Abstract

In this work, two different solvers are used for Model-based Predictive Control (MPC) of an Inverter with LC Output Filter. The aim is to regulate the output voltage of a Distributed Energy Resource (DER) LC filter to achieve its behavior as a controlled oscillator while complying with the converter's physical limitations. The MPC makes it possible to impose constraints on the state and the input, but at a relatively high computational cost compared with other control methods. Nevertheless, the restrictions applied to the converter modulated voltage and the converter current allow for improving the control performance while achieving a safe response to contingencies. The two alternative solvers, namely Quadprog and Operator splitting solver for quadratic programs (OSQP), are evaluated through simulations to identify efficient alternatives for future real-time embedded implementations. Additionally, a state observer that estimates the DER's output current and input disturbances is included in the control to achieve reference tracking. The electrical simulation software used is PLECS integrated with Matlab/Simulink for implementation of the control. This simulation test bench is used to tune the control parameters, including the horizon, and estimate the potential gain in numerical efficiency achievable using OSQP in embedded control of the application of interest.