Data-driven inertia estimation based on frequency gradient for power systems with high penetration of renewable energy sources

摘要

The time-dependent inertia can trigger rapid frequency variations, which have become main concerns in power system stability. Prior understanding of the system inertia will help operators to mitigate the stability issues by applying appropriate measures and executing suitable control schemes. Therefore, this paper proposes a datadriven method to estimate the time-dependent inertia in the system based on the frequency gradient of an estimated model of the system. In this approach, phasor measurement units (PMUs) are used for measuring the system data, which is then used to estimate a dynamic model representing the system. Next, a system identification approach is applied to estimate the power system model, from which the inertia will be estimated. The benefits of the proposed method include reduction of the model order to avoid computation burden, extraction of the inertia from the reduced order model using the gradient mapping on RoCoF of the system response and estimation of the inertia constant of the system using normal operating conditions. The use of normal operating conditions to estimate the inertia makes the method unique. The method has been verified using the simulated IEEE 39-bus system in DIgSILENTTM PowerFactory? tool and real data from the New Zealand power system.