Artificial Intelligence Applications in the Aviation Industry: From

Design to Maintenance for Aircraft

Jiahao Chen

Chengdu Foreign Languages School, Chengdu, Sichuan, China

Keywords: Artificial Intelligence, Aviation Engineering, Aircraft Lifecycle, Digital Twins, Smart Manufacturing.

Abstract: Recently, the application of artificial intelligence in the aviation industry has received widespread attention.

However, the existing literature review in this field remains fragmented and lacks a systematic and holistic

approach. To fill this gap, this paper focuses on the whole life cycle of an aircraft, starting from the four key

phases of design, manufacturing, operation, and maintenance, and systematically combs through the typical

applications and representative technologies of AI in the aerospace industry, covering the optimization of

deep learning in aerodynamic layout, the combination of digital twins and intelligent manufacturing, the

intelligent analysis of flight operation data, and the predictive maintenance system and other key cases.

Meanwhile, this paper also focuses on the core challenges of data quality, model credibility, system

integration, and security faced in the current implementation process and proposes a preliminary outlook on

future research directions such as credible AI, human-machine collaboration, and full-process data closure.

As a review study, this paper builds a framework for aircraft life cycle by structurally organizing the existing

results and practice cases, which provides a reference for promoting the deeper application of AI technology

in the aviation field.

1 INTRODUCTION

Traditional engineering processes in the highly

complex and safety-critical aviation industry are

facing unprecedented challenges. The increasing

technical sophistication of aircraft design structures,

rising manufacturing costs, real-time operational

scheduling requirements, and the trend toward

precision in maintenance and assurance have made it

difficult for traditional methods relying on

experience-driven and linear processes to meet the

combined requirements of efficiency, precision, and

adaptability of modern aviation systems.

Artificial Intelligence (AI), as a technology with

strong data processing, pattern recognition, and

optimization capabilities, is gradually becoming an

important driving force for the intelligent

transformation of the aviation industry. In the design

phase, AI technology can be used for aerodynamic

layout optimization, structural topology

reconstruction, and multi-objective decision support,

effectively shortening the design cycle and improving

design quality. In the manufacturing stage, by

combining computer vision, digital twin technology,

and intelligent robots, AI realizes efficient

identification of defects in parts, real-time monitoring

of equipment status, and optimal configuration of

production processes. At the operation level, AI is

widely used in route scheduling optimization, fuel

consumption prediction, and flight data anomaly

detection, helping airlines improve operational

efficiency and reduce costs. In maintenance, the

predictive maintenance system models and analyzes

the remaining life, potential failures, and health status

of equipment based on AI, realizing the

transformation from “planned maintenance” to

“dynamic maintenance based on status”, thus

ensuring flight safety and reducing the waste of

maintenance resources at the same time. This process

reduces the waste of maintenance resources while

ensuring flight safety.

Although AI has demonstrated its application

value in several key nodes of aviation, most current

research and practices still focus on a single link or a

localized scenario and lack a systematic sorting and

integration of the complete lifecycle of the aircraft.

This fragmented application model not only limits the

synergistic potential of AI technology but also does

not help to evaluate its overall impact on efficiency,

safety, and cost structure at the system level.

496

Chen, J.

Artiﬁcial Intelligence Applications in the Aviation Industry: From Design to Maintenance for Aircraft.

DOI: 10.5220/0014361800004718

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 2nd International Conference on Engineering Management, Information Technology and Intelligence (EMITI 2025), pages 496-503

ISBN: 978-989-758-792-4

Therefore, this paper will start from the four core

phases of “design-manufacturing-operation-

maintenance”, systematically review the main

application paths and key technologies of AI in the

whole life cycle of aircraft, and deeply analyze the

challenges faced during its actual deployment. It also

analyzes the challenges faced in the actual

deployment process, including the difficulties in data

acquisition, low credibility of models, and the

complexity of engineering system integration. At the

same time, this paper will also explore the future

development trends, such as the construction of the

full-process digital twin system, the realization of the

autonomous decision-making system, and the key

role of Explainable AI in the aviation safety system.

Through the systematic analysis of these issues, we

aim to construct a set of prospective and sustainable

AI application research frameworks for the field of

aviation engineering and provide theoretical support

and methodological guidance for subsequent

academic research and industrial practice.

2 APPLICATION OF ARTIFICIAL

INTELLIGENCE

TECHNOLOGY IN AIRCRAFT

DESIGN

Aircraft design is one of the most complex and critical

aspects of aeronautical engineering, involving

multiple perspectives of aerodynamics, structural

mechanics, materials science, flight control system

design, and comprehensive economic evaluation.

2.1 Deep Learning and Machine

Learning Enabling Aerodynamic

Layout Design

In aerodynamic layout design, AI has been widely

used to quickly predict the aerodynamic

characteristics of aircraft configurations. Although

traditional Computational Fluid Dynamics (CFD)

methods are highly accurate, they are expensive and

time-consuming, making it difficult to meet the

demand for efficient evaluation in the preliminary

design stage.

Modern methods for aerodynamic data surrogacy

modeling based on deep neural networks (DNN),

support vector regression (SVR), and other models

have been developed. Researchers have created a

simpler model for changing Mach number unsteady

flow using long short-term memory networks

(LSTM), which can manage large amounts of training

data. This model can accurately predict aerodynamic

load characteristics across a wide range of Mach

numbers. Through predictive experiments on the

aerodynamic forces and aerodynamic elastic

responses of the NACA 64A010 airfoil under

transonic conditions, the results show that this DL

model can accurately capture linear and nonlinear

aerodynamic characteristics at different Mach

numbers, vibration amplitudes, and frequencies,

significantly improving the generalization

performance and prediction accuracy of the model

compared with traditional aerodynamic analysis

methods (Zou & Sun, 2021). It is particularly

noteworthy that the LSTM-ROM model reduces

errors by 77% and improves computational efficiency

by 90% in the prediction of transonic flutter in the

NACA 64A010 airfoil. This DL-based method uses

fewer computer resources than traditional

computational fluid dynamics simulations while still

providing more accurate models.

“Aerodynamic Intelligent Topology Design

(AITD)-A Future Technology for Exploring the New

Concept Configuration of Aircraft” proposes an AI-

driven aerodynamic topology design method that

combines topology parameterization with DL to

break through the limitations of traditional empirical

design and achieve efficient prediction of transonic

flow fields (Liao et al., 2023).

In the AIAA Journal, Brunton et al. highlighted

that using data-driven methods instead of detailed

CFD models for quick estimates is a key way to

enhance efficiency, particularly for modeling

complex boundary layer behavior and predicting

trans-Mach number issues (Brunton et al., 2021).

The application of DL algorithms has, to some

extent, improved the efficiency and quality of the

design of aircraft aerodynamics.

2.2 Application of Multi-Objective

Optimization and Generative

Design in Aircraft Configuration

Exploration

The preliminary design stage of an aircraft requires a

comprehensive evaluation of multiple performance

objectives, such as aerodynamic efficiency, structural

strength, weight, manufacturability, and cost. This

process typically involves multi-objective, multi-

constraint optimization problems. Traditional

optimization methods are limited by computational

resources and design space search capabilities,

making it difficult to converge quickly on high-

quality solutions under complex coupled conditions.

With the introduction of artificial intelligence,

Artiﬁcial Intelligence Applications in the Aviation Industry: From Design to Maintenance for Aircraft

497

especially the development of machine learning and

generative modeling technologies, new opportunities

have emerged for aircraft configuration exploration.

Multi-objective optimization (MOO) is a key

component of design automation. Researchers have

combined swarm intelligence algorithms such as

genetic algorithms, particle swarm optimization, and

Bayesian optimization with agent-based modeling to

achieve more efficient design space search

capabilities. For example, Zou and Sun summarized

AI algorithms in the multidisciplinary design

optimization of aircraft aerodynamics and structures,

pointing out that AI is particularly suitable for

nonlinear optimization problems with obvious

conflicting objectives and can assist engineers in

quickly providing feasible solutions by balancing lift-

drag ratio, stability, and structural weight (Zou &

Sun, 2021).

Generative design has also been a hot topic of

research in the aviation industry in recent years. Its

core idea is to use AI to automatically generate

multiple design solutions based on objective

functions and design constraints for engineers to

select from. Unlike traditional parametric design, the

generative method does not rely on manually set

design templates and has strong divergent and

innovative capabilities. On this basis, Liao et al.

proposed the Aerodynamic Intelligent Topology

Design (AITD) method, which combines topology

optimization with DL to automatically generate

configurations with novel structural features through

multi-objective mapping learning of the

configuration space, which can be used for high lift-

to-drag ratio design exploration (Liao et al., 2023).

AITD significantly expands the configuration search

space and breaks through the limitations of human

experience in the traditional design process.

2.3 Application of More Advanced

Digital Prototyping Technology

In aircraft development, digital prototype (DP)

technology is widely used to validate the

manufacturability of structural, electrical, and

assembly designs, particularly playing a critical role

during the design phase of complex customized

configurations. By constructing multi-dimensional

product models, engineers can simulate production

processes early in the design phase, identify potential

assembly conflicts and manufacturing bottlenecks in

advance, thereby reducing the cost of later

modifications and improving design iteration

efficiency.

Moenck et al. noted that in fields with significant

customization requirements, such as aircraft interiors,

DP supports the synchronized development of

product configurations and manufacturing processes,

breaking down traditional barriers between design

and manufacturing (Moenck et al., 2023). Compared

to document-based development methods, digital

prototypes provide a visual, interactive design

platform that enables different engineering

departments to collaborate on optimizing design

parameters and process paths within the same

environment. Although this technology does not yet

have real-time feedback and synchronized

operational status capabilities, as a precursor to

digital twin systems, it has laid an important

foundation for the intelligent and closed-loop control

of subsequent manufacturing processes.

3 APPLICATION OF AI IN

AIRCRAFT MANUFACTURING

Currently, the international aviation industry is

gradually introducing AI-driven systems to improve

manufacturing efficiency, precision, and safety. AI

has made the manufacturing process increasingly

dependent on data-driven methods, enabling real-

time status monitoring, predictive quality control, and

comprehensive optimization of the production

process.

3.1 Practical Application of Digital

Twin Technology in Aircraft

Manufacturing

The concept of digital twins was first proposed by

NASA for the remote monitoring of their spacecraft.

Recently, this technology has expanded from

aerospace engineering to multiple fields such as

industrial manufacturing, energy, and power. Digital

twin technology has garnered significant attention in

the aerospace manufacturing industry since the 21st

century because it uniquely addresses three critical

requirements: high complexity, high customization,

and high reliability. With the synergistic development

of hardware and software, this technology has

evolved into a virtual mapping system spanning the

entire product lifecycle, playing a pivotal role in

driving intelligent transformation within aircraft

manufacturing processes.

In “Digital Twins in Aircraft Production:

Challenges and Opportunities”, Moenck et al.

summarize several specific applications of digital

EMITI 2025 - International Conference on Engineering Management, Information Technology and Intelligence

498

twin technology in aerospace manufacturing,

covering various stages such as assembly, logistics,

and quality inspection. In flexible assembly

scenarios, the digital twin system achieves path

planning and dynamic guidance for aircraft cabin

interior components through real-time modeling and

on-site data fusion. For example, by combining with

a laser projection system, operators can complete the

installation of complex foam honeycomb structures

based on virtual models, effectively improving the

consistency and efficiency of human-robot

collaborative assembly (Moenck et al., 2023).

In the production and supply chain process, the

aviation manufacturing industry often faces

challenges in identifying a large number of unlabeled,

irregularly shaped components. Traditional methods

relying on manual labor or physical labels are prone

to failure in scenarios such as modification or

refurbishment, and manual identification is costly and

inefficient. Researchers have pointed out that AI

training images can be synthesized using existing 3D

models in digital twin systems to construct visual

recognition models and achieve automatic

recognition of unlabeled parts (Alexopoulos et al.,

2020; Manettas et al., 2021; Schoepflin et al., 2021;

Schoepflin et al., 2022). This method has shown

excellent adaptability in production environments

with a wide variety of parts and small batches with

high variability, improving the flexibility and

intelligence of aviation logistics.

Digital twin technology also plays a key role in

the quality inspection process. Based on previous

research, Moenck et al. combined key technologies

such as multi-model databases (Koch et al., 2022),

IoT platforms (Nițulescu & Korodi, 2020), and

optical assistance systems to construct a “digital

quality assurance twin” framework suitable for

aviation manufacturing scenarios and introduced the

“as-inspected twin” concept proposed by Kwon et al.

as the theoretical basis for quality data modeling and

closed-loop traceability (Kwon et al., 2020)

3.2 AI Intelligent Optimization System

and Automated Production Case

Study

In aircraft manufacturing, complex processes often

require extremely high demands on process stability

and precision control. Traditional production relies on

empirical judgment and post-production testing,

which not only results in delayed responses but also

makes it difficult to promptly identify potential

deviations. Brunton et al. pointed out that by

integrating embedded sensors into the manufacturing

system to collect real-time process data and

combining it with ML models trained on historical

production samples, predictions and early warnings

can be made before deviations exceed specifications,

thereby significantly improving the robustness (the

ability of a system to maintain stable operation or

achieve expected performance in the face of

disturbances, uncertainties, or anomalies) and

feedforward control capabilities of the production

process (Brunton et al., 2021). This approach reduces

reliance on manual quality inspections and provides a

data foundation for intelligent adjustments in

complex assembly processes, highlighting the critical

role of AI technology in “pre-emptive prediction” and

“process adaptation”.

In addition, with the help of embedded sensors

and data modeling technology, CNC machines can

monitor their operating status in real time and use

machine learning methods to predict tool wear trends,

effectively reducing equipment failure rates and

maintenance costs. Computer vision technology is

also widely used in automatic thread laying and

composite manufacturing processes, combining

image recognition and thermal imaging methods to

achieve online detection and intelligent classification

of process defects. In the assembly process,

researchers have also explored robotic systems that

integrate visual feedback and path planning

algorithms to improve automatic alignment accuracy

and human-machine collaboration efficiency through

dynamic recognition of the relative positions of parts

and trajectory optimization. The integrated

application of the above technologies is gradually

promoting aviation manufacturing from an

experience-driven process to a higher level of data-

driven and intelligent manufacturing (Brunton et al.,

2021).

4 APPLICATION OF ARTIFICIAL

INTELLIGENCE IN AVIATION

OPERATIONS AND

MAINTENANCE

After an aircraft leaves the production line and is

delivered, the focus of its lifecycle shifts to the

dimensions of operation and maintenance. Every day,

tens of thousands of flights take off and land

worldwide, supported by complex air traffic control

and flight scheduling systems, extensive health

monitoring networks, and high-intensity maintenance

and support mechanisms. Faced with ever-increasing

aviation demand and safety standards, traditional

Artiﬁcial Intelligence Applications in the Aviation Industry: From Design to Maintenance for Aircraft

499

methods are becoming increasingly inadequate in

terms of efficiency, response speed, and accuracy.

Against this backdrop, AI has begun to be used in key

areas related to flight efficiency and safety, such as

flight path optimization, flight status sensing, fault

prediction, and maintenance strategies.

4.1 Route Optimization and Scheduling

Decisions

Airlines face highly complex dynamic decision-

making challenges at the scheduling level, requiring

the simultaneous coordination of multiple resource

constraints across dimensions such as route networks,

fleet utilization, passenger demand, weather changes,

and ground support. With the accumulation of data

and improvements in computing power, AI,

especially algorithms represented by Reinforcement

Learning (RL) and DL, is gradually becoming an

important tool for route planning and operational

optimization. Tafur et al. pointed out that RL methods

can effectively address airspace resource fluctuations

and path congestion issues and, through continuous

iterative training, obtain optimal route strategies,

thereby improving route utilization and flight

punctuality (Tafur et al., 2025).

In practical applications, the Operations Process

Support and Decision Suite (OPSD) developed by

Lufthansa has been deployed in operations. The

system integrates flight schedules, weather data,

maintenance status, and passenger information, with

AI models generating scheduling plans for

operational staff to adopt. The acceptance rate of its

recommended plans exceeds 90%, demonstrating

outstanding performance in improving resource

allocation efficiency and system robustness

(Lufthansa Group, 2023).

Qantas Airways has introduced a cloud-based

flight path simulation system called Constellation for

route optimization. The system was developed by

Qantas in collaboration with the University of

Sydney's Field Robotics Center over a period of five

years. It can simulate thousands of possible routes

based on millions of data points, such as weather and

wind speed, and select the most fuel-efficient route.

According to public statements by Qantas CEO Alan

Joyce, the system saves the company approximately

40 million Australian dollars in fuel costs annually

and reduces carbon emissions by approximately 50

million kilograms (Bice, 2013). The system has

already achieved practical results on routes such as

Sydney to San Diego, with fuel savings of up to one

ton per flight. The full deployment of Constellation

marks the large-scale application of AI in commercial

aviation scheduling, and its concept of “pre-flight

route optimization” also provides the industry with an

innovative paradigm that combines energy

conservation and profitability (Bice, 2013). This

instance demonstrates that AI enhances aircraft

operational efficiency while also exhibiting

significant promise for energy sustainability and cost

management. AI enhances the operational efficiency

of flight scheduling and serves as a crucial tool for

airlines to attain environmental protection objectives,

save operating expenses, and augment revenues while

maintaining safety.

At present, many international airports use AI-

related technologies to assist in management, control,

and decision-making. For instance, Singapore's air

traffic control department uses a gradient boosting

algorithm to optimize runway allocation, reducing

takeoff delays by about 15%; Frankfurt Airport uses

a DL model to predict airflow disturbances, helping

flights adjust their flight paths in real time and

significantly improving punctuality (Tafur et al.,

2025).

In addition, AI has been used in Air Traffic

Management (ATM) systems for critical tasks such as

flight trajectory prediction and conflict avoidance. As

mentioned in the literature, strategy models based on

Markov Decision Processes (MDP) can dynamically

select the optimal flight path, effectively alleviating

scheduling pressure in high-density airspace (Tafur et

al., 2025).

In summary, flight route optimization and

scheduling decisions are gradually transitioning from

rule- and experience-based manual processes to data-

driven, model-assisted human-machine collaborative

systems.

4.2 Flight Data Monitoring and

Anomaly Identification

Modern aircraft generate vast amounts of operational

data during each flight, such as flight altitude, speed,

throttle position, engine vibration, and attitude angle

changes. This data is continuously recorded by the

Flight Data Recorder (FDR) and other sensor

systems. Extracting meaningful information from this

multidimensional, dynamic, and unstructured data

has become one of the key challenges for flight safety

monitoring and fault warning systems. Currently,

researchers have begun to gradually introduce ML

methods into flight data analysis, using models to

learn “normal operating conditions” so that potential

anomalies can be identified in real time during flight.

Some algorithms use unsupervised learning strategies

to extract “deviation from trajectory” patterns from

EMITI 2025 - International Conference on Engineering Management, Information Technology and Intelligence

500

historical data to provide early warnings of possible

faults or manipulation errors (Tafur et al., 2025).

The primary application of such AI models lies in

automatically identifying invisible trends and early

failure signals from complex flight data, particularly

excelling in engine health assessment and the

identification of abnormal flight attitudes. For

instance, research has utilized LSTM networks to

model multiple time-series flight parameters,

enabling effective prediction of risk states several

minutes into the future and providing pilots with

intelligent assistance. Additionally, convolutional

neural networks (CNNs) and clustering methods have

been applied to flight data visualization and multi-

category fault classification, enhancing the system's

sensitivity and response speed under complex

operational conditions (Tafur et al., 2025).

With developments in modeling capabilities and

real-time data processing technology, AI's role in

flight data analysis is shifting from “post-event

attribution” to “in-flight identification and

prediction”, enabling more timely decision assistance

for risk management.

4.3 Predictive Maintenance and Health

Management

For aircraft with long operational cycles and complex

structures, extending service life and reducing

maintenance costs without compromising safety is a

critical task in aviation operations. Compared with

traditional periodic maintenance and post-accident

repairs, Predictive Health Management (PHM)

emphasizes continuous monitoring of equipment

operating conditions and trend analysis so that

interventions can be taken before failures occur. This

method relies on large amounts of flight and sensor

data and uses ML and statistical modeling methods to

achieve a shift from “planned maintenance” to

“condition-driven maintenance”.

NASA's public C-MAPSS engine dataset has

become a benchmark testing platform in this field,

with numerous studies using it as a basis for

developing Remaining Useful Life (RUL) prediction

models (NASA Ames Prognostics Data Repository,

n.d.). For example, researchers have constructed a

regression structure based on Convolutional Neural

Networks (CNN) to extract temporal patterns from

engine sensor data and predict the remaining

operating cycles of the equipment in its current state

(Dong et al., 2021); other methods utilize LSTM

networks to model vibration and thermal data,

enabling early identification of potential failures, with

some models capable of issuing warnings 10 to 20

seconds before a failure occurs.

In addition to life prediction, condition diagnosis

and root cause analysis are also critical components

of PHM systems. Brunton et al. point out that health

management should not only be able to predict “when

problems will occur” but also explain “what problems

will occur” and “why they will occur” (Brunton et al.,

2021). They propose feeding the results of anomaly

identification back into the design and manufacturing

stages to form a cross-stage maintenance closed loop.

This two-way path from data monitoring to design

optimization is changing traditional maintenance

logic, making aircraft operation and maintenance

processes more transparent, efficient, and forward-

looking.

5 CHALLENGES, LIMITATIONS,

AND FUTURE PROSPECTS

Despite the extensive integration of artificial

intelligence in the design, manufacture, and

maintenance of airplanes, its comprehensive

deployment continues to encounter several practical

obstacles. AI must resolve several critical challenges,

including the construction of trust mechanisms, data

quality, model flexibility, and system integration, to

attain persistent and profound use in this high-safety,

high-dependence industry.

5.1 Reliable Foundations for AI in

Aviation Systems

In fields such as aviation, where there is zero

tolerance for safety issues, artificial intelligence must

earn trust by simultaneously focusing on model

transparency and engineering safeguards. Research

indicates that “explainable artificial intelligence”

(XAI) is one of the key pathways to establishing trust

mechanisms. In recent years, various methods have

been employed to enhance model reviewability,

including feature weight analysis, visualization

modules, and rule embedding (Kobayashi & Alam,

2022).

In current engineering practices, AI is more often

embedded as a “recommendation system” rather than

a direct controller. For example, in Lufthansa's OPSD

system, AI provides flight scheduling

recommendations, which are then adopted by human

operators (Lufthansa Group, 2023); similarly, Qantas'

Constellation system assists in pre-flight planning by

simulating optimal flight paths, but the final decision

Artiﬁcial Intelligence Applications in the Aviation Industry: From Design to Maintenance for Aircraft

501

remains with the pilot or control center (Bice, 2013).

This “recommendation-first, human-in-the-loop”

model effectively reduces the risk of loss of control

caused by algorithmic misjudgment.

Additionally, the industry widely adopts

technologies such as multi-model verification, fault-

tolerant redundancy, and safety envelope design to

enhance system stability and recovery capabilities in

extreme scenarios (Kobayashi & Alam, 2022). AI is

no longer a traditional “replacement” but rather a

“third eye” capable of detecting subtle anomalies in

advance, thereby improving the system's data fusion

efficiency, operational organization capabilities, and

forward-looking judgment.

5.2 Challenges Encountered: Multiple

Constraints on Data, Models, and

Systems

The implementation of AI in aviation still faces

numerous practical challenges. First, high-quality

labeled data is scarce, particularly in critical tasks

such as engine fault prediction and PHM, where

abnormal samples are extremely limited (Brunton et

al., 2021). Second, the interpretability and stability of

deep learning models still need improvement, making

it difficult to completely replace traditional

mechanisms in high-safety scenarios (Kobayashi &

Alam, 2022).

In addition, AI systems must operate in

conjunction with traditional aviation information

architectures, facing engineering constraints such as

system compatibility, real-time performance, and

computing resources. In practical applications, data

drift, input anomalies, or attacks can all pose systemic

risks. These issues determine that the evolution of

aviation AI requires multi-party collaboration

covering algorithms, data, systems, and security,

rather than just improving model accuracy.

5.3 Trend Outlook and Future

Research Directions

As algorithms advance and aviation's digital

infrastructure enhances, the function of AI in aviation

is progressively transitioning from a "auxiliary tool"

to a "system backbone”. Future research will

emphasize multi-source data fusion, continuous

learning processes, and human-machine

collaboration frameworks. Joint modeling that

integrates sensor data, maintenance records, and

flight logs is anticipated to improve the system's

condition recognition skills and resilience.

At the same time, the verifiability and traceability

of AI will become prerequisites for its in-depth

application in mission-critical scenarios. AI will not

only be a technical tool for improving efficiency but

will also become a long-term driving force for the

intelligent transformation of aviation systems in

terms of system adaptability, operational resilience,

and risk prevention and control.

6 CONCLUSIONS

AI is becoming increasingly embedded in the core

processes of the aviation industry. From its initial role

as an assisting tool, AI has gradually developed into

an intelligent component in key areas such as design,

manufacturing, operation, and maintenance. AI is no

longer just a technical means of improving efficiency,

but is also influencing the design concepts and

engineering logic of aviation systems.

This article systematically reviews typical

application paths for AI in aerodynamic design

optimization, intelligent manufacturing systems,

flight operation scheduling, predictive maintenance,

and health management throughout the entire life

cycle of aircraft. These technologies have already

shown initial success in shortening development

cycles, improving reliability, and reducing operating

costs. At the same time, AI technology is also driving

the industry toward higher levels of digitization,

automation, and intelligence.

The in-depth application of AI in aviation is not

limited to improving algorithm capabilities, but also

depends on the systematic construction of its

“reliability, transparency, and usability”. Future

technological advancements require the

establishment of a complete closed-loop system in

areas such as standardization, data governance, model

validation, and system integration, driving AI from

“black-box reasoning” toward “explainable decision-

making” and from “support tools” toward “decision-

making partners”. This is not only a technical

challenge but also necessitates the restructuring of

engineering systems, organizational structures, and

management mechanisms.

Artificial intelligence is not destined to be a short-

term technological “stimulus” in the aviation

revolution but rather a strategic partner deeply

involved in the long term, continuously influencing

the operational rules and value judgments of aviation

systems. It will build a bridge of balance between

efficiency and safety, establish a platform between

complex machine systems and human intelligence,

and gradually reshape the form and structure of the

EMITI 2025 - International Conference on Engineering Management, Information Technology and Intelligence

502

aviation industry. Future aviation engineering will no

longer be a simple combination of mechanical and

electronic systems but rather a cross-integration of

human intelligence and AI. Building an efficient,

reliable, intelligent, and controllable future aviation

system is not only an exploration of technology itself

but also a profound response to how humans can

coexist and co-create with AI.

REFERENCES

Alexopoulos, K., Nikolakis, N., & Chryssolouris, G. (2020).

Digital twin-driven supervised machine learning for the

development of artificial intelligence applications in

manufacturing. International Journal of Computer

Integrated Manufacturing, 33(5), 429–439.

Bice, C. (2013, October 21). Qantas cloud-based flight sim

saving millions in fuel. CIO. Retrieved from

https://www.cio.com/article/213940/qantas-cloud-

based-flight-sim-saving-millions-in-fuel.html

Brunton, S. L., Kutz, J. N., Manohar, K., Aravkin, A. Y.,

Morgansen, K., Klemisch, J., Goebel, N., Buttrick, J.,

Poskin, J., Blom-Schieber, A. W., Hogan, T., &

McDonald, D. (2021). Data-driven aerospace

engineering: Reframing the industry with machine

learning. AIAA Journal, 59(8), 2819–2847.

Dong, Y., Wang, J., Xiang, Y., & Yu, D. (2021). Deep

learning in aircraft design, dynamics, and control:

Review and prospects. Aerospace Science and

Technology, 116, 107127.

Kobayashi, T., & Alam, M. (2022). Explainable AI for

safety-critical aerospace applications: Challenges and

prospects. IEEE Transactions on Aerospace and

Electronic Systems, 58(3), 1510–1524.

Koch, J., Lotzing, G., Gomse, M., & Schüppstuhl, T. (2022).

Application of multi-model databases in digital twins

using the example of a quality assurance process. In A.-

L. Andersen et al. (Eds.), Towards Sustainable

Customization: Bridging Smart Products and

Manufacturing Systems (Lecture Notes in Mechanical

Engineering, pp. 364–371). Springer, Cham.

Kwon, S., Monnier, L. V., Barbau, R., & Bernstein, W. Z.

(2020). Enriching standards-based digital thread by

fusing as-designed and as-inspected data using

knowledge graphs. Advanced Engineering Informatics,

46, 101102.

Liao, F., Zheng, Y., He, H., et al. (2023). Aerodynamic

intelligent topology design (AITD)—A future

technology for exploring the new concept configuration

of aircraft. Aerospace, 10(1), 46.

Lufthansa Group. (2023). How AI can reduce fuel

emissions in aviation. Retrieved from

https://www.lufthansagroup.com/media/downloads/en/

responsibility/LHG-CTH-

Sustainability_Whitepaper_Content_EN_FINAL.pdf

Manettas, C., Nikolakis, N., & Alexopoulos, K. (2021).

Synthetic datasets for deep learning in computer-vision

assisted tasks in manufacturing. In Proceedings of

CIRP Global Web Conference (Vol. 103, pp. 237–242).

Moenck, C., Behrendt, R., Weiland, J. E., & Dumstorff, H.

(2023). Digital twins in aircraft production: Challenges

and opportunities. CEAS Aeronautical Journal, 14,

1001–1018.

NASA Ames Prognostics Data Repository. (n.d.). C-

MAPSS: Turbofan engine degradation simulation data

set. Ames Research Center, CA, USA. Retrieved from

https://www.nasa.gov/intelligent-systems-

division/discovery-and-systems-health/pcoe/pcoe-

data-set-repository/

Nițulescu, I.-V., & Korodi, A. (2020). Supervisory control

and data acquisition approach in Node-RED:

Application and discussions. IoT, 1(1), 76–91.

Schoepflin, D., Holst, D., Gomse, M., & Schüppstuhl, T.

(2021). Synthetic training data generation for visual

object identification on load carriers. Procedia CIRP,

104, 1257–1262.

Schoepflin, D., Iyer, K., Gomse, M., & Schüppstuhl, T.

(2022). Towards synthetic AI training data for image

classification in intralogistic settings. In Annals of

Scientific Society for Assembly, Handling and

Industrial Robotics 2021 (pp. 325–336). Springer,

Cham.

Tafur, C., Stachon, A., & Scholz, S. B. (2025). Applications

of artificial intelligence in air operations: A systematic

review. Results in Engineering, 19, 101117.

Zou, T., & Sun, K. (2021). Application and prospect of

artificial intelligence in aircraft design. In 2021

International Conference on Networking Systems of AI

(INSAI) (pp. 201–205). IEEE.

Artiﬁcial Intelligence Applications in the Aviation Industry: From Design to Maintenance for Aircraft

503