Ethiopia Institute of Technology- Mekelle
Permanent URI for this communityhttps://repository.mu.edu.et/handle/123456789/58
Browse
101 results
Search Results
Item Distributed Power Flow Controller Based Power Quality Improvement for Grid Connected Wind Farm- Case Study Ashegoda Wind Farm(Mekelle University, 2025-09-23) Ashenafi SelemaGrid penetration level of renewable energy is growing and merging dramatically. However, poor power quality creates a major integration and operation problems. Ashegoda wind farm represents Ethiopian first large scale wind power installation, having a total rated capacity of 120 MW. The substation is equipped with two 230kv buses that interconnect Lachi and Alamata substations. In the analyzed system, both transmission lines experienced high reactive power flow approximately 43MVAr and more than 3% current total harmonic distortion (THD), which leads to additional power loss, voltage drop, equipment overheating, and network congestions. On the other hand, the on-load tap changer (OLTC) transformer used to support voltage sag, swell, under-voltage, and over-voltage has slow response time up to 10 seconds per tap, and creates transformer overheating and mechanical fatigue. To solve such integration challenges, the incorporation of a distributed power flow controller (DPFC) with in the system is necessary for improved power quality and reliability. This study analyzed and modeled the integration, impact, cost benefit analysis, and Genetic Algorithm (GA) based optimal sizing and placement of DPFC for Ashegoda wind farm. The objective was to model, simulate, and asses its impact on voltage stability, reactive power compensation, and steady state and dynamic performance. The result indicated that, without DPFC the 230kv system operated at under-voltage of 0.89 pu with 5% current THD at bus 690v. After integration of 8MVA series and 25MVA shunt DPFC controller, the voltage profile is improved to 0.96 pu and current THD is minimized to 1.8%. Integrating DPFC exhibited excellent performance in maintaining voltage stability and limiting short-circuit current levels under different fault scenarios. The cost benefit analysis was carried out over a 20-year period. The total net present value (NPV) is estimated around $15,310,316.0 US dollars, while the total investment cost amounts $7,883,000.00 US dollars. By implementing the system, Ethiopian electric utility is expected to gain $7.5 million US dollars profit without considering scrap value. Generally, the researcher proved that DPFC is technically and economically feasible. MATLAB/Simulink 2018a and Excel-2013 was employed to model, simulate, and analyze the proposed DPFC system.Item Artificial Neuro-Fuzzy Inference System (ANFIS) Based Speed Control of Separately Excited DC Motor for Load Torque Variations(Mekelle University, 2025-09-09) Kibrom ZerauSeparately excited direct Motor (SEDCM) is an electromechanical actuator used to power different loads across several industrial and domestic applications. One fundamental characteristic for controlling when driving is the motor's speed. The external load linked to the drive negatively impacts speed if the controller is weak and the load varies. Objective of this thesis work is to design an Artificial Neuro-Fuzzy Inference System (ANFIS)-based speed control mechanism for a separately excited DC motor under varying load torque conditions. The ANFIS controller integrates neural networks and fuzzy logic to improve motor speed regulation, ensuring robust performance despite disturbances in load torque. Additionally, this this work explores the effectiveness of armature voltage control (source voltage adjustment) for dynamically regulating motor speed, comparing its performance with conventional control strategies. ANFIS, fuzzy logic controllers, PID controllers, and open loop (without controller) have all been used to measure the speed of an independently stimulated SEDC motor. At first, the motor speed can be regulated and modified by changing the armature voltage (the input supply voltage).When the torque of the load grows and the transient and steady state faults rise, the motor's speed falls in the absence of a controller. The motor performs poorly as a result, and its speed will not maintain its rated level. A PID controller improves the motor's speed over an open loop, but the overall performance is still poor and there are still some transient and steady state issues. Although fuzzy logic controllers perform better than PID controllers in terms of system performance, the speed still fluctuates as the torque of the load changes. However, ANFIS better than fuzzy, PID, and open loop control systems, operates at its rated speed, has low steady state and transient errors, and keeps the motor's speed constant as the load increases. In conclusion, NFIS is superior to fuzzy and PID controllers due to its zero overshoot and reduced 38.31% settling time compared to fuzzy and 50.93 % compared to PID, reduced 37.12% rise time compared to fuzzy and 44.87 % compared to PID, and reduced 85% steady-state error compared to fuzzy and 98.5 % compared to PID. Additionally, by resolving the motor's nonlinear properties, the system's overall performance will improve.Item Voltage Control of DC-DC Boost Converter Using Lyapunov Rule Based Model Reference Adaptive Controller(Mekelle University, 2024-10-24) : Mebrahtu NgusseThe switched mode DC-DC converters are the simplest power electronic circuits that facilitate the conversion of electrical voltage from one level to another through a switching process efficiently. DC-DC boost converters are utilized for step-up voltage in various applications. The output voltage of boost converter has, oscillation, overshoot, undershoot, and steady-state error. PID controllers have been usually applied to the converters because of their simplicity to obtain the desired voltage. But, PID controller could work well in one operating condition and cannot continuously adapt the changes in the process dynamic. To overcome this problem an advanced controller is required. The proposed Lyapunov Rule Based MRAC is adaptive and non-linear controller designed to overcome the uncertainties and nonlinearities for DC-DC boost Converter under Continuous Conduction Mode (CCM) operating condition. Using MATLAB/Simulink the performance of the proposed Lyapunov rule based MRAC is compared with that of the PID controller based on the dynamic response of the system. Using PID controller, overshoot and settling time have been improved by reducing from 67.5229% to 7.7717% and 0.6985 Sec to 0.277 Sec respectively. In the case of the proposed controller (Lyapunov rule based MRAC), overshoot, settling time, and undershoot have been improved by reducing from 67.5229% to 4.5993%, from 0.6985 Sec to 0.1458 Sec, and from 0.0036% to 0.0% respectively. To test the performance of the DC-DC boost converter, it is assumed that, the input voltage has been decreased and increased from its operating point by 25% and 41.67% respectively. Also the load resistance is assumed to be decreased and increased from its operating load resistance by 25% and 20% respectively. An external disturbance is applied to the system to check how the controller handles to uncertainties and PID controller has shown deviation from the desired value but, the controller MRAC maintained the desired value.Item Comparative Analysis of Passive, PID And Fuzzy_PID Controller For Quarter Car Model(Mekelle University, 2025-09-15) Atsede GebreyohansThis paper presents a comparative analysis of passive, PID, and fuzzy PID control strategies for a quarter-car suspension system utilizing a fuzzy PID controller to enhance ride comfort and stability by effectively managing displacement and velocity. The quarter-car model serves as a simplified representation of vehicle dynamics, where the primary objective is to minimize the vertical displacement of the vehicle body and control the relative velocity between the body an the wheel. Additional PID controllers often struggle with non-linearity and uncertainties present in suspension systems; therefore, it is proposed a fuzzy logic approach to tune the PID parameters dynamically based on real-time system states. The fuzzy-PID controller integrates the benefit of fuzzy logic’s ability to handle uncertainty and the robustness of classical PID control. Simulation results demonstrate that the proposed fuzzy-PID versus passive controller significantly reduces body displacement by 88％of overshoot and 56％of settling time with bumpy input road1. Fuzzy-PID versus passive controller significantly reduces velocity response by 24％of overshoot and 55％of settling time with bumpy input road1.Item POLITICAL STANCE DETECTION AND CLASSIFICATION ON TIGRIGNA TEXT USING DEEP LEARNING APPROACHES(Mekelle University, 2025-07-08) Ngsti GebrehiwotThe rise of social media has transformed public discourse, providing platforms for individuals to express their opinions on various topics, particularly political issues. Political stance detection, which identifies an individual's position on specific topics, has become increasingly important for policymakers, researchers, and organizations aiming to navigate complex social landscapes and make informed decisions. Despite its significance, most research in this area has focused on English and other European languages, with limited attention to Amharic and virtually none to Tigrigna, a language spoken by millions in Eritrea and Ethiopia. This gap is particularly critical given the ongoing socio-political challenges, such as unemployment and civil unrest, in Tigrigna speaking communities. This study addresses the lack of research on political stance detection in Tigrigna by analyzing comments from the TPLF Facebook page. Data was collected us in the Face ¬pager tool, and two feature extraction strategies—Bag of Words (BOW) and Skip¬ gram from Word2Vec—were employed to convert textual data into numerical representations suitable for machine learning. Advanced deep learning algorithms, including Gated Recurrent Unit (GRU), Transposed Gated Recurrent Unit (T¬GRU), and Long Short¬ Term Memory (LSTM), were applied to classify political sentiments toward the TPLF party. The results demonstrate that the Transposed GRU model combined with the Skip¬ gram strategy achieved an accuracy of 82% and an F1¬score of 0.8822, representing a significant advancement in political stance classification for low¬ resource languages. These findings highlight the effectiveness of deep learning approaches in analyzing Tigrigna text and provide a foundational methodology for future research. This study addresses a gap in the existing literature by providing a nuanced analysis of the socio political dynamics within Tigrigna¬ speaking communities, which have been largely overlooked in political discourse research. By utilizing advanced techniques in stance detection, this research enhances our understanding of public sentiment and sets a precedent for scholarly inquiry into underrepresented languages. The contributions are threefold: it establishes a foundational dataset specifically tailored to Tigrigna ¬speaking contexts; it employs innovative natural language processing methods, such as transfer learning and alternative word embed dings; and it considers idiomatic expressions and the role of emojis, offering a more granular understanding of public sentiment. Looking ahead, future research should broaden the dataset to encompass a wider array of political topics and explore advanced machine learning techniques, thereby enriching the findings. This research lays the groundwork for subsequent studies and contributes to a more inclusive understanding of political discourse across diverse linguistic landscapes, ultimately fostering greater engagement with marginalized voices in the political arena.Item Numerical Analysis of Pile Group Bearing Capacity Using Finite Element Method(Mekelle University, 2025-06-02) Lchya GebrePile group foundations are widely used to support heavy structures where individual piles cannot provide adequate bearing capacity or settlement control. This study focuses on the numerical analysis of pile group bearing capacity under axial loading using PLAXIS 3D v. 2013 a finite element software for geotechnical applications. The analysis investigates the load settlement behavior of pile groups embedded in clay, considering homogeneous and layer soil profiles, using the Mohr Coulomb constitutive model to represent soil behavior. To analyze the influence of a group of piles on bearing capacity and load transfer mechanism among them, the research includes the use of different pile configurations such as changes in the pile spacing, geometry, and even material properties. In behavior of the structures tested, contact elements have been used to model the integration of piles with soil whereas the model boundary conditions are properly placed so as to avoid displacement that is deemed arbitrary. Numerical results obtained from the simulations are validated against results obtained from traditional analytical methods like Meyerhof’s and Vesic’s equations. The key findings state that the pile group bearing capacity is highly associated with the multiplier of pile spacing, and interaction between the pile and soil. The investigation also underlines the necessity to employ modern numerical methods, like PLAXIS 3D, to solve complex interaction of soil and structure that is common place in the analytical approach. This research offers some crucial information to the geotechnical engineers as aid in development and analysis of pile foundations to be more precise and more reliable. The results highlighted the importance of combining classical techniques with numerical ones in order to achieve more secure and more competitive design concepts for the foundation.Item Assessment of the Performance and Challenges of Public Procurement of Works in Semera, Afar National Regional State(Mekelle University, 2025-09-25) Ali Amin IbrahimPublic procurement is a pivotal mechanism linking project aspirations to the delivery of quality infrastructure. In Semera, Afar National Regional State, despite substantial investment in public works, procurement outcomes have often fallen short of expectations. This study assessed thirty public works projects implemented between 2018 and 2023, revealing systemic inefficiencies, delays, and quality shortfalls. Only 58% of projects complied with national procurement guidelines, procurement cycles averaged 6.5 months, and cost overruns affected 40% of projects, with an average escalation of 18% above contract values. Time and quality performance were also suboptimal, with less than half of projects completed on schedule and only 60% meeting technical standards. The analysis identified several institutional and market-related challenges undermining procurement performance. Capacity constraints were significant, with 72% of procurement staff lacking formal training and high turnover reducing continuity. Planning weaknesses were widespread, as 65% of procurement plans were delayed or incomplete, often without adequate market analysis. Transparency and accountability were insufficient: only half of tenders were openly advertised and nearly a third lacked complete bid evaluation reports. Market limitations further constrained performance, with an average of only three bidders per tender and contractor capacity often insufficient, resulting in project suspensions or abandonment. The consequences of these challenges were evident in persistent cost overruns, project delays, and compromised quality, particularly in essential infrastructure such as schools, roads, and health centers. Stakeholder confidence in procurement was low, with 68% rating performance as ineffective. These systemic weaknesses reduce the value for money and delay the delivery of public services, highlighting the urgent need for institutional reforms and operational improvements in public procurement practices in Semera. To address these gaps, the study recommends capacity building, improved planning, enhanced transparency, stronger oversight and accountability, and support for local contractors. Implementing these reforms will enhance efficiency, cost-effectiveness, and quality in public works, promoting sustainable economic development in Afar.Item EMOTION DETECTION AND CLASSIFICATION ON TIGRIGNA SOCIAL MEDIA TEXTS USING TRANSFORMER MODELS(Mekelle University, 2025-09-23) Rahel GebruThe rapid growth of social media has reshaped emotional expression, producing large-scale digital data for social, cultural, and political analysis, thereby highlighting the importance of reliable automated emotion detection tools. Despite advances in Natural Language Processing (NLP), Tigrigna remains underrepresented, with existing multilingual models often underperforming due to limited annotated data, lack of tailored resources, and linguistic complexity. To address this gap, this study introduces transformer-based models tailored for emotion detection and classification in Tigrigna social media texts, focusing on four emotion categories: happiness, sadness, neutral, and disgust. A total of 4,000 Tigrigna sentences were collected from Facebook and YouTube and manually annotated with a high Inter-Annotator Agreement. To expand and balance the corpus, 6,000 additional sentences were generated using data augmentation techniques, including backtranslation and synonym replacement, resulting in a final dataset of 10,000 sentences. Following preprocessing, including normalization, tokenization, and cleaning, the data was split into training (8,000), validation (1,000), and testing (1,000) subsets. Three transformer-based models namely XLM-RoBERTa, tiBERT, and the Tigrigna-specific tiRoBERTa were fine-tuned and evaluated using Macro-F1, precision, and recall metrics to address class imbalance. The results demonstrated progressive improvements across models: XLM-R achieved an F1-score of 81%, tiBERT 84.4%, and tiRoBERTa 88%, with tiRoBERTa outperforming the others across all emotion categories, particularly in distinguishing subtle distinctions between sadness and happiness. Misclassifications between neutral and disgust persisted, reflecting data-related issues, model-specific challenges, and the low-resource nature of Tigrigna. Data augmentation improved F1-scores by 2–10% across models, underscoring its crucial role in enhancing performance in low-resource NLP tasks. The study concludes that transformer models, when culturally and linguistically adapted, are highly effective for Tigrigna emotion detection. Future research should expand Tigrigna-specific pretraining corpora, explore advanced augmentation, investigate hybrid architectures, and integrate multimodal data (e.g., combining text with images or videos). Applying these findings via APIs and dashboards can support researchers, policymakers, and organizations in leveraging Tigrigna social media for informed decision-making.Item Performance Evaluation of CNN, ViT, and Hybrid Models in CT Based Brain Stroke Classification(Mekelle University, 2025-08-25) Angesom BerhaneMedical image classification is the use of AI tools to automatically detect disease, and segment affected body section which ultimately assist medical experts diagnose disease easily. So far, CNNs have played a major role in the field of medical image processing. Recently, another deep learning approach known as transformers have evolved, and outperformed CNN models; especially, when trained with enough amount of data. These models were originally developed for natural language processing. But, they have shown promising results in the computer vision also. This research investigates the performance of the standalone CNN and vision transformer models, hybrid these deep learning approaches to entertain the advantage of both at the same time, evaluate performance of the hybrid models, and ends up comparing the performance of the standalone and hybrid models. The research was conducted by gathering CT scan image data of stroke disease from an online repository i.e. kaggle. Eight models were trained and evaluated, including two CNN (ConvNext, EfficientNet), two transformers (Swin, ViT), and four hybrid models (ConvNext + Swin, ConvNext + ViT, Efficient + Swin, Efficient + ViT). Several metrics such as accuracy, precision, recall, f1 score, and ROCAUC plot were used to compare the performance of the models. In general, the hybrid models have outperformed the standalone models. Specifically, the ConvNext + Swin outperformed all with an accuracy of 95.42%, and AUC of 0.99. Overall, the findings shows that hybrid models are more preferable for better accuracy in classification tasks, and Swin based architectures are less prone to over fitting.Item Feasibility Study and Energy Management System of Mini Grid Hybrid Systems for Energy Intensive Industries: A Case Study of Industries in Mekelle(Mekelle University, 2025-05-30) Hiwot NigusieHybrid systems integrate renewable energy sources with battery storage to supply energy in offgrid or on-grid setups. Many studies on hybrid power generation focus primarily on rural electrification, the socio economic benefits for households and local communities and remote areas, often overlooking the impact on industrial development. This literature gap limits our understanding of how reliable electricity access could drive industrial growth, enhance productivity, and foster economic diversification. This study focused on the techno-economic feasibility of a mini hybrid power generation system for electrification of three energy intensive manufacturing industries that are located in Mekelle city of Tigray namely Mesfin Industrial Engineering, MOHA soft drinks industry, and Desta Alcohol & Liquor Factory. The aim was to study the feasibility of a hybrid renewable energy solution to make industries energy independent and provide sufficient power and tied them with reliable power system by avoiding their grid dependency. The software packages utilized is used to design, analyze, and optimize the hybrid power system were HOMER Pro modeling tool. The mini grid has a peak capacity of 230 kW requires 3005 kWh/day. The Generic PV system has a nominal capacity of 720 kW. The annual production is 1,321,381 kWh per year for Mesfin Industrial Engineering. The electric needs for MOHA soft drinks industry are met with 720 kW of PV, 320 kW of generator capacity, 330 kW of wind generation capacity with operating costs for energy of $388,003 per year without battery storage. An addition of 1,000 kWh of battery capacity is proposed. This will reduce the operating costs to $458,636 per year. A 50 kW of generator capacity, 1,000 kWh of battery capacity and for Desta Alcohol & Liquor Factory 50 kW of wind generation capacity, with operating costs of $154,451 per year. It is proposed that adding 110 kW of hydropower generation capacity would reduce operating costs to $154,421/yr.