Proportional resonant control of three-phase grid-connected inverter during abnormal grid conditions

Abstract

The development of using grid-connected three-phase inverter has augmented the standing of realizing muted distortion along with high-quality current waveform. The standard three-phase grid-connected inverter is the full-bridge voltage source inverter. This inverter is usually controlled by proportional integral (PI) controller in order to ensure sinusoidal current injection to the grid. Although the PI controller is well established and easy to use under normal grid conditions, it leads to system instability under abnormal grid conditions. When abnormal grid conditions are likely to occur, the control system with PI controller can be configured to include two separate PI controllers for the positive and negative sequence components of the grid current. However, this increases control complexity and total harmonic distortion (THD). More recently, the proportional resonant (PR) controller started to replace PI controller in a different application including grid-connected current control. In this thesis, a comprehensive theoretical and experimental comparison between the PI and PR controllers is presented. The comparison shows that the PR controller offers lower total harmonic distortion (THD) in the current signal spectrum and is simpler to implement as it uses only the positive sequence component of the grid current and consequently only one PR controller is needed. For these reasons, the PR controller is adopted in this thesis. Despite the PR controller offering enhanced functioning under abnormal grid conditions compared to PI controller, a sudden change in the grid voltage could additionally raise the error between the reference signal and the controlled signal which results in causing significant divergence from its ostensible value. In this case, the performance of the conventional PR controller will not keep up with the increase in the error which weakens controller performance. To overcome this problem, a new design concept for controlling the current of the three-phase grid connected inverter during normal and abnormal conditions is presented in this thesis. The proposed technique replaces the static control parameters by adaptive control parameters based on a look-up table. This adaptive PR, controller has been investigated and demonstrated with different normal and abnormal grid conditions. The proposed control technique is capable of providing low THD in the injected current even during the occurrence of abnormal grid conditions compared with PI and PR controllers. It also achieves lower overshoot and settling time as well as smaller steady-state error. Proportional Resonance Control of Three-Phase Grid-Connected Inverter II Additionally, despite the fact that both PI and PR controllers are relatively straightforward to tune, and are sometimes capable of dealing with many time-varying grid conditions. This research also presented an adaptive controller tuned using advanced optimization techniques based on particle swarm optimisation (PSO). PSO is presented to optimize the control parameters of both PI and PR controllers for the three-phase grid-connected inverter. There are many advantages of using PSO, such as no additional hardware being required. Thus, it can be extended to other applications and control methods. In addition, the proposed method is a self-tuning method and can thus be suitable for industrial applications where manual tuning is not recommended for time and cost reasons. Simulation and experimental test were carried out to investigate the performance of the proposed techniques. In the simulation, the system was tested under 100 kW model using Matlab/Simulink environment. In addition, the system was also investigated through a practical implementation of the control system using a Digital Signal Processor (DSP) and grid-connected three-phase inverter. This practical system was demonstrated a 300 W scaled-down prototype. As a result, the comparisons between experimental and simulation results show the behaviour and performance of the control to be accurately evaluated.