In recent years, the study of three-phase inverter controls has become important with the rising use of renewable energy sources (RES) in the form of distribution generation (DG). Many control types have been developed for DG inverters and others were traditional controls for the generation of the main grid power that were adapted for a system with less inertia. Among these controls is the model predictive control (MPC) which allows for a fast transient response and good reference tracking. One disadvantage of the MPC is that it does this prediction and optimization online which can limit the applications due to computational loading. Although there are some solutions to this problem in the form of a finite control-set MPC (FCS-MPC) which takes advantage of the only two states of a switch mode converter to reduce complexity, this still takes the form of a nonlinear online optimization problem. However, compared with the continuous control set (CCS) MPC, using the FCSMPC may result in poor performance due to the degradation of the switching frequency. The high computation of CCS-MPC prevents it from being implemented in the resources limited digital signal processor (DSP). To reduce the computational burden, machine learning (ML) methods such as artificial neural networks (ANN) are used for learning the input and output of the MPC. This thesis compares the ANN-MPC, and support vector machine (SVM) based MPC in a three-phase inverter. A comparison of total harmonic distortion (THD), and reference tracking during different scenarios will be provided.
Date of Degree
Master of Science (MS)
Program of Study
Electrical and Computer Engineering and Technology
Science, Engineering and Technology
De La Cruz, Arturo. (2023). Machine-Learning Based Model Predictive Control for a Three-phase Inverter. [Master’s thesis, Minnesota State University, Mankato]. Cornerstone: A Collection of Scholarly and Creative Works for Minnesota State University, Mankato. https://cornerstone.lib.mnsu.edu/etds/1390/
Creative Commons License
This work is licensed under a Creative Commons Attribution-Noncommercial-No Derivative Works 3.0 License.