How to Achieve Efficient Production in the Era of IoT

How to Achieve Efficient Production in the Era of IoT ☞ This is the 13237th article published by Metal Processing (mw1950pub).

Editor’s Note

Academician of the Chinese Academy of Engineering, and head of the 04 Special Project Group, Lu Bingheng, emphasized sternly at the Fifth World Intelligence Conference: “Producing machine tools like producing cars is absolutely wrong!” He also stated that there is still a certain gap in the processing efficiency, dynamic accuracy, and advanced levels of high-end CNC machine tools, and this area needs to strengthen basic research, including the deep development of machine tool dynamics, electromechanical control, servo systems, and intelligent control systems. He noted that intelligent control is an important means to solve these problems and is the main route, which will be our future focus.“

Among the vast crowd of professionals, there are definitely many experts in factory management. I wonder if everyone has this consensus: there is a deep gap between Information Technology (IT) and Operational Technology (OT) in traditional industrial systems, which makes the execution of key processes on the production line, as well as the collection and feedback of important data, quite inefficient and chaotic.

The smart manufacturing widely promoted in our factories aims to achieve higher interoperability, flexibility, and seamlessness in the production process. In simple terms, it means gaining insights from the massive amounts of data generated by IoT devices, making data transmission more efficient.

Let’s look at a practical case:

As 5G becomes increasingly popular, many believe that simply applying 5G in enterprise production will enhance performance in various aspects. However, this is not the case. In the era of IoT, in addition to improving network speed, it is also necessary to enhance the handling capacity of corresponding spare parts to meet the demands of smart manufacturing production. For example, focusing on the field of smart manufacturing for many years, Foxconn Industrial Internet (hereinafter referred to as “Industrial Fulian”) collaborates with Shenzhen Zhuoxin Chuangchi Technology Co., Ltd. (hereinafter referred to as “Zhuoxin Chuangchi”) to leverage the strong processing capabilities provided by hardware devices based on Intel® architecture processors, Intel® Ethernet controllers, and Intel® FPGAs (Field Programmable Gate Arrays), as well as their good support for Time Sensitive Networks (TSN), to introduce smart manufacturing solutions based on TSN networks, achieving good results in actual production scenarios.

How to Eliminate Data Communication Barriers and Achieve Network Integration in the Era of IoT?

How to Achieve Efficient Production in the Era of IoT

During the transformation of enterprises towards digitalization and networking, do we often encounter common interconnection methods in traditional industrial OT systems, such as buses and industrial Ethernet, which lead to compatibility issues due to different interface protocols, making it necessary to perform complex protocol conversions for different vendor devices on the production line to connect with each other? Moreover, for the sake of system robustness and reliability, OT systems are often separated from IT systems, resulting in data isolation between the two?

In fact, in the current network era, the issues mentioned above can be resolved. For instance, manufacturing companies like Industrial Fulian deploy a wealth of control, sensing, and data collection devices at the edge of the production line to provide effective data support for applications such as defect detection and predictive analysis using Artificial Intelligence (AI).

For example, in our daily processing production, how can we solve the problems of high defect rates and low production efficiency caused by traditional offline testing methods in Computer Numerical Control (CNC) processing?

Taking the aforementioned Industrial Fulian factory as an example, it has built an automated defect detection system based on Intel products and technologies, combining Computer Vision (CV) systems and Motion Control (MC) systems. The typical workflow is shown in Figure 1. When a product is captured by an industrial camera, the image is sent to a visual processing system deployed on an edge server or cloud platform for real-time processing. The detection system, built on CV algorithms or deep learning methods, quickly determines whether the product meets quality requirements and sends control instructions to the production line to execute actions such as stopping the line.

Figure 1 Automated Equipment Based on Computer Vision System and Motion Control System

In Figure 1, if data congestion, network delay, etc., occur due to the separation between the OT system (robot arm control, production line adjustments) and the IT system (high-definition image acquisition, transmission, and analysis), it will lead to the inability of devices to achieve precise synchronization, which may increase the defect rate of products and reduce production efficiency.

However, we can solve this problem by leveraging TSN, which can promote the integration of IT and OT systems by optimizing solutions, allowing real-time data from different production line devices to be centralized through a unified network, processed and analyzed on a unified data platform, and the results sent to various production line devices for execution.

It has the following advantages:

1) TSN networks, as an open protocol standard (IEEE 802.1), can eliminate compatibility issues between devices from different manufacturers, making industrial IoT connectivity more convenient and also providing advantages in data transmission bandwidth.

2) TSN networks break the boundaries between control systems and analysis systems in traditional manufacturing production lines, making it easier to deploy AI applications on the production front lines. At the same time, the priority control and data security mechanisms provided by hardware devices can ensure the high reliability required by the production line.

However, integrating TSN networks into existing production processes to create a new generation of industrial IoT smart manufacturing factories presents significant challenges. First, deploying a large number of controllers, sensors, and data collectors in the relatively confined environment of production lines and achieving interconnectivity will inevitably increase the difficulty of operation and maintenance. Therefore, detailed integration is required to enhance the sensitivity around the production line. Secondly, to fully showcase the advantages of TSN networks, the various hardware infrastructures selected, such as processors and Ethernet controllers, need to possess workload integration and data acceleration capabilities, while I/O flexibility and functional safety must also meet the needs of the manufacturing industry.

Leveraging Advanced Hardware Infrastructure to Achieve Workload Integration and AI Acceleration Based on TSN Networks

How to Achieve Efficient Production in the Era of IoT

In the aforementioned Industrial Fulian factory, a series of products based on Intel® architecture were introduced, further enhancing the efficiency of implementing anomaly detection and predictive analysis in production, thereby effectively improving yield and production efficiency.

Taking CNC processing as an example, as shown in Figure 2 (the TSN network symbol below the equipment in the figure indicates that the device supports the TSN network), by installing various sensors at key positions such as the tool holder, spindle, and power supply, it addresses the frequent occurrences of tool breakage and misalignment during CNC processing. This solution can utilize data collectors for real-time data collection, upload to a Programmable Automation Controller (PAC), and perform rapid inference based on algorithm models, then carry out corresponding reverse control for anomalies, forming a closed-loop network for anomaly detection.

Figure 2 Intelligent Manufacturing Solution Based on TSN Network from Industrial Fulian

In the production process shown in Figure 2, due to TSN networks providing a series of time synchronization and traffic scheduling mechanisms, the information transmission of the above systems can be made more precise, efficient, and controllable. When the closed-loop network may experience congestion due to excessive data volume, the data flow transmission between devices can be prioritized according to established rules, ensuring that high-priority data, such as control instructions, can be transmitted more quickly. If more devices on the production line can support TSN networks, the advantages of TSN will become even more apparent. Additionally, in Figure 2, through devices such as switches, data from edge devices can be uploaded to a dedicated Fog AI server for model training, and the resulting model can then be sent back to the edge for inference and execution via the switch.

For instance, in CNC processing, one critical parameter that plays a key role in product accuracy is the tool’s “golden lifespan.” When the tool exceeds its golden lifespan, it can no longer guarantee the accuracy of product processing. The golden lifespan of the tool depends on a series of factors, including processing methods, line temperature, current, etc. Therefore, we can use the aforementioned solution to collect data on the tool’s vibration curves, temperature, current, etc., through a series of sensors, and upload it in real-time to the Fog AI server for training to obtain a predictive model for the tool’s lifespan. This predictive model can then be used to make predictive analyses of the tool’s lifespan and usage in CNC equipment. By making these predictive judgments, the tool can be replaced promptly upon reaching its golden lifespan, thereby ensuring product processing accuracy and enabling engineers to design optimal replacement schedules based on the different conditions of thousands of tools on the production line, maximizing the avoidance of production efficiency declines due to downtime for tool replacement.

Moreover, when dealing with heavy traffic loads, TSN networks provide essential support for high-speed communication by performing operations such as timing, scheduling, synchronization, forwarding, queuing, seamless redundancy, and flow reservation on massive amounts of data. Therefore, TSN networks place higher demands on hardware infrastructures such as processors, Ethernet controllers, and FPGAs. The series of hardware products provided by Intel, with their excellent workload integration performance, strong data processing capabilities, good scalability, and reliable security features, can effectively support these needs.

Taking the PAC device, which plays a core role in Figure 2, as an example, it not only undertakes tasks such as data collection, preprocessing, and model inference but also needs to perform real-time control of devices such as robotic arms based on AI inference results. At the same time, for TSN networks to function, they must ensure time-sensitive traffic, requiring PAC devices to handle tasks such as configuration information parsing, status information reporting, clock synchronization, and IP data frame fragmentation and reassembly. To endow the next-generation PAC devices with robust performance, Intel® Core™ processors can be chosen as their power engine. For example, the latest 11th generation Core processors from Intel not only meet the demands of PAC devices for high-speed data preprocessing and low-latency deterministic computing in AI model inference with their third-generation 10nm microarchitecture, higher single-thread/multi-thread performance, and more advanced graphics, media, and display capabilities, but they are also specifically optimized for harsh environments such as production lines.

More importantly, regarding TSN networks, the new Intel® Core™ processors (see Figure 3) have incorporated Intel® Time Coordinated Computing (TCC) technology, which minimizes network or system latency to meet the manufacturing industry’s demands for critical real-time computing applications.

Figure 3 Intel® Core™ Processor Series with Good Support for TSN Networks

Balancing Efficiency and Reliability to Promote Manufacturing Industry Performance Enhancement

How to Achieve Efficient Production in the Era of IoT

Based on the excellent performance of Intel® architecture hardware products, the TSN network solution can effectively promote the integration of industrial OT and IT systems, ensuring that industrial IoT can achieve high-efficiency data transmission while also ensuring industrial-grade robustness and reliability, thereby bringing the following advantages to Industrial Fulian:

• Improve the yield rate of manufacturing production lines: By deploying IoT devices to the production front line and leveraging the efficient transmission capabilities provided by TSN networks, effective control can be exercised when anomalies occur on the production line, and predictive analysis methods can be utilized to forecast equipment usage. In some scenarios, the new solution has improved the yield rate by about 8% through the collection and analysis of temperature or pressure data.¹

• Enhance industrial IoT transmission efficiency: In some scenarios, issues such as network congestion, data packet loss, or retransmission due to excessive data volume were common, resulting in data transmission efficiency below 90%. After adopting the new solution based on TSN networks, data loss issues have been effectively controlled, and data transmission efficiency has increased to over 99%.²

• Reduce costs and increase efficiency, improving production efficiency: The good support of TSN networks for anomaly detection, predictive analysis, and other AI applications allows the production line to effectively reduce the frequency of downtime for inspections and enhance production efficiency based on more scientific equipment management solutions. In some scenarios, previously each machine required a dedicated operator, but after introducing the new solution, one operator can manage multiple machines.

^1、2Data sources are from internal tests, evaluations, and validations of Industrial Fulian. For details, please consult Industrial Fulian at https://www.fii-foxconn.com. Intel does not control or audit third-party data. Please review this content, consult other sources, and verify the accuracy of the mentioned data.

About Zhuoxin Chuangchi

Shenzhen Zhuoxin Chuangchi Technology Co., Ltd. (under the brand: Future Robotics) is committed to providing reliable, convenient, and cost-effective hardware and system solutions in the fields of smart manufacturing and artificial intelligence. With rich project experience and superb R&D capabilities in industrial computers, machine vision, motion control, and other fields, it is market-driven, providing a wealth of embedded products for global enterprises and individual users, applied in industrial automation, robotics, machine vision, medical applications, intelligent transportation, finance, education, and IoT fields.

About Intel

Intel (NASDAQ: INTC) as an industry leader creates technologies that change the world, driving global progress and enriching lives. Inspired by Moore’s Law, we are constantly committed to advancing semiconductor design and manufacturing, helping our customers tackle the most significant challenges. By embedding intelligence into cloud, network, edge, and various computing devices, we unlock the potential of data, making business and society better. For more information on Intel’s innovations, please visit Intel China News Center newsroom.intel.cn and the official website intel.cn.

Related posts

Leave a Comment Cancel reply