Discrete-time modeling and tracking control of pulse-width modulated systems

Pulse-width modulation (PWM) is a widely used strategy for operating switch controlled power electronic circuits and systems. This thesis develops a new model for PWM systems which can be used for simulation or for control design. The new model provides the discrete-time response of the one-cycle-average value of any signal in the PWM system, for any choice of switch period or control period.; The new model, like other discrete-time large-signal models, possesses two features which make control design quite challenging: typically unstable zero dynamics, and a nonlinear dependence on the duty ratio input. This thesis develops a new approach to discrete-time nonlinear approximate tracking control. The new model is approximated by one which has stable zero dynamics, using scheduled output redefinition, and input-output linearization is applied to this approximate design model. Although the new approach leads to approximate rather than perfect output tracking, it guarantees bounded and non-oscillatory responses.; Both the modeling and control design efforts of this thesis are enhanced by numerical implementations based on piecewise-linear approximations of various non-linearities. Fast simulation is achieved by storing pre-computed values of the state equation nonlinearities in memory and evaluating them at run-time using linear interpolation. For real-time control, scheduled output redefinition is implemented using linear interpolation from pre-computed output equation nonlinearities, and iterative computation of the control input is implemented using a piecewise-linear version of Newton's method.