Electrical and Computer Engineering
Permanent URI for this collectionhttps://repository.mu.edu.et/handle/123456789/426
Browse
2 results
Search Results
Item Design of an Adaptive Neuro-Fuzzy Inference System Controller for Temperature and Concentration Control in a MIMO Continuous Stirred Tank Reactor (CSTR)(Mekelle University, 2025-04-07) Ybrah ZemchealThe Continuous Stirred Tank Reactor (CSTR) is a critical unit in chemical processing industries, where precise control of process variables is essential for optimal product quality and efficiency. Among the key variables, temperature and concentration are particularly important to regulate. However, chemical processes often exhibit nonlinear and multivariable behavior, making conventional controllers less effective in real-time operation (PID control in CSTR exhibits sluggish or oscillatory responses for feed concentration, slow response to variable water flow, Poor robustness for uncertain parameters (e.g., reaction rates, heat transfer coefficients), perform poorly due to cross-coupling effects, less nonlinearity handling due to reaction kinetics. Challenges such as dynamic process variations, reactant nonlinearities, fluctuating environmental conditions, and diverse disturbances further complicate control. Additionally, most industrial chemical processes are multi-input multi-output (MIMO) systems, necessitating advanced control strategies and decoupling techniques. In this study, an Adaptive Neuro-Fuzzy Inference System (ANFIS) is proposed as an advanced control approach to enhance system performance and accuracy compared to conventional controllers. ANFIS integrates the structured knowledge representation of fuzzy logic with the adaptive learning capabilities of neural networks, offering improved control for complex systems. The performance of ANFIS is evaluated against a traditional PID controller through offline simulations using MATLAB/Simulink. The results demonstrate that ANFIS outperforms PID control in key performance metrics, including overshoot, rise time, settling time, and system stability. Furthermore, ANFIS exhibits superior disturbance rejection capabilities, making it a more robust solution for CSTR control in industrial applications.Item Load Frequency Control of Small Hydro Power Plant Based on Artificial Neural Network with Tuned PID Controller(Mekelle University, 2025-04-10) Rigbey HailetsionElectricity propels the advancement of society and economy, fueling sectors such as healthcare, education, and industries. Real-time adjustments in power generation are made by load frequency control to stabilize frequency and voltage, a critical factor for uninterrupted power supply. An innovative approach suggests the integration of artificial neural networks and PID controllers for enhancing the performance of small hydropower plants. The core objective of the research is to create an ANN combined with tuned PID for regulating load frequency in small hydro power plants. The thesis elaborates on the limitations of conventional PID controllers and the flexibility of the ANN-based strategy, outlining the process of plant modeling and controller configuration. Both Proportional-Integral-Derivative (PID) controllers and ANN with Tuned PID controllers are commonly employed techniques. While PID controllers offer stability, the ANN with Tuned PID controllers exhibit superior adaptability and quicker responses to dynamic variations, thereby enhancing the efficiency of the SHP. Using the ANN-tuned PID controller results in significant improvements in several areas. The settling time is notably enhanced, decreasing by 74.36% compared to the untuned PID controller and 50.88% compared to the tuned PID controller. The overshoot is greatly reduced, showing a decrease of 96.29% compared to the untuned controller and 90.26% compared to the tuned controller, indicating much better stability. Additionally, the peak time increases slightly by 2.78% compared to the untuned controller and 2.14% compared to the tuned controller, demonstrating minimal delay in reaching the maximum value. These changes highlight faster, more accurate, and stable system responses with advanced tuning techniques.