Repository@USM
Welcome to closed access digital repository of Universiti Sains Malaysia (USM).
This repository contains multiple types of scholarly materials, especially USM theses and exam papers.
To access the full text, please log in with your USM email account.

Research outputs
4750
Projects
0
People
0
Recent Additions
- PublicationRisk assessment of lead, cadmium and copper among workshop car spray painter in Kota Bharu, Kelantan.(2016)Exposure to chemical hazardous at car spray workshop will lead painter to be exposed to lead, cadmium and copper. Such exposure can cause adverse health effects to car painters due to neglect of regulation and guidance of safety and health. This study was aimed to assess the risk level of lead, cadmium and copper among car paint worker in Kota Bharu. 13 subjects for exposed group and non-exposed were selected from spray painter at car and in Campus Universiti Sains Malaysia (USM) respectively .Urine samples were collected at the post-shift work for both exposed group and for nonexposed group. They were asked to answer questionnaire related to their work. Urine sample of exposed and non-exposed group were digested with Nitric Acid (HNO3), Hydrochloric Acid (HCI) and Hydrogen Peroxide (H2O2) and lead, cadmium and copper were analyzed by using the Atomic Absorption Spectrometer (AAS)-Perkin Elmer Analyst 800 to measure the level of concentration. Factors socio-demographic such as smoking, duration of exposure to chemicals, duration of employment, and intake of seafood and alcohol can cause presence of lead, cadmium and copper present in urine. In conclusion, lead, cadmium and copper concentration levels in urine can detect the presence of lead, cadmium and copper among exposed and non-exposed groups.
- PublicationFlood Prediction Based On Deep Learning Networks With Variational Mode Decomposition(2024-09)Climate change increases the frequency of extreme weather events, causing river overflow floods that threaten human safety and ecosystems. Traditional flood prediction models face challenges due to fluctuations in water levels from topography and rainfall, leading to less accurate forecasts. This thesis aims to enhance flood prediction accuracy by developing and evaluating three new machine learning models that incorporate data decomposition, feature selection, and parameter optimization. The first two models use water level data for each hour. The first model utilizes hydrological data by integrating the Variational Mode Decomposition (VMD) method to reduce disturbances, along with Directional Bidirectional Long Short-Term Memory (BiLSTM) optimized with attention for forecasting purposes. The second model enhances prediction effectiveness by incorporating meteorological data specifically rainfall, humidity, and wind speed. This model emphasizes the benefits of VMD component classification and feature selection by considering water level changes to categorize Intrinsic Mode Functions (IMFs) obtained from the VMD method and using feature selection through the Pearson correlation method. The third model uses an optimized Gated Recurrent Unit - Temporal Convolutional Network (GRU-TCN) to forecast daily data at point estimates and confidence intervals. This model improves Kernel Density Estimate (KDE) predictions to assess forecast uncertainty more accurately and enhance model reliability. These three proposed models can overcome the weaknesses of traditional methods by utilizing real data from the Yangtze River station.
- PublicationMagnet segmentation to reduce torque ripple in permanent magnet synchronous motor(2024-07)Magnet segmentation is a common technique to reduce the eddy current loss and overcome manufacturing issues regarding the large permanent magnet (PM) machines, which intend to have large magnets. Magnet segmentation in Permanent Magnet Synchronous Motors (PMSM) plays a crucial role in reducing torque ripple and improving motor efficiency, addressing the primary target of energy efficiency for electric motor designers. Torque ripple refers to variations in torque output during each electrical cycle, which can lead to undesirable vibrations, noise, and reduced motor performance. Eddy current losses in the rotor can contribute to torque ripple by causing fluctuations in magnetic flux and inducing additional losses in the motor. Segmenting the magnets in a PMSM helps mitigate eddy current losses by breaking up the continuous magnetic circuit and reducing the magnitude of circulating currents induced in the rotor. This segmentation minimizes the skin effect phenomenon, where currents tend to concentrate near the surface of the magnets, further reducing eddy current losses. Additionally, segmented magnets allow for better control of magnetic flux distribution, leading to smoother torque output and reduced torque ripple. Furthermore, magnet segmentation helps overcome manufacturing challenges associated with large permanent magnet machines, as it allows for the use of smaller magnets that are easier to handle and assemble. Despite its advantages, magnet segmentation may affect motor performance, particularly in terms of back electromotive force (EMF) and developed torque. Finite Element Analysis (FEA) using software like ANSYS Maxwell 2D enables to study the effects of magnet segmentation on PMSM performance. By simulating different segmentation patterns and configurations, it can evaluate their impact on torque ripple and other performance metrics. Research has shown that magnet segmentation in PMSMs reduces electromagnetic torque ripple, leading to smoother motor operation and improved efficiency. Simulations comparing PMSM with different segmentation structures, such as one magnet per pole, two magnets per pole, and three magnets per pole, demonstrate that segmentation reduces torque ripple and improves motor performance.
- PublicationHazard classification using artificial neural network(2024-08)This research focuses on enhancing the landslide susceptibility mapping on Penang Island that targeting the electrical infrastructure using an Artificial Neural Network (ANN) model. The study integrates remote sensing data, Geographic Information Systems (GIS), and various normalization methods to create a comprehensive dataset of landslide conditioning factors including elevation, slope, aspect, curvature, rainfall, NDVI, and land use cover. By applying a frequency ratio method, the research identifies high-probability landslide areas. The samples were labelled as '1' for the landslide-prone regions and '0' for the non-landslide areas. The dataset of 26,294 samples is the result from balancing the 13,147 samples of the landslide occurrence data. Different normalization techniques such as mean-standard deviation and min-max are used to evaluate the ANN model's performance. The results show that the ANN model achieved a test accuracy of 86.66% with min-max normalization and 85.76% with mean-standard deviation normalization. This indicates that min-max normalization slightly outperforms mean-standard deviation normalization in this context. The results demonstrate the model's ability to accurately predict landslide-prone areas and providing valuable insights for risk mitigation and infrastructure protection. This approach underscores the importance of precise data preprocessing and balanced datasets in improving the reliability of landslide susceptibility models.
- PublicationDevelopment of internet of things based power distribution unit(2024-08)The rapid growth of data centers and the increasing demand for the efficient power management have highlighted the limitations of traditional PDU. Current PDU lack remote control and monitoring capabilities, and integration with IoT cause to poor power usage. The main objectives of this study are to develop an IoT based PDU by designing and implementing a smart PDU with an integrated sensor module, PZEM-004t instead of using a standalone module to reduce overall size of the system and to develop an interface system for remote monitoring and control of the PDU. This study involves hardware design, software development, and testing. The hardware design steps involve selecting and integrating the suitable components to develop the system such as PZEM-004t sensor, Tmega328p microcontroller, and CP2102 USB-to UART communication module. A simulation-based project is conducted using NI Multisim software to test the circuit design before printing the system on a PCB. For the software development, the firmware for the control and monitoring system is created using Arduino IDE. Lastly, the system to tested to evaluate the performance and effectiveness of the IoT based PDU. The analysis of this studies, PDU system can compacted into desired size of the board, and the system can have serial communication with the connected PDU for the control and monitoring features. The development of an IoT-based PDU helps data center management by providing an efficient solution for control and monitoring.
Most viewed
- PublicationAn approach to assembly facilities layout design problems using simulation(2005-03-01)This project focuses on designing an assembly line. Besides, the effects of different model configuration and number of station reduction on the assembly line are studied. In order to design an assembly line, minimum number of station is use to ensure high station’s and worker’s utilization. The assembly line model is developing using WITNESS simulation software. Four assembly line models with different configuration are built using WITNESS. The aim of the simulation is to find the best design that will be later implemented in reality. Statistical analysis, ANOVA is used to find any significance or insignificance of performance measures for different layout model. Results from the experiment have revealed that model’s configuration influence the line’s performance measure.
- PublicationCDS502 – Big Data Storage and Management (Storan dan Pengurusan Data Raya)(2023-02)Pusat Pengajian Sains Komputer (School Of Computer Sciences): Exam Papers/Teaching Resources
- ItemE-Government And Trust In Public Administration From The Perspective Of E-Notifikasi Users(Universiti Sains Malaysia, 2021-05)Trust in public administration has been declining over the years. Inefficiency and poor service delivery were identified among the factors that contributed to this problem. In this case, e-government related activities were proposed to rebuild citizens’ trust levels in public administration. This research was carried out to examine the relationship between e-government related activities and trust in public administration by targeting the users of the e-notifikasi system Ministry of Health, Malaysia. A conceptual framework was developed and validated through 382 screened sets of self-administered questionnaire. Data analysis using SmartPLS software revealed that trust in public administration was influenced by e-service quality and public administration communication. The attitude towards e-government mediated the relationship between independent variables and trust in public administration. Gender only moderated the relationship between e-service quality and trust in public administration. The results contributed to a new body of knowledge in examining the relationship between e-government activities and trust in public administration. The study demonstrated gender imbalance between male and female respondents for the moderator analysis. The study results suggested that more studies need to be conducted to discover the associations between all the variables, particularly in the Malaysian context.
- ItemCustomer Experience Management: Topology, Antecedents, And Outcome(Universiti Sains Malaysia, 2018-09)Academic research on Customer Experience Management (CEM) is still inconclusive although it is an important element in exploring customer experience. This limitation calls for a systematic theorization and operationalization of CEM. To this end, this study employed a sequential mixed-method methodology to identify the CEM topology and its antecedents and outcome in the Malaysian hotel industry. An exploratory study was first conducted to develop this model based on the triangulation between data from a set of best-practice reviews, interviews, and observations. The proposed model was then tested through a survey and analyzed by using structural equation modeling via Smart PLS. This study reconceptualized customer experience in the context of the service industry based on the experiential values that are detectable, memorable, manageable, distinguishable, and personalizable. Accordingly, CEM topology was operationalized as an organizational competency to manage experiential values co-creation (emotional, sensorial, behavioral, intellectual, relational, and interactional values). Customer relationship management, employee experience management, innovation management, and experiential marketing were identified as the key antecedents and marketing performance as the main outcome. The findings revealed the relative importance of the CEM antecedents. Apparently, CRM drives behavioral, relational and interactional experience management whilst innovation management drives emotional, sensorial and intellectual experience management. Employee experience management enables relational and interactional experience management, and experiential marketing is a necessary prerequisite for managing sensorial, intellectual, behavioral, and relational experiences. The study also revealed that emotional experience management and interface experience management are better predictors of marketing performance. The implications of this study suggest that hotels are more likely to achieve greater marketing performance if they invest in their customers’ emotional and interactional experiences. Although this study used the Malaysian hotel industry as a proxy for the service industry, the proposed model is an inclusive and non-industry-specific framework that is useful for future investigations of service experience management
- ItemEffects Of Gender And Age On Risk Preferences(Universiti Sains Malaysia, 2016-04)Women are commonly stereotyped as more risk averse than men in financial decision. Young age group are believed to be more willing to take more risk than elder age group. The present study intended to look into the effects of gender and age on risk preference. This study explained why a subject behaved as risk taker or risk averse in risk decision through three aspects, certainty equivalent; Probability Weighting Function; and Value Function. By looking on how a subject evaluated probability, we knew his or her risk behavior through the underweighting or overweighing of probability. Payoff valuation helped us to determine whether a subject valued a payoff higher or lower than the expected payoff. The study conducted a field study by recruiting working adults from different gender and different age groups. They were asked to make decision on lotteries and the data collected were used to estimate and plot Probability Weighting Functions and Value Functions. The findings showed that women were more risk averse than men as they overweighed small probabilities less (more) and underweighted large probabilities more (less) than men in gain (loss) domain. At the same time, elder adults were found to be more risk averse in gain domain but more risk seeking than young adults in loss domain.