KMITL

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

Browse

Search Results

Now showing 1 - 10 of 16

Some of the metrics are blocked by your
consent settings
Deep learning-based prediction models for the vertical total electron content using GNSS satellite observations
(2026-08-01)
Mutasov, Gleb
;
Myint, Lin Min Min
;
Budtho, Jirapoom
;
Perwitasari, Septi
;
Nishioka, Michi
Ionospheric Total Electron Content (TEC) is a crucial parameter for characterizing the state of the ionosphere and assessing its impact on satellite-based navigation systems and on communication technologies. In equatorial and low-latitude regions, ionospheric irregularities, particularly equatorial plasma bubbles (EPBs), pose significant challenges for satellite navigation and communication due to their capacity to cause rapid TEC fluctuations and signal degradation. Since these effects are especially pronounced during ionospheric and geomagnetic disturbances, making accurate TEC prediction is an essential task for improving the reliability of GNSS-based positioning and space weather applications. This study presents a machine learning-based framework for one-day-ahead prediction of TEC with a 30-min resolution over the magnetic equator and low-latitude regions, with a focus on Southeast Asia. Unlike global models, our approach is tailored to local GNSS observations and directly predicts TEC values along specific satellite-receiver paths, defined by geographic location and satellite visibility. We integrate ionospheric pierce point (IPP) coordinates, geomagnetic indices, and solar activity indicators as features to enhance temporal and spatial forecasting accuracy. To address the nonlinear and nonstationary nature of TEC variations, we investigate and compare three deep learning architectures: a Transformer-based time-series model, a Temporal Kolmogorov–Arnold Network (TKAN), and a Long Short-Term Memory (LSTM). Additionally, the predictions are benchmarked against the empirical IRI-2020 model and a persistence baseline. The results demonstrate that both the Transformer and TKAN models outperform the LSTM and empirical approaches, particularly during different geomagnetic and ionospheric conditions, showing improved robustness and generalization. The proposed framework highlights the potential for accurate, resource-efficient TEC prediction in low-latitude regions and opens a pathway for further improvements by integrating multi-GNSS observations and additional space weather parameters.
Some of the metrics are blocked by your
consent settings
Integrated analysis of atmospheric and ionospheric precursors using SARIMAX, NARX, and LSTM approaches for the 2024 Mw 7.4 Taiwan earthquake
(2026-08-01)
Tahreem, Azka
;
Shah, Munawar
;
Jamjareegulgarn, Punyawi
Earthquakes are among the most damaging natural hazards, highlighting the need for improved monitoring frameworks and rigorous analysis of potential precursory signals. The Mw 7.4 Taiwan earthquake provides a relevant case for evaluating methodologies to identify and interpret atmospheric and ionospheric anomalies in seismically vulnerable regions. In this study, satellite-based Remote Sensing (RS) products and Global Navigation Satellite System (GNSS) observations are integrated to examine candidate precursors, including Outgoing Longwave Radiation (OLR), Relative Humidity (RH), Air Temperature (AT), Air Pressure (AP), and Total Electron Content (TEC). Using statistical approaches, including the standard deviation (STDEV) method and the Seasonal AutoRegressive Integrated Moving Average with Exogenous Variables (SARIMAX) model, together with machine-learning frameworks such as the Nonlinear AutoRegressive model with eXogenous inputs (NARX) and Long Short-Term Memory (LSTM) networks, this study identified synchronized anomalies approximately 5–6 days prior to the event. In addition, geomagnetic perturbations were observed approximately nine days before the event, coinciding with a pronounced geomagnetic storm (Kp > 8; Dst < −120 nT; ap > 225 nT). To limit the influence of background variability and potential false alarms, a historical comparative analysis was performed using atmospheric parameters from the same region and comparable time window across the preceding five years, which further supported the robustness of the observed anomalies. By integrating statistical detection, spatial screening, and time-series forecasting models, this work contributes to a more detailed understanding of atmospheric–ionospheric signals associated with seismic activity and highlights the value of multi-parameter monitoring for seismic hazard assessment and risk-reduction planning.
Some of the metrics are blocked by your
consent settings
IoT-based Water Quality Monitoring Station and Forecasting System with Machine Learning
(2025-01-01)
Jomjaiekachorn, Thanart
;
Anuwongpinit, Thanavit
;
Purahong, Boonchana
This paper presents an IoT-based water quality monitoring and forecasting system designed for real-time and continuous assessment of water resources. The system integrates Siemens SIMATIC IOT2050 as an Industrial IoT Gateway, which collects data from sensors measuring conductivity, pH, dissolved oxygen, and temperature using RS485 Modbus RTU communication. Data processing occurs at the edge using Node-RED and is transmitted to AWS Cloud via MQTT for storage and visualization on a dashboard. Predictive analysis employs machine learning models, including XGBoost with Optuna parameter tuning and Long Short-Term Memory (LSTM) networks, for water quality forecasting. Results indicate superior performance of LSTM for most parameters, while XGBoost excels in pH prediction. This system demonstrates scalability, reliability, and potential for enhanced water quality management in diverse environments.
Some of the metrics are blocked by your
consent settings
Improving the Sound Classification Accuracy Using CNN-LSTM and MFCC with Audio Augmentation for Diagnosing Respiratory Disease
(2025-01-01)
Phankokkruad, Manop
;
Wacharawichanant, Sirirat
Audio is vital information data for understanding various situations. A multitude of sound features can be explained by analysis through the audio signals. Numerous classification methods have been developed to study audio classification. This work studies the improvement of audio classification for the diagnosis of respiratory disease through the integration of audio data augmentation and CNN in conjunction with LSTM (CNN-LSTM). Furthermore, this paper focuses on audio data augmentation and feature extraction in the deep learning approach. This study proposed the CNN-LSTM model to diagnose respiratory disease by learning from the different audio datasets. The results reveal that the CNN-LSTM model attained an accuracy of 81.48%, precision of 0.8340, sensitivity of 0.6948, and F1-score of 0.7225. Considering the achieved F1-score, the CNN-LSTM model demonstrates a high level of diagnotic accuracy. Therefore, all evaluation evaluation parameters collectively indicate the robust performance of the proposed disease classification model.
Some of the metrics are blocked by your
consent settings
Evaluating the effectiveness of facial actions features for the early detection of driver drowsiness in driving safety monitoring system
(2025-01-01)
Rahmawati, Yenny
;
Woraratpanya, Kuntpong
;
Ardiyanto, Igi
;
Nugroho, Hanung Adi
Traffic accidents caused by drowsiness remain a serious threat to driving safety. Many of these accidents can actually be prevented with an early warning system that detects the early signs of driver drowsiness. This study proposes a non-invasive system to detect drowsiness based on visual features extracted from videos recorded by a dashboard camera. The system uses facial landmarks generated by a facial network detector to identify key areas such as eyes, mouth, and head. The eye aspect ratio (EAR), mouth aspect ratio (MAR), and head rotation angle were calculated as the main features. These features were fed into three classification models: 1D-CNN, LSTM, and BiLSTM. Evaluation was conducted using 87 videos from the YawDD dataset for training and 20 videos from custom data for testing. During training, the 5-fold cross-validation was used to ensure model generalization and reduce the risk of overfitting. In addition to accuracy, other metrics such as precision, recall, and F1-score were used to provide a more comprehensive overview of the system performance. The results showed that the combination of the three facial features (EAR, MAR, and head rotation) provided a better performance than did the use of a single feature or a combination of two features, with an accuracy improvement of 5–8%. The BiLSTM model showed the best performance, with a training accuracy of 99% on the YawDD dataset and a testing accuracy of 98% on the custom data.
Some of the metrics are blocked by your
consent settings
Comparative Analysis of Deep Learning Models for Daily Solar Indices Forecasting in Solar Cycle 25
(2025-01-01)
Min Myint, Lin Min
;
Mutasov, Gleb
;
Supnithi, Pornchai
;
Budtho, Jirapoom
Accurate forecasting of solar activity indices, particularly the Sunspot Number (SSN) and the F10.7 solar radio flux index (F10.7), is essential for effective space weather monitoring, as severe solar and ionospheric disturbances can significantly impact satellite operations, radio communications, and navigation systems. This paper presents a comparative analysis of deep learning models - Long Short-Term Memory (LSTM), Temporal Convolutional Networks (TCN), and encoder-only Transformer architectures - for daily forecasting of SSN and F10.7 up to 14 days ahead based on past 27 days. Considering relatively simple model structures, both single-step and multi-step prediction strategies are explored to evaluate the models' capability in handling short-and long-term dependencies in time series data. Daily solar activity data spanning seven solar cycles (Cycles 19-25), obtained from the GFZ Helmholtz Centre for Geosciences, are used for model training and evaluation. Experimental results show that LSTM consistently achieves the best performance across most forecast horizons, particularly in short-to medium-term predictions. The Transformer model delivers competitive and stable results, while TCN performs relatively less effectively, indicating the need for more complex architecture and optimization strategies. These findings highlight the strengths and limitations of each architecture for solar activity forecasting applications.
Some of the metrics are blocked by your
consent settings
Predicting Equatorial Ionospheric Total Electron Content Using the Transformer-based Model with Observations From Ground GNSS Receivers and COSMIC-2 Satellites
(2025-01-01)
Mutasov, Gleb
;
Supnithi, Pornchai
;
Budtho, Jirapoom
;
Perwitasari, Septi
;
Nishioka, Michi
Ionospheric Total Electron Content (TEC) is a key parameter for monitoring and studying the ionosphere, which induces significant delays in radio signals. Equatorial ionospheric irregularities, such as Equatorial plasma bubbles (EPB), can severely disrupt satellite navigation and communication. Predicting TEC is, therefore, essential for space weather monitoring and high-precision positioning applications. This study employs a Transformer-based model to predict TEC 24 hours in advance for specific satellites based on observations from a ground station and COSMIC-2 satellites. Unlike other approaches, our model directly forecasts TEC values for visible satellites within predefined longitude-latitude ranges. To enhance predictive accuracy, we also integrate additional features: ionospheric pierce points (IPP), geomagnetic (HP60), and solar activity indices, utilizing time-series Transformer architecture, and we consider a long-short-term memory (LSTM) model as a baseline. The proposed approach provides promising results for local TEC forecasting in the specific coverage area, with potential for further enhancements using additional GNSS or TEC measurements.
Some of the metrics are blocked by your
consent settings
Improving Water Salinity Forecasting in Bang Pakong River with Attention Mechanism
(2024-12-02)
Saksopit, Thongthai
;
Khawne, Amnach
Seawater intrusion in the Bang Pakong River estuary poses a significant threat to freshwater resources used for agriculture, municipal consumption, and industrial applications. Accurate prediction of salinity fluctuations is crucial for effective water management strategies. This study proposes an enhanced univariate salinity prediction method utilizing a Long Short-Term Memory (LSTM) model augmented with an Attention Mechanism. The Attention Mechanism empowers the LSTM to selectively focus on crucial information within extended historical salinity data sequences. The optimal input sequence length for the model is determined through a training process, aiming for the most accurate predictions. Here, the model forecasts salinity values 24 hours ahead and is evaluated against actual measurements. Performance metrics demonstrate that the Attention-LSTM model achieves the lowest error (MAE: 0.007834, MSE: 0.000094, RMSE: 0.009697, MAPE: 0.048736) and the highest accuracy (R<sup>2</sup>: 0.782927) at an input sequence length of 504 hours. These findings highlight the potential of the Attention-LSTM model for improved salinity prediction in the Bang Pakong River estuary, aiding water resource management strategies.
Some of the metrics are blocked by your
consent settings
Thai Question-Answering System Using Similarity Search and LLM
(2024-07-01)
Jearanaitanakij, Kietikul
;
Srithongdee, Chananchida
;
Ketkham, Sirinoot
;
Ardsana, Onwanya
;
Kullawan, Tiwat
A question-answering (QA) system is essential to an organization where numerous QA pairs respond to customer queries. Choosing the right pair corresponding to the query is a complex task. Although the QA system from a commercial product like ChatGPT provides an excellent solution, it is costly, and the fine-tuned Large Language Model (LLM) cannot be downloaded for private use at the local site. In addition, the cost of using such LLM may significantly increase when the number of users grows. We propose a Thai QA system that can swiftly respond and correctly match the user query to the reference answer in the QA dataset. The proposed system encodes both QA pairs and a query into individual embeddings and finds a couple of QA pairs that are most related to the query by using the fast similarity search called Faiss (Facebook AI Similarity Search.) Afterward, the relevant QA pairs and the query are fed to the fine-tuned LLM (WangchanBERTa-pretraining multilingual transformer-based) to choose the single best match QA pair. The fine-tuned WangchanBERTa can retrieve the correct answer and respond to the query naturally. The experiment conducted on the Thai Wiki QA dataset indicates the superior ROUGE values, precision, recall, F1-score, and runtime of the proposed system against other strategies.
Some of the metrics are blocked by your
consent settings
Atmospheric precursors associated with two Mw > 6.0 earthquakes using machine learning methods
(2024-06-01)
Khalid, Zaid
;
Shah, Munawar
;
Riaz, Salma
;
Ghaffar, Bushra
;
Jamjareegulgarn, Punyawi
The advancements in remote sensing (RS) satellite applications have revolutionized natural disaster surveillance and prediction in the earthquake monitoring by delineating various precursors at the Earth’s surface and in atmosphere. In this paper, the earthquake precursors comprising land surface temperature, outgoing longwave radiations, relative humidity, and air temperature for both the daytime and nighttime are investigated for two Mw > 6.0 events in USA. Interestingly, we noticed surface and atmospheric parameters anomalies in 6–8 days window prior to both the events by using standard deviation method. Moreover, these abrupt deviations are also validated by the recurrent neural networks like autoregressive network with exogenous inputs and long short-term memory inputs. The findings of this study demonstrate the potential of using modern analysis tools to further develop our knowledge of the linked dynamics of the lithosphere and atmosphere preceding seismic occurrences. This study implements substantially the developing of natural hazard surveillance and earthquake prediction capabilities for future researches as a valuable addition of reference in the field of RS.

KMITL

Browse

Filters

Settings

Sort By

Results per page

Search Results