M.Phil / MS
Permanent URI for this collectionhttps://repository.cuilahore.edu.pk/handle/123456789/42
This collection archives the complete set of theses produced by students of the COMSATS University Islamabad, Lahore Campus.
Browse
16 results
Search Results
Item Performance Improvement of Smart Grid by Implementing Hybrid Load Forecasting Method Using Neural Networks(COMSATS University Islamabad Lahore Campus, 2024) Muhammad Asad,; SP21-REE-004; Dr. Muhammad Farooq-i-Azam Assistant Professor Assistant Profesor [Supervisor]; LHR TP 8065The increasing use of renewable energy resources in power systems worldwide has led to the development of smart grids, which allow for bidirectional communication between supply and demand. Accurate load forecasting is crucial for maximizing the efficiency of these smart grids, as it allows for the switching between renewable and other energy resources, reducing overall costs and transmission losses. Neural network-based load forecasting models are performing better than traditional statistical models. The hybridization of neural network-based models with optimized learning algorithms is the focus of our research. In this work, we propose a novel approach for load forecasting in smart grids. The approach includes four key components: refactoring the smart grid dataset for load forecasting, gathering a highly relevant set of parameters affecting domestic load, constructing four hybrid models with appropriate tuning parameters, and comparative analysis of these models. We devise proper format and structure the data used for load forecasting, ensuring that it is in the best possible form for accurate predictions. We carefully select the most important parameters for medium and short-term load forecasting, ensuring that the model is able to take into account the most relevant factors. We use an empirical approach to determine the best set of parameters for our models, ensuring that they are optimized for accurate predictions.We have modelled three machine learning hybrid models with ANN, hybridized with genetic algorithm, particle swarm optimization, and Levenberg Marquardt, along with one deep learning model RNN-LSTM. Our comparative results show that ANN-GA and ANN-PSO performed better in medium-term forecast with accuracy rates of 93.7 and 98.26 percent, respectively. While ANN-LM and RNN LSTM performed better in short-term forecast achieving accuracy rates of 97 and 99 percent, respectively. Our results demonstrate that our proposed approach is effective in achieving highly accurate load forecasts, making it a valuable tool for optimizing the performance of smart grids. iItem Fuzzy Model Predictive Controller for Attitude Control of Satellite(COMSATS University Islamabad Lahore Campus, 2023) Rabiya Noor,; SP20-REE-008; ,Dr. Mujtaba Hussain Jaffery Assistant Profesor; LHR TP 7894The technological advancement in satellite has tremendously improved in recent years. Generally, satellite applications have been classified into four categories i.e., navigation, communication, weather and earth observation missions. The successful execution of these applications heavily relies on the performance of attitude determination and control subsystem of satellite. Therefore, researchers have been continuously working to improve the satellite attitude regulation under challenging space environment by applying various control strategies i.e., PD, PID, LQR, MPC. MPC is a famous control technique, as it can handle the operational constraints of the system effectively. In satellite, actuator torque has limited capability and it can result in the saturation problem. MPC can regulate the satellite attitude without violating the actuator constraints hence ensuring system stability. However, the mathematical model of satellite is highly nonlinear and for nadir pointing satellite, it can be treated as simple linear system for a constrained operating range. Although, it impacts the accuracy of the nonlinear model when a model-based controller is used. This results in a compromise between simplicity and accuracy that is faced by every control engineer. Therefore, in this work a systematic approach to improve the attitude accuracy has been explored while deriving linear subsystems of the nonlinear satellite model for the controller design. In this approach, the mathematical model of the plant is represented by various linear subsystems in terms of Takagi-Sugeno Fuzzy rules which are independent of each other and a control law is designed for each rule using MPC. The output of each MPC is accumulated based on Parallel Distribution technique hence providing command to the actuator. Eventually, the torque produced by the actuator is applied on nonlinear plant. Comparison of the response specifications (transient and steady state response) between PID, MPC and Fuzzy-MPC controllers under different cases such as external disturbance, actuator constraints, large reference angles and initial angles has been discussed. The simulation results indicated that proposed Fuzzy-MPC exhibits consistent and improved performance in all four cases in comparison to other controllers. The pointing accuracy is successfully achieved while catering to the challenge of actuator constraints under external disturbances of LEO.Item Nowcasting of RSSL in wireless communication channel over the sea using machine learning algorithms(COMSATS University Islamabad Lahore Campus, 0023) Farwa Jafar,; FA20-REE-005; Dr. Khurram Zaidi; LHR TP 7896The presence of naturally occurring evaporation duct (ED) phenomenon is very high in the tropical/equatorial regions of the world. Although, refractivity estimation of EM and Radio waves in ED is well studied in the literature, still, the signal propagation through ED over-the-horizon needs to be thoroughly researched to help determine the received-signal-strength-level (RSSL) for a reliable wireless communication link. In order to accurately predict RSSL in ED, we have acquired RSSL (avg.) per-minute data for three months over-the-horizon distance of 50 km (Tx-Rx) from onshore-to-offshore Oil & Gas Platform. This data was collected using fixed antenna heights in ED. Applying deep learning algorithms on real-time RSSL data, we have nowcasted the future RSSL values for next 5 seconds timescale in this thesis. A thorough comparison is made between the CNN and LSTM deep learning methods for real-time series prediction analysis. These deep learning networks are linked with numerous convolution layers to grasp the nonlinear mapping between measured and future RSSL values. Coding and Simulation work is performed in Python 3.9 environment and results are generated in Kaggle Notebook. CNN and LSTM networks have never been used earlier for predicting “signal strength” over-the-horizon and over-the-sea under ED environment. The contribution of this research is to bridge this gap and examine the accuracy of LSTM and CNN for nowcasting RSSL data. According to what we've discovered, both of these neural network models are capable of achieving adequate to high prediction power given that the "datasets" are suitably big. Both methods, when taken as a whole, are reliable with regard to their hyperparameters. However, with increasing number of training courses, LSTM didn’t improve its performance, whereas CNNs performed correspondingly more accurate each time. For 3rd training, CNN has given the most optimal fitting of training data as compare to test data. The RMSE achieved for CNN after third training was 4.47 which is the least of all simulations. Hence, CNNs proved to be superior, since they operate one order of magnitude quicker than LSTM. We proposed that the early predictive capability, speed, and resilience of CNN open its door to nowcasting’s future.Item Propulsion control of aerial vehicle using online control allocation approach(COMSATS University Islamabad Lahore Campus, 2020) Javairia Mehar,; FA18-REE-004; Dr. Mirza Tariq Hamayun; LHR TP 7886Propulsion Control of Aerial Vehicle using On-Line Control Allocation Approach Control Allocation (CA) is one of the reconfiguration approaches to manage actuator redundancy in over actuated systems in an effective way, so that the fault tolerance can be achieved. There is a wide range of algorithms and approaches which include linear as well as quadratic programming-based CA schemes. Such schemes can deal with constraints on the control input, but these methods cannot be represented in explicit form.Reliability is an important property in Fault Tolerant Control domain and managing redundant actuators optimally plays a vital role to achieve this goal. In this thesis to handle the issue happening in actuators channels through Integral sliding mode FTC scheme by taking advantage of redundant actuators through online control allocation scheme designed for Propulsion Control application, (where only the engines thrust is available) for a safe landing. It is presumed that the health information of actuators is available for the proposed FTC scheme. Here, both lateral and longitudinal models of a large airplane are considered to design separate controller for each modes of operation. The viability level of the actuators is utilized by the control distribution conspire to reallocate the control signs to adjust engines push whenever actuators-failures happens without reconfiguring the controller. The main objective in this thesis is safe landing when system is totally dependent upon engines thrust. The effectiveness of the proposed propulsion control scheme against deficiencies is tried in simulations using MATLAB in view of an enormous vehicle airplane model. The basic purpose to design this scheme is emergency safe landing in case of total hydraulic failures. The proposed scheme using online control allocation is compared with direct control allocation to show its effectiveness.Item Module level power electronics in distrubted power system for solar PV application(COMSATS University Islamabad Lahore Campus, 0022) Muhammad Talha Naveed; , SP20-REE-015; Dr. Muhammad Yaqoob Javed; LHR TP 7895hotovoltaic (PV) solar energy is as promising as other renewable energies. Different researchers and engineers are attempting to increase the efficiency of solar PV system. As a result, for PV modules, this enhancement may be accomplished at almost the same level. As is well known, solar PV systems are less efficiency as a result of changing climatic conditions. Module-level power electronics (MLPE) do this by providing the performance improvements of a distributed transmission system in both partial and full shading conditions. As a result, MLPE successfully harvests the distributed maximum power point (DMPP) from solar to accept DC from PV or the grid. Each PV module is connected to the power system via a separate dc/dc converter with Maximum Power Point Tracking (MPPT) capabilities in the DMPP scheme. Each PV panel has a built in power optimization or micro-inverter that helps it work better in partial shade. In order to design the MLPE the efficiency of contemporary string inverters are comparing. The efficiency may be measured in a variety of situations, including uniform irradiation and partial shade. As a result, a DC-DC converter that is attached to each PV module is required to offset shading losses. The optimizer identifies the Local peak using a DC-DC converter from the unit, shuts down the modules during fire situations, troubleshoots, and monitoring a module in a highly efficient manner in this work. On the Matlab software tool, the effectiveness of the proposed power optimizer is displayed. To calculate efficiency, several firms such as SMA, Solar Edge, Huawei, Tigo, and Enphase can use helioscope to compare production and loses data. The results suggest that MLPE produces superior outcomes.Item Analysis and modeling of lithium iron phosphate (LiFePO4) batteries for state of health estimation(COMSATS University Islamabad Lahore Campus, 2022) Hamza Bashir,; FA19-REE-016; , Assistant Profesor [Supervisor]Li-ion batteries have been chosen as the best candidate for portable, mobile and high power applications due to their high energy densities, long lifetimes, and high efficiencies compared to other battery types, like lead-acid and nickel-based batteries recently. On the other hand, these batteries are prone to failure due to charge imbalance in batteries linked in series or parallel, which may be catastrophic. Therefore, they must be carefully monitored in real-time. So, the Li-ion cells chemistry batteries have exceptional performance compared to other chemistries, but only if treated well (cell balancing, protection from overcharge, over-discharge, and short circuit conditions). Despite the advantages of lithium-ion batteries, these batteries still have a short life due to particular aging processes inside the batteries and improper cell balancing, which become evident under defined conditions. Reliable battery calendar ageing prediction, on the other hand, remains a crucial yet complex issue for improving the performance of related storage devices. So, manufacturers and researchers need to investigate the cell balancing process and aging processes in lithium-ion batteries to determine the aging effects that occur and the factors that cause a rapid decrease in their lifetimes. Nowadays, Lithium iron phosphate (LiFePO4) based chemistry batteries are considered to be one of the most valuable lithium-ion batteries in the market because of their high energy density, lack of memory effect, lower self-discharge, long lifetime, large cycle life number, inherently safe cathode structure under critical conditions, and non polluting characteristics. In this thesis, 3S-1P battery pack cell balancing system is implemented by using passive cell balancing technique including the thermal effects and an aging prediction model has been developed for lithium iron phosphate (LiFePO4) batteries using MATLAB/Simulink©, which has been able to predict the aged effect in terms of capacity fade and internal resistance increase. Thevenin equivalent circuit model is utilized and SOH is estimated. Parameter estimation is done using soft computing technique. Overall, different models of SOH estimations are studied and reviewed. The results are developed in MATLAB Simscape and further extended in MATLAB scripts. Possible effects of cell balancing and internal resistance has been discussed in the conclusion.Item Blood glucose forecasting in type 1 diabetes mellitus patients using machine learning techniques(comsats university Islamabad lahore campus, 2022) Hatim Butt,; FA19-REE-009; LHR TP 7889Diabetes Mellitus is a metabolic disease that causes the body to lose control over blood glucose regulation. Patients with Type 1 diabetes completely rely on insulin therapy by themselves or using some automated insulin delivery systems. In both the cases, it is pertinent to have good estimate of future blood glucose levels. An efficient diabetes management demands accurate prediction of future blood glucose levels, failure of which results in short and long term health complications. With the modern exordium of quantified-self such as continuous glucose monitoring(CGM) systems, a patient can have access to their personalized glycemic profile which can be utilized for accurate prediction of future blood glucose levels. In recent years, machine learning methodologies have sparked a lot of interest in predicting glucose levels in diabetic patients, leading to the development of a variety of methods and techniques. However, the prediction accuracies of these methods are not good enough to be declared them as reliable predictors for evaluating glycemic conditions. In this research work we utilized multi-layered Long Short Term Memory(LSTM) network a famous deep learning technique based on recurrent neural network(RNN) for making prediction of blood glucose levels in patients with type 1 diabetes. The proposed framework predicts the future blood glucose level using Ohio T1DM dataset at prediction horizon(PH) of 30 and 60 minutes. Experimentation was also carried out on better feature representation to model in order to achieve higher prediction accuracy. The effect of different input feature sets, towards improvement of prediction accuracy was also been investigated. The results on Ohio T1DM Dataset (2018), that contain eight weeks’ worth of data shows that our method achieves the lowest RMSE score of 14.76mg/dL and 25.48mg/dL for prediction horizon of 30min and 60min respectively. The obtained results are the best known as per our knowledge using this dataset. The proposed methodology can be utilized in closed loop systems for precise insulin delivery to patient for their better glycemic control.Item Mitigation Of Harmonics Using Soft Computing(COMSATS University Islambad Lahore Campus, 2020) Muhammad Waleed Rafique,; SP18-REE-008; Dr. Mujtaba Hussain Jaffery, Assistant Profesor [Supervisor]; LHR TP 7467Excessive use of non-linear devices in industries results in current harmonics that degrade the power quality and has an unfavorable impact on the entire performance of power system. In this research, a Hybrid Shunt Active Power Filter (HSAPF) is implemented for compensation of reactive power and harmonic current component for balanced load by improving the power factor, total harmonic distortion (THD), and performance of the system. For extracting three-phase reference current for HSAPF, a novel control technique for harmonics mitigation based on Particle Swarm Optimization (PSO) and Fuzzy logic Controller (FLC) is proposed in this research. SIMULINK is used to implement Pq0 and Id-Iq control approaches for harmonics mitigation, with a PI controller for voltage regulation and a hysteresis controller for the reference current generation. Following a comparative analysis of both techniques i-e Pq0 and Id-Iq control techniques, implement the Soft computing technique (PSO) on PI controller technique on any of these techniques that perform better for harmonics mitigation. As the performance of the PI controller is dependent on its gains, the best gains produce better results and improve system response. The given results show that PSO technique is an efficient technique that helped to give better performance as compared to the conventional PI controller by improving steady-state response by giving the best gains to PI controller. Because of the system's robustness and non linearity, a single PI controller does not provide better performance So, by using gains obtained from PSO adaptively tuning of PI controller has been carried out by using FLC that can help to improve the dynamic performance of HSAPF. Therefore, from the comparative analysis, it can be inferred that PSO based Adaptive Fuzzy PI system has more efficient results with a minimum THD, improved stability time, and a power factor nearer to unity as compared to other techniques.Item Output Feedback FTC Scheme For Linear Parameter(Publisher COMSATS University Islambad Lahore Campus, 2016) Izhar Ul Haq; , FA13-MSEE-006; Dr. Mirza Tariq Hamayun, Assistant Profesor [Supervisor]; LHR TP 6820The proposed research has been carried out for an Active Fault Tolerant Control (FTC) of Linear Parameter Varying (LPV) systems using output feedback mechanism. Output feedback is a practical approach due to the fact that all the plant states are not accessible or measurable, therefore estimating the unknown states and maintaining the closed loop stability in an LPV framework is closer to the control of nonlinear systems. At the same time, it utilizes the mature nature of linear control theory. In this research, LPV observers in LMI framework have been designed for affine LPV plant, using two different LPV approaches namely affine and polytopic, in order to estimate the unknown states. These estimated states are used as a feedback to LPV controller. Integral sliding mode controller in LPV framework has been designed along with control allocation scheme to control the controlled states as well as to counter for actuators faults and failures. It has been assumed that Fault Detection and Isolation unit, also called FDI, is providing updated actuator faults or failures information to the control allocation (CA) scheme. Detailed performance and stability of the proposed control scheme has been checked in fault-free case, i.e. nominal condition, as well as in the situation when actuator fault occurs. To validate the proposed output feedback FTC mechanism, an LPV model of longitudinal plant of aircraft, taken from the literature, has been considered as a benchmark in the simulation. The simulation results show that the system is quadratically stable and give good tracking capability by the LPV controller-observer pair in nominal as well as in actuator’s fault or failure situation.Item Human Activity Recognition System For Long Term(Publisher COMSATS University Islambad Lahore Campus, 2020) Shan E Ali,; SP18-REE-026; Dr. Ali Nawaz Khan, Assistant Profesor [Supervisor]; LHR TP 6444The study of Human Activity Recognition (HAR) for Long Term Health Monitoring (LTHM) has gained significant importance for its wide range of applications. These applications range from sports and rehabilitation sciences to assisted living for older people. In addition to that LTHM is an efficient solution for the prevention of lifestyle diseases like stroke, heart failure, and various health problems that occur due to prolonged inactivity. With the increased availability of accelerometer sensors embedded in mobile phones, we can efficiently explore the Activities of Daily Living (ADLs) of an individual. This research aims to develop a LTHM system for evaluating ADLs of a person using a mobile phone-based accelerometer sensor and the ‘MyNeuroHealth’ application. Data collected in an unconstrained environment by various individuals throughout the day to create templates of ADLs. Collected data is prepared and preprocessed by assigning hourly labels to the ADLs, encoding categorical values and random sampling of data. This data is used for training the machine learning model and for classifying activities according to their energy expenditure or user exhaustion levels. Collected dataset further extended to daily, weekly, and monthly basis to provide long-term health profiling (LTHP). 23 types of basic, complex and transitional activities were evaluated for each day. The results show that an Artificial Neural Network (ANN) can efficiently identify and detect ADLs with more than 90% accuracy. Person independent ADLs templates for weeks 1, 2, 3 and 4 achieved an accuracy of 89, 96, 93 and 89 percent correspondingly. On the other hand, person dependent ADLs templates from various walks of life achieved on average 94% accuracy. Person independent ADLs templates for weeks 1, 2, 3 and 4 achieved an accuracy of 89%, 96%, 93% and 89% percent correspondingly.