Konghuayrob, Poom

Preferred name

Konghuayrob, Poom

Main Affiliation

School of Engineering

poom.ko@kmitl.ac.th

Now showing 1 - 10 of 17

Some of the metrics are blocked by your
consent settings
Comparison of Reduced-Length FFT-Based Feature for Induction Motor Fault Classification
(2025-01-01)
Taweewat, Pat
;
Suwan-Ngam, Warachart
;
Songsuwankit, Kanoknuch
;
Konghuayrob, Poom
This research presents a comparison of FFT-based features which can be used for classifying induction motor faults via neural network. In this paper, the misalignment and rotor bar damage faults are investigated by using stator current as input data only. As the length of the full FFT can include both informative data corresponding to the faults and uninformative data such as noise from environment or electrical supply, only relevant magnitude from FFT bins should be selected and used instead. This paper proposed to use threshold level determined from the magnitude of FFT bins in dataset as a criterion for the selection. From experimental results, an input feature vector created by proposed method can create short input feature vector length to be used by neural network efficiently. The trained neural network performs classification task at 99.98% in accuracy. Comparing to using dimension reduction by PCA, thresholding method needs basic computation, and yields result close to PCA method.
Some of the metrics are blocked by your
consent settings
Microprocessor based fuzzy MPPT for PV-AC module DCM-flyback Inverter
(2013-01-01)
Konghuayrob, Poom
;
Kaitwanidvilai, Somyot
This paper proposes a maximum power point tracking using fuzzy based perturb and observe (P&O) algorithm for a photovoltaic (PV) system (an AC module). In AC module flyback inverter, modulation index (Δma) are adopted as a control variable to track the maximum power point (MPP) of PV array. In the conventional technique, step size of modulation index (Δma) is adopted in the simple P&O technique; Although this technique is easy to be implemented but there is some problems regarding large oscillation around the MPP and slow tracking when improper step size is selected. The proposed technique, fuzzy based P&O technique, is adopted to provide non-equal step size of Δma. The proposed fuzzy system was programmed on microcontroller which is a low cost processing device. Experimental result confirms that the proposed system can effectively track the power from PV system, and is better than the conventional technique.
Some of the metrics are blocked by your
consent settings
Robust H∞ Mixed-sensitivity PID structural based on PSO considering input constraint
(2017-01-01)
Konghuayrob, Poom
;
Kaitwanividvilai, Somyot
Robust control is one of the potential design methods to maintain system performance and system stability from the familial uncertainties. In addition, the limitation of performance as an input energy constraint needs to be considered to prevent the system stability degradation as well. This paper focuses on the design of H<inf>∞</inf> mixed sensitivity based on structural PID controller for high accuracy hard disk drive servo system which examines the input saturation constraint. Particle swarm optimization (PSO) is utilized in the proposed design to maximize the system stability index called stability margin (), which consists of three norms of H<inf>∞</inf> mixed sensitivity control. In order to confirm the effectiveness of the proposed controller, full order H<inf>∞</inf> mixed sensitivity, the proposed PID design with considering the input constraint and trial-error tuning based PID are compared in the simulation section. It is clearly illustrated that the considering input saturation constraint is important to design the controller for sustaining the system performance. Moreover, the results confirm the robustness and performance of three controllers under the repeatable runout (RRO) disturbance. The structure of the proposed PID controller based on PSO is more simple and appropriate to apply into the actual application than the conventional H<inf>∞</inf> mixed sensitivity full order. In addition, the results of the proposed PID controller and conventional H<inf>∞</inf> are quite similar which can reduce the effect of RRO by 50 times and gains 50% more effective than the normal PID in terms of output error.
Some of the metrics are blocked by your
consent settings
Mitigating Racial Bias in Skin Lesion Classification With a Novel Deep Learning-Driven Dataset
(2025-01-01)
Joshi, Priyanka
;
Ruangrit, Chanet
;
Konghuayrob, Poom
(1) Background: A critical issue in the application of Machine Learning (ML) in dermatology is the presence of racial bias in training datasets, which can lead to disparities in diagnostic performance across different skin tones. This paper proposes a mitigation strategy for the BCN20000 dataset, addressing its limitations regarding darker skin lesions by augmenting it with a custom dataset. (2) Method: This was achieved by augmenting the BCN20000 dataset with a custom-developed dataset of dark skin lesion images, generated by combining serial style transfer and latent diffusion-based upscaling. The impact of this augmented dataset was assessed on the accuracy of several established Convolutional Neural Network (CNN) architectures, including DenseNet, ConvNeXt, EfficientNet, RegNet and ResNet. (3) Results: Our findings underscore the necessity of diverse datasets for AI-driven dermatology tools with the best performing model achieving 92% accuracy when trained on the augmented dataset containing additional dark-skin images compared to baseline models trained solely on BCN20000. These results align with prior studies emphasizing the critical role of representative training data in mitigating racial bias in medical AI systems.
Some of the metrics are blocked by your
consent settings
Autonomous Rescue Hexapod Robot with AI Human Detection and Tracking
(2023-01-01)
Saengsint, Chissanupong
;
Juntacheevakul, Kidtipod
;
Chalermkitpaisan, Kirawut
;
Suthapintu, Napak
;
Tomaneenilrat, Pavares
The occurrence of catastrophic caused tons of thousands of lives. To enhance the efficacy of search and rescue operations in challenging to access region, we have introduced an autonomous hexapod robot equipped with a partial human detection and tracking system that is capable to identify fragments of the human body. The chosen hexagonal arrangement of six legs enables the robot to navigate through rough terrains. This configuration provides stable mobility, with an average acceleration of 0.8 m/s<sup>2</sup> and a peak acceleration of 3.1 m/s<sup>2</sup> during walking, making the camera of the robot remains steady for human identification. Due to the employed DeeplabV3+ architecture for the training process, partial human bodily structure detection has achieved 94% accuracy. Moreover, the robot can track the detected individual by maintaining 1m distance.
Some of the metrics are blocked by your
consent settings
The Development of Adaptive Gray Level Mapping Combined Partical Wwarm Optimization for Measuring the Dimeter Size of Automotive Nut
(2020-04-01)
Pondech, Wichai
;
Saenthon, Anakkapon
;
Konghuayrob, Poom
In line assembly process, it is necessary to use the accurate tools to inspect the work piece. In order to measure the width, thickness or depth of the nut used in the commercial car, previously in Thai Steel Cable Company used the caliper to measure size of the automotive nut operated by human as an sample test inspection. Even the result on this technique is high accuracy, however; there are some disadvantages based on this method such as human error, idle time and also fatigue by the human. Therefore, this research focus on the development of the new measurement technique that utilized industrial camera together with the image processing algorithm to measure the entire nut with 100% inspection test. Circular Hough Transform (CHT) is applied to be the basic concept used for finding the circle position and measuring nut diameter. Although the CHT technique can measure diameter of the interested nut, but the result's accuracy is not acceptable due to measuring error from the various range of light condition. This research proposed the new technique, Adaptive Gray Level Mapping algorithm (AGLM) to increase the quality of the input picture before measuring the radius by CHT technique. Moreover, the beta in AGLM is optimized by particle swarm optimization (PSO) technique that applies 50% of nut data and other is used for validate. The results show the effectiveness of the proposed AGLM combined PSO that increase the accuracy of the visual measuring method via compare to the conventional threshold with CHT technique.
Some of the metrics are blocked by your
consent settings
Two-Tier AI Surveillance: Enhancing Weapon Detection Through Edge and Cloud Collaboration
(2026-01-01)
Sun, Sanchai
;
Songsuwankit, Kanoknuch
;
Thamrongrongveerachart, Chawapon
;
Yanyong, Sarucha
;
Sri-on, Jiramate
This research presents a novel two-tier AI-based surveillance system designed for real-time weapon detection to enhance security and prevent potential robberies. The system leverages the computational capabilities of both edge and cloud resources, integrating a YOLOv5s model on a Jetson Nano for initial detection and a YOLOv8l model on a server equipped with an Nvidia A100 for refined analysis. The initial detection performed by the Jetson Nano rapidly identifies potential threats and forwards compressed images to the server, optimizing bandwidth usage and transmission speed. Upon receipt, the server applies image enhancement techniques to restore and upscale the images from 640 pixels back to 1280 pixels before further verification. Utilizing Python Django, the server processes the enhanced images with a more sophisticated model to ensure high accuracy in detection. The integration of edge computing optimizes the system’s performance by enabling real-time processing and reducing latency. This hybrid approach enhances the efficiency and scalability of the surveillance system, ensuring robust and timely detection. Upon confirming the presence of a weapon, the system sends immediate alerts via LINE Notify to both users and local law enforcement, thereby enabling prompt responses to potential security threats. Additionally, the Jetson Nano hosts a Flask application, allowing users to download previously recorded videos for further review and evidence collection. The proposed solution combines edge computing, cloud processing, and image optimization techniques to provide an efficient, scalable, and high-performance solution for real-time weapon detection which enhancing accuracy and reliability in real-world surveillance.
Some of the metrics are blocked by your
consent settings
The Cattle Rut Behavior Detection Base on AI Deep Neural Network
(2022-01-01)
Phaechaiyaphum, Phatsarut
;
Konghuayrob, Poom
This paper proposes an application model of camera object detection to observe the cattle rut through its pose. This research is based on deep neural network method along with transfer learning technique. This paper employs four different deep convolutional neural network architectures 1) SSD Mobile-Net V2 2) SSD ResNet 101 V1 FPN 3) YOLOv5s and 4) Mask R-CNN. All the network mentioned above are pre-trained on COCO dataset. To observe the cattle rut behavior, the images of cattle behavior were taken in natural open environment, the data obtained are split into training and testing datasets. In this research cattle behaviors are categorized mainly in two classes, (1) normal behavior and (2) the cattle rut behavior. The result obtained after training the models shows that the YOLOv5s model obtained the highest mean average precision which is 95.5% and least training time. Thus, this paper proposes that YOLOv5s model can be applied to detect the cattle behavior for precise artificial breeding in order to increase the cattle population in farm to serve higher consumption demand in the near future.
Some of the metrics are blocked by your
consent settings
Maximum power point tracking using neural network in flyback MPPT inverter for PV systems
(2012-12-01)
Konghuayrob, Poom
;
Kaitwanidvilai, Somyot
Generally, perturb and observe (P&O) technique is widely adopted in photovoltaic (PV) system to maximize the output power. In flyback inverter, the modulation index needs to be adjusted based on the P&O algorithm. However if the changing step size of modulation index (Δma) is too large, the fast MPP (Maximum Power Point) tracking can be achieved but the power oscillation around the MPP will be large. In contrary, the small changing step size results in long tracking time and small oscillation. Consequently, this paper proposes a technique to adjust the changing step size (Δma) of Flyback inverter to achieve both acceptable tracking time and low power oscillation. In the proposed technique, irradiance is adopted as the input of a neural network which is used to estimate the appropriate modulation step size. Simulation results confirm that the proposed neural network based inverter can find the appropriate changing step size (Δma) which is adequate for any irradiance conditions. © 2012 IEEE.
Some of the metrics are blocked by your
consent settings
Optimization of 3D SLAM and Localization for Quadruped Robot Dogs Using Autoencoder-based Techniques
(2025-01-01)
Thamrongveerachart, Chawapon
;
Thepsit, Thitipong
;
Ningnoi, Paphop
;
Konghuayrob, Poom
This paper presents an optimized 3D SLAM and localization framework for quadruped robot dogs using a custom autoencoder-based point cloud compression technique. Our approach combines denoising, voxelization, self-attention, and PointNet modules to reduce the size of the point cloud by more than 80% while removing outliers and maintaining structural clarity. We use a 3D laser-scan matcher for localization, which combines onboard odometry from the Unitree Go2 quadruped robot with LiDAR odometry from a Livox Mid-360 sensor. Our approach maintains accurate pose estimation and considerably reduces processing time by working directly on compressed point clouds. Comparative tests show that the system is appropriate for real-time applications in unstructured contexts due to its enhanced map clarity and decreased localization delay. The proposed framework offers an effective and computationally efficient solution for deploying 3D SLAM in agile legged robots operating in dynamic or hazardous settings.

Konghuayrob, Poom

Filters

Settings

Sort By

Results per page