacturing have emerged as answers to the question of how the manufacturing industry should adapt. A smart factory is not limited to simple automation of production, such as an unmanned factory through automation that was dreamed of in the past. To be called a smart factory, automation must be achieved in all areas, from product development to ordering and production, inspection, and inventory management.

[0003] And this smart factoryization is developing in the direction of maximizing management capabilities in the manufacturing industry, making it possible to transition from the existing mass production of small varieties to small-batch production of many varieties and to product personalization. This makes it possible to produce and supply bespoke products, which have recently become a trend. In particular, with the non-face-to-face environment approaching as the new normal due to the novel coronavirus infection (COVID-19) and the emergence of new initiatives such as the 4th Industrial Revolution and the metaverse, this trend is expected to accelerate further.

[0004] The reason why the mass production system of small varieties was popular in the past was because it could provide universal and uniform quality products to everyone while significantly reducing costs. However, it had limitations in that its goal was to provide products that the majority could accept, rather than to satisfy everyone.

[0005] While this pod system is advantageous for lowering prices, it's difficult to respond flexibly to environmental changes. For example, if a product is developed but cannot be produced in quantities exceeding a certain level, profitability will be unattainable. Furthermore, producing new products requires extensive modifications to existing production lines.

[0006] Additionally, in this process, producing different products requires a significant amount of time and money, starting with product development, market analysis, supply chain restructuring, and production line modifications conducted by skilled experts.

[0007] Smart factories emerged as a solution to these existing problems. Their goal is to build a flexible production system by sharing data in real time through automation, optimization, and intelligence across all processes, from product development to mass production, from market demand forecasting to customer orders, product production, and finished product shipment.

[0008] Smart factories are broadly divided into three areas: application systems, control automation, and field automation. To implement these, they must be equipped with foundational technologies such as cloud computing, cyber-physical systems (CPS), and artificial intelligence (AI).

[0009] Application systems cover everything from product development to SCM (Supply Chain Management), production management, inventory management, and even business processes and information systems (ERP, MES, CRM, KMS, etc.). Control automation is the area that collects and controls data on production line equipment and robots, while field automation is the area that manages the production process through direct automation of the product manufacturing process.

[0010] Furthermore, by establishing a foundation for collecting, managing, and analyzing all data through the cloud, and sharing this data in real time, more efficient and organic processes can be implemented. Through this, simulation, management, and control technologies such as CPS, as well as analysis and learning capabilities using AI, a foundation can be established for flexibly responding to changes in the manufacturing environment in real time.

[0011] Of course, a smart factory requires a wider range of elements and technologies to be applied appropriately. For example, machine vision, various sensors, collaborative robots, security, and wired / wireless network technologies must be closely interconnected and managed.

[0012] Recently, due to the development of various electronic, communication, and mechanical technologies, automation systems are being used in industrial fields to achieve high production volumes with less manpower, and the technology for this purpose, the Programmable Logic Controller (PLC), is being introduced in many fields.

[0013] PLC is a highly autonomous control device that replaces the functions of relay timers, counters, etc. equipped in the control panel of a conventional system with semiconductor devices such as ICs and transformers, and adds numerical calculation functions to the basic sequence control functions to enable program control. The National Electrical Manufacturers Association (NEMA) defines it as an electronic device that uses programmable memory to perform special functions such as logic, sequencing, timing, counting, and calculation through digital or analog input / output modules and controls various types of machines or processors.

[0014] In factories, sensing data is a critical issue, and is often hidden within the machines. The term "PLC" (Process Control Unit) is used to control machine processes and maintain machine data in a factory environment. Furthermore, the data generated by PLCs is used to monitor machine status using visualization tools such as HMIs and SCADA. Most of these tools are monitoring and visualization tools, and are not suitable for data analysis and prediction.

[0015] Accordingly, the present invention has been devised to solve the above-described conventional problems, and according to an embodiment of the present invention, the purpose is to provide a data conversion system using an on-premise generative AI that can generate result data so that it can be applied to a model to be trained and can be connected to a learning model by inputting an on-premise command through a user terminal linked to a server data cloud and allowing the learning model of an AI server to learn and derive the target learning data, and to connect the result data to the learning model.

[0016] According to an embodiment of the present invention, the purpose is to provide a heterogeneous PLC and edge platform linked cloud system that collects, stores, and analyzes heterogeneous PLC (programmable logic controller) data on an edge platform, stores the stored data and analyzed data in a time-series manner on a server data cloud, and allows a user terminal to monitor and analyze PLC data and analysis data by linking with the cloud through a web application.

[0017] Meanwhile, the technical tasks to be achieved in the present invention are not limited to the technical tasks mentioned above, and other technical tasks not mentioned can be clearly understood by a person having ordinary knowledge in the technical field to which the present invention belongs from the description below.

[0018] The first object of the present invention is to provide a data conversion system for diagnosing the status of production facilities, comprising: an edge platform; a server data cloud data-linked to the edge platform; and an AI server that generates diagnostic data for the status of the production facilities using a learning model; wherein the server data cloud includes an on-premise generative AI that receives a user command, processes the data so that it can be applied to a learning model that the user wants to learn, and outputs the data, which can be achieved as a data conversion system using an on-premise generative AI.

[0019] And, the data is monitored by linking with the server data cloud through a user terminal on which a web application is installed, and when a user inputs a user command through the user terminal, the on-premise generative AI processes the data so that it can be applied to a learning model that the user wants to learn and outputs the result data to the AI ​​server.

[0020] In addition, the user command may be characterized by including data processing request command data and command data linking to a learning model to be learned.

[0021] And the above learning model may be characterized by including a production facility failure diagnosis learning model, a maintenance learning model, and an exchange cycle learning model.

[0022] The second object of the present invention is to provide a system for collecting, processing, and analyzing data from a plurality of heterogeneous PLCs (programmable logic controllers), comprising: an edge platform for collecting and storing various heterogeneous PLC data; a server data cloud linked to the edge platform and having a time series database (TSDB) for linking data between edge platforms; and an AI server for generating diagnostic data on the status of the production facility using a learning model; wherein the server data cloud monitors data by linking with the server data cloud through a user terminal having a web application installed thereon, and the server data cloud includes an on-premise generative AI that receives a user command, processes the data so that it can be applied to a learning model that the user intends to learn, and outputs the data, thereby achieving the above-mentioned heterogeneous PLC and edge platform linkage cloud system using on-premise generative AI.

[0023] And when a user inputs a user command through the user terminal, the on-premise generative AI processes data so that it can be applied to a learning model that the user wants to learn and outputs the result data to the AI ​​server and the user terminal, and the user command may be characterized by including data processing request command data and command data connecting to the learning model that the user wants to learn.

[0024] In addition, the edge platform may include a PLC gateway, a communication interface having various communication protocols for communicating with the PLC gateway, an administrator terminal, and the PLC, and the edge platform may be characterized by including a PLC data collection unit that analyzes models of various PLCs, collects common data and exclusive data of heterogeneous PLCs, classifies them, and preprocesses them through data tagging and sensing data extraction; a PLC data analysis unit that analyzes PLC data collected and preprocessed by the PLC data collection unit; a local storage unit that records, stores, and scans the common data and exclusive data of the PLC, the preprocessed PLC data, and the analyzed PLC data; and a GUI that enables setting, monitoring, and analyzing data stored in the local storage unit through the administrator terminal.

[0025] And the PLC data analysis unit performs statistical analysis through data change amount over time, data mathematical function derivation, sorting, and data operation on the preprocessed PLC data, and the server data cloud is linked with user terminals on which a web application is installed, and the web application monitors data stored in the server data cloud through the user terminal, and may include a data monitoring unit that monitors diagnostic data generated by the AI ​​server.

[0026] Additionally, the multi-communication protocol may be characterized by at least one of OPC-UA Client, Ethernet / IP, TCP / IP, Modbus, Profinet, and DeviceNet.

[0027] According to a data conversion system using on-premise generative AI according to an embodiment of the present invention, the purpose is to provide a system that can generate result data so that it can be applied to a model to be trained and can be connected to a learning model by inputting an on-premise command through a user terminal linked to a server data cloud and allowing the learning model of an AI server to learn and derive target learning data through the on-premise generative AI.

[0028] According to a heterogeneous PLC and edge platform linked cloud system according to an embodiment of the present invention, heterogeneous PLC (programmable logic controller) data is collected, stored, and analyzed on an edge platform, and the stored data and analyzed data are stored in a time series manner on a server data cloud, and a user terminal can monitor PLC data and analysis data by linking with the cloud through a web application, and the purpose is to provide a system that can analyze PLC data.

[0029] Meanwhile, the effects that can be obtained from the present invention are not limited to the effects mentioned above, and other effects that are not mentioned can be clearly understood by a person having ordinary skill in the technical field to which the present invention belongs from the description below.

[0030] Figure 1 is a schematic diagram of a data conversion system using on-premise generation AI according to an embodiment of the present invention.

[0031] Figure 2 is a schematic diagram of a heterogeneous PLC and edge platform linked cloud system according to an embodiment of the present invention.

[0032] Figure 3 is a block diagram of a heterogeneous PLC and edge platform linked cloud system according to an embodiment of the present invention.

[0033] Figure 4 is a block diagram of an edge platform according to an embodiment of the present invention.

[0034] Figure 5 is a block diagram of a data conversion system using on-premise generation AI according to an embodiment of the present invention.

[0035] Figure 6 is a block diagram of an AI server according to an embodiment of the present invention.

[0036] Figure 7 illustrates a web application block diagram according to an embodiment of the present invention.

[0037] Below, the configuration and function of a data conversion system using on-premise generative AI according to an embodiment of the present invention will be described.

[0038] First, FIG. 1 is a schematic diagram of a data conversion system using on-premise generation AI for diagnosing the status of production equipment according to an embodiment of the present invention.

[0039] It consists of a server data cloud linked to the edge platform and data.

[0040] It is configured to include an AI server that generates diagnostic data on the status of the above production facility using a learning model.

[0041] The server data cloud according to an embodiment of the present invention comprises an on-premise generative AI that receives user commands, processes the data, and outputs it so that it can be applied to the learning model the user wishes to train. "On-premise" refers to the direct installation and operation of IT systems and software in a physical space.

[0042] Data is monitored by linking with the server data cloud through a user terminal on which a web application is installed.

[0043] And, when a user according to an embodiment of the present invention inputs a user command through the user terminal, the on-premise generative AI is configured to process data so that it can be applied to a learning model that the user wants to learn and output the result data to the AI ​​server.

[0044] In a heterogeneous PLC and edge platform linked cloud system, the edge platform and server acquire and analyze data from production facilities, organize the data, and use AI learning to predict the fault diagnosis / maintenance / replacement cycle of production facilities.

[0045] To this end, the original data of the factory equipment stored in the system must be processed and trained on an AI learning model that predicts the fault diagnosis / maintenance / replacement cycle.

[0046] This presents an inconvenience as the user must go through a series of data processing tasks.

[0047] In an embodiment of the present invention, the on-premise generative AI is characterized by performing a function of extracting, preprocessing, processing, and / or connecting the processed result data to a desired learning model by converting the factory equipment data stored in the server data cloud into a desired data format as simple interactive command data.

[0048] A user command according to an embodiment of the present invention may include data processing request command data and command data linking to a learning model to be learned.

[0049] And the learning model can include a production equipment failure diagnosis learning model, a maintenance learning model, and an exchange cycle learning model.

[0050]

[0051] And, Fig. 2 is a schematic diagram of a heterogeneous PLC and edge platform linked cloud system according to an embodiment of the present invention.

[0052] As illustrated in FIG. 2, the heterogeneous PLC and edge platform linked cloud system (100) according to an embodiment of the present invention is configured to include an edge platform (20) that communicates with a plurality of heterogeneous PLCs (1) through a communication interface (10) and collects, analyzes, and stores data, a server data cloud (30) that stores PLC data collected and analyzed in the edge platform (20) in the cloud, and a web application (40) that is downloaded to a user terminal (3) to enable access to the server data cloud (30).

[0053] Figure 3 illustrates an overall block diagram of a heterogeneous PLC and edge platform-linked cloud system according to an embodiment of the present invention. Figure 4 illustrates a block diagram of an edge platform according to an embodiment of the present invention.

[0054] As shown in FIGS. 3 and 4, the edge platform (20) is installed on site to collect, store, and analyze various types of heterogeneous PLC data, and may be configured to include a PLC gateway (21), a PLC data collection unit (22), a local storage unit (23), a PLC data analysis unit (24), a data cloud gateway (25), a GUI (26), etc.

[0055] An edge platform (20) according to an embodiment of the present invention includes a PLC gateway (21) and is configured to communicate with the PLC gateway (21) of the edge platform (20), an administrator terminal (2), and a plurality of heterogeneous PLCs (1) through a communication interface (10) having a variety of communication protocols.

[0056] The multi-communication protocol according to the embodiment of the present invention may be OPC-UA Client, Ethernet / IP, TCP / IP, Modbus, Profinet, and DeviceNet.

[0057] The PLC data collection unit (22) is configured to analyze various PLC models for each of the heterogeneous PLCs being communicated with and collect common and dedicated data for the PLCs. Furthermore, the PLC data collection unit (22) can classify and sort the collected data and perform preprocessing through data tagging. Furthermore, sensing data can be extracted from the collected data.

[0058] And the PLC data analysis unit (24) according to the embodiment of the present invention is configured to analyze PLC data collected and preprocessed by the PLC data collection unit (22).

[0059] In addition, the PLC data analysis unit (24) can be configured to perform statistical analysis on preprocessed PLC data through data change over time, data mathematical function derivation, sorting, and data operations. The data can be processed to enable data operations, graphing, and visualization, such as standard deviation, summation, subtraction, and average.

[0060] And the local storage unit (23) is configured to record, store, and scan data collected from the PLC data collection unit (22), preprocessed data, and data analyzed from the PLC data analysis unit (24).

[0061] In addition, the edge platform according to an embodiment of the present invention can monitor data stored in the local storage unit through the administrator terminal (2) by means of the GUI (26). In addition, the edge platform (20) can be linked through the administrator terminal (2) so that the administrator can directly set and analyze data through the PLC data analysis unit (24) within the edge platform (20).

[0062] And the data cloud gateway (25) of the edge platform (20) is configured to transmit data stored in the local storage (23) to the server data cloud (30).

[0063]

[0064] Figure 5 illustrates a block diagram of a data conversion system using on-premise generative AI according to an embodiment of the present invention. Figure 6 illustrates a block diagram of an AI server according to an embodiment of the present invention.

[0065] Server data cloud Basically, data stored in the local storage (23) is transmitted from the data cloud gateway (25) of the edge platform (20) and stored in a time series database (TSDB, 31).

[0066] And it is linked to the web application downloaded to the user terminal through the web connection part (32, blazor setver (.NET)).

[0067] As previously mentioned, the server data cloud according to an embodiment of the present invention comprises an on-premises generative AI that receives user commands, processes data applicable to the learning model the user wishes to train, and outputs the data. Data is monitored by linking with the server data cloud through a user terminal equipped with a web application.

[0068] And, when a user according to an embodiment of the present invention inputs a user command through the user terminal, the on-premise generative AI is configured to process data so that it can be applied to a learning model that the user wants to learn and output the result data to the AI ​​server.

[0069] In a cloud system that links heterogeneous PLCs and edge platforms, the edge platform and server acquire and analyze production equipment data. Furthermore, they organize this data and train AI to predict equipment failure diagnosis, maintenance, and replacement cycles. To achieve this, raw factory equipment data stored in the system must be processed and trained on an AI learning model that predicts failure diagnosis, maintenance, and replacement cycles. This approach requires users to undergo a series of data processing tasks, which can be inconvenient.

[0070] In an embodiment of the present invention, the on-premise generative AI is characterized by performing a function of extracting, preprocessing, processing, and / or connecting the processed result data to a desired learning model by converting the factory equipment data stored in the server data cloud into a desired data format as simple interactive command data.

[0071] A user command according to an embodiment of the present invention may include data processing request command data and command data linking to a learning model to be learned.

[0072] And the learning model can include a production equipment failure diagnosis learning model, a maintenance learning model, and an exchange cycle learning model.

[0073] Specifically, as illustrated in FIG. 5, the on-premise generative AI can be used to process data so that it can be easily applied to a learning model that the user wants to train. According to the first embodiment, if a user commands, “Collect the 1 rpm value of the motor and the pump water level sensor value of Factory A and create a .csv file of one year’s worth of data,” the on-premise generative AI can receive the command and output a .csv file as the result data of the on-premise generative AI to the user. Accordingly, the user can receive this result data and apply it to the learning model that the user wants to train to generate diagnostic data for the production facility.

[0074] Also, according to the second embodiment, when a user commands, “1) Create one year’s worth of data for the pump water level sensor values ​​of factory A and 2) connect that data to the exchange cycle learning model and train it,” the on-premise generative AI receives the command, outputs the result data of the on-premise generative AI, connects that result data to the exchange cycle learning model, and trains the exchange cycle learning model on the result data to predict the exchange cycle of the water level sensor of factory A and provide the diagnostic data.

[0075]

[0076] Figure 7 illustrates a web application block diagram according to an embodiment of the present invention.

[0077] As shown in FIG. 3 and FIG. 7, it can be seen that the server data cloud (30) is linked with each of the user terminals (3) on which the web application (40) is installed.

[0078] A web application (40) according to an embodiment of the present invention may be configured to include a cloud linkage unit (41, Blazor WebAssembly), a data monitoring unit (42), a data analysis unit (43), etc., as shown in FIGS. 3 and 7.

[0079] That is, through the cloud linkage (41), the user can run the web application (40) downloaded to his / her user terminal (3) and link it to the server data cloud (30).

[0080] The web application (40) is equipped with a data monitoring unit (42) and is configured to monitor data stored in the server data cloud through a user terminal.

[0081] Accordingly, the user can monitor heterogeneous PLC collection data, preprocessing data, and data analyzed in the PLC data analysis unit (24) stored in the server data cloud (30), i.e., stored in the local storage unit (23) of the edge platform.

[0082] Additionally, the web application (40) may be equipped with a data analysis unit (43) so that the user can directly analyze data stored in the server data cloud.

[0083] The data analysis unit (43) according to an embodiment of the present invention can access the server data cloud (30) via the web application (40) of the user terminal (3) and perform statistical analysis, signal processing analysis, mathematical signal processing analysis, multiple tag function, historical large-scale data statistical analysis, and data visualization on data stored in a time-series database (31). In addition, through such analysis data, it is possible to monitor various smart factory components such as sensors and AI-based devices, as well as to determine and predict failures.

[0084]

[0085] In addition, the devices and methods described above are not limited to the configurations and methods of the embodiments described above, and the embodiments may be configured by selectively combining all or part of each embodiment so that various modifications can be made.

Claims

1. As a data conversion system for diagnosing the status of production equipment, Edge platform; Server data cloud linked to the above edge platform; and An AI server that generates diagnostic data on the status of the production facility using a learning model; The above server data cloud is a data conversion system using on-premise generative AI, characterized in that it includes on-premise generative AI that receives user commands, processes data so that it can be applied to a learning model that the user wants to learn, and outputs the data.

2. In paragraph 1, Data is monitored by linking with the server data cloud through a user terminal with a web application installed. A data conversion system using on-premise generative AI, characterized in that when a user inputs a user command through the user terminal, the on-premise generative AI processes the data so that it can be applied to the learning model that the user wants to learn and outputs the result data to the AI ​​server.

3. In paragraph 2, A data conversion system using on-premise generative AI, characterized in that the user command includes data processing request command data and command data connected to a learning model to be trained.

4. In paragraph 3, The above learning model is a data conversion system using on-premise generative AI, characterized in that it includes a production facility failure diagnosis learning model, a maintenance learning model, and an exchange cycle learning model.

5. A system for collecting, processing, and analyzing data from multiple heterogeneous PLCs (programmable logic controllers). Edge platform for collecting and storing various heterogeneous PLC data; and A server data cloud for linking data between edge platforms and having a time series database (TSDB) linked to the above edge platform; and An AI server that generates diagnostic data on the status of the production facility using a learning model; and monitors data by linking with the server data cloud through a user terminal on which a web application is installed. The above server data cloud is a heterogeneous PLC and edge platform linked cloud system using on-premise generative AI, characterized in that it includes on-premise generative AI that receives user commands, processes data to be applied to a learning model that the user wants to learn, and outputs the data.

6. In paragraph 5, When a user inputs a user command through the user terminal, the on-premise generative AI processes the data so that it can be applied to the learning model that the user wants to learn and outputs the result data to the AI ​​server and the user terminal. A heterogeneous PLC and edge platform linked cloud system using on-premise generative AI, characterized in that the above user command includes data processing request command data and connection command data to a learning model to be trained.

7. In paragraph 6, The above edge platform includes a PLC gateway, and includes a communication interface having various communication protocols for connecting the PLC gateway, an administrator terminal, and the PLC. The above edge platform is, A heterogeneous PLC and edge platform linkage cloud system using on-premise generative AI, characterized by comprising: a PLC data collection unit that analyzes models of various PLCs, collects common data and exclusive data of heterogeneous PLCs, classifies them, and preprocesses them through data tagging and sensing data extraction; a PLC data analysis unit that analyzes PLC data collected and preprocessed by the PLC data collection unit; a local storage unit that records, stores, and scans the common data and exclusive data of the PLC, the preprocessed PLC data, and the analyzed PLC data; and a GUI that enables setting, monitoring, and analyzing data stored in the local storage unit through the administrator terminal.

8. In paragraph 7, The PLC data analysis unit performs statistical analysis on the preprocessed PLC data through data change over time, data mathematical function derivation, sorting, and data operations. A heterogeneous PLC and edge platform linked cloud system using on-premise generated AI, characterized in that the above server data cloud is linked with user terminals on which a web application is installed, and the web application monitors data stored in the server data cloud through the user terminal, and includes a data monitoring unit that monitors diagnostic data generated by the AI ​​server.

9. In paragraph 7, A heterogeneous PLC and edge platform linkage cloud system using on-premise generation AI, characterized in that the above multi-communication protocol is at least one of OPC-UA Client, Ethernet / IP, TCP / IP, Modbus, Profinet, and DeviceNet.

Images