Data-driven Oscillation Mode Prediction Model based on Eigensystem Realization Algorithm and Extreme Learning Machine
Jiong Ding, Jiebei Zhu, Xiaorui Cui, Lujie Yu · 2023
This paper proposes a data-driven oscillation mode prediction model based on the measured multi-channel signals and operating scenario information from power system. Firstly, eigensystem realization algorithm (ERA) is introduced to identify the system oscillation modes based on the measured multi-channel signals after disturbance. Then, the oscillation mode prediction model is established based on artificial intelligence algorithm: extreme learning machine (ELM). And the prediction model is trained by historical data with scenario information as inputs and ERA identified oscillation modes as outputs. To evaluate the accuracy of the prediction model, mean absolute percentage error (MAPE) and correlation coefficient $(R^{2})$ are adopted in this paper to access the total biases between test value and predicted value. As verified by the case studies using IEEE 39-bus benchmark test system, after trained by historical data, the proposed data-driven prediction model can accurately predict system oscillation modes under new operating scenarios.