Navigating Energy Market Dynamics: A preliminary framework for anomaly detection in balancing utilities

Balancing groups in power systems are responsible for maintaining the equilibrium between electricity generation and consumption and thereby ensuring power grid stability against unprecedented fluctuations. Integration of renewables consequently introduces uncertainties in the system, therefore, these balancing services play a pivotal role to maintain a functioning stable power network. Additionally, to foster prosumer behavior for increased renewable share and fair market competition, grid management requires these services to exhibit fair trading practices and the daily schedules that the services are supposed to provide to management be as least anomalous as possible. Our proposed framework aims to serve as an initial step to detecting anomalies and unexpected variations by considering schedules transmitted to Transmission system operators and billing data generated that is used to make payments to the provided balancing services. Feature engineering is used to extract features indicating possible deviations from multiple perspectives and interquartile range-based thresholds are used to calculate tolerance boundaries for a base model that functions both as a verification tool and a method for cross-referencing results obtained through the application of One class Support vector machines (SVM). The approach proved effective in identifying potential anomalies and is intended to serve as an initial point of inquiry into energy market discrepancies.