Dynamic analysis of fuel cell/microturbine generation scheme with neural network control for peak power shaver applications
MetadataShow full item record
This dissertation describes the development, design, modeling and small-scale experimental evaluation of two distributed generation schemes employing a microturbine/fuel cell. The first of these schemes used a microturbine connected in parallel to a fuel cell through the dc-link of a rectifier-inverter conversion unit, without a battery bank. The second scheme added a battery bank to the dc-link. These generation schemes have been evaluated for supplying both isolated and grid connected loads. The operation of these schemes as a peak power shaver is simulated and results are presented in this work. The full dynamic behavior of the system is investigated in this research effort. The research contribution lies in the treatment of the microturbine/fuel cell schemes as unified systems with full modeling, simulation and controller design techniques. The simulation results are validated through a laboratory model of the system. The laboratory prototype of the generation schemes is built using microturbine and fuel cell emulators to represent the dynamic and steady state characteristics of the actual microturbine and fuel cell. With the computer-based data acquisition and controller, it has been established that different system parameters can be easily changed and their effects on system performance monitored and analyzed. The complex and nonlinearity nature of the proposed system makes it difficult to control by using conventional controllers. On the other hand, neural networks offer several key advantages over the conventional control techniques including its ability to learn from examples and their ability to approximate any complex system by using a set of input/ output data sets. In this dissertation, a multi-layer perceptron neural network is trained to represent the microturbine/ fuel cell inverse model and is employed in the direct inverse control scheme to control the flow of real and reactive powers from the system.