IRTSD: Open-Source Data and Toolset for Electromagnetic Transient Analysis of Disturbances and IBR Control Malfunctions in Transmission Systems

Following a series of unusual transmission system disturbances involving unexpected IBR behaviors, there has been a significant need for monitoring, control, and protection functions that can help system operators more rapidly understand and respond to these disturbances. The IBR-rich transmission system datakit (IRTSD) is focused on transient disturbances in IBR-rich transmission systems and consists of a dataset, power system model, and automation scripting. The data consists of 5500 events and roughly 1.4M labelled signal recordings of power system disturbances produced using an electromagnetic transient model of a transmission system with roughly 40% IBR penetration. While the transmission system is relatively small topologically, detailed models are used for IBR controls, traditional generators, and phasor measurement units, resulting in a system that provides high-fidelity representation of a wide variety of power system phenomena from 0.5 Hz to 4 kHz. Events simulated include both the commonplace (e.g., stepping of voltage controls) and the emergent issues (e.g., forced oscillations or unexpected momentary cessation at an IBR plant).

The dataset contains three resources:

1)      The dataset itself, which is a collection of labelled tabular data representing currents, voltages, powers, and frequencies for 5500 power system events.

2)      A PSCAD model of a transmission system with a high penetration of inverter-based resources (IBR).

3)      Python code that was used to procedurally generate the 5500 events from the PSCAD model.

The tabular data is synthetic—it is of simulation results obtained from the PSCAD model. The PSCAD model is designed to be relatively small topologically but has very detailed representations of generator and IBR controls. The model also contains detailed phasor-measurement unit (PMU) models which can also act as protective relays. Thus, the value of this model lies in its ability to provide high-fidelity representations of protection and control related power system phenomenon.

The dataset can be used and analyzed without any use of (2) or (3), but the purpose of including (2) and (3) is so that users can easily modify and extend the dataset to be more tailored for their specific needs. Resource (2) is also standalone, and users who only wish to study a small number of events (e.g., a dozen or two), may be better served by simply manually running simulations and making no use of (1) or (3).

The Python code was developed on Python 3.12 and the only dependency, other than the PSCAD API (which comes with PSCAD), is Pandas.