KMITL

Permanent URI for this communityhttps://dspace.kmitl.ac.th/handle/123456789/1

Browse

Search Results

Now showing 1 - 10 of 20

Some of the metrics are blocked by your
consent settings
Enhancing deep learning–based railway inspection via PSO-guided brightness–contrast optimization
(2026-03-01)
Songthai, Maethinee
;
Kaitwanidvilai, Somyot
Reliable operation of electrified railway systems depends critically on the pantograph–catenary system (PCS). Existing inspection practices are largely manual or periodic and remain vulnerable to low illumination, background clutter, and thin structural components, limiting robustness and scalability. Although vision-based deep learning is promising, performance often degrades in low-light and complex scenes, while conventional enhancement (e.g., CLAHE) provides limited and inconsistent improvements. This study proposes an illumination-aware PCS inspection framework that integrates Particle Swarm Optimization (PSO)–guided brightness–contrast optimization with a lightweight YOLO detector. A version benchmark selected YOLOv9t as the backbone, achieving the best overall performance (Precision = 0.947, F1-score = 0.913) compared with YOLOv8n and YOLOv11n. Although YOLOv11n has lower FLOPs, PCS inspection is accuracy-critical because detection errors propagate to event-frequency counting and lateral-deviation assessment; therefore, YOLOv9t was fixed for subsequent experiments. PSO is employed to estimates a single dataset-level global enhancement parameter set, improving robustness while maintaining computational efficiency for early-stage field deployment under limited annotated data and edge hardware constraints. The framework was evaluated on a small yet diverse visible-light dataset spanning day/evening/night conditions and challenging locations (station roofs, bracket regions, and transition sections). Comparative evaluation across three configurations—YOLOv9 baseline, YOLOv9 with CLAHE, and YOLOv9 with PSO-tuned brightness/contrast—achieved detection rates of 90%, 79%, and 99%, respectively, with the largest gains on thin, low-contrast structures (contact and messenger wires). The optimized pipeline further enables reliable estimation of lateral wire deviation in compliance with EN 50367 (≤200 mm), supporting safety-critical inspection in low-light and complex-background conditions.
Some of the metrics are blocked by your
consent settings
Decentralised and Centralised Fixed Structure H∞ Robust Loop Shaping for the MIMO Microsurgical Manipulation Based on PSOGSA
(2026-01-01)
Kaitwanidvilai, Somyot
;
Konghuayrob, Poom
;
Aoyama, Hisayuki
This paper presents decentralised and centralised fixed-structure H∞ robust control methods optimised by the Particle Swarm Optimisation and Gravitational Search Algorithm (PSOGSA) for a coupled multi-input multi-output (MIMO) microsurgical manipulator. The design framework explicitly considers uncertainties and disturbance constraints. Conventional H∞ loop-shaping controllers are typically of high order, complex, and difficult to implement in practice. To address this limitation, Proportional-Integral-Derivative (PID)-structured decentralised and centralised H∞ controllers are proposed, providing compact structures while retaining robustness. The novelty of this work lies in embedding H∞ robustness criteria into practical PID-based frameworks, bridging the gap between theoretical robust design and experimental implementation in microsurgical applications. The proposed controllers are evaluated against a reduced-order H∞ controller derived from Hankel norm approximation and a Ziegler–Nichols tuned PID controller, using both simulation and experimental studies. Results demonstrate that the proposed controllers achieve improved stability margins (0.449–0.521 compared with 0.436 for the reduced-order design), maintain low root-mean-square errors (≈0.067–0.089) and remain robust under voltage disturbances where conventional PID control fails. These findings confirm the contribution of a practical and efficient robust control strategy for enhancing the precision and reliability of microsurgical manipulators.
Some of the metrics are blocked by your
consent settings
Smart surveillance with conversational alerts for wild elephant early warning
(2026-01-01)
Kongyim, Nattaya
;
Kaitwanidvilai, Somyot
Human-elephant conflict continues to threaten communities near forested areas, particularly in rural regions where wild elephant intrusions endanger lives and agricultural livelihoods. This study presents a real-time AI-powered surveillance and alert system integrating YOLOv8-based object detection, a localized CCTV network, and a human-like conversational agent on the LINE platform. Unlike traditional systems relying on expensive cloud computing and commercial messaging services, this solution employs a free communication channel and a custom-built bot that delivers timely alerts—complete with images, GPS location, and timestamp—directly to local residents and authorities. The object detection model, trained on localized datasets, achieved a precision of 98.9%, recall of 97.3%, F1-score of 98.1%, and mAP@0.5 of 98.6%. The end-to-end response time—from detection to alert delivery—averaged just 3.8 s. Enhanced by strong Wi-Fi antenna deployment, the system enables rapid, wide-area data dissemination without relying on stable internet or costly mobile networks. This significantly reduces data charges, a critical barrier in the deployment of smart city technologies in rural areas. Comparative analysis with conventional manual patrols and cloud-based AI systems confirms that the proposed approach is not only faster and more accurate but also far more cost-effective. Field tests validate its robustness in diverse lighting conditions, with rare false positives primarily under visually ambiguous scenarios. Feedback from local stakeholders further highlights its practical utility in improving situational awareness, response time, and long-term conflict mitigation. This system demonstrates a scalable and sustainable model for smart, AI-driven wildlife monitoring in developing regions.
Some of the metrics are blocked by your
consent settings
Experimental realization of PSO-based hybrid adaptive sliding mode control for force impedance control systems
(2025-06-01)
Yanyong, Sarucha
;
Kaitwanidvilai, Somyot
This paper presents a practical solution for an adaptive impedance force controller with online learning capabilities, designed to mitigate the effects of inaccuracies in system identification models. The proposed hybrid algorithm addresses the challenges associated with online learning in real-world machines. Additionally, the system demonstrates the ability to adapt to environmental changes, maintaining high-quality performance despite variations. A sliding surface guarantees system stability, while Particle Swarm Optimization (PSO) optimizes impedance parameters, reducing the risk of local minima. The hybrid algorithm also reduces overshoot and undershoot, resulting in faster system responses. Simulation and experimental results demonstrate that the proposed technique outperforms conventional force control systems in terms of learning ability and overall performance.
Some of the metrics are blocked by your
consent settings
Optimizing Product Quality Prediction in Smart Manufacturing Through Parameter Transfer Learning: A Case Study in Hard Disk Drive Manufacturing
(2025-04-01)
Kaitwanidvilai, Somyot
;
Sittisombut, Chaiwat
;
Huang, Yu
;
Bom, Sthitie
In recent years, the semiconductor industry has embraced advanced artificial intelligence (AI) techniques to facilitate intelligent manufacturing throughout their organizations, with particular emphasis on virtual metrology (VM) systems. Nonetheless, the practical application of data-driven virtual metrology for product quality inspection encounters notable hurdles, such as annotating inspections in highly dynamic industrial environments. This leads to complexities and significant expenses in data acquisition and VM model training. To address the challenges, we delved into transfer learning (TL). TL offers a valuable avenue for knowledge sharing and scaling AI models across various processes and factories. At the same time, research on transfer learning in VM systems remains limited. We propose a novel parameter transfer learning (PTL) architecture for VM systems and examine its application in industrial process automation. We implemented cross-factory and cross-recipe transfer learning to enhance VM performance and offer practical advice on adapting TL to meet individual needs and use cases. By leveraging extensive data from Seagate wafer factories, known for their large-scale and high-dimensional nature, we achieved significant PTL performance improvements across multiple performance metrics, with the true positive rate (TPR) increasing by 29% and false positive rate (FPR) decreasing by 43% in the cross-factory study. In contrast, in the cross-recipe study, TPR increased by 27.3% and FPR decreased by 6.5%. With our proposed PTL architecture and its performance achievements, insufficient data from the new manufacturing sites, new production lines and new products are addressed with shorter VM model training time and smaller computational power with strong final quality prediction confidence.
Some of the metrics are blocked by your
consent settings
PSO-Optimized Deep Learning for Ultra-Precise Corrosion Detection on HDD Read/Write Heads
(2025-01-01)
Punyammaree, Chaiwat
;
Kaitwanidvilai, Somyot
This paper presents a novel deep learning approach for automated detection and counting of corrosion pits on Hard Disk Drive (HDD) read/write heads using Scanning Electron Microscopy (SEM) images. A U-Net model optimized via Particle Swarm Optimization (PSO) is developed to enhance segmentation performance by automatically tuning hyperparameters. The methodology includes optimized SEM image acquisition, preprocessing (patch-based subdivision and expert annotation), PSO-driven hyperparameter selection, and post-processing with thresholding and connected component analysis for pit counting. Experimental results demonstrate that the PSO-optimized U-Net significantly outperforms standard U-Net, SegNet, and LinkNet models, achieving an F1-score of 79.60%, an IoU of 86.51%, and an accuracy of 99.77%. Additionally, the proposed method achieves 86.9% counting accuracy, surpassing human experts (72.7%) while processing images 15 times faster (180 seconds vs. 2700 seconds per image). These findings highlight the potential of PSO-optimized deep learning for improving HDD quality control by providing an accurate, efficient, and standardized solution for corrosion pit detection, ultimately reducing the risk of HDD failure and data loss.
Some of the metrics are blocked by your
consent settings
An Adaptive Machine Learning Framework Integrating AutoML and MLOps for Two-Stage Classification in Hard Disk Drive Manufacturing
(2025-01-01)
Rungtalay, Natthakritta
;
Kaitwanidvilai, Somyot
This study aims to predict hard disk drives (HDDs) that pass initial testing but fail during reliability testing, using historical data from 8968 records with 218 features, such as head position and flying height of the read/write head. Since reliability testing is time-intensive, early failure prediction can significantly accelerate problem detection and resolution. The research focuses on detecting fly height modulation, a key symptom of HDD failure, and introduces an adaptive machine learning (ML) framework integrating AutoML for optimised model selection and hyperparameter tuning with MLOps for deployment, monitoring and continuous updates. Building on a previously proposed dual-stage classification framework that combines novelty detection and supervised learning, the proposed framework addresses the inefficiencies of manual hyperparameter tuning inherent in the earlier methods. The proposed framework achieves 92% accuracy in novelty detection and 100% in supervised learning, outperforming prior approaches. This integration of AutoML and MLOps offers a scalable, robust solution for early failure prediction, enabling real-time adaptability with minimal human intervention. Future work will focus on enhancing computational efficiency and responsiveness to data shifts and drifts, advancing data-driven decision-making in reliability testing.
Some of the metrics are blocked by your
consent settings
Calibration of Bi-prism Stereo Systems: A Model Free Approach
(2024-07-01)
Dissanayaka, Supun
;
Sooraksa, Pitikhate
;
Kaitwanidvilai, Somyot
;
Morris, John
A simple stereo system can be constructed from a single camera using a prism in the optical path to provide the required two views of a system. The simplicity of these systems has several advantages, particularly if the target is an underwater robot, where compact size and ability to seal the optical components are key factors. However, dispersion by the prism, in addition to the lens distortion, makes calibration challenging. By using a model-free approach, we were able to calibrate a prism-based stereo system effectively. We also aimed to use readily available 45° prisms, which present significant dispersion in the system, but retain simplicity and reduce cost, compared to custom low angle prisms. Modern LEDs provide high intensity, low bandwidth light sources and we used a set of three sources, roughly centered on the RGB channels of a readily available commercial camera. Our system used a circular target pattern covering the binocularly visible region in the scene and collected sets of images at known distances, using three separate light sources. From these images, we generated two look-up tables, one for each pixel in the image and a disparity derived by matching corresponding points, Cp(u,v,du), which has three dimensions, and another look-up table, which has a single dimension, Cz(z), so are not quite large, and not beyond the memory capability of even small modern camera systems, but provide fast, O(1), lookup times, suitable for real-time systems. Our calibration strategy enables a simple stereo system built from a single camera to measure depths in a scene: the single camera requires no electronic synchronization and is built from a single, inexpensive, and readily available optical component – a right-angle prism.
Some of the metrics are blocked by your
consent settings
Dual-stage Classification Framework for Detecting Rare or Unseen Patterns Based on Novelty Detection and Supervised Learning
(2024-01-01)
Rungtalay, Natthakritta
;
Kaitwanidvilai, Somyot
In this article, we propose a dual-stage classification framework designed for identifying rare or unseen patterns in the hard disk drive (HDD) industrial test process. The proposed framework integrates novelty detection and supervised learning methodologies to effectively address the challenges associated with imbalanced datasets and the detection of infrequent or unseen patterns within continuously changing environments. By employing novelty detection as the first-stage classifier followed by supervised learning as the second-stage classifier, the proposed method demonstrates an increased capacity to adapt to fluctuating environments, consequently enhancing the overall accuracy of process classification in practical manufacturing settings. To strengthen the robustness of novelty detection methods, an ensemble model technique is employed. Notably, the accuracy of the novelty detection methods in the first stage can be further enhanced with the incorporation of supervised learning techniques, particularly when a sufficiently large number of labeled samples are amassed. The proposed method consistently maintains accuracy, even in the face of changing environments, as it demonstrates the ability to adapt to data drift without necessitating the acquisition of new labeled data in the initial stage. This adaptability makes it particularly well suited for managing imbalanced datasets, rendering it highly practical for industrial applications. In a comprehensive case study conducted within the HDD industry, the framework exhibits immediate adaptability to rapidly changing environments while preserving high accuracy. This highlights the practical effectiveness of the proposed dual-stage classification framework in addressing the unique challenges posed by industrial scenarios.
Some of the metrics are blocked by your
consent settings
Sensor Fusion of Light Detection and Ranging and iBeacon to Enhance Accuracy of Autonomous Mobile Robot in Hard Disk Drive Clean Room Production Line
(2023-01-01)
Yanyong, Sarucha
;
Parichatprecha, Rattapoohm
;
Chaisiri, Punyavee
;
Kaitwanidvilai, Somyot
;
Konghuayrob, Poom
In this paper, the adaptive Monte Carlo localization (AMCL) error in terms of similar data detected by light detection and ranging (LiDAR) in different locations is investigated. This localization causes a robot to move to the incorrect location temporarily. We propose the fusion of landmark-based localization using an iBeacon device combined with the AMCL algorithm. This technique can solve the probabilistic localization problem of the conventional techniques applied in mobile robots by fusing the timed elastic band (TEB) and scan-matching algorithms, which reduces the error from 7 cm to less than 3 cm. The proposed technique is implemented on a clean-room-type mobile robot with 100 kg payload certificated by the SOP39 standard.

KMITL

Browse

Filters

Settings

Sort By

Results per page

Search Results