Training data for effect of preventive actions

  • Matthias Noebels (Creator)
  • Robin Preece (Creator)
  • Mathaios Panteli (Creator)



The provided Matlab file contains simulation data for the effect of preventive actions during extreme events in the German transmission network (489 buses, 852 lines). The data can be used for training and testing of machine learning classifiers identifying the optimum preventive action given a set of event parameters. The data contains 2000 events, and for each event and preventive action the results of 100 randomly created event outcomes. Cascading failures were simulated using AC-CFM (

The following preventive actions are considered:
1 - no preventive action
2 - isolating a vulnerable area
3 - islanding into 5 islands
4 - islanding into 10 islands

The file contains the following variables:

- network is a struct describing the network and can be used as a network case struct for Matpower. network has an additional field "coordinates", which contains the geographic coordinates of each bus.

- scenario_details is a table describing the event properties, such as location, radius, intensity and network load. It also contains the load supplied after pre-event load reduction for the preventive actions.

- faulty_lines is a 2000x100x852 matrix describing the lines damaged by each event and for each event outcome. The matrix contains a "1" if the respective line has been damaged, and "0" if not.

- load_after_cascade is a 2000x100x4 matrix describing the load supplied after the event for each event, event outcome and preventive action.
Date made available24 Jun 2021
PublisherUniversity of Manchester Figshare


  • Power networks
  • Transmission system
  • Cascading failure
  • islanding
  • Preventative action
  • Resilience assessment
  • Power and Energy Systems Engineering (excl. Renewable Power)

Cite this