Advanced AI Boosts Multi-Failure Detection Precision in Aerospace Systems

By Muhammad OsamaReviewed by Susha Cheriyedath, M.Sc.Oct 31 2024

In a recent paper published in the journal Aerospace, researchers explored a non-contact approach to detect multiple failures in avionic systems. They introduced a novel, non-destructive algorithm for diagnosing failures in printed circuit boards (PCBs) in modern aircraft, supporting advancements in control and electrification.

This algorithm leverages advanced technology to improve the reliability and efficiency of PCB diagnostics, particularly in the aerospace field. It combines a self-attention mechanism with an adaptive neural network to enhance diagnostic accuracy. The goal was to address the complexity of multi-failure detection in PCBs, which is important for ensuring the safety and performance of avionic systems.

Advancement in the Aerospace Industry

Advances in aerospace and electronic information technology have transformed avionic systems. Today’s aircraft rely on complex electronic modules for essential functions like engine control, communication, and servo coordination. The move toward More Electric Aircraft (MEA), which replaces hydraulic and pneumatic systems with electrical ones, aims to reduce energy loss and improve efficiency, placing a high demand on reliable electronic systems.

PCBs, especially those used in radio frequency (RF) applications, are crucial for high-frequency tasks but can be challenging to diagnose due to their sensitivity to electromagnetic interference and complex structures. Effective diagnostic methods are essential to keep these systems stable and functioning properly.

Traditional contact-based diagnostics, while useful, can damage sensitive components and introduce interference that distorts measurements. As a result, there is growing interest in non-contact diagnostic methods that avoid these issues while ensuring accurate results.

Introduction of Novel Diagnostic Algorithm

This study presents a fast, accurate, and non-destructive algorithm for diagnosing multiple failures in complex PCBs used in aerospace applications. The algorithm consists of three main components: PCB feature extraction, feature enhancement, and model training.

The process begins with spatial dimension reduction, where two-dimensional (2D) spatial data is converted into a one-dimensional (1D) format. This transformed data is then combined with frequency data to create a unique 2D spatial/frequency coordinate system.

For feature extraction, the researchers utilized convolutional neural networks (CNNs) with a residual network (ResNet) architecture. To enhance feature analysis and reduce overfitting, they integrated a self-attention mechanism (SAM) with an adaptive convolutional graph neural network (AGCN). This approach enabled effective mapping of extracted features to various failure types, with a binary labeling system where each bit represents a specific failure.

The experimental setup included a dual-phase amplitude boosting circuit (DPAB) to simulate different PCB failure modes, allowing for comprehensive testing of the algorithm. Additionally, a near-field scanning technique was employed to collect electromagnetic data from the PCB under various operational conditions, supporting robust data collection and analysis.

Key Findings and Insights

The results showed that the proposed diagnostic technique performed well, particularly in detecting multiple simultaneous failures. The algorithm achieved high performance across several key metrics: overall precision (99.08 %), per-class precision (98.50 %), overall recall (98.78 %), per-class recall (98.01 %), overall F1 score (98.93 %), and per-class F1 score (98.25 %).

The study also found that using a 34-layer ResNet (ResNet34) for feature extraction struck an ideal balance between model complexity and diagnostic accuracy. Integrating the adaptive convolutional graph neural network (AGCN) further enhanced the model’s diagnostic capability. The researchers also highlighted the benefits of using electromagnetic data in PCB diagnostics, emphasizing its versatility compared to traditional approaches.

This non-contact method improved detection efficiency and minimized risks associated with contact-based diagnostics, making it especially suitable for high-frequency, low-noise circuits. Additionally, the algorithm proved effective in managing complex multi-failure scenarios, underscoring its robustness for real-world applications.

Applications

The diagnostic technique introduced in this study has huge potential for the aerospace industry, where fast, reliable PCB failure diagnosis is essential for the safety and performance of avionic systems. Its non-contact design is particularly beneficial in settings where direct PCB contact is challenging or undesirable.

With its high accuracy and efficiency, this algorithm is well-suited for other high-reliability applications, including satellite communications and space probes. It effectively diagnoses failures across various electronic systems, such as communication modules, radar systems, and other critical components in aircraft and spacecraft, making it ideal for environments where physical contact is impractical or could cause damage.

Beyond aerospace, this diagnostic approach could also benefit industries that rely on complex electronic systems, including telecommunications, automotive, and consumer electronics. Its capability to detect multiple failures in intricate PCBs improves maintenance efficiency and reduces downtime. Additionally, this technology could support advancements in automated diagnostics, contributing to the development of smart systems capable of self-monitoring and self-repair.

Conclusion and Future Directions

In summary, this novel diagnostic algorithm provides an effective solution for diagnosing multiple failures in complex electronic systems, with promising applications across aerospace and other high-reliability sectors. Its approach has the potential to transform PCB diagnostics by enhancing accuracy and efficiency in these critical fields.

Future work could expand the algorithm’s capabilities by incorporating multimodal feature fusion techniques to further improve diagnostic accuracy and adaptability. This research not only contributes to the reliability of aerospace systems but also establishes a foundation for future advancements in electronic diagnostics.

Journal Reference

Liu, C.; Ferlauto, M.; Yuan, H. A Non-Contact AI-Based Approach to Multi-Failure Detection in Avionic Systems. Aerospace 2024, 11, 864. DOI: 10.3390/aerospace11110864, https://www.mdpi.com/2226-4310/11/11/864

Disclaimer: The views expressed here are those of the author expressed in their private capacity and do not necessarily represent the views of AZoM.com Limited T/A AZoNetwork the owner and operator of this website. This disclaimer forms part of the Terms and conditions of use of this website.