Simulating the Frequency Converter Inverter’s Self-Learning Control System for Suppressing Higher Harmonic Components

Abstract

Simulation of a three-phase frequency converter based on three single-phase full-bridge voltage inverters with a sine-wave output filter is considered. A new self-learning control system for the converter’s inverters is proposed. The previously developed inverter control system uses the discrete Fourier transform (DFT) to suppress the output voltage higher harmonic components. Individual regulators are used to control each harmonic component (including the fundamental and higher-order ones). The DFT algorithm produces signals proportional to the amplitudes of the fundamental and a number of higher-order harmonic components, which are applied to the regulators as feedbacks. The main drawback of the existing system is that it poses rather high demand to the microcontroller computation capacities. The newly proposed system is based on the self-learning principle and uses a periodic integrator (a P-integrator) as the output voltage regulator. The P-integrator is represented in discrete form by an array of discrete integrators, each of which operates only in the 1/64 part of the fundamental harmonic component cycle. The new control system's software code is much smaller in size and contains a fewer number of “slow” arithmetic operations, due to which it is easier to implement and less demanding with respect to computation resources. The control system analysis results demonstrate its good performance in terms of suppressing higher harmonic components: the output voltage total harmonic distortion factor does not exceed 5% for the cases of operation on linear and nonlinear loads. The article describes the modernized control system structure and its implementation in MATLAB Simulink software package. Advantages of the self-learning algorithm as compared with the algorithm based on the discrete Fourier transform are demonstrated. The simulation results confirm the applicability of the new control method. The algorithm is implemented in the C language in the MATLAB S-function block and is suitable for being implemented in real-time control systems based on dedicated microcontrollers without the need of significantly modifying the software code.