Neural network controlled optimal pulse-width modulation for a voltage source converter

This study is aimed at the application of feed-forward artificial neural networks to the on-line calculation of the optimal firing angles of a voltage source converter. Optimal pulse width modulation (OPWM) is a method to determine the switching instants of the power electronic switches in a voltage source converter (VSC) such that the current harmonies at its ac terminals are minimized while the number of switching actions per cycle remains constant. To achieve this, the converter switches must operate (or fire) at certain angles, which are dependant on the magnitude of the fundamental frequency component of the output voltage. Practical implementation of this technique is impeded partially due to the difficulties of the on-line calculation of the firing angles. Artificial neural networks (ANN) with their ability to provide nonlinear input-output relationships can serve as an appropriate tool to determine the firing angles in real time as the required magnitude of the fundamental varies. In this study a feed forward ANN is employed for this purpose. To verify the basic ideas, in addition to the extensive number of simulations, a hardware prototype was built. This prototype includes the inverter, the firing circuits and the controller and was successfully used as an induction machine drive.; This thesis is organized in six chapters. After a short introduction in Chapter 1, the basic ideas related to the ac to dc and dc to ac converters are presented in Chapter 2, with a clear focus on the harmonic reduction methods. Chapter 3 briefly introduces the artificial neural networks. The focus here will be on the particular methods used in the following chapters. Chapter 4 contains the theoretical background for the optimal PWM voltage source converter, the proposed method for controlling the firing angles and details of the hardware implementation of a prototype converter. Chapter 5 explains the application of the same technique to an ac to dc converter and a STATCOM. The conclusions and suggestions for further study are presented in Chapter 6.