Data-Driven Control of LVDC Network Converters: Active Load Stabilization

Abstract

This paper addresses the (model-free) data-driven control of power converters, acting as distributed generators, in low voltage direct current (LVDC) networks (e.g. DC microgrids, DC distributed power systems, DC buses wit multiple sources and loads, etc.). Since traditional stand-alone control design, cannot guarantee stability when converters are connected to a network, it is proposed a deterministic solution that does not require the network model-an approach purely based on measurement data. This is a suitable way to overcome common issues when using a model-based approach, e.g. the use of an excessive number of variables and equations, the presence of un-modeled dynamics, unknown parameters and/or the lack of first principle model equations. To corroborate the advantages of the proposed approach, the present work addresses an extreme but also realistic scenario: weak networks with active loads, such as constant power loads (CPLs). It is also shown that the proposed scheme guarantees stability in a rigorous deterministic way-using a Lyapunov approach based on coefficient matrices directly constructed from data. Simulation results using a multibus LVDC distribution network, based on PSCAD/EMTDC, are provided as proof of concept. Index Terms-Data-driven control, distributed generators, LVDC networks, DC microgrids, DC buses, oscillations, stability, constant power loads. Recent contributions of model-based approaches include: [11]-[13], where feedback control techniques are proposed to stabilize power converters in the presence of CPLs. In [2],

Voltage Control of DC Microgrids: Robustness for Unknown ZIP-Loads

“…The adaptive controller in [12] requires the knowledge of some system parameters for the observer design, while the adaptive negative impedance strategy in [13] does not establish stability guarantees. A datadriven controller is proposed in [14], where the errors of the output voltage and input current w.r.t. a desired steady-state are required.…”

Employing Machine Learning for Enhancing Transient Stability of Power Synchronization Control During Fault Conditions in Weak Grids

“…Related applications of machine learning for power electronic systems have been presented in [16]- [25]. For instance, long-short term memory (LSTM) technique has been applied in [17] and [18] to predict the stability of the smart grid and to specify the active power fluctuations.…”

“…Similarly, machine learning has been used in protection of power electronic based systems and determining the possible outage of grid components in [21] and [22]. Further, some data-driven control has been presented to improve frequency regulation and lowvoltage ride through (LVRT) performance of converters in [23] and [24], and the data-driven control of DC power converters has been developed in [25].…”

Active Stabilization Techniques for Cascaded Systems in DC Microgrids

“…In terms of loading, besides passive (resistive) loads, active loads are mainly connected to the DC bus through tightly regulated point-of-load converters (POLCs) [9]. Therefore, the cascaded interconnection of source-and load-side subsystems is a prominent configuration in such networks, where switching power converters are usually designed and tested individually with only resistive loads [9]- [13] and [15]. A generic model of such configuration in DC MGs is depicted in Fig.…”

“…From the point of view of the upstream circuit, the CPL characteristic behaves as a negative resistance [9]. Though the inherent incremental negative resistance feature of the CPL guarantees the fastdynamic performance of the converter of the load-side, it will jeopardize the stability of cascaded systems [9]- [12], [15]. This instability effect on cascaded systems can cause reduced system damping, limited cycle oscillation and voltage collapse on the DC bus, degraded stability margins, and in the worst case, blackout of the whole microgrid.…”