Globally, artificial intelligence (AI) technology is rapidly becoming a powerful driver of economic growth, injecting the power of change into all walks of life. In the civil aviation industry, AI is regarded as a new generation of "invisible engine" that promotes the development of the industry - those invisible big data are becoming a new "fuel" that leads people to become intelligent and start a better travel. Especially in the field of aerospace engines, AI, the "wings of wisdom", is driving the wave of change at an unprecedented speed, showing unlimited potential.
According to industry experts, the application of AI technology in the field of aero-engines is expanding to the entire life cycle from design, testing to manufacturing and operation and maintenance. While accelerating the aero-engine research and development process, it also enables collaboration between the upstream and downstream industry chains. More connected and efficient.
Designing a more powerful aircraft "heart"
As the "heart" of the aircraft, the aeroengine integrates a large number of advanced technologies, materials and processes and is a key component of the aircraft. It has extremely high requirements for precision, stability and reliability in design, manufacturing and use.
Xiao Hong, a professor at Northwestern Polytechnical University, said that the characteristics of aeroengines can be summarized as "three highs and one long". In terms of performance, aero engines have the characteristics of high flight envelope, high thrust (power) to weight ratio, high reliability and long service life. Currently, globally, the longest aircraft engine life span reaches 50,000 hours. In terms of operating environment, aerospace engines face the challenges of high pressure, high speed, high temperature and long life cycle operation. Economically, aviation engines are products with high investment, high threshold, high returns, and long cycle. Xiao Hong said that the development of a typical aerospace engine takes 10 to 20 years, but after finalization, the return on investment is very high due to the long service life.
Nowadays, with the "joining" of AI technology, this highly precise operation of equipment has also entered a safer and more efficient intelligent era. The impact of AI technology on aerospace engines can be traced back to the “source.” In other words, in the product design process, AI technology is already contributing to its model construction. As we all know, aeroengine is a concentrated expression of human wisdom and technological power. Its design process involves structural mechanics, fluid mechanics, aerodynamics, combustion and other engineering sciences, and it relies heavily on basic equations, basic models and calculation methods. In the field of engineering science, AI technology has taken the lead in realizing widespread application, and with increasingly sophisticated machine learning (Machine Learning), it has greatly improved the work efficiency and accuracy of related industries.
As part of AI technology, machine learning allows computer systems to find patterns in large amounts of data by learning from existing experience and data, promoting the development of automation, data-driven decision-making, and intelligent systems. Compared with traditional models, models built using machine learning methods have efficient and cross-scale physical feature description capabilities, and have potential advantages in calculation accuracy and efficiency. This ability is very important for building models of aeroengines.
In addition to building models, AI technology can also play a role in aero-engine performance prediction, design model optimization, test verification and other aspects. Take aviation engine testing as an example. Liu Daxiang, an academician of the Chinese Academy of Engineering, once mentioned in a public speech that a certain type of aerospace engine requires thousands or even tens of thousands of hours of testing from design to finalization, which can last up to 10 years. With the development of AI technology, people are trying to transfer some experiments to the digital space. In the digital space, digital engines with one-to-one corresponding performance are developed through digital twin technology, and experiments are carried out on the digital engines to significantly save manpower and material resources. , financial resources, and speed up the development process.
Improving the research and development efficiency of aerospace engines through AI technology is not just a good wish of academic circles. Many companies have already joined in the exploration of this cutting-edge field. For example, GE Aerospace in the United States has developed an AI-driven design tool-DT4D (Digital Thread for Design). This is a digital thread system covering the entire product life cycle. It aims to unify the data of aerospace engines from conceptual design to actual operation by connecting multiple links such as design, manufacturing, supply chain and service, making the entire R&D and manufacturing process more efficient. Efficient and transparent. The system not only allows engineers, manufacturers, supply chain and other stakeholders to access the latest product design and performance data in real time, but also integrates simulation, design and manufacturing data into the same digital thread system, effectively reducing friction in the product development process. Repetitive labor and manual data transmission errors not only reduce product costs but also improve product reliability while speeding up product development.
Break the bottlenecks that restrict efficiency
Good design requires a high level of production to achieve.
In the manufacturing process, AI technology has proven its capabilities and value in many industries. For example, in the field of automobile manufacturing, in June this year, the BMW Group introduced the general-purpose robot Figure01 at its Spartanburg plant in South Carolina, USA. This robot is driven by an AI vision model and can accurately place metal parts and automatically correct errors during execution through neural network learning. In the field of aviation manufacturing, Airbus has integrated AI technology and computer vision technology into its production process, significantly improving the accuracy of aircraft assembly. At the same time, AI technology can also automatically record the installation of key components by analyzing video data and detect whether there are installation problems.
Although such AI robots are still in the exploratory stage, in the view of some high-end manufacturing company executives, the focus of future AI technology is not whether to use it, but how to use it. A broader application scenario of AI technology in the manufacturing process is intelligent production line monitoring and intelligent product quality control. By using AI technology to monitor the production process, manufacturing companies can optimize process parameters and adjust the operating status of the production line in real time. When AI image recognition technology is used for product inspection and quality control, some minor production defects can be discovered in time, thereby improving the accuracy of the product. Therefore, it is more suitable for chip manufacturing, aviation parts production, etc. that have extremely high accuracy requirements. industry.
Aero-engine components have complex structures and have high requirements for production accuracy. Combining AI technology with emerging technologies such as industrial robots, digital twins, virtual reality (VR), augmented reality (AR), additive manufacturing (3D printing), and data The integration of industrial software such as collection monitoring and production execution systems (MES) can reduce human participation in complex and harsh environments, improve the accuracy and production efficiency of the manufacturing execution process, and thereby improve product quality. In this regard, engine manufacturer Rolls-Royce is actively exploring the use of AI technology in engine design and manufacturing to predict and solve potential problems in the production process to ensure that each component can meet strict accuracy requirements.
At present, some of the more fascinating explorations are to combine 3D printing technology with AI technology to break through the manufacturing bottleneck in the aerospace field. In the field of aerospace engines, the application of 3D printing technology is gradually expanding. As the aerospace engine field that currently uses 3D printing technology most extensively, GE Aerospace used more than 300 3D printing technology parts in the development of the GE9X engine. Through the flexible use of a variety of new materials, 3D printing technology not only improves the production accuracy of complex parts, but also effectively reduces the weight of parts and greatly shortens the production cycle. However, although 3D printing technology is favored by engine manufacturing companies, it is still restricted by printing efficiency. In this regard, some technology companies have proposed that combining AI technology with 3D printing technology is expected to significantly shorten material research time and improve the manufacturing efficiency of engine parts.
“AI surgical lamp” that serves as “airplane doctor”
In the service guarantee link, by sensing the product usage status in real time, AI technology can quickly formulate maintenance and repair plans, build spare parts prediction and optimal configuration models, realize predictive maintenance of aero engines, and improve service guarantee capabilities. At present, many large aviation manufacturing companies have used AI technology as a blade inspection tool for aero-engines, shortening the original inspection time from 3 to 4 hours to 30 to 45 minutes, which can save companies hundreds of millions in inspection costs.
In fact, AI technology not only empowers large enterprises. During the on-site maintenance of aerospace engines, detection tools driven by AI technology help standardize operations, improve personnel work efficiency and work quality, and shorten maintenance and repair time. And some "aircraft doctors" who pay attention to new technologies, that is, aircraft maintenance personnel, have begun to build their own "AI surgical lights."
At Guangzhou Baiyun Airport, Luo Chenggong, a "post-90s" maintenance staff of Guangzhou Aircraft Maintenance Engineering Co., Ltd. (GAMECO), used China's first self-developed industrial-level deep learning platform "Flying Paddle" to create a "Flying Paddle" for maintenance staff. "AI surgical light" - aircraft defect recognition model. In the process of building this model, training the model is the first step, which requires importing a large amount of collected data and images into the system to help it perform machine learning.
In traditional post-flight work, aircraft maintenance personnel need to spend about an hour visually inspecting the aircraft to ensure that all facilities and equipment, including the aircraft engine, are normal and meet operational requirements. After completing the training of the aircraft defect identification model, Luo Cheng began to test whether it could improve the efficiency and accuracy of visual inspections in actual work. The results showed that the model successfully identified that a screw in the aircraft was loose and made a "didi" sound, indicating that AI technology also has great potential in front-line work.
McKinsey & Company pointed out in "Generative Artificial Intelligence Opportunities in Aviation Maintenance" released in August this year that without the aircraft maintenance, repair and overhaul (MRO) services that operate behind the scenes, the civil aviation industry would not be able to complete safe transportation around the world every day. It is an astonishing feat that it has carried nearly 10 million passengers and flown more than 20 billion kilometers. But today, the industry is facing unprecedented challenges. The rapid growth in demand for business aviation travel, the global shortage of aircraft, and the maintenance backlog caused by the COVID-19 epidemic have continued to increase airlines' demand for MRO services. As airlines strive to meet the growing demand for passenger travel when the supply of new aircraft is limited, the MRO industry must ensure the availability and reliability of existing aircraft and extend their service life. With the rapid development of science and technology, the key to solving these problems and seizing these opportunities points to artificial intelligence.
As AI technology continues to mature, its application in the field of aerospace engines will become more extensive and in-depth. From predictive maintenance to improving fuel efficiency to intelligent fault diagnosis, AI technology innovation provides strong support for improving the performance and operational reliability of aero engines. Looking to the future, with the support of new technologies such as artificial intelligence, aircraft engines will evolve in a more intelligent, environmentally friendly and efficient direction, not only laying the foundation for the sustainable development of the aviation industry, but also improving the safety and economy of the global aviation industry. bring new breakthroughs. (China Civil Aviation News reporter Wang Yichao)
Experts talk
The entire life cycle of the aviation manufacturing industry will be closely connected with data
Liu Yi
AI technology empowers aviation manufacturing and optimizes the entire life cycle of "design-manufacturing-maintenance". It is a cutting-edge topic and an important manifestation of the expansion of AI technology in the digital aviation field. From the current point of view, the impact of the development of AI technology on aviation manufacturing is mainly reflected in three links.
In the design process, the new generation of AI technology has the ability to further integrate with CAD (computer-aided design)/CAE (computer-aided engineering). This is essentially a transition from the "third paradigm (computing science) to the fourth paradigm (data intensive science)” shift. For example, in the simulation analysis of digital models with multiple physical fields such as structural strength, vibration noise, and heat flow coupling, AI technology can enable the flow and integration of aircraft design knowledge and data across time, space, fields, and units, and is used for Automatically generate high-quality mesh models to optimize solver parameter settings and improve simulation efficiency and accuracy.
The “troika” of the new generation of AI technology—computing power, algorithms, and data—are all driving innovation in productivity in the field of aircraft design. Among them, the rapid growth of computing power meets the performance requirements of large-scale numerical calculations in design simulation; intelligent algorithms on the one hand enable the design to adapt to higher dimensions and more variables, and on the other hand give rise to the relationship between professional design and generative design. A new situation of integration; and the core ability of large models to process, understand and create large amounts of data information exactly meets the highly segmented and highly professional data processing needs in aviation manufacturing.
In the manufacturing process, the aviation industry has extremely high requirements for manufacturing accuracy and processing quality. Compared with the traditional post-sampling statistical quality control scheme, the new generation of AI technology can meet the real-time requirements of quality control. For example, multi-modal capabilities based on AI technology can integrate different sensor data to control and optimize the production process in real time; apply AI technology to each inspection link of production and manufacturing, and use the knowledge and data sets of domain experts for training, which can be adaptive It can continuously learn various sensor data and feedback information to improve the accuracy of product defect detection; and for personalized and flexible intelligent manufacturing needs, AI technology can continuously learn and understand the patterns and characteristics of production and manufacturing data in the current environment, and then provide Develop implementation plans for specific tasks. In addition, AI technology can empower robots with perception, analysis, and decision-making capabilities, such as understanding human instructions based on natural language processing models, judging position information based on machine vision models, and realizing processing path planning based on intelligent decision-making algorithms. Repeat in part Significantly reduce manpower in highly sensitive, standardized or high-risk scenarios, and efficiently coordinate personnel to carry out work.
In the maintenance process, AI technology will efficiently optimize the core link of aircraft maintenance from "manuals" to "work orders", that is, realize automation and intelligence from "professional knowledge base" to "work list". When an aircraft malfunctions, AI based on the fault diagnosis expert system and knowledge graph technology can intelligently analyze the fault phenomenon and operating data, quickly determine the cause of the fault, directly generate maintenance plans and decision-making suggestions, and guide maintenance personnel to perform precise maintenance and component replacement. Fault diagnosis and maintenance knowledge can also be backfilled into the knowledge base to continuously improve and optimize the diagnostic model. Using machine learning algorithms, such as deep neural networks and long-short-term memory networks, to perform feature extraction and pattern recognition on aircraft maintenance knowledge and operating data, we can build prediction models for equipment degradation and fault evolution. Through model training and verification, we can accurately predict equipment remaining service life and potential failure risks to achieve proactive predictive maintenance and optimization of aircraft.
BD (big data) and AI are the core of technological progress in new productivity. In the future, the entire life cycle of aircraft design, manufacturing, and maintenance will be closely connected with BD and AI, which requires the joint efforts of basic civil aviation data governance, data to domain knowledge transformation, and intelligent vertical application scenario research. In addition, we should note that on the one hand, the aviation manufacturing scene is characterized by complexity, professionalism, and dynamics. The interpretability and safety of the current new generation of AI technology are issues that need to be solved for large-scale production applications; on the other hand, With the deepening of intelligent transformation, more and more unmanned scenarios will emerge, and decisions should be made based on comprehensive consideration of factors such as efficiency and cost, which tasks should be completed by machines and which should be completed by humans. This is a process of human-machine collaboration. Blindly pursuing unmanned technology may deviate from the origin of intelligent manufacturing. (The author is the director of the Big Data and Artificial Intelligence Department of the Civil Aviation Administration of China and the executive director of the Key Laboratory of Data Governance and Decision Optimization of the Civil Aviation Administration of China)