Theses & Dissertations
Permanent URI for this collection
Browse
Browsing Theses & Dissertations by Title
Results Per Page
Sort Options
-
Publication3D localization of moisture distribution in rice silo using RTI based on Wi-Fi signal( 2022)Abd Alazeez Al MaleehRice is a staple food which conveys a sign of local culture throughout Asia, particularly in South Asia, and it is consumed daily, either as cooked rice or indirectly as rice flour. Several measures are aimed at raising national rice production for the next few years as expected to see rising demand and falling supply. Researchers discovered that, in order to ensure an adequate supply of domestic output, appropriate silo facilities together with good agriculture practices should be addressed to resolve this prolonged issue in the agricultural industry. A silo's main purpose is to protect grain crops from the environment, especially moisture build-up, pest infestation, and fungal load. Therefore, grain storage is part of an important aspect of quality assurance in post-harvest activities. One of the main issues is the moisture content of the rice during storage. The current state of the art for moisture measurement of rice in a silo is based on grab sampling or very much relies on single rod sensors placed at random locations during measurement. The moisture content of stored rice is directly dependent on the surrounding and environmental factors which in turn affects the quality and economic value of the rice. In addition, the moisture content needs to be measured frequently for prompt action. Until today, the current sensor is very localized and the continuous measurement microwave sensor is very costly. There is also no commercially available 3D volumetric measurement of rice moisture content in the silo. This research reported preliminary work using a off-the-shelf wireless device i.e., esp8266 which can be placed around the silo to measure the change of moisture. A new technique has been proposed in this research, which uses a more accurate reconstruction of the image generated by radio tomography based on Wi-Fi signals. The technique is based on the Hybrid Tikhonov-LASSO (HTL) combines the advantage of Tikhonov and the LASSO method, which achieved the reconstructed image is cleaner. Also used Regression-based machine learning (ML) on RF Tomographic Imaging which can provide 3D moisture content measurements to localize the moisture distribution in storage. This proposed technique can detect multiple levels of localized moisture distributions in the silo with high accuracies, depending on the size and shape of the sample under test. Unlike other approaches proposed in open literature or employed in the sector, the proposed system can be deployed to provide continuous monitoring of the moisture distribution in silos.
-
PublicationA hybrid human sensory mimicking approach: an integrated e-nose and e-tongue system( 2012)Nazifah Ahmad FikriTaste and smell are two of the five human senses that are common among mammalians. These two senses are usually used together to make up the brain’s perception of flavour. This has lead to the study of data fusion of multiple artificial sensory systems such as electronic nose and electronic tongue. However, the data fusions performed by these studies are based on separate single-modality systems. Presented is the development of a hybrid system which combines an electronic nose and electronic tongue in a single system. Both sub-system uses off-the-shelf components and developed using rapid prototyping techniques. The hybrid system combines two sensor arrays of MOS gas sensors and ion-selective electrodes. It also consists of a signalcollecting unit and pattern recognition software applied to a computer. The system uses qualitative analysis which is similar to the human sensory system, implementing Principal Component Analysis (PCA) and Artificial Neural Network (ANN). Three tests were performed representing agricultural, environmental and food production applications. The performance of the single-modality systems were compared to the hybrid system. The results show that the hybrid system performed better than the both single sub-systems when appropriate fusion method was used, and able to archive up to 98.67% accuracy. This proved that the multi-modality system performed better in samples discrimination than single-modality system which mimics more closely the human sensory system.
-
PublicationA modified retinex illumination normalization approach for infant pain recognition system( 2014)Pains in newborn babies are monitored in a Neonatal Intensive Care Unit (NICU) for medical treatment. Pain in newborns can be detected by studying their facial appearance. Even though the outcome is acceptable, it is not adequately vigorous to be used in unpredictable, non-ideal situations such as noise and varying illumination environment. First, to improve the noise cancellation robustness an adaptive median filter (AMF) is proposed. Mean and variance of median values are selected to generate a weight for each window part of the images such as 3x3, 5x5 or 7x7. Various linear and nonlinear filters are adopted to eliminate the noise in the images. Quantitative comparisons are performed between these filters with our AMF in terms of Peak Signal-to-Noise Ratio (PSNR), Mean Square Error (MSE), Image Enhancement Factor (IEF) and Mean Structural SIMilarity (MSSIM) Index. The average results show improvement in terms of 40.63 db for PSNR, 6.01 for MSE, 258.09 for IEF and 0.97 for MSSIM respectively. In this work a novel method of illumination invariant normalization known as Modified Retinex Normalization (MRT) for preprocessing of infant face recognition is proposed. This is based on a modified retinex model that combines with histogram normalization for filtering the illumination invariant. The proposed method is compared to other methods like Single scale Retinex (SSR), Homomorphic method (HOMO), Single Scale Self Quotient Image (SSQ), Gross and Brajovic Technique (GBT), DCT-Based Normalization (DCT), Gradientfaces-based normalization technique (GRF), Tan and Triggs normalization technique (TT), and Large-and small-scale features normalization technique (LSSF) for evaluation with Infant Classification of Pain Expressions (COPE) database. Several experiments were performed on COPE databases. Single PCA, LBP and DCT feature extraction information yielded a good recognition result. However, by summing these three, it gives more robustness to noise and illumination classification rate because the sum rule was the most resilient to estimate errors and gives higher than 90% accuracies of pain and no pain detection. The new illumination normalization and combination of features gives higher results of more than 90% on five different classifiers with various algorithms such as k-nearest neighbors (k-NN), Fuzzy k-nearest neighbors (FkNN), Linear Discriminat Analysis (LDA), Feed Forward Neural Network (FFNN), Probabilistic Neural Network (PNN), General regression Neural Network (GRNN), SVM Linear kernel (SVMLIN), SVM RBF kernel (SVMRBF), SVM MLP kernel (SVMMLP) and SVM Polynomial kernel (SVMPOL) with different performance measurement such as Sensitivity, Specificity, Accuracy, Area under Curve (AUC), Cohen's kappa (k), Precession , F-Measure and Time Consumption .
-
PublicationA new hybrid control algorithm design and simulated for Longitude and latitude movements stabilization of nonlinear fixed-wing UAV( 2014)Faizan Ahmed WarsiUAVs (Unmanned Aerial Vehicles) have enabled a number of new mission capabilities and are frequently used in many applications. There are a few sorts of UAVs configuration available in the market, but fix-wing UAVs is the most popular among them. It is mostly used in surveillance and rescue type applications by militaries as well as business organizations .This makes UAV design and controlling as one of the most sizzling subject for the researchers. The troublesome undertaking for the scientists in UAVs design is to develop its efficient control algorithm which makes UAV flight settle under typical and instability or irritated conditions. Like other UAVs, fix-wing UAVs are also non linear in nature and its stabilization during flight is troublesome task. It has two major movements that are, longitudinal and lateral movement, which must be control legitimately to make Fix Wing UAV flight stable. There are several control techniques available that are used to control its flight movements. These accessible control techniques have a few pros and cons, and have their own working impediments. This research exploration deals with the designing of control system for small size fixed-wing UAV to enhance the flight performance under uncertainties condition. Generally these UAV countenances unpredicted problems during flight such as, heavy wind gust, alter in wind current course, sensors commotions or sensors noises. These impacts may float the UAV from it sought direction and makes it unstable. The available traditional control techniques are not robust enough to handle these perturbed circumstances. In this thesis a new hybrid control algorithm is presented for longitudinal and lateral movements controlling of small fixed-wing UAV. The proposed control technique is developed by joining the PID algorithm with PD-LQG algorithm to stabilize the small fixed-wing UAV flight under sensor noisy conditions and external disturbance circumstance. For verifying the performance of proposed control strategy it is simulated on ‘Yardstick’ type small fixed wing UAV. The simulation are performed and analyzed under different windy and noisy conditions. MATLAB Simulink with its Aerosim block set is used to execute all the simulation. The simulation results demonstrates that the proposed control technique performed exceptionally well under perturbed conditions and its performance is much better than available traditional algorithms under uncertainty conditions.
-
PublicationA novel human blood-plasma separation efficiency measurement method based on ultrasonic technique( 2013)Muhammed Sabri SalimClinical diagnosis of disease relies on the preliminary conclusions provided by blood testing. Accurate disease diagnosis exemplifies the crucial importance of proper separation of the blood plasma, which contains antibodies and proteins, from the platelets and red and white blood cells before testing. The most common method of clinical blood separation is via centrifugation, based on sedimentation theory. Inaccuracy of blood-plasma separation occurs not only because cells vary from one person to another in their size and density but also due to variations in the human age, cellular environment and measurement technique. In addition, the sedimentation rate is affected by changes in the viscosity and density of the sample, which can lead to incomplete separation. The separation efficiency of a laboratory centrifuge device is enhanced by combining acoustic theory with sedimentation theory. The behaviour of waves propagating through a liquid during centrifugation is studied, and the acoustic transfer function model is derived. The derived acoustic transfer function facilitates the process of system performance assessment based on an off-line test of how changes in acoustic impedance affect a solution. This assessment will help to avoid costly experimental validation studies. Low concentrations of blood cells suspended in plasma samples were detected and measured via a new ultrasound pulse technique called Power-Pulser Decay (PPD). The PPD generates multi-decay pulses that are injected into the blood sample for a discrete centrifugation process to yield the detection ability for low concentrations of particles in a liquid. As a result, a new mathematical model for the separation efficiency is derived. This model is based on experimental work and defines the separation time of a constant working spinning velocity, which results in an optimised separation time for a predefined input of the separation efficiency. Based on these results, a new centrifugation controller is designed, and its performance is simulated. The results of the centrifugation time model demonstrate that centrifugation for 3 minutes at 3000 rpm for a 0.35-ml volume is sufficient to produce plasma with separation efficiency range of (95-100)%. The volume of 0.35 ml of 1ml blood sample was selected because this volume is sufficient for multiple types of disease tests. New centrifuge controller is designed based on a Fuzzy Logic Controller. The optimisation time model reveals that this controller saves approximately 2 minutes with 95% separation efficiency compared with the PID centrifuge controller. The new centrifuge controller allows for adjustable separation accuracy, a shorter separation time and low power consumption in comparison with the commercial centrifuge device. The controller successfully saves 18kWh monthly for assuming that the centrifuge is operated 100 times per day. The improved separation efficiency will lead to broad improvements in the centrifugation process, including the ability to define the percentage of separation and the separation accuracy for the contents of a solution. In medical applications, this technique will result in more accurate test results that can be obtained more quickly, which will improve the ability of physicians and patients to make important clinical decisions.
-
PublicationA novel hybrid fuzzy PID controller for attitude stabilization of a remote operated quadrotor unmanned aerial vehicle( 2012)Zul Azfar AhmamThis thesis presents a new attitude control development to be implemented in the flight control board (FCB) of quadrotor unmanned aerial vehicle (UAV). A simple structure of quadrotor was developed to test the attitude stabilization control. The cross “+” shaped structure of quadrotor make it is very easy to develop. A remote operated quadrotor consists of four brushless DC motors (BLDC) with fixed pitch propeller attached on it, the FCB equipped with an inertial measurement unit (IMU) sensors, four electronic speed controllers (ESC), a set of remote controller transmitter and receiver and a high discharge lithium polymer (Li-PO) battery. Quadrotor have six degree of freedom (DOF) of flight control. The quadrotor flight behavior is same as helicopter but can fly as fast as fixed wing aircraft. However, only four movements are produced from 6 DOF flight control which are take-off/landing, roll, pitch and yaw. These movements are performed by varying speed of four propellers to produce different amount of thrust. The differences of thrusts will produce different quadrotor flight direction. A mathematical modeling was done to analyze the effectiveness of a control system to the real quadrotor. This mathematical model is used to represent the real quadrotor which was simulated using Simulink in Matlab Software. The new attitude control involved a hybrid controller of proportional-integral-derivative (PID) and a fuzzy logic controller (FLC). This new hybrid Fuzzy-PID (FPID) controller is developed to improve the performance of traditional PID controller. The approach to hybrid both of these controllers is using the parallel technique. All hybrid Fuzzy-P, Fuzzy-I and Fuzzy-D structures are combined together to form a new FPID controller. The purpose of designing the hybrid system is to use FLC as an automatic tuner for PID controller. The well-tuned PID gain of PID controller is combined with FLC to get a better performance compared to using the PID controller alone. Both controllers are simulated in Matlab software and then implemented to the real quadrotor to compare the performance. A test flight is conducted to observe the differences in controlling the quadrotor in flight using the new FPID controller instead of using PID controller. The result showed that the new FPID controller is better than PID controller in term of response and stability. The FPID controller is very quick to achieve the desired target and produce less overshoot than the PID controller and thus proof that the FPID controller is more stable compared to the conventional PID controller.
-
PublicationA numerical study on photovoltaic cell performance at high temperature( 2013)Abdulkareem Naser MahmoodThis research work presents the improvement of photovoltaic cell characteristics above room temperature. The effect of temperature on different characteristics of the photovoltaic cell has been analyzed mathematically. Among the characteristics of photovoltaic cell first of all, the light transmission and absorption characteristics were investigated. Then the open circuit voltage, short circuit current and the output power characteristics of the photovoltaic cell were investigated. Finally the effect of temperature on the rate of change of all these characteristics was analyzed using Si and GaAs in the active layer of the photovoltaic cell. The numerical results obtained were compared. The comparison results revealed that the absorption characteristics, open circuit voltage, short circuit current and the output power have been increased but the variation of these characteristics have been reduced significantly by using GaAs. Therefore GaAs can be considered as the best alternative material to fabricate solar cell in upcoming decades.
-
PublicationA study of a vision-based lip movement analysis for hearing-impaired person( 2014)Muhamad Kamil Syahid TalhaThe study of vision based lip movement is the interpretation of human lip movement while speaking. Deaf and hard of hearing people often have a problem being able to understand what being talking in the conversation. Sign language may be useful for them to communicate but not everyone may understand the sign language. To make the communication more interesting and does not have any obstacle to talk to anybody, the lip reading is the best way to communicate. Deaf and hard hearing person can talk like the normal person, however, they do not hear the word spoken by themselves. On the other hand, there is no sound feedback system for them. This is a reason they could not speak with the correct pronunciations. To teach them pronounce the word correctly, fluently and to understand others in the conversation, the lip reading systems that could train them pronounce the word correctly needs to be developed. Previous researcher proposed many methods to recognize the spoken word based on the lip movement. One of the method is attach the colour marker onto the lip surface and a vision sensor is used to track the moving marker. Other researchers extract the lip region by using Active contour Model (ACM) or a snake method to draw the keypoint around lip edges. In this studies, a system to track and recognize the spoken word based on lip movements is proposed. A camera is used as a vision sensor and an image processing technique is employed to extract and track the lip. The lip horizontal and vertical distances of the lip are used to measure the ellipse while the systems track the lip movements. The trajectories of the ellipse are resampled to 10 point called the features point. The word database contains 10 spoken word frequently speak at the hospital are designed based on the distribution of the 10 feature points. Two types of word database have been designed which are the individual word database and the universal word database. The individual word database is defined as the distributed feature points of each subject by speaks the words with repetition. Meanwhile, the universal word database is defined as the distributed feature point of all word regardless who is the subjects. The recognition experiments are conducted and the system recognizes the unknown words with the recognition rate 92.47% accuracy by using the individual database. Meanwhile, the system recognizes the unknown words with the recognition rate 90.39% accuracy by using the universal database.
-
PublicationA surface roughness based visual and analysis system for surface quality improvement in fused deposition modeling rapid prototype machine( 2006)Khairul Fauzi KarimIn rapid prototyping (RP), part deposition orientation and surface finish are two significant concerns, but they are contradicting with each other. In model building in RP, a concession is commonly made between these two features to get good quality surface roughness at a short build time. A concession among these two contradicting concerns can be achieved via an adaptive slicing method; on the other hand, selection of an appropriate part deposition orientation will further provide an improved solution. In this thesis, an effort towards determining an optimum part deposition orientation and adaptive slicing method for Fused Deposition Modeling (FDM) process for enhancing part surface finish, and hence, reducing build time (repeating process in RP cycle) is proposed. The quality of the surface roughness is determined by using visual and analysis. This Surface Roughness Based Visual and Analysis (SRVA) system is obtained based on the calculation of surface roughness (Ra). In this present work, the Region Based Adaptive Slicing method is applied in building the model in FDM. The proposed methodology allows the RP user to observe and analyze the prototype model before fabricating the prototype model in the FDM. A program based on fuzzy logic is also used to verify the input and output parameters obtained from the proposed method. The developed SRVA system has successfully improved the surface finish and minimized the build time in fabricating the prototype model in FDM. The result showed that increasing part deposition orientation would decrease the Ra value of the model. For 00 and 900 part deposition orientation, the Ra from measurement are closed to the Ra output from fuzzy logic with percentage differences 1.78% and 1.52% respectively. Therefore, the Ra values calculated from the SRVA system are acceptable for these orientations. However, for 450 part deposition orientation, it is 2.26% higher than the Ra output from fuzzy logic because during fabrication process, the surrounding support model affects the surface finish of the prototype model. However, this value is also acceptable because the effect of surrounding support model to the surface finish has not been the focus of the present work. The result also shows that the adaptive slicing method has improved the surface roughness of the prototype model. The inspected Ra obtained by this method is 1.22% lower than that obtained without adaptive slicing method, but 0.56% higher than that obtained by fuzzy logic. This result is obtained without the necessity to repeatedly fabricate the model or piecework in FDM for good quality surface roughness as the proposed method in this thesis successfully managed to optimize RP cycle; hence the build time in RP is reduced.
-
PublicationAdaptive neural controller for attitude stabilization of quadrotor unmanned aerial vehicle (UAV)( 2011)Shaiful Zairi Ahmad SubhiIn line with the rapid development of the world science and technology, our country has matured in the exploration of new knowledge. Our country has given the very bright opportunity in an engineering field. The researchers intend to produce a useful technology to the community. Now on, air to space technology become famous for them, especially in the unmanned aerial vehicle (UAV) area. Therefore, this thesis also will present the development of an UAV that uses a new concept named quadrotor which are a combination of helicopter and airplane. This quadrotor capable of takeoff and landing vertically like a helicopter and can maneuver like an airplane. Quadrotor built with a fourrotor placed symmetrically like an added shape, Electronic Speed Controller (ESC), Inertial Measurement Unit (IMU) sensor and a Lithium Polymer (Li-Po) battery. Simple configuration causes quadrotor preferred used as unmanned vehicles. Quadrotor movement is controlled by the thrust produced by the four-rotor which means to move quadrotor the four rotor speed must be controlled independently. The rotor can be controlled separately through the programming provided in the control system. The main problem faced is to control the quadrotor attitude during flight time. Adaptive Neural Controller (ANC) is used to control the quadrotor attitude. This system was adopted because it can respond quickly and accurately following the reference model. Before that, the quadrotor kinematic and dynamic analysis must be done. Through the analysis, we can determine quadrotor modeling. Modeling quadrotor used to find the quadrotor plant so that the system will be able to identify the quadrotor characteristics. Constant parameters are getting from calculation and experimental test used in quadrotor modeling. Simulation tests carried out using Matlab software to find quadrotor stability in roll, pitch and yaw axis. After the controller gives a good response in simulation, the code converts into C programming and implement in the microcontroller and should synchronize with the hardware configuration. The actual test flights conducted in an indoor to test for stability. Disturbance test also will be conducted. For the future development several of sensors will be added to increase the ability quadrotor to fly autonomously without any guidance from the ground station and also human.
-
PublicationAdaptive neuro-controller design for nano-satellite attitude control( 2012)Norhayati Mohd NazidThe motivation of this research is to bring the technology of spacecraft control into university education and to bring the possibility of developing our own satellite that will put us of equal standard with other developed nations. The purpose of this research is to develop the control scheme for three axes stabilization of nano-satellite system namely Innovative Satellite (InnoSAT). An adaptive neuro-controller (ANC) is applied as a controller in many application such as in robotics, power system, industries and etc. There are many successfully applications of ANC in controlling the satellite attitude control have been proposed. In this regards, four types of ANCs using two different control scheme and using two different algorithm for nano-satellite attitude control have been introduced in this research. These are ANC based on Model Reference Adaptive Control (MRAC) scheme trained by Back-Propagation (BP) algorithm, ANC based on MRAC scheme trained by Recursive Least Square (RLS) algorithm, ANC based on Internal Model Adaptive Control (IMAC) scheme trained by BP algorithm and ANC based on IMAC scheme trained by RLS algorithm. These two different control schemes are used by the ANC to adjust the output response of InnoSAT to follow the desired target. In this research, BP and RLS algorithms were used as an adjustment mechanism to update the parameters of the ANC. A multilayer perceptron (MLP) network with one hidden layer has the capability to approximate any continuous function up to certain accuracy. It is a very powerful technique in the discipline of control systems, especially when the controlled systems have large uncertainties and strong non- linearities. MLP network is used for ANC in this research. The design of ANC is initially started with design of ANC based on MRAC scheme using BP algorithm. Then, the ANC based on MRAC using RLS algorithm is designed and the performance for both ANCs based on MRAC were compared in term of convergence speed and possible divergence for certain conditions. The design is continued by designing the ANC based on IMAC scheme using BP algorithm and the last part of designing is designed the ANC based on IMAC scheme using RLS algorithm. The performance for both ANC based on IMAC scheme are also compared in term of convergence speed and possible divergence for certain conditions. The simulation results for all ANCs indicated that ANC using RLS algorithm have faster convergence speed compared to the ones trained by BP algorithm. The best ANC based on MRAC and ANC based on IMAC are compared with a conventional proportional, integral and derivative (PID) controller. Simulations have been carried out and for several reference inputs namely unit step, square wave and Y-Thompson. The simulation results are presented and the output responses show that the ANC based on MRAC performance is acceptable even in the case of the InnoSAT is subjected to varying gain, measurement noise, time delay and disturbance. Then, the ANC based on MRAC scheme is simulated with two axes cross coupling system and the simulation results show that the InnoSAT system is stable. The final simulation is tested the ANC with real time attitude reference which is Y-Thompson input reference. The results showed that the ANC based on MRAC scheme can stabilized the InnoSAT system even the system is subjected with varying gain, measurement noise, time delay and disturbance.
-
PublicationAn assessment of the orbital elements of RazakSAT for attitude determination using extender Kalman Filter( 2020)Malaysia has successfully developed and launched RazakSAT, the first mini satellite at the Near Equatorial Orbit (NeqO). The mission was focus on a technology demonstrator and one of the payloads is a high-resolution camera to capture images at the equatorial region. The purpose of RazakSAT’s Attitude Determination System (ADS) is to ensure that the orientation is relative to Earth. Any misalignments or disturbances on the ADS can affect the orientation of the satellite in terms of the orbital position. Therefore, this research focuses on determining the attitude of a satellite in order to determine the orbital position for control requirement purposes. However, conducting experiments in the space environment to determine the accuracy and computation time is impossible due to cost constraints and the fact that facilities are scarce resources. As RazakSAT data is NEqO based orbits, so it proposed to use as reference to the actual data to substitute for the experiments in the space environment. The main aim is to give accurate information and computation time for the attitude estimation of ADS. In this thesis, RazakSAT data are used as a reference to give accurate information and computation time for the attitude estimation of ADS. Keplerian orbit model is implementing as an orbital model and compares with the NEqO of RazakSAT data. Besides that, Satellite Tools Kit (STK) software were used to conduct and validate the reliability analysis for orbital elements of RazakSAT such as the Earth-Centered Inertial (ECI), Earth Centered Earth Fixed (ECEF) and Latitude Longitude Altitude (LLA) based on the Two-Line Element (TLE) provided by Astronautic Technology Sdn Bhd (ATSB). The sun vector and magnetic field vector in the field in the orbit frame were analysed using the sun model, and International Geomagnetic Reference Field (IGRF). The sun vector in the body frame is the measurement from the sun sensor and magnetometer. The result on ECI and ECEF was analysed as well using the STK. The Keplerian orbit model was found to be more accurate than STK for LLA where the error yield is less compared to the Keplerian orbit model. The overall result for ECI, ECEF, and LLA is less than 5%, which also meets the error requirement of ATSB for RazakSAT orbit. For the sun model analysis, due to the difficulties in obtaining information about the RazakSAT sun sensor from ATSB, the analysis for the sun model utilized the results from STK and mathematical model developed in this thesis. The result shows that the percentage error for the sun in the orbit frame and body frame is acceptable, less than 5%. The magnetic field in the body frame and orbit frame from STK is more accurate compared to IGRF model. From the overall orbital results, RazakSAT was found are able to fulfill the requirement of the orbital specification while entering its orbit. For the attitude estimation of the satellite, a recursive approach known as an Extended Kalman Filter (EKF) is used to estimate the attitude. In this thesis, the kinematic and dynamics models for the EKF method are derived and analyzed in terms of controllability, observability, and stability. The result proved that the model could be controllable, observable, and marginally stable condition where it can be used for estimation by using EKF.
-
PublicationAn automated cells counting system for Malaria based on thick blood smear samples( 2023)Thaqifah Ahmad ArisMalaria is one of the most serious blood infection disease that remains to be a global public health challenge especially in African region. An estimated total of 241 million of malaria cases happened worldwide which leads to 627 000 deaths in 2020, found on statistic reported by World Health Organization (WHO). Malaria is caused by plasmodium parasites carried through contaminated female Anopheles mosquito. There are five types of plasmodium parasites, yet Plasmodium Falciparum and Plasmodium Vivax are the main species that most commonly detected worldwide. Based on the high number of malaria occurrence, it is crucial to do medical inspection every year. Currently, microscopy test using thick blood smear still be the standard method for malaria detection. However, this procedure is time consuming and prone to human error. Nowadays, image processing is recognized as a quick ways to analyze a lot of blood samples. Therefore, there is an urge to develop an automated cells counting system for malaria based on thick blood smear samples. Thus, this research has established an automated cells counting system for malaria detection by using several image processing such as image enhancement, image segmentation and intelligent classifiers. Starting with image enhancement, there are various contrast enhancement techniques and colour constancy techniques that were applied to enhance the malaria images. Next, the malaria images was segmented by using several thresholding and clustering techniques. Here, phansalkar technique able to achieve good segmentation performance in terms of accuracy, specificity and sensitivity with value of 99.86%, 99.87% and 92.47%, respectively. In addition, phansalkar technique able to segment malaria parasites as well as obtain the fully segmented malaria parasite region with clean segmented malaria images. After that, size, shape, texture and colour based features were extracted from the segmented parasite to be used as inputs to the three different types of classifiers namely multi-layered perceptron (MLP) trained by Levenberg-Marquardt (LM), single-hidden layer feed forward neural network (SLFN) trained by extreme learning machine (ELM) and online sequential extreme learning machine (OS-ELM). Overall, the automated cells counting system for malaria that has been developed using MLP network trained by LM algorithm is capable to perform the classification between malaria and non-malaria parasites by utilizing a total of 7500 segmented cells extracted from 300 malaria images with validation accuracy of 86.78%. To conclude, the proposed automated cells counting system for malaria based on thick blood smear samples is capable to perform the detection of malaria parasites using thick blood smear images by producing high counting accuracy with total counting of 1736 parasites from 300 total images and achieve accuracy of 86.78%.
-
PublicationAn improved mobile robot based gas source localization with temperature and humidity compensation via slam and gas distribution mapping( 2016)This research is concerned with the problem of localizing gas source in indoor environment using a mobile robot. The problem could be seen as similar to the event of hazardous gas leak in a building. Since the environment is often unknown to the robot, the Simultaneous Localization and Mapping (SLAM) operation is required. Two open source SLAM techniques (i.e. Gmapping and Hector SLAM) were implemented to provide this crucial information. Extensive experiments and analysis on both SLAM techniques yielded that the Hector SLAM is more suitable for gas distribution mapping (GDM) application due to the improved robot pose estimation, less computational requirement and only performs map correction locally. Therefore, the Hector SLAM is combined with Kernel DM+V algorithm to achieve real-time SLAM-GDM for predicting gas source location. Rigorous real-time experiments were conducted to verify the performance of the proposed SLAM-GDM method in an uncontrolled office building with the presence of ethanol emission. The experimental results showed that the prediction of gas source location is often accurate to 0.5 to 2.0m. Furthermore, an Epanechnikov based Kernel DM+V algorithm was also introduced to limit extrapolation range in GDM computations. The observed advantages were lower computational requirement and slightly more accurate prediction on gas source location. More importantly, it was found that the maps produced were able to indicate the areas of unexplored gas distribution and therefore could be used for the robot‘s path planning. The final and the main part of the thesis deals with the effect of ambient temperature and humidity on metal oxide gas sensor (i.e. TGS 2600) response; which could affect the GDM results. Linear regression processes were conducted to create a model to correct the temperature and humidity drift of the gas sensor response. The model (i.e. function) was tested in various configurations and was found to minimize the effects of the two environmental factors on the gas sensor response in different gas concentrations. Finally, two versions of Kernel DM+V/T/H algorithms were proposed and coupled with the drift model to compensate for temperature and humidity variation during the GDM task. The experimental results showed that the Kernel DM+V/T/H algorithms were able to produce more stable gas distribution maps and improve the accuracy of gas source localization prediction by 34%.
-
PublicationAn integrated indoor air quality monitoring system with pollutants recognition and enhanced indoor air quality index( 2016)Shaharil Mad SaadPoor indoor air quality (IAQ) may pose threats to human’s health. The concentration level of harmful gases and contaminants in polluted indoor air is up to five times higher than in normal indoor air. In order to ensure that people breathe-in safe air comfortably in the indoor air environments, continuous IAQ monitoring is deemed important. The main objective of this study is to develop an integrated indoor air quality monitoring system (IAQMS) with pollutants recognition and Enhanced Indoor Air Quality Index (EIAQI). The wireless IAQMS adopts an array of sensors including gas sensors, particle sensors and thermal sensors to detect multiple pollutant parameters at a relatively low cost as compared to the professional sensing devices. Overall, this study uses eight sensors to measure nine indoor air pollutants which are Oxygen (O2), Carbon Dioxide (CO2), Carbon Monoxide (CO), Ozone (O3), Nitrogen Dioxide (NO2), Volatile Organic Compounds (VOCs), Particulate Matter (PM), Temperature (Temp) and Relative Humidity (RH). This IAQMS has successfully recognized five sources of indoor air pollution with classification rate of 100%. These five sources of indoor air pollution are: ambient air, human activity, presence of chemical, presence of fragrance and presence of food and beverage, are successfully classified by Multilayer Perceptron (MLP) and KNearest Neighbour (KNN) using Vector Array Normalization (VAN) before Principle Component Analysis (PCA) feature. Finally, the last objective of this study is to integrate the IAQMS with EIAQI. This study proposes an EIAQI which comprises of three different indices: Indoor Air Quality Index (IAQI), Thermal Comfort Index (TCI) and Smell Index (SI). IAQI utilized the seven air parameters to measure the quality of indoor air and shows the status of IAQ whether it is “Good”, “Moderate”, “Unhealthy” or “Hazardous”. This IAQI is developed using Air Quality Index (AQI) from the United States Environmental Protection Agency (US EPA) as its main reference. TCI applied the same principle with IAQI. The TCI used Temp and RH to indicate the thermal comfort level of a room. Therefore, TCI status is shown either as “Most Comfort”, “Comfort”, “Less Comfort” or “Least Comfort”. In contrast with the IAQI and TCI which generate their index based on single pollutant parameter, SI is generated based on an array of pollutant parameters. For example, IAQI is determined based on single pollutant that gives the lowest rate. SI on the other hand, generates the smell perceptions based on all nine pollutants input. The final result would be classified as either the smell is “Neutral”, “Pleasant” or “Unpleasant”. After all individual index has been obtained, an EIAQI is formulated which combines all the previous three indices. This EIAQI informs the users about the overall comfort status in the room.
-
PublicationAn intelligent gesture recognition system( 2012)Wan Mohd Ridzuan Wan Ab MajidInformation and knowledge are expanding in quantity and accessibility. However, people with functional limitations, such as hearing impaired, often left out of conversation where there are wide communication gaps between them with the ordinary people. The sign language is the fundamental communication method between people who suffer from hearing defects. In order for an ordinary people to communicate with hearing impaired community, a translator is usually needed to translate the sign language into natural language. This project presents a simple method for converting sign language into voice signal using features obtained from the hand gestures. Using a camera, the system receives sign language video from the hearing impaired subject in the form of video streams in RGB (red-green-blue) colour with a screen bit depth of 24-bits and a resolution of 320 x 240 pixels. For each frame of images, two hand regions are segmented and then converted into binary image. Feature extraction model is then applied on each of segmented image to get the most important feature from the image. Artificial Neural Network (ANN) provides alternative form of computing that attempts to mimic the functionality of the brain. A simple neural network model is developed for sign recognition directly from the video stream. An audio system is installed to play the particular word for the communication between the ordinary people and hearing impaired community.
-
PublicationAnalysis of crosstalk in Mechanomyographic signals from elbow flexors during forearm flexion, pronation and supination tasks( 2020)Irsa TalibMuscle assessment has diverse applications including sports, athletics, medicine and prosthetic control. Mechanomyographic (MMG) signals have their significant importance for assessment of muscle function. Crosstalk is the contamination of MMG signals coming from target muscle by the signals coming from adjacent muscles. MMG signals do exhibit crosstalk, thus measurement of crosstalk is of vital significance in muscle function study. This research examined crosstalk, root mean square (RMS) and mean power frequency (MPF) for MMG signals from three elbow flexor muscles including biceps brachii (BB), brachialis (BRA) and brachioradialis (BRD). Three sustained isometric tasks including forearm flexion, pronation and supination were performed. Further this study analyzed crosstalk and MMG signal parameters in three different directions to muscle fiber axes, at five different submaximal to maximal torque levels, with variation in anthropometric parameters including skinfold thickness (ST), inter-sensor distance (ISD), length (LA) and circumference (CA) of arm. During each task, three microelectromechanical systems (MEMS)-based tri-axial accelerometers were used to obtain the MMG signals from the longitudinal, lateral and transverse directions with respect to muscle fibers. Peak cross-correlation coefficients at zero-time lags were used for quantification of the crosstalk. Crosstalk values showed statistical significance among all nine possible axes pairs (p<0.05, η2=0.160-0.510). The transverse axis showed lowest MMG RMS values and the transverse axes pair generated the lowest mean crosstalk values (2.160-9.140%). Submaximal to maximal (20%, 40%, 60%, 80% and 100% maximal voluntary contraction) torque levels were observed to have strong positive correlations with crosstalk values and MMG RMS (r>0.700) while strong negative correlations (r<-0.900) with MMG MPF. Negligible correlations were observed for crosstalk, MMG RMS and MMG MPF with the four anthropometric parameters. The results may be used to improve our understanding on mechanics of the elbow flexor muscles during sustained isometric forearm flexion, pronation and supination tasks using the MMG technique.
-
PublicationAnalysis of grounding rods size in varying soil resistivities using COMSOL multiphysics( 2024)Ong Shen YoongThis thesis discusses the performance of various grounding rod sizes in reducing grounding resistivity. Malaysia is a country that constantly has lightning strikes all year round, therefore grounding systems should be improved to provide better safety features. Small modifications could be made to any existing grounding grid to improve its efficiency in conducting electricity. Grounding rods of different lengths and diameters were designed and simulated in COMSOL Multiphysics to understand how it will alter the overall grounding resistivity and performance. The results have shown that the thicker and longer rods had a higher conductivity and was able to disperse electricity more efficiently. This method can be further improved by simulating other variables for future work.
-
PublicationAnalysis of time-frequency features for classification of asthma severity level using computerized wheeze sounds( 2019)Fizza Ghulam NabiIn asthma patients wheeze sounds are produced due to obstruction in lung sounds. Any medication or management of patients is done according mild, moderate and severe condition of asthma patients. Literature review indicates that analysis and classification of wheeze sounds according to severity levels of asthma patients using time-frequency features in different datasets according to location and phase is required to explore more. The objective of this study is to investigate and classify wheeze sounds according to the severity levels (mild, moderate and severe) of asthma patients using time-frequency features. This study focusses on the self-monitoring and self-management of asthma patients using tidal breathing. Segmented and validated wheeze sounds were collected from the trachea and lower lung base (LLB) of 111 asthmatic patients during tidal breathing. The collected data was split into 9 datasets based on the auscultation location, and/or breath phases. For every segment, the frequency-based, spectral integrated (SI) and integrated power (IP) features were computed. Subsequently, a univariate and multivariate statistical analysis were performed on the features to investigate the significant difference of features in details. Classification was then performed using the ensemble, support vector machine (SVM) and k-nearest neighbor (KNN) methods. In addition, two classification frameworks introduced to identify most effective classification of severity levels. Most of the selected individual features and feature vectors frequency-based, SI, IP observed indicated significant difference (p < 0.05) in majority of datasets. Overall, the best PPV for the mild, moderate and severe samples were found to be 100% (KNN), 92% (SVM) and 94% (ensemble) respectively were obtained with IP features. The μ(SD) values of features have not indicated any specific and continues behavior with respect to severity level in all nine datasets. The findings of research illustrate that the distribution of frequency and spectral energy in the recorded signal varies depending on the auscultation location (trachea and LLB), phase (inspiratory and expiratory) and severity levels (mild, moderate and severe). With the consideration of auscultation location trachea-related datasets produce higher effect size than that of LLB-related datasets. For SI and IP features in most comparisons, the ensemble classifier produced best performance in terms of sensitivity, specificity and positive predictive value (PPV). However, frequency-based features indicated highest performance with the KNN and SVM classifier. Trachea-related datasets samples produced the highest classification performance than all other datasets in all type of combinations. The results of validation also have been found above on average. In future acoustic features, deep learning classification technique and feature optimization can be implemented.
-
PublicationAnalysis on network performance by considering doppler shift effect for connected car wireless mobile communication( 2022)Azarul Fahmin Ab HamidConnected car technology ought to be a keystone of revolutionizing the automotive area. The concept of a so-called connected car has recently emerged as one of the distinctive designs of new-generation automobiles, with the ability to provide drivers with a new dimension if services through wireless communication. The performance of the car's engine can now be monitored from basic information like location and security. Connected cars consist of some important components, namely the vehicle itself, communication link and application layer. This research focuses on communication component, where the environmental effects such as doppler shift and area density are analyzed. A few problems have been identified including the doppler shift effect which will cause a communication delay, high speed movement which will cause the vehicle to frequently change cellular areas, consequently causing delays during handover process. The majority of high-speed transportation uses a private network of its own for monitoring purposes. This study conducted an experiment on a network for 2G, 3G, and LoRA communication modules by using SIM900, SIM800L and SIM5360E in order to study the impact of doppler shift on path loss. The experiment demonstrated that there is a doppler shift effect, and that the effect increases noticeably as one moves from a rural to an urban environment. The experiment on handover delay is run in this study on each communication delay as well. Analysis of the data from that experiment shows that the doppler shift effect has a negative impact on handover delay. The study concludes that SIM5360E, a communication module with a 3G network, is the best communication module with 10% to 20% better performance in each analysis and that Fast Base Station Switching (FBSS) is the best handover technique based on all of those findings with 12% better result.
6 10