"Industry 4.0 has greatly changed the traditional production line, and the new era intelligent factory came into being. Smart factory will face the technical development trends of terminal intelligence, ubiquitous connection, marginalization of computing, flat network, service platform and security improvement. It needs to solve the problems of intelligent perception control, comprehensive interconnection and other aspects. This paper compares the sensor to the sensory organ of the intelligent factory, the robot to the limbs of the intelligent factory, and the AI system to the computing brain. It introduces the key technologies required to realize the intelligent factory, and points out that the data security of the edge and cloud will be the biggest challenge faced by the intelligent factory.
With the promotion of national strategies such as "industry 4.0" and "made in China 2025", intelligent manufacturing has gradually developed from concept publicity to implementation stage. Advanced technologies such as industrial Internet of things (iiot), artificial intelligence, big data, robotics, cloud computing and augmented reality are also emerging. The integration of digital and physical fields has further promoted the rise and development of intelligent manufacturing.
"The main line of intelligent manufacturing is intelligent production, and the main carrier of intelligent production is intelligent factory." Mr. Zhou Ji, President of the Chinese Academy of engineering, concisely and clearly expounded the relationship among intelligent production, intelligent manufacturing and intelligent factory. Smart factory is a fully connected and flexible manufacturing system, which can integrate supply chain and production line according to dynamic market demand, and produce personalized customized products when needed.
According to the prediction of authoritative market institutions, the global intelligent manufacturing market will reach US $395.2 billion by 2025. Altizon's 2017 intelligent manufacturing report pointed out that in the industrial manufacturing industry, intelligent manufacturing is likely to reduce direct labor costs and indirect costs by 5 ~ 10%. It can be seen that the value of intelligent manufacturing and future market development have great potential. So, what technologies are needed to realize the smart factory?
The realization of intelligent factory needs six key technologies
The ultimate goal of building an intelligent factory is to realize intelligent manufacturing, and the key technology of intelligent manufacturing is the core technology of building an intelligent factory. The information from Vero solutions illustrates six innovative technologies that drive and enable intelligent manufacturing.
Figure 1: six top innovative technologies that drive and enable intelligent manufacturing (source: Vero solutions)
1. 3D printing
In the past, it took several months to design and assemble products and communicate with suppliers. With 3D printing technology, the manufacturing process became faster and the time required was greatly reduced.
In addition to traditional 3D printing materials, such as plastics, glass and ceramics, there are also electronic components and environmental protection materials. The flexibility and advantages of 3D printing can reduce the total cost caused by the use of new materials, and because the products are produced on demand on site, there is no need to worry about insufficient inventory space.
2. Augmented reality and wearable devices
By monitoring the level of fatigue, managers can improve employee safety, reduce the risk of injury and improve productivity. In order to further improve safety, wearable technology can also be used to operate the machine to keep employees away from dangerous areas at work.
Wearable technology can track the working speed of employees and detect the efficiency of the system.
3. Cloud computing platform
Using cloud computing platforms such as the industrial Internet of things (iiot), manufacturers realize interconnection and internal exchange of information, and suppliers and distributors can also enhance understanding. If the intelligent factory is compared to the human body, iiot is equivalent to the nervous system.
The system based on cloud computing realizes the coordination of the whole development process, so that employees can prioritize more important tasks, so as to speed up the development process of the new system.
The quality management panel connects manufacturing sites around the world and reports sales results directly to the site administrator.
4. Forgery detection technology
Manufacturing ranks third in terms of the severity of forgery in different industries. The manufacturer's average loss is US $194000. In addition, forgery will also affect the company's total productivity, reduce the trust of employees and lose the trust of customers.
Deep neural network (DNN) can be used to identify forgery in transactions, collect data from online transactions and model them, so that manufacturers can mark and predict the possible forgery risk in future transactions.
5. Position detection technology
Tracking and tracing asset location and status has become a key link in manufacturing. Since the beginning of industry 4.0 and smart factory, the manufacturing industry has adopted advanced positioning technology.
Commodities and goods will be continuously tracked through various systems, such as supply chain management, ERP, MES and other information technology systems, so as to improve the overall efficiency.
Radio frequency identification (RFID), satellite navigation system and bar code technology can monitor the movement of goods. Location based services such as GPS or mobile ID can provide geolocation solutions for outdoor applications.
6. Intelligent sensor
Still taking the human body as an example, intelligent sensors are equivalent to various sensory organs of the human body, which are used to collect information such as temperature, humidity, pressure, spatial movement and position in the industrial environment. As early as 2013, the industry predicted that more than 20 million sensors would be used in the supply chain. Gartner predicts that up to 25 billion sensors will be used by 2025.
Intelligent sensors produce very complex data, which become the basis of automatic and intelligent control. With intelligent sensors, there is no need for remote processing, which improves the productivity and efficiency of intelligent machines.
Intelligent sensors are particularly useful for warehouse management. They can monitor the temperature, check whether there are problems with the goods, and record some problematic parameters.
Industrial Internet of things (iiot) constitutes the nervous system of intelligent factory
The development of industrial Internet of things provides new opportunities for manufacturing automation and intelligence.
The intelligent factory pursues the high coordination and efficient utilization of the intelligence of human, machine, material, method and environment. In addition to intelligent equipment and intelligent production and execution system, it is also necessary to consider factors such as environment and factory facilities. A large number of sensors are used to control or monitor water, electricity, gas, dust particles, temperature and humidity. However, various sensors are scattered, heterogeneous, data formats are not unified, or even mobile. Using the architecture of industrial Internet of things for intelligent coordination is a good way.
The industrial Internet of things platform continuously integrates various acquisition or control sensors or controllers with sensing and monitoring capabilities, as well as ubiquitous technology, machine to machine (M2M) communication, machine learning and cloud based analysis into all links of industrial production, so as to improve efficiency and obtain competitive advantage by collecting and analyzing data from New and existing equipment. Through the powerful management operation function, it can:
Improve the transparency of production process. Use data recording and communication functions to make the equipment status clear at a glance, so as to improve productivity;
Prolong equipment life and reduce cost;
Expand control functions. The control function at the equipment end is used to replace the central control in the central control room to realize decentralized control.
For example, the cloud IOT platform of bistel company, an AI engineering and automation system solution provider, combines software and hardware, and can provide storage, computing, security, development tools and many other common functions, improve quality and engineering productivity, and improve operation efficiency.
Sensors constitute the sensory network of intelligent factory
The key component of modern industrial system is sensor, which sends data to controller, monitor and other equipment to make the factory operate. Sensors play a variety of roles in modern factories. In addition to providing data for process control, they also help in quality evaluation, asset tracking, and even ensure the safety of workers. In order to meet different purposes, there are many kinds of sensors. The most common types are lighting, temperature, motion, position, existence, vision, force, flow and chemical composition. In fact, each different process or environmental condition has a type suitable for its measurement.
Micro electro mechanical system (MEMS) for data acquisition and processing is a huge category of sensors from 3D gyroscopes, accelerometers, magnetometers and oscillators to heat, pressure, humidity and other sensing elements. MEMS temperature and humidity sensors can be used to detect environmental conditions, and MEMS accelerometers can be used to monitor the vibration and rotation speed of industrial equipment.
For example, ADI MEMS accelerometer integrates embedded RF transceiver in digital MEMS vibration sensor and other devices, which is a solution integrating signal processing and communication functions. The functions of such programmable devices include:
Automatically wake up and capture time domain vibration data regularly;
Performing fast Fourier transform (FFT) on the data record;
User configurable spectrum analysis of FFT results;
Provide simple pass / fail results through efficient wireless transmission lines;
Access data and results, and then return to sleep.
High precision MEMS accelerometers and gyroscopes can provide accurate position information for the navigation and rotation of industrial robots. These MEMS sensors feed data to a neural network for visual and other perceptual processing to help factory robots "understand" their environment and whether there are some changes in the surrounding world.
Figure 2: various sensors used in industry need to be connected to the Internet of things（ Image source: postscapes)
Figure 3: introducing sensor systems into iiot requires a complex hardware and software ecosystem（ Image source: disruptive technologies)
Robots form the limbs of an intelligent factory
Industrial robot is an important part of intelligent manufacturing.
In an intelligent factory, there is no need for large-scale production facilities, huge supply lines and a large number of labor. Just using more compact and efficient industrial robots can reduce the transportation cost and delivery cycle.
Through more integrated manufacturing processes, industrial robots can make stable improvement and adjustment by using built-in sensors, data recording and cloud based real-time inspiration and calculation, so as to improve production. Using enhanced machine learning and artificial intelligence (AI) systems, intelligent manufacturing technology can learn how to produce products faster, reduce waste and avoid complex settings.
Figure 4: new ways of human-computer collaboration and interaction enable employees to focus on creative work and value-added tasks（ Image source: Infineon)
The integration of industrial robot in intelligent factory not only refers to the integration of equipment and system itself, but also includes the coordination and integration of machine and human. It is also one of the goals of intelligent factory to promote the change of interaction mode through innovative HMI (man-machine interface) and remote assistance.
In more and more application scenarios of modern industry, traditional industrial robots are being replaced by cooperative robots or supported by cooperative robots. Cooperative robots work with humans in the production process, and are no longer separated from human colleagues equipped with protective devices like traditional industrial robots. Compared with traditional industrial robots, cooperative robots are smaller, more flexible, and easier to program or even self-learning programming.
Experts from Boston Consulting Company believe that the use of robots in the future will improve the work efficiency of employees by 30%. Cooperative robots will not replace human work, but as a supplement, so that staff can perform new tasks in new activity areas, such as machine programming and quality control at the end of the automatic production process.
At present, there are many robot manufacturers, and the integration has a certain heterogeneous type, but the relevant industry specifications reduce the integration complexity a lot. Many equipment manufacturers follow industry standards, such as secs / gem protocol. For example, Infineon's factory takes the requirements of SECs / gem interactive protocol and message flow content as one of the procurement standards when purchasing equipment, and establishes a real-time message distribution and subscription mechanism Yoda based on Fast Ethernet internally. When the new equipment is programmed into the equipment number and Unicom Network, it can be directly integrated into the system. It can communicate with the system and other machines through the real-time message system to realize status report, parameter report, result data report, alarm information report, instruction acceptance, parameter modification, etc.
AI system is the computing brain of intelligent factory
In the future smart factory, AI will be ubiquitous, ubiquitous and invisible. AI will be used to manage the entire input and output, run the robot, and run the assembly line at full load through utilization management and predictive maintenance.
In intelligent factories, digitization is the key, but its purpose is not to generate massive data, but to use massive data. Cao Hong, information technology director of Infineon semiconductor (Wuxi) Co., Ltd., believes that digitization is to obtain knowledge through data generation and analysis, so as to make prediction and automatic decision-making, so as to achieve the efficient utilization of people, machines, materials and methods of manufacturing, Reduce operating costs and maximize value chain benefits.
With the support of big data, everything will become more accurate through AI analysis methods and algorithms, such as machine learning, predictive analysis or multivariable statistics. For example, the alarm content of the equipment is usually non-SQL data. AI integrates the production execution system to process structured, unstructured and massive distributed real-time information, realize intelligent predictive maintenance, improve product quality and release labor force through automatic X-ray image recognition.
Figure 5: Using AI to analyze data, generate required knowledge and trigger actions to optimize the process and efficiency of the whole plant（ Image source: Infineon)
Machine learning is a branch of artificial intelligence. It is ubiquitous in practical applications, especially in intelligent factories. Repetitive human tasks will eventually be handled by intelligent machines, instead of requiring deep emotional connections between employees.
In an interview with the reporter of electronic technology design, W.K. Choi, CEO of bistel company, introduced that the intelligent manufacturing applications supported by bistel AI are divided into three core areas: detection, analysis and prediction. Detection includes health monitoring, predictive maintenance, fault detection, etc. In the field of analysis, bistel's edatalyer suite has four core applications. Customers can use AI to quickly analyze wafers and panels to determine the root causes affecting wafer production or engineering productivity.
Figure 6: bistel's vision for intelligent manufacturing（ Image source: bistel)
In addition, Renesas Electronics' e-ai solution can also be embedded in existing manufacturing devices. It does not need to upload huge sensor data to the network, but judge whether the equipment is abnormal through e-ai at the terminal. Dai Qihong, manager of Renesas Electronics China industrial solution center, introduced that the solution only informs the production management system of the judgment results. Therefore, It can realize real-time autonomous control without occupying more network bandwidth.
AI technology is pushing more intelligence and data analysis to the edge. At present, the processing of voice and human language by AI is more similar to human functions, and more AI applications will appear in the field of software based on cloud and network.
Some new AI applications are developing and creating a new niche in the Internet of things, but not all terminal nodes need advanced functions. Chen xiongji, senior marketing manager of silicon lab in Asia Pacific, introduced in the interview that silicon Labs' expertise in mixed signal processing, low-power technology and wireless connection may bring more benefits.
Many applications will continue to use gateway and cloud resources to handle AI like tasks. The system architect needs to make a comprehensive evaluation in terms of characteristics, cost, performance, latency, security and the overall value orientation of these functions.
Edge and cloud data security is the biggest challenge faced by smart factories
A large amount of data will be generated in industry 4.0, and some data need to be entered closer to the interconnected system (i.e. "edge")
Our other product:
