http://hdl.handle.net/123456789/113 Sat, 04 Apr 2026 12:02:59 GMT 2026-04-04T12:02:59Z http://hdl.handle.net/123456789/11435 reactive power management and OPTIMIZATION OF TRANSMISSION LOSSES FOR TRANSMISSION NETWORK Rehman, Muhammad Fawwad ur Enroll # 02-244181-008 Reactive power (RP) plays a substantial role in transmission network efficiency. Though, the management of reactive power is extremely challenging for utility companies with diverging loads and deviating seasonal conditions. To cope with the reactive power flow ofthe transmission network, that is a dense network of lines (transmission) deals with different voltage levels, substations or grid stations and, different types oftransformers.. In this research, the Transmission network of a utility company (radial system) selected to analyze the reactive power management (RPM) and optimizing their transmission losses for this Transmission losses were simply calculated by using the load flow method of a power system, while technical losses were calculated by suitable load-flow studies simulated under the Power System Simulator for Engineering (PSSE) Software. The model presented in this study was simple, work on different phases, and easily applicable to a large transmission network comprising several buses, After the simulation studies, it was observed that the control and management of reactive power by adding the different sizes of capacitors locate on different buses at the transmission network of a utility company is adequate to improve the voltage profile in the selected case study. It not only improves the transformer’s loading but also provides system stability to the case study along with the reduction ofstress levels on the entire network of a utility company. Supervised by Dr. Haroon Rasheed Fri, 01 Jan 2021 00:00:00 GMT http://hdl.handle.net/123456789/11435 2021-01-01T00:00:00Z http://hdl.handle.net/123456789/11434 PERFORMANCE ANALYSIS OF MASSIVE MIMO ON TERAHERTZ COMMUNICATION Abdi, Abdinisar Hirsi Enroll # 02-244191-015 Growing mobile traffic demand, wireless technologies must expend mobile wireless capacities. 6G is designed to provide a high standard infrastructure enabling a va riety of technologies including fully Artificial Intelligence (AI), Augmented Reality (XR), the Internet of Things (IoT), e-health, smart vehicles and mobile broadband communication. The terahertz band (0.1-10 THZ) communications are expected to be essential to assist in making 6G wireless technology vision conceivable of the next decade. Although THz supporting a high data rate but there are many practical challenges that needs to be addressed. This thesis focuses performance analysis of massive Multiple Input Multiple Output (MIMO) antennas on tera hertz bands for 6G wireless technology. We analyze the outdoor performance of massive MIMO systems for two ultra-high frequencies (73 GHz and 100 GHz). The simulation was performed with the Matlab based NYUSIM statistical simulator which gave us an accurate analysis and measurements at millimeter and terahertz wave bands with the carrier frequency range from 500 MHz to 100 GHz. For designing our model, we considered a physical model because of cost, suitable for simulation and low computational complexity. We specially selected Geometry Based Stochastic Model (GBSM) because of accuracy. Modeling of massive MIMO channels used GBSM model also called Statistical Special Channel Model (SSCM) and the simulation was realistic scenarios. This thesis investigates different trans mitters and receivers, LOS and NLOS propagation environment with lower and upper bound T-R separation distance at 50 m to 500 m by using two Ultra high frequencies (73 GHz and 100 GHz) in Urban Micro (UMi) cellular scenario. The result parameters including omnidirectional and directional Path Loss (PL), omnidirectional Power Delay Profile (PDP), Directional PDP with strongest power, Angle of Arrival (AoA) power spectrum and Angle of Departure (AoD) power spec trum for both LOS and NLOS environment in UMi scenario were compared. The results shows that line of sight environments provides better received power due to power is concentrated in one direction rather than distributing the same power in different directions and PL are too high in NLOS environments. The AoA and AoD we found different lobes of signal power spectrum which help wireless design ers to get an information of antenna arrays and beam steering algorithms to select maximum strength of the signal. The investigation frequencies (73 GHz and 100 GHz) are a key component for technologies on 6G wireless communication. Supervised by Dr. Haroon Rasheed Wed, 01 Jan 2020 00:00:00 GMT http://hdl.handle.net/123456789/11434 2020-01-01T00:00:00Z http://hdl.handle.net/123456789/11429 FINE-GRAINED CLASSIFICATION OF VEHICLES BY USING CONVOLUTIONAL NEURAL NETWORKS Khairi, Danish ul Reg # 48863 Machine Learning has a practical and profound application in intelli j gent traffic management systems. ITS is a very broad terminology in which includes vehicle detection, classification, monitoring, surveillance, license plate recognition, etc. Vehicle classification playing a vital role in the intelligent transportation sys tem for traffic management and monitoring. This study is aimed at the fine-grained classification of vehicles using convolutional neural networks. To accomplish the task there are lots of challenges involved in which the biggest challenges are Inter class and Intra-class similarities between the make and models of vehicles, lightning conditions, background, shape, pose, a viewing angle of the camera, speed of the vehicle, the size of the vehicle, color occlusion and environmental conditions. There are three different datasets are used in this research BMW-10, Stanford Cars, and PAKCars.The BMW-10 and Stanford Cars datasets are available open-source, while PAKCars dataset is self-generated especially for fine-grained classification of cars in Pakistan to analyze the implementation of research. The system will work on ma chine learning which is further divided into two steps namely training and testing. Initially, the system will be trained on the training dataset and afterward, the per formance of the system will be tested using the test dataset. In the training part of the system, four different DCNN models are Mobilenet, InceptionV3, VGG-19, and ResNet-50 used. Each model is trained on all three datasets (BMW-10, Stanford Cars, and PAKCars). A total of 10 classes are evaluated in the BMW-10 dataset having a total of 511 images whil 196 classes are evaluated in Stanford Cars datasets having 8144 training images and 44 classes evaluated in PAKCars datasets which have total 1000 images. To perform the classification of the fine-grained vehicle DCNN models are used. The result acquired after processing reveals the results under the performance of true classification ResNet-50, VGG-19, inception-V3, and Mobilenet respectively. Mobilenet and InceptionV3 models consume less computational power and are less accurate, but VGG19 and Resnet50 are more accurate, because of their higher num bers of layers and architecture that make them complex and more computational power consuming as compared to Mobilenet and InceptionV3. Some false classifica tions occur due to inter-class and intra-class similarities Supervised by Umair Arif Wed, 01 Jan 2020 00:00:00 GMT http://hdl.handle.net/123456789/11429 2020-01-01T00:00:00Z http://hdl.handle.net/123456789/11428 INTEGRATION OF RENEWABLE ENERGY TO THE DISTRIBUTION NETWORK Hassan, Muhammad Qadeer ul Enroll # 02-244172-005 The advanced era needs continues supply of electricity. Since the depletion of fossil fuels could not fulfill the increased demands, hence Integration of Renewable Energy sources with the Distribution Network can provide sustainable and continuous supply of electricity. Depending on the scale there are various issues in integration that is needed to be resolved. Both technical and non-technical aspects of integration serious concerns that leads to voltage instability in distribution network, fluctuation in the frequency of the system, protection of the system from faults. In Sindh region of Pakistan, currently RE is integrated to only two areas that is Gharo and Jhimpir. Although there is potential of DG penetration to sindh region yet no such power plant is established. Many researches are done in this area but most of them are analyzing economical and regional benefits. This thesis comprises of selecting optimal DG placement and it’s size required for penetration at that selected region represented as buses. Load flow at the buses analyzed the voltage instability and it’s mitigation with and without hourly loading conditions also with respect to the valuation in season for optimal improvement in certain challenges. Sensitivity analysis is done to calculate the voltage deviation at different loading factors to find out the appropriate place for interconnecting DG to the network. Size of the DG is also a main factor to maximize the system voltage. DG’s are placed at selected location of different size and their voltage improvement is analyzed. The size of the DG that gave maximum output voltage are chosen for integration. The network condition is analyzed with and without load by simulating it in Power factory tool. Open circuit network is simulate in the first step to analyze the losses of the present network. Some of the buses that are away from the grid showed power losses. From observation it was analyzed that voltages are improved after integrating DG units but some buses are also showing over voltage condition violating 70 % voltage limit considering it not suitable for connection point. The proposed method is validated by implementing it on CIGRE Task force and IEEE-33 bus system. Both network validates the method by selecting suitable site and size decreasing the network losses and improving bus voltages Supervised by Dr. Muhammad Raza Wed, 01 Jan 2020 00:00:00 GMT http://hdl.handle.net/123456789/11428 2020-01-01T00:00:00Z