The project aims to develop a real-time sub-synchronous oscillation (SSO) monitoring system to detect and analyse early signs of SSO events, enabling preventive actions. The project will span two years and include three main stages:
· Stage 1: Analyse historical data to build an event database.
· Stage 2: Develop a software prototype of early-warning system to detect and track SSO early signs.
· Stage 3: Test and validate the tool with simulated and real-world data.
Expected outcomes include a prototype software tool for detecting SSO risks and aiding in mitigation. Challenges involve identifying early signs accurately and managing data-sharing complexities among partners, which could affect project progress.
Benefits
The key benefit of the project is to enable NESO and TOs with a real time SSO monitoring and mitigating system, purely based on existing data sources and infrastructures, thus providing a realistic and powerful solution to address the SSO challenges during the energy transition process. More specifically, the benefits include:
· Early identification of SSO precursory signs and risks well before they occur, providing sufficient time to respond.
· Enhancing the understanding of the impact of operational actions (e.g. variation in windfarm outputs, line outages) on the system behaviour and stability.
· Enabling preventive actions with real time feedback on the effect of implemented actions on the changes in SSO risks.
· Reduced cost of system operation by reduced expensive defensive measures that are unnecessary.
Learnings
Outcomes
As the project is still in its early delivery stage, final project outcomes have not yet been delivered. The expected project outcomes remain aligned with the original PEA, including a processed historical event dataset, extracted feature set, early-warning prototype functions, and subsequent testing and validation evidence.
Interim outcomes from this reporting period are mainly linked to Task 1. The project has started to establish an interpretable workflow for historical PMU data processing and event characterisation using Strathclyde historical PMU data. Initial analysis has been used to test the workflow and to identify candidate feature outputs, including signal-level, spectral-level, rolling window, multi-band, and event-level descriptors.
These interim outputs provide a basis for the next stage of work, which will scale the workflow to wider Strathclyde historical PMU records, develop feature tables and reusable processing scripts, and prepare detector-ready inputs for later Task 2 development. The project has not yet delivered a final detector, prototype demonstration, or validation results. No change in TRL is claimed at this stage.
Lessons Learnt
As the project is still at an early stage, the lessons learnt are preliminary. However, the initial Task 1 work and project discussions have provided several useful observations for the next stages of this project and for similar future projects.
First, the early work confirms that a structured historical-data workflow is needed before developing detector or prototype functions. The current Task 1 approach uses Strathclyde historical PMU data to establish a reproducible workflow for signal processing, event characterisation, and feature extraction. This approach helps ensure that features are interpretable and linked to observable PMU behaviour rather than being treated only as model inputs.
Second, the early analysis highlights the importance of using multiple complementary views. Time-domain inspection, detrending, PSD analysis, time-frequency analysis, segment comparison, and multi-band descriptors each provide different information. No single indicator should be treated as sufficient on its own for early-warning purposes.
Third, the technical discussions highlighted the importance of PMU measurement metadata, including reporting rate, measurement window, PMU configuration, filtering, signal definitions, and channel availability. These factors affect the frequency range and signal behaviour that can be reliably interpreted from available PMU data.
Finally, further work is required to test whether candidate features are robust across wider Strathclyde historical PMU records and, later, across partner PMU datasets once data-sharing arrangements allow. This will be a key focus of the next stage of Task 1.
