学术文献库

Keeping the power system stable is becoming more challenging with the growing share of renewable energy sources of low or negligible inertia. Inertia constants for individual power plants are generally not known and are roughly estimated by considering the type of power generation technology and the nameplate capacity of plants. More accurate knowledge of inertia constants of individual power plants would give greater transparency to decisions of the transmission system operator (TSO) and to auctions for the procurement of inertia services. Additionally, a more accurate forecast or estimation of system inertia would improve the price signals to the power market well in advance of balancing actions taken by the TSO. We develop methods based on a combination of mathematical optimisation and machine learning that reverse-engineer the inertia constants of individual power plants from the historical values of their power production and from aggregate values of inertia in a power system. We demonstrate the methods for the power system of Great Britain (GB), where historical values for aggregate inertia are published by the TSO. We show that the recovered inertia data is crucial in understanding certain individual balancing decisions by the TSO. We use the reverse-engineered inertia constants to predict system inertia which gives valuable information to the power market.