Research

PhD defence by Mengfan Zhang

Time

16.06.2022 kl. 08.30 - 11.30

Description

Mengfan Zhang, AAU Energy, will defend the thesis "Artificial Intelligence Enabled Impedance Identification and Stability Estimation of Grid-converter System"

TITLE

Artificial Intelligence Enabled Impedance Identification and Stability Estimation of Grid-converter System

PHD DEFENDANT

Mengfan Zhang

SUPERVISOR

Professor Xiongfei Wang

CO-SUPERVISOR

Professor Mads Græsbøll Christensen

MODERATOR

TBD

OPPONENTS

Associate Professor Amin Hajizadeh, Aalborg University, Denmark (Chairman)
Frans Dijkhuizen, Hitachi Energy Research Center
Associate Professor Kai Sun. Tsinghua University

ABSTRACT

Moving towards a carbon-neutral society, traditional power systems are going through a shift with the wide integration of renewable energy resources, which require power-electronics-based systems as an alternative to synchronous generator systems. However, the interaction between the grid and the internal control systems of power electronics converters may result in the instability of the grid-converter system, which will threaten the security of grid operation and limit the further integration of renewables.
To address the grid-converter stability issue caused by grid-converter interaction in power converter dominated power systems, the small-signal stability analysis method that uses interface impedances is widely adopted. However, there are several issues that may undermine the performance of impedance-based stability estimation method in industrial applications. First, although the impedance modeling method is mature nowadays, some nonlinear blocks (e.g. dead-time, dc-link control loop, etc.) in the power-electronics converter are ignored, which decrease the accuracy of the impedance modeling methods and stability estimation results. Second, due to the nonlinearities in the power-electronics converter, the impedance model varies with the operating point. Thus, the impedance model obtained in a stiff operating point cannot be used for stability estimation considering the uncertainties of renewables. Moreover, many vendors are unwilling to share details of their products, making it difficult to get the full information required for impedance modeling. Therefore, to accurately estimate the stability of gird-converter system, it is significant to reveal the effect of the nonlinear block on the impedance model of power-electronics converters and to obtain the accurate impedance model in global operating point’s vision without access to the detailed information of the inner control systems of converters.
To address the issues mentioned above, this Ph.D. project focuses on the converter system modeling and stability estimation and develops a series of artificial intelligence-based impedance identification methods and “black-box” stability estimation approach. The thesis consists of four research topics. First, an analytical impedance model of single-phase voltage source converter (VSC) is derived, where the dead-time effect is considered in this model with the double input describing function (DIDF) method. It is revealed that the impedance model is highly dependent on the operating point.  Second, an artificial intelligence helped ‘black-box’ impedance identification method is developed for the impedance modeling that can cover multiple operating points. Third, to reduce the data amount requirement, a physical informed impedance identification method is developed. A combination of physics and artificial neural network (ANN) techniques is used to improve the accuracy of the impedance model with limited data. At last, an ANN-based stability estimation is developed to give a guideline on how to use the ANN-based model to achieve stability estimation.
The outcome of this thesis is expected to give a guideline of AI-based stability estimation of electrical power systems with large-scale integration of renewables, which will add significant value to the industrial applications and sustainable society.

 

 

THE DEFENCE will be IN ENGLISH - all are welcome.

 

 

Events

See the list