Van Su Luong
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Research Interests
Van Su Luong has a total of 71 co-authors. Here are the top co-authors by number of shared publications:
Chih-Cheng Lu
11 shared publicationsJen-Tzong Jeng
8 shared publications- Anh Tuan Nguyen
Hanoi Univ Sci & Technol HUST, Int Training Inst Mat Sci ITIMS, 1 Dai Co Viet Rd, Hanoi 100000, Vietnam
6 shared publications - Minhhuy Le
Phenikaa University
6 shared publications Jen-Hwa Hsu
5 shared publications- Jen-Tzong Jeng
Department of Mechanical Engineering, National Kaohsiung University of Science and Technology, Kaohsiung, TAIWAN
4 shared publications
Latest Publications
- 9 January 2026
- Published by IOP Publishing in Measurement Science and Technology
- Vol. 37 (3) , 035109
High-fidelity embedded gas sensing: a unified deep learning framework for simultaneous identification and quantification
Electronic nose systems are a transformative technology, enabling low-cost, real-time gas analysis for critical applications in environmental monitoring, industrial safety, and medical diagnostics. However, their practical performance is fundamentally limited by the computational challenge of analyzing time-series data from their sensor arrays. Conventional machine learning (ML) pipelines rely on lossy feature engineering and employ separate, inefficient models for gas classification and concentration regression, a methodology that fundamentally limits predictive accuracy. This paper presents a unified multitask deep learning model that solves these problems through three key innovations. First, the model learns features directly from the complete, raw sensor time-series using a convolutional neural network, avoiding information loss from pre-processing. Second, a novel attention mechanism links the classification and regression tasks, allowing the model to use the identified gas type to improve its concentration estimate. Third, a custom loss function was developed to ensure high accuracy across a wide range of gas concentrations, a critical challenge in chemical sensing. Using 5-fold cross-validation, the model achieved 100% accuracy in gas classification and a mean absolute percentage error (MAPE) of 6.61% for concentration regression, results that significantly outperform classical ML benchmarks. We also demonstrated the model’s efficiency by deploying it on 10 different STM32 microcontrollers, confirming its suitability for real-time embedded applications. This work provides a robust framework for developing high-performance, AI-based chemical sensors and offers a foundational methodology for future analysis of complex gas mixtures. The source code and data are provided in this GitHub repo: https://github.com/quenx-rd/gas_identifier_public .
- 9 January 2026
- Published by IOP Publishing in Measurement Science and Technology
- Vol. 37 (3) , 035109
High-fidelity embedded gas sensing: a unified deep learning framework for simultaneous identification and quantification
Electronic nose systems are a transformative technology, enabling low-cost, real-time gas analysis for critical applications in environmental monitoring, industrial safety, and medical diagnostics. However, their practical performance is fundamentally limited by the computational challenge of analyzing time-series data from their sensor arrays. Conventional machine learning (ML) pipelines rely on lossy feature engineering and employ separate, inefficient models for gas classification and concentration regression, a methodology that fundamentally limits predictive accuracy. This paper presents a unified multitask deep learning model that solves these problems through three key innovations. First, the model learns features directly from the complete, raw sensor time-series using a convolutional neural network, avoiding information loss from pre-processing. Second, a novel attention mechanism links the classification and regression tasks, allowing the model to use the identified gas type to improve its concentration estimate. Third, a custom loss function was developed to ensure high accuracy across a wide range of gas concentrations, a critical challenge in chemical sensing. Using 5-fold cross-validation, the model achieved 100% accuracy in gas classification and a mean absolute percentage error (MAPE) of 6.61% for concentration regression, results that significantly outperform classical ML benchmarks. We also demonstrated the model’s efficiency by deploying it on 10 different STM32 microcontrollers, confirming its suitability for real-time embedded applications. This work provides a robust framework for developing high-performance, AI-based chemical sensors and offers a foundational methodology for future analysis of complex gas mixtures. The source code and data are provided in this GitHub repo: https://github.com/quenx-rd/gas_identifier_public .
- 5 October 2025
- Published by Springer Nature
- p. 103-109
Designing Circuitry to Enhance Sensitivity of Fluxgate Sensors Through High-Order Harmonic Response
- 5 October 2025
- Published by Springer Nature
- p. 103-109
Designing Circuitry to Enhance Sensitivity of Fluxgate Sensors Through High-Order Harmonic Response
- 26 September 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Sensors Journal
- Vol. 24 (21) , 35572-35579
Detection of Corrosion on Aircraft Structures Using an Electromagnetic Testing Sensor and a Spiking Convolutional Neural Network
Nondestructive testing (NDT) of aircraft intake structures is an important task to ensure the safety and integrity of the aircraft. Corrosion often appears at rivet locations and at the far-side surface or at hidden layers, presenting a challenge for NDT methods. To tackle this issue, we present a spiking NDT system that includes an electromagnetic testing probe with a spiking convolutional neural network (SCNN). The magnetic image of the rivet and corrosion are captured by a high spatial resolution Hall sensor array while scanning on the surface of the intake’s structure. The SCNN model is developed to analyze the obtained magnetic image and detect the presence of corrosion. The SCNN model, inspired by the principles of biological neurons, utilizes discrete and asynchronous spikes, resulting in remarkable energy efficiency. Our proposed SCNN model demonstrates a remarkable detection accuracy of approximately 93.67%, surpassing the performance of both the prior support vector machine model (89.48%) and a conventional CNN model (92.86%) on the same dataset. Importantly, when evaluated on the μBrain chip, our SCNN model exhibits power consumption approximately four orders of magnitude lower than that of a conventional CNN model running on an edge device (Jetson Nano). This substantial reduction in power consumption underscores the practicality and efficiency of our approach for concealed corrosion detection within aircraft intake structures, holding promises for enhanced maintenance practices in the aviation industry.
- 26 September 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Sensors Journal
- Vol. 24 (21) , 35572-35579
Detection of Corrosion on Aircraft Structures Using an Electromagnetic Testing Sensor and a Spiking Convolutional Neural Network
Nondestructive testing (NDT) of aircraft intake structures is an important task to ensure the safety and integrity of the aircraft. Corrosion often appears at rivet locations and at the far-side surface or at hidden layers, presenting a challenge for NDT methods. To tackle this issue, we present a spiking NDT system that includes an electromagnetic testing probe with a spiking convolutional neural network (SCNN). The magnetic image of the rivet and corrosion are captured by a high spatial resolution Hall sensor array while scanning on the surface of the intake’s structure. The SCNN model is developed to analyze the obtained magnetic image and detect the presence of corrosion. The SCNN model, inspired by the principles of biological neurons, utilizes discrete and asynchronous spikes, resulting in remarkable energy efficiency. Our proposed SCNN model demonstrates a remarkable detection accuracy of approximately 93.67%, surpassing the performance of both the prior support vector machine model (89.48%) and a conventional CNN model (92.86%) on the same dataset. Importantly, when evaluated on the μBrain chip, our SCNN model exhibits power consumption approximately four orders of magnitude lower than that of a conventional CNN model running on an edge device (Jetson Nano). This substantial reduction in power consumption underscores the practicality and efficiency of our approach for concealed corrosion detection within aircraft intake structures, holding promises for enhanced maintenance practices in the aviation industry.
- 2 July 2024
- Published by Elsevier in NDT & E International
- Vol. 146
Enhancing corrosion detection in pulsed eddy current testing systems through autoencoder-based unsupervised learning
- 2 July 2024
- Published by Elsevier in NDT & E International
- Vol. 146
Enhancing corrosion detection in pulsed eddy current testing systems through autoencoder-based unsupervised learning
- 23 September 2023
- Published by Springer Nature in Journal of Superconductivity and Novel Magnetism
- Vol. 36 (7-9) , 1767-1775
Fluxgate-Based Displacement Sensor Design
- 23 September 2023
- Published by Springer Nature in Journal of Superconductivity and Novel Magnetism
- Vol. 36 (7-9) , 1767-1775
Fluxgate-Based Displacement Sensor Design
- 1 September 2023
- Published by IOP Publishing in Measurement Science and Technology
- Vol. 34 (12) , 125907
Development of magnetic image super-resolution model for nondestructive testing
Electromagnetic testing (ET) is a method in the nondestructive evaluation (NDE) methodology for detecting and evaluating cracks in engineering structures by measuring the distribution of the electromagnetic field. With the development of advanced deep learning (DL) techniques, DL is getting attention in crack detection problems through the measuring signals of NDE systems. However, obtaining the necessary big data for such experiments is a time-consuming and expensive task. Moreover, the quality of the crack detection results heavily depends on the quality and spatial resolution of the input image data. In this research, we solve the above-mentioned problems by proposing a DL-based method on the magnetic image for both the image super-resolution and image denoise. We first build a magnetic image simulation framework for the ET system, and thus, it is possible to build a large dataset for training the DL model. Second, we build a DL model for enhancing the quality of the magnetic image in both the spatial resolution and the denoise signal. This approach helps to reduce the cost of experiments (i.e. reducing the number of sensors for each experiment), sensor fabrication process, and noise removal of the ET system. We evaluate the proposed approach on several cracks with different sizes and shapes of aluminum specimens. The evaluation metrics, such as the structural similarity index measure (SSIM) and root mean square error show good performance of the proposed approach and compared to traditional methods.
- 1 September 2023
- Published by IOP Publishing in Measurement Science and Technology
- Vol. 34 (12) , 125907
Development of magnetic image super-resolution model for nondestructive testing
Electromagnetic testing (ET) is a method in the nondestructive evaluation (NDE) methodology for detecting and evaluating cracks in engineering structures by measuring the distribution of the electromagnetic field. With the development of advanced deep learning (DL) techniques, DL is getting attention in crack detection problems through the measuring signals of NDE systems. However, obtaining the necessary big data for such experiments is a time-consuming and expensive task. Moreover, the quality of the crack detection results heavily depends on the quality and spatial resolution of the input image data. In this research, we solve the above-mentioned problems by proposing a DL-based method on the magnetic image for both the image super-resolution and image denoise. We first build a magnetic image simulation framework for the ET system, and thus, it is possible to build a large dataset for training the DL model. Second, we build a DL model for enhancing the quality of the magnetic image in both the spatial resolution and the denoise signal. This approach helps to reduce the cost of experiments (i.e. reducing the number of sensors for each experiment), sensor fabrication process, and noise removal of the ET system. We evaluate the proposed approach on several cracks with different sizes and shapes of aluminum specimens. The evaluation metrics, such as the structural similarity index measure (SSIM) and root mean square error show good performance of the proposed approach and compared to traditional methods.
- Open Access
Two-Dimensional Position Tracking Using Gradient Magnetic Fields
In this work, a two-dimensional (2D) position-detection device using a single axis magnetic sensor combined with orthogonal gradient coils was designed and fabricated. The sensors used were an induction coil and a GMR spin-valve sensor GF807 from Sensitec Inc. The field profiles generated by the two orthogonal gradient coils were analyzed numerically to achieve the maximum linear range, which corresponded to the detection area of the tracking system. The two coils were driven by 1-kHz sine wave currents with a 90° phase difference to generate the fields with uniform gradients along the x- and y-axis in the plane of the tracking stage. The gradient fields were detected by a single-axis sensor incorporated with a digital dual-phase lock-in detector to retrieve the position information. A linearity correction algorithm was used to improve the location accuracy and to extend the linear range for position sensing. The mean positioning error was found to be 0.417 mm, corresponding to the relative error of 0.21% in the working range of 200 mm × 200 mm, indicating that the proposed tracking system is promising for applications requiring accurate control of the two-dimensional position.
Open AccessTwo-Dimensional Position Tracking Using Gradient Magnetic Fields
In this work, a two-dimensional (2D) position-detection device using a single axis magnetic sensor combined with orthogonal gradient coils was designed and fabricated. The sensors used were an induction coil and a GMR spin-valve sensor GF807 from Sensitec Inc. The field profiles generated by the two orthogonal gradient coils were analyzed numerically to achieve the maximum linear range, which corresponded to the detection area of the tracking system. The two coils were driven by 1-kHz sine wave currents with a 90° phase difference to generate the fields with uniform gradients along the x- and y-axis in the plane of the tracking stage. The gradient fields were detected by a single-axis sensor incorporated with a digital dual-phase lock-in detector to retrieve the position information. A linearity correction algorithm was used to improve the location accuracy and to extend the linear range for position sensing. The mean positioning error was found to be 0.417 mm, corresponding to the relative error of 0.21% in the working range of 200 mm × 200 mm, indicating that the proposed tracking system is promising for applications requiring accurate control of the two-dimensional position.
- Open Access
- 20 May 2022
- Published by MDPI AG in Applied Sciences
- Vol. 12 (10) , 5175
Auto-Detection of Hidden Corrosion in an Aircraft Structure by Electromagnetic Testing: A Machine-Learning Approach
An aircraft is a multilayer structure that is assembled by rivets. Under extreme working conditions, corrosion appears and quickly propagates at the rivet sites of the layers; thus, it threads the integrity and safety of the aircraft. Corrosion usually occurs at the hidden layer around the rivet, making it difficult to detect. This paper proposes a machine learning approach incorporating an electromagnetic testing system to detect the hidden corrosion at the riveting site effectively. Several machine learning methods will be investigated for the detection of different sizes and locations of corrosion. The training strategy of the machine-learning models on the small numbers of data will also be investigated. The result shows that the proposed approach could effectively detect 89.48% of the hidden corrosion having from 2.8 to 195.4 mm3 with only 20% of training data and could be increased to 99.0% with 60–80% of the training data.
- 20 May 2022
- Published by MDPI AG in Applied Sciences
- Vol. 12 (10) , 5175
Open AccessAuto-Detection of Hidden Corrosion in an Aircraft Structure by Electromagnetic Testing: A Machine-Learning Approach
An aircraft is a multilayer structure that is assembled by rivets. Under extreme working conditions, corrosion appears and quickly propagates at the rivet sites of the layers; thus, it threads the integrity and safety of the aircraft. Corrosion usually occurs at the hidden layer around the rivet, making it difficult to detect. This paper proposes a machine learning approach incorporating an electromagnetic testing system to detect the hidden corrosion at the riveting site effectively. Several machine learning methods will be investigated for the detection of different sizes and locations of corrosion. The training strategy of the machine-learning models on the small numbers of data will also be investigated. The result shows that the proposed approach could effectively detect 89.48% of the hidden corrosion having from 2.8 to 195.4 mm3 with only 20% of training data and could be increased to 99.0% with 60–80% of the training data.
- 28 January 2021
- Published by Elsevier in Journal of Magnetism and Magnetic Materials
- Vol. 527, 167790
Pick-up coil free gating-flux magnetometers powered by differential excitation signals
- 28 January 2021
- Published by Elsevier in Journal of Magnetism and Magnetic Materials
- Vol. 527, 167790
Pick-up coil free gating-flux magnetometers powered by differential excitation signals
- 29 September 2020
- Published by Springer Nature in Journal of Superconductivity and Novel Magnetism
- Vol. 34 (2) , 339-345
High-Resolution Pinning GMR Sensors for Extremely Low Frequencies Powered by a Simple Alternating Current–Biased Scheme
- 29 September 2020
- Published by Springer Nature in Journal of Superconductivity and Novel Magnetism
- Vol. 34 (2) , 339-345
High-Resolution Pinning GMR Sensors for Extremely Low Frequencies Powered by a Simple Alternating Current–Biased Scheme