Intelligent Control of a Fuel Cell Power Plant as a Distributed Generation Source

Open Access
Choi, Tae-Il
Graduate Program:
Electrical Engineering
Doctor of Philosophy
Document Type:
Date of Defense:
November 26, 2007
Committee Members:
  • Kwang Yun Lee, Committee Chair
  • Heath F Hofmann, Committee Member
  • Jeffrey Scott Mayer, Committee Member
  • Robert M Edwards, Committee Member
  • Setpoint Reference Governor
  • Distributed Generation
  • Fuel Cell
  • Intelligent Control
  • Neural Network Modeling
  • Utility Grid
Environmental problems caused by fossil fuel energy have become a focus of attention. Deregulation has accelerated the development of various types of alternative technologies as inventors see the opportunity to compete in market niches that did not exist a few short years ago. Specific market needs include producing reliable and voltage sag-free power in a manner that is less disruptive to the environment as well as the ability to supply power where it is too expensive to string power lines. Fuel cells will start out as a high-cost technology, supplying electricity (and heat) to these niches and gradually become more attractive to mainstream electricity users as they improve in capability and decrease in cost. The Molten Carbonate Fuel Cell (MCFC) stack dynamic model was developed to analyze a spectrum of dynamic responses. A simplified process flow diagram of the fuel cell power plant is presented. The massive parallelism, natural fault tolerance, and implicit programming of neural network (NN) computing architectures suggest that they may be good candidates for implementing real-time controllers for complicated, nonlinear dynamic systems. To reduce the complexity of the NN model, a fuel cell power plant can be divided into three subsystems. With fewer weights for each NN model, the random influence on the fuel cell model can be lowered. A new concept of an intelligent setpoint reference governor (I-SRG) using a heuristic algorithm will be developed to find the optimal setpoints based on system constraints and performance objectives. The NN Identification system realizes an adaptive NN identifier for the control of the MCFC. A fuel cell power plant can be interfaced with the utility system via a three-phase inverter, controlling real and reactive power. Some of operating conflicts and the effects of distributed generation on power quality are also addressed and possible solutions are suggested. Simulation results of the matlab/simulink model of a fuel cell power plant with an inverter controller proved load tracking capability following the real and reactive power change of the load.