Model Dependent Heuristic Dynamic Programming Approach in Virtual Inertia-Based Grid-Connected Inverters

Abstract

A model dependent heuristic dynamic programming (MDHDP) is presented in this paper to regulate the voltage and frequency of a virtual inertia-based grid-connected inverter. To overcome the inertialess and fast-responding characteristics of the conventional direct power inverter, the virtual synchronous generator (VSG) concept has introduced. The negative points of the VSG approach including: (i) weak performance in inductive grids, (ii) disability to perform under uncertainties, and (iii) its dependency to the operating point, motivates us to implement a neurocontrol reinforcement learning technique. The simulation results illustrate that a well-designed and well-trained MDHDP can perform optimally in tracking the active power while facing changes in system parameters.

Meeting Name

52nd North American Power Symposium, NAPS 2021 (2021: Apr. 11-13, Tempe, AZ)

Department(s)

Electrical and Computer Engineering

Comments

This material is based upon work supported by the U.S. Department of Energy, "Enabling Extreme Fast Charging with Energy Storage", DE-EE0008449.

Keywords and Phrases

Grid Connected Inverter; Machine Learning; Model Dependent HDP; Reinforcement Learning; Virtual Synchronous Generator

International Standard Book Number (ISBN)

978-172818192-9

Document Type

Article - Conference proceedings

Document Version

Citation

File Type

text

Language(s)

English

Rights

© 2021 Institute of Electrical and Electronics Engineers (IEEE), All rights reserved.

Publication Date

13 Apr 2021

Share

 
COinS