i-nergy

Energy Commodities Networks: AI for electricity networks optimised operation

Over the last years, the amount of data streams and data sources in the energy systems have been increasing.

This has shown the need for improved data analytics tools, in particular to derive predictive models for assets maintenance. These tools are expected to facilitate the extension of assets lifetime and/or the reduction of O&M assets costs, and hence enable in a later stage the integration of asset management with grid improved network operation.

Furthermore, the massive penetration of distributed renewable energy resources and an expected more proactive behaviour of consumers by taking part in demand response initiatives are, among others, key developments that significantly increase the complexity in which the power systems will be operated and planned.

System Operators need to develop methodologies to assess both current and expected system behaviours. In that regard, load prediction is of paramount importance for the overall assessment of the network.

I-NERGY is addressing these two challenges related to the electricity network in the Portuguese pilot managed by R&D Nester. It includes two use-cases, which are further described below, namely:

  • AI for enhanced network assets predictive maintenance, integrating off-grid data with condition-based monitoring
  • AI for network loads and demand forecasting towards efficient operational planning.

Furthermore, I-NERGY has two Open Calls to include services from the enlarged external community in the I-NERGY platform. In particular, R&D Nester held an Info Corner about the I-NERGY 2nd Open Call at the AI & Big Data Expo 2022.

AI for enhanced network assets predictive maintenance, integrating off-grid data with condition-based monitoring

The network asset considered in the project is the Circuit Breaker. The results obtained so far include:

  • Machine Learning based architecture to develop data-driven COMTRADE files analysis
  • Fully automated data processing, analysis and reporting system, enabling the development of condition-based maintenance strategies to support O&M optimisation.

Automated Circuit Breaker’s fault detection and identification

img

The next steps include the definition of the survival function and the design of the corresponding data-driven condition-based maintenance policy.

AI for network loads and demand forecasting towards efficient operational planning

This use case is considering the forecasting of the National load for the day-ahead operational planning. The results obtained so far include:

  • Time-series processing and feature engineering
  • Development of stacking methodologies to aggregate the members Pattern Sequence Forecasting with k-Means, Agglomerative clustering and SOMs, together with new XGBoost members, resulting in a MAPE of 2.7%.

AI Forecasting for National load

imgimg

 

The next steps include the adaptation of the developed forecasting methods to the TSO-DSO interfaces.