A Data-Physics Fusion-Based Approach to Power System Equivalent Inertia Assessment

In order to solve the error problem of the traditional inertia assessment method, a power system equivalent inertia assessment method incorporating machine learning technology is proposed. First, a physical model of single-machine equivalence is used, combined with a convolutional neural network (CNN) with self-adaptive and learning capabilities as a machine learning model. Second, the strongly correlated feature quantities are selected as inputs by analyzing the model structure and calculating the correlation between the inertia time constant and other feature quantities. In addition, the computational results of traditional evaluation methods are incorporated into the data model inputs, and the model parameters are optimized using the particle swarm optimization (PSO) algorithm. The simulation results prove that the proposed method has certain advantages in improving the accuracy of inertia estimation compared with the traditional method.