Utility interface design and mutually coupled switched reluctance machine study for a wind generation system

Wind power generation is gaining more attention and assuming more importance. The recent evolution of power semiconductors and variable frequency drives technology has aided the acceptance of variable speed wind generation systems. In spite of the additional cost of power electronics, variable speed generation provides such advantages as larger energy capture and more powerful control. This thesis studies the utility interface design for an 18kW variable speed wind generation system, the mutually coupled switched reluctance machine (MCSRM), which is used as the generator in this system, and the application of the artificial neural network (ANN) in position estimation and torque smoothing for the MCSRM.; The utility interface adopted is a voltage-source PWM controlled single-phase full-bridge converter. Its functions are to regulate the DC bus voltage within the wind energy conversion system at the reference value and present the utility with unity power factor operation with low harmonic distortion. Modelling and controller design issues are discussed. A detailed design of this utility interface is provided. The converter is built and a DSP is used to implement the control functions. Both simulation and experimental results are provided.; The MCSRM is a new type of the switched reluctance machine (SRM). This thesis studies its electromagnetic characteristics in detail using the finite element method, with its relationship with the conventional SRM and its peculiarities highlighted. Control principles for different excitation patterns and operation modes are derived. The modelling difficulties associated with the MCSRM are emphasized and the finite element-circuit coupled simulation method is used to study the interactions between the MCSRM and the power electronics and to check the control principles derived from the nonlinear analysis. Both simulation and experimental results are provided.; This thesis studies position estimation and torque smoothing for the MCSRM using the ANN. Special aspects associated with the use of the ANN in the case of the MCSRM are discussed. The principles of using the ANN in these applications are detailed. Simulation results are provided to verify the feasibility of these methods.