Data-Driven Risk Assessment Early-Warning Model for Power System Transmission Congestions

With the continuous development of renewable energy in modern power systems, the balance between the power supply and demand side become more volatile, which may cause potential power system transmission congestions. Traditional risk assessment in the power system static security analysis area always uses the power flow model-based method, which cannot address all the possible operation scenarios. Therefore, a novel machine learning (ML) based data-driven risk assessment model for early-warning of power system transmission congestion is proposed in this paper. The proposed model can make full use of the power system historical operation data as well as the measurement of the current time step, which can be used in real-time for early-warning of the power system transmission congestion in advance. A feature selection method called Max-Relevance and Min-Redundancy (mRMR), is adopted to reduce the calculation burden of the ML model. Numerical tests are performed on a regional power grid of France. The proposed data-driven risk assessment model can accurately predict the risk conditions under normal operation, single and multiple component outage scenarios, over 93.3%. The result validates that our model can be used for real-time early-warning of power system transmission congestions.