The concept of "twin-oriented" manufacturing model is new!
The concept of "twin-oriented" manufacturing model is new!
The concept of "twin-oriented" manufacturing model is new!
Xiao Wenlei, founder of Grep and associate professor of Beijing University of Aeronautics and AstronauticsIn recent years, the rise of artificial intelligence technology has brought a new round of scientific and technological revolution, and many attempts have been made in the industrial field. Xiao Wenlei, founder of Grep and associate professor of Beijing University of Aeronautics and Astronautics, said in an exclusive interview with the reporter of China Electronics News that it is necessary to innovate the core links of the machining (hereinafter referred to as "machining") industry, build an intelligent machining workshop, and create a large-scale service platform for the machining industry, so that artificial intelligence technology can truly bring changes to the manufacturing process. In view of the pain points of the machining industry in terms of intelligence, Xiao Wenlei's team proposed a "twin-oriented" manufacturing model, introduced artificial intelligence and digital twin technology into CNC machine tools, digitized the existing operation experience, and independently developed the industrial manufacturing simulation engine GPCORE, the edge twin system of CNC machine tools GrapeSim, and the cloud twin system GrapeSaaS, forming an edge-cloud collaborative CNC machine tool digital twin solution. It is understood that this solution has six patents, can be closed-source commercial, has been in China Commercial Aviation, AVIC Chengfei, Jinhang Digital, China Academy of Engineering Physics, Chinese Academy of Sciences Institute of Automation and other scientific research institutes and enterprises have been delivered and applied, can achieve rapid and large-scale replication and promotion. From automation to intelligence, how to develop industrial software that meets the characteristics of industry is a problem that Xiao Wenlei has been thinking about: "Industrial Ethernet is not equal to industrial plus Ethernet, and industrial Internet is not equal to industrial plus Internet. Industry has particularities, and it cannot be done directly by an addition. By reviewing the software crisis in the computer industry in the 60s of the 20th century, Xiao Wenlei put forward the concept of "manufacturing crisis", that is, backward manufacturing methods can not meet the rapidly changing design needs, resulting in a series of serious problems in the process of product design and production management. He said that if the digital upgrade is blindly carried out in the backward manufacturing method, it is easy to cause the digital links to be unable to be interconnected, forming an "intelligent island".
The concept of "twin-oriented" manufacturing model is new!
Edge-cloud collaborative digital twin system deployed in actual production lines
For the machining industry, Xiao Wenlei believes that there are problems in three aspects: simplicity, compatibility and intelligence. Complex process details and frequent process adjustments lead to long preparation times for machining operations and a lack of simplicity in the manufacturing process. There are differences between different manufacturers and different versions of CNC machine tools and systems, which makes the operation experience not universal, the case is difficult to migrate, and the compatibility is insufficient. The traditional automated manufacturing system integrates too much industrial software, with high production costs, high dependence and requirements for operators, which is easy to cause artificial lag, and it is also difficult to meet the current demand for flexible and intelligent manufacturing. These problems lead to the binding of machining manufacturing to hardware and a high degree of dependence on labor, which hinders the transformation of the machining industry from automation to intelligence. "Physical work can be greatly liberated by automation, but the level of automation of mental work is very low. The role of people in the manufacturing process has not been fully understood, and it is difficult to truly liberate from the complex production process. The solution of various problems and the final decision still rely on human wisdom to a large extent. Xiao Wenlei said, "The purpose of intelligent manufacturing is to reduce people's mental work in manufacturing. "From experience to data" let the information run more and let people run less", this is the intelligent CNC machine tool system that Xiao Wenlei expects. Data is particularly important in intelligent manufacturing, and it is inseparable from process adjustment and optimization and training of artificial intelligence. However, there is a lack of data in the machining industry. Xiao Wenlei said: "Some factories process a part, if it is cut diagonally and unevenly, it will flutter and shake, and the surface will not be smooth. What I care about is what the signal is at this time, what parameters are used, what is the current, and what is the cutting force. This data cannot be retrieved at all, because this machine does not have this information. So I can't optimize if I want to. In the interview, the reporter learned that the traditional CNC machine tool can not export data or the exported data is not labeled, which makes the operation process difficult to trace and the process difficult to optimize. When the process needs to be adjusted, due to the lack of historical record, the operation relies heavily on the experience of old workers, which can easily lead to problems such as technical disconnection and unstable quality. "Only with big data can there be big models, and artificial intelligence without data is all small models." He added.
The concept of "twin-oriented" manufacturing model is new!
The edge-cloud collaborative digital twin system actually deployed by the Jiangxi Research Institute of Beijing University of Aeronautics and Astronautics is now the digital twin solution of CNC machine tools developed by Xiao Wenlei's team, which can simulate the machine tool and the processing process, perceive the machine tool status in real time and collect relevant data, digitize the operation process, and form an operation record that can be traced, so as to achieve "processing is experiment". With enough "big" data, there is more room for artificial intelligence to be used. It is understood that this solution includes a variety of artificial intelligence algorithm models such as intelligent algorithms and functions generated by processing technology, intelligent recognition models of CAD drawing processing features, processing cost accounting models based on cutting rates, automatic reasoning and prediction models of machining processes, and process parameter optimization models based on machining status data of machine tool parts. With the help of these algorithms, the machine tool operating system learns from the worker's operation and past debugging data, and automatically adjusts to different needs, so that the task is hardware-agnostic. Taking tool management as an example, in traditional production, the process of machining is a given, and a set of tools needs to keep the initial value consistent. When a knife is worn out and needs to be replaced, the entire set of knives needs to be replaced. This not only creates waste, but also binds production to hardware, making it difficult to adjust flexibly. The intelligent machine tool system will generate a digital twin for each tool, record the status of the tool in real time, including usage, wear, and service life, and carry out full life cycle management. A knife is replaced, and the artificial intelligence algorithm of the intelligent machine tool system automatically adjusts the process program to isolate the manufacturing from the hardware. Through the artificial intelligence-driven manufacturing system, the processing workshop can also achieve intelligence in many aspects such as CNC machine tool task execution, cost accounting, process optimization, processing services, etc., reduce the dependence of the process on information systems and personnel before, during and after processing, improve the utilization rate of machine tools, processing efficiency and processing quality, and achieve cost reduction and efficiency increase. From machine tools to platforms, the Internet has long been a necessity of modern life, but in industrial production, it is a different picture. Traditional machine tools are not networked, and each machine runs independently in production. In the context of intelligent manufacturing, IT and OT tend to converge. The networked machine tool not only makes data sharing and large model training more convenient, but also allows the edge twin system and the cloud twin system on the machine side to work together, providing the possibility for the construction of large-scale smart factories. It is understood that the cloud twin system GrapeSaaS is divided into four layers, the meta-factory layer, the infrastructure layer, the twin layer, and the business layer. The meta-factory layer is automatically programmed based on requirements, providing service-based STEP-NC (an object-oriented new international standard for NC programming data interface), the infrastructure layer provides unified data storage, data communication and other services for manufacturing-related entities and resources, the twin layer forms intelligent manufacturing microservices, and the business layer forms manufacturing industry solutions in combination with business scenarios. At present, Xiao Wenlei's team has completed the development of the second layer, that is, the infrastructure layer. In Xiao Wenlei's complete vision of the edge-cloud collaborative CNC machine tool digital twin solution, human participation has been greatly reduced, the workload of engineers and developers has also been reduced, and the manufacturing process is mostly decided, communicated and executed by machines, and unmanned manufacturing has begun to take shape. In the past few decades, CNC technology has evolved, but the language used for programming CNC machine tools today is still G-code invented in the 50s of the 20th century. This is a language similar to computer assembly, where engineers design the movement route and feed rate of the tool through G-code, i.e., programming it for the manufacturing process. The simulation system can generate a digital twin of each entity, such as tools, parts, and machine tools, that is, it is object-centric. Xiao Wenlei summarizes these two as "process-oriented" and "object-oriented", respectively. He hopes that the transformation from "process-oriented" to "object-oriented" can solve the "manufacturing crisis" in the machining industry, and in the future, this set of edge-cloud collaborative CNC machine tool digital twin solutions will comprehensively innovate the CNC system and production methods of machine tools, and create a large-scale service platform for the machining industry.
