Exploration and Application of Cloud-Network Integration Control Architecture in Industry

Cloud-network integration(CNI) is an inevitable choice to enhance the supply capacity of industrial Internet. In traditional industrial control systems, there are problems such as insufficient depth of perception, insufficient connectivity and insufficient analytical predictability. Firstly, an industrial CNI control architecture is proposed in this paper. Then a software-defined PLC integrated development & runtime environment based on the cloud framework is developed and deployed on a private cloud platform. An efficient 5G private network is created to facilitate wireless omnidirectional connectivity of cloud PLC. The problems of complex industrial field circuits and difficult to unify protocols are solved through 5G intelligent terminal devices. Finally, the cloud control architecture and software platform are successfully applied to the remote centralized control of the tundish preparation system in a steel enterprise’s two steelmaking plants. The results shows that the innovative system architecture meet the requirements for high-performance access, transmission, computing and storage in industrial plants, which achieves the cloud control of entire process industrial automation production.


Introduction
In recent years, the industrial economy has shown a steady and sustained recovery.However, under the influence of multiple factors such as the rising cost of materials in the upstream of the industry, the rising cost of human resources and the rising demand of customers for high-quality products and services, enterprises are constantly required to adopt new technologies for transformation and upgrading to close to intelligence [1].With the implementation of industrial Internet, software products and open architecture, the automation architecture of process industry is redefined, supported by the new generation of network and IT technology [2].The industrial control system architecture will be more interoperable, more open, networked and multi-function.Therefore, the software-defined control mode appears, and the control capability is expanded with virtualization technology.Cloud computing is considered to be an effective solution for industrial automation in the future.5G communication network has the characteristics of low latency, large bandwidth, and wide connections.Combined with the requirements and characteristics of industrial automation, the new generation of control architecture will stride forward in a more open direction driven by the integration of 5G and cloud computing [3].
In 2020, Midea's microwave oven production line adopted a centralized deployment of PLC based on 5G.In June 2021, Huawei, Zijinshan Laboratory, Shanghai Jiao Tong University and Baosight Software jointly released the world's first wide-area cloud PLC technology test results.The experiment was based on deterministic WAN technology and the next generation industrial control edge computing architecture.Between Shanghai and Nanjing, with a transmission distance of nearly 600 kilometers, the wide area cloud based PLC industrial control system was deployed and operated stably.In October 2021, Supcon Technology and China Unicom jointly released the innovative application achievement "5G+MEC Cloud Control System".In December 2021, China Unicom, Kyland Technology, and Sany Heavy Industry jointly launched a massive 5G cloud PLC application for global lighthouse factories.The technology transformed traditional industrial control hardware into a 5G MEC edge cloud software system.Beijing University of Science and Technology constructed a new generation 5G-TSN fusion network with end-to-end low latency and deterministic guarantee [4].Based on independently developed cloud PLC, material recognition, positioning and sorting action testing was achieved through machine vision in the laboratory.The above research results and applications are based on the deployment architecture of edge cloud industrial control systems, without maximizing the expansion of PLC computing power and considering perceptual parameters that carry visual, audio, and other dimensions.In November 2022, Ansteel released the first 5G cloud PLC application in the steel industry in China.Adopting the CNI architecture, the PLC system was truly deployed on the industrial private cloud platform.A high-quality 5G network system with cross domain integration, integrated computing & networking and cloud edge collaboration was constructed.The application of 5G and cloud control in the industrial field represents the extreme and deterministic network capabilities of 5G private network introduced into control systems [5].The deep integration of 5G and industrial environment enables all industrial applications in the IT and OT domains to be carried by 5G networks.

5G expands to industrial production
In-depth infrastructure reform and application innovation are required to realize industrial intelligence.The Industrial Internet as a key infrastructure builds a comprehensive service system for massive data collection, aggregation and analysis to provide protection for cutting-edge innovation [6].The rapid launch of top-level industrial applications and the integration of control systems and big data systems will bring more significant efficiency improvements.

The current situation of industrial control scenarios networks
Industrial control needs to meet the real-time task requirements.Especially the adoption of industrial communication network has realized distributed real-time control.In order to limit the communication delay of information transmission between task processes, the deterministic dedicated real-time communication network structure has been adopted.The most typical network used in production scenarios is fieldbus.Decentralized control devices are converted into network nodes and connected into networked control systems.Wired network is the most widely deployed in the traditional field network.With the continuous upgrading and updating of system structure and automation products, The situation of complex bearer networks and various industrial protocols has gradually formed [7].Relying on the complex network architecture, systems, equipment and protocols cannot be compatible.The difficult integration of data flow and business flow will gradually lead to the subsystem becoming an information island.Moreover, multiple protocol conversions are required in the communication configuration between traditional controllers and different devices.Some operating devices that do not support online configuration need to be configured offline and then reconnected to the network.Such operations not only take time, but also affect the overall stability of the network.The interconnection of the whole process industrial production is also invalid.
The new generation of smart sensors and applications have higher demands on network performance.However, the existing network structure cannot meet the demand.Building additional network facilities leads to coexistence of multiple heterogeneous networks.Even the phenomenon of independent networking of a workshop or a link appears, which further increases the difficulty of operation and maintenance.There are large numbers of early dedicated computing nodes in some control links, carrying traditional industrial control hardware such as PLC, DCS and industrial computers.The limited single-point computing power of dedicated nodes makes it more difficult to deploy new applications.In addition, real-time processing data output control instructions are required, and offline/online data is used to establish a data-driven control model.Various network connection modes make the control system increasingly complex.Therefore, high capacity network bandwidth, high performance computing resources, adaptive software architecture and intelligent computing framework are required for the control system.

"5G + Industrial Internet" applied in production
In the field of industrial control automation, there are currently a variety of real-time industrial Ethernet, such as EtherCAT, Profinet, Ethernet/IP, etc.Although the machine motion control with the highest real-time requirements has been met, there are some shortcomings in usability, interoperability, bandwidth and equipment cost.To ensure the real-time and certainty of big data transmission in the industrial interconnection environment, the existing real-time Ethernet protocol is insufficient.Industrial Internet is the application ecology formed by the deep integration of the new generation of Internet information technology and industrial system.It is also the key information infrastructure for the intelligent development of industrial enterprises [8].The industrial Internet platform has been utilized, which can quickly respond to market demand and efficiently integrate resources to promote production and operation.New models such as networked collaboration and personalized customization will be formed.
In the context of IT and OT convergence, new network technologies such as big data and cloud computing, time-sensitive networks (TSN), software-defined networks (SDN), and OPC UA have gradually entered the field of industrial control, especially 5G communication technology.With the gradual improvement of 5G standards and industry chain, MBB, URLLC and mMTC are designed and introduced.The R16 standard released by 3GPP for 5G focuses on the technical features of URLLC, providing the ability of low latency and high reliability.This provides a technically feasible path to ensure the large-scale application of 5G technology in the field of industrial control [9].The 5G fully interconnected factory is being actively constructed, which accelerated the in-depth expansion of "5G+industrial Internet" to all sectors and links of industrial production.
With the emergence of new applications such as digital twins and digital driven production process control, the whole process data needs to be collected.Therefore, more kinds of sensors will be connected to the network.Relying on the 5G large-bandwidth network, the collected raw data can be uploaded to the cloud without complicated preprocessing.Through the integration of the new generation of information and communication technology represented by 5G, the new industrial Internet infrastructure has been built successfully [10].In the future, industrial production will gradually form an advanced factory with extensive connection of production units, deep integration of information operation, full utilization of data elements, and innovative application of high efficiency empowerment [11].

Cloud-network integration creates a new form of industrial production
Cloud-network integration(CNI) is a transformation of network architecture driven by business requirements and technological innovation.Now more and more enterprises are gradually building cloud network integration infrastructure platforms.The platform is deeply integrated with network links, edge computing and industrial protocol analysis, which effectively supports the digital transformation and intelligent application of enterprises.

CNI function architecture
Through the construction of the three functional systems of network, platform and security, the industrial internet creates the comprehensive interconnection of people, machines and things, and forms the application mode of intelligent development.Enterprise applications are gradually evolving from "device-centric" to "user-centric".The network will become more concise and intelligent, and the upper business and strategy are no longer required [12].Network applications are gradually transitioning from "interconnection" to "cloud+network+ICT services" or "cloud+network+business".The interconnection between clouds is only a process, and the ultimate goal is the high integration of cloud network and actual business.The cloud network integration architecture is formed through software, virtualization, intelligence and service, shown in figure 1. Multi-source and heterogeneous resource systems are provided to meet the service requirements for cloud platforms, network connections, data and computing resources in diversified business scenarios.The key elements of 5G promoting cloud network integration are edge computing and network slicing.Edge computing has great challenges for large bandwidth capability, low latency capability and high security.There are different practical methods in function customization, resource isolation and quality assurance about network slicing .In the industrial Internet scenario, 5G network provides a secure and reliable communication environment with low latency, high speed, wide connectivity.Highquality cloud services achieve resource integration and collaboration of production factors across the network.From the optimization of material flow and information flow, the one-stop intelligent service of the whole process will greatly improve the production management efficiency of enterprises.

Industrial CNI control architecture
The industrial Internet platform is to overlay the Internet of Things, big data, artificial intelligence and other technologies on the cloud platform, so as to quickly respond to market demand and efficiently integrate resources to promote production and operation.Then new models such as networked collaboration and customized features will be created [13].At present, the equipment data of many production links and various resources in the supply chain have been connected to the industrial Internet platform.However, if the "industrial automation control system" as the core of production and manufacturing cannot be accessed to a flexible and extensible platform, there will be a large number of individual equipment and PLC distributed deployment.Such an industrial service system seriously restricts the development of the entire industrial value chain in the future.
The industrial CNI control architecture integrating 5G and cloud computing is an industrial automation architecture integrating cloud services and control modes.Hardware resources in elastic compute service(ECS) are virtualized to realize dynamic allocation and cross-domain sharing of cloud resources.The software-defined control system is installed in the ESC.The control instructions as services are transmitted to the field equipment through the 5G network, making the control core directly connected with the industrial Internet platform to realize the CNI control, as shown in figure 2.

System composition of softward-defined PLC
The core of softward-defined PLC is an integrated platform for system programming and configuration, which includes a integrated development environment (IDE) complying with IEC61131 standard and a runtime environment (RTE).The IDE includes core components such as device configuration, virtualization configuration, algorithm configuration, compilation and debugging, as shown in figure 3.
. Compared to the traditional PLC, softward-defined PLC has many advantages.In addition to the traditional four major sensing signals of temperature, pressure, flow rate and liquid level, the softwaredefined PLC also processes visual, voice and other signals.At the same time, it adopts a relatively open and unified communication protocol to adapt to the future development direction of intelligent manufacturing.The software-defined PLC realizes hardware resource configuration and definition through software, which is easy to migrate and software reuse.And it has lower costs.It contains multi-core processor.A single CPU creates multiple soft PLC controllers.Not only does PLC have powerful computing power, but it also achieve collaborative control of multiple devices.The softwaredefined PLC is interconnected with the Industrial Internet platform.From then on, PLC will no longer be used as a standalone device.The control signals and data analysis results are uploaded to the cloud.The application of API and ecosystem Industrial Internet platform has been expanded.

Construction of security defense strategies
As the key support of the new industrial revolution, the industrial Internet focuses on the three aspects of network, data and security.The network is the foundation, the data is the core, and security is the guarantee.CNI has opened up the original independent security architecture of cloud, network, edge and end, which provides integrated linkage monitoring and control of security risks in the process of data storage, cloud computing, network transmission, etc [14].Through the integration and innovation of cloud-network security, data security and other technologies, the unique ability to deal with various security risks has gradually formed [15].
In industry, the CNI control architecture not only succeed at the business level, but also change the security architecture and security governance model of the industrial control system.The virtualized PLC deployed on the cloud platform has unique advantages in network security.For example, virtualization network I/O is built on the SDN network.Through fine control of network traffic, full traffic detection, full function forwarding and safe network links in the "zero trust network" environment can be finely constructed [16].Perhaps blockchain-based industrial security block system can be considered to be constructed.In addition, the industrial control software firewall will also be set to provide dual security protection for the flow information of the virtual PLC.In short, the CNI control architecture can be trusted in the field of industrial control from the perspective of network security.

5G private network solution
The 5G private network adopts a soft switch network with control and user separation, mainly including core network, IP bearer network, SPN network and UPF network element.The core network has mobility management, session management and data transmission functions, mainly including access and mobility management function (AMF), session management function (SMF), unified data management (UDM), authentication server function (AUSF), policy control function (PCF), network opening function (NEF), network storage function (NRF), and network slice selection function (NSSF).SPN network based on SDN architecture is a slice packet network with the characteristics of time division multiplexing (TMD), efficient routing allocation, multi-channel aggregation, etc., to achieve high security and low latency of services.The UPF network element is mainly responsible for the routing and forwarding of user-side data packets, which are forwarded by UPF to the application terminal.The production business data is processed and forwarded locally, which greatly reduces the transmission delay and ensures that the data does not leave the park.
Advantages of 5G private network: (1) The reliability of wireless transmission delay is guaranteed: 10~20ms control delay; (2) Wireless network slice is adopted to provide differentiated service guarantee: priority is given to transmission of control signals to ensure delay, and non-real-time service isolation; (3) Multi-frequency dual-network redundancy technology ensure network reliability: 1+1 channel redundancy protection can achieve highly reliable transmission.

Remote control application
In this paper, in a large iron &steel enterprise, the control application is designed for the tundish turnover equipment of continuous casting in the steelmaking plant.The CNI control architecture is used to design logic programs to complete the packet switching operation.Two steelmaking plants with an interval of 200 km were selected as the control scenario.The specific control process is as follows: the crane moves the ladle to the position of the ladle turnover mechanism and locks the ladle.The waste truck is controlled to stop directly under the bag tipping equipment, and the bag tipping machine is driven to rotate 180°.After the hydraulic pump drives the ejector rod above the tipping machine, the telescopic movement of the ejector rod pushes the steel slag inside the ladle out of the casting mouth and falls into the waste truck below.Finally, the waste truck carrying steel slag is controlled to move to the next process for subsequent treatment.
Before the transformation, all operations were completed by Siemens PLC, and its communication adopted industrial bus protocol.At present, the 5G network with the required communication quality was built in the selected application scenario.The private cloud platform is the infrastructure, and the on-site collected signals are uploaded to the cloud platform through 5G communication at high speed.The control system defined by the software is deployed on the cloud platform to realize remote centralized control between the two places, as shown in figure 4. Firstly virtualization technology is used to separate some CPU resources in the cloud server.The virtual PLC controller is deployed on it to replace the CPU part of the traditional PLC hardware functionally, in which the hardware driver and system interface are designed abstractly.Then the virtual PLC establishes a wireless communication connection with the I/O of the ladle turnover equipment in two different steel plants.The control logic is written and debugged in the cloud programming software, and the control instructions are executed.The signal is transmitted to the I/O modules at the two steel-making sites through 5G communication.Finally a set of control system deployed on the cloud platform can realize remote control of the equipment in the two steel plants at the same time.A multi-dimensional industrial data and multi-form visualization interface are created to quickly locate industrial control problems.It also solves the problem of insufficient intelligence in the data dimension for massive industrial applications.After adopting the industrial CNI control mode, the number of workers in the preparation phase of the subcontracting service in the past has reduced from 4 to 2 per shift, and the work efficiency has been improved by more than 50%.At the same time, the labor intensity and operation risk of the continuous casting site staff have been effectively reduced.The software-defined PLC based on cloud architecture includes three levels: edge connection layer, operation layer and application service layer.It achieves close interaction and collaborative work among the three core elements of "connectivity, cloud and application".The multisource data integration and edge device intelligent transmission control method based on 5G solves the problems of data acquisition, translation, and fusion.And safety regulations and data integration standards have been proposed, making it possible to demonstrate applications for steel product manufacturing.After this application attempt of CNI, there is a very good inspiration for achieving the goal of "stable production and quality improvement, cost reduction and efficiency increase" in the future.

Conclusion
In this paper, the industrial CNI control architecture has been applied to achieve enterprise level remote collaborative control capabilities.After the application, the 5G network reduces the number of intermediate cables and equipment, and solves the problems of complicated wired network, high maintenance difficulty, and difficulty in carrying large bandwidth, which provides help to support new industrial applications.The control system defined by software is installed on the cloud platform.Multiple virtual PLC are loaded, which replace the traditional control hardware in function and reduce the maintenance cost.The computing, storage, network and other resources of each virtual PLC are allocated as required, so that the resources are exclusive, isolated from each other and not affected by each other.The overall reliability of the control system is high.Based on cloud computing, the effective combination of automatic control and intelligent application has played a key role in building cloud intelligent control system under 5G high-quality network.

Figure 1 .
Figure 1.Diagram of CNI architecture 3.2 5G facilitates CNI The implementation of 5G network service architecture promotes the coordinated development of CNI.The deep integration of 5G and vertical fields of various industries has led to new application models.The key elements of 5G promoting cloud network integration are edge computing and network slicing.Edge computing has great challenges for large bandwidth capability, low latency capability and high security.There are different practical methods in function customization, resource isolation and quality assurance about network slicing .In the industrial Internet scenario, 5G network provides a secure and reliable communication environment with low latency, high speed, wide connectivity.Highquality cloud services achieve resource integration and collaboration of production factors across the network.From the optimization of material flow and information flow, the one-stop intelligent service of the whole process will greatly improve the production management efficiency of enterprises.

Figure 2 .
Figure 2. Diagram of industrial CNI control architecture

Figure 3 .
Figure 3. System component of softward-defined PLC The device configuration module supports graphical drag drop adding of devices, business buses, and remote I/O devices.And thumbnail windows are provided to quickly locate redundant configurations of industrial servers.The bus protocols of the software include Modbus TCP, Modbus RTU, EtherCAT, CANopen, as well as custom TCP and custom serial ports.The virtualization configuration of softward-defined PLC includes graphical creation of virtual PLCs, allocation of interface resources, redundancy settings and support for virtual network port configuration.Each physical network port can be used by multiple virtual PLCs.At the same time, virtualization of PLC operation control and status monitoring is supported.The algorithm configuration function supports programming languages such as SFC, LD, FBD, etc.It supports creating C++libraries, importing C++source files and importing C++files generated by Matlab/Simulink.The compilation debugger includes three major functions: compilation, uninstallation, and debugging.The compilation function supports full engineering compilation, compilation according to the selected controller, HMI compilation and configuration correctness check during compilation.The download function supports full engineering download, selection of target controllers and HMI site download algorithms.The debugging function supports breakpoint debugging and algorithm page display of real-time values of variables.Compared to the traditional PLC, softward-defined PLC has many advantages.In addition to the traditional four major sensing signals of temperature, pressure, flow rate and liquid level, the softwaredefined PLC also processes visual, voice and other signals.At the same time, it adopts a relatively

Figure 4 .
Figure 4. Remote centralized control diagram of tundish equipment in two steelmaking plants This application solution meets the requirements for real-time and accurate data transmission in actual industrial production.A multi-dimensional industrial data and multi-form visualization interface are created to quickly locate industrial control problems.It also solves the problem of insufficient intelligence in the data dimension for massive industrial applications.After adopting the industrial CNI control mode, the number of workers in the preparation phase of the subcontracting service in the past has reduced from 4 to 2 per shift, and the work efficiency has been improved by more than 50%.At the same time, the labor intensity and operation risk of the continuous casting site staff have been effectively reduced.The software-defined PLC based on cloud architecture includes three levels: edge connection layer, operation layer and application service layer.It achieves close interaction and