Machine Learning Assisted Inertia Estimation Using Ambient Measurements

With the increasing penetration of converter-based renewable resources, different types dynamics have been introduced to power system. Due complexity and high order modern system, mathematical model-based inertia estimation method becomes more difficult. This paper proposes two novel machine learning assisted methods based on long-recurrent convolutional neural (LRCN) network graph (GCN) respectively. Informative features are extracted from ambient measurements collected through phasor measurement units (PMU). Spatial structure with dimensional graphical information then incorporated improve accuracy estimation. Case studies conducted IEEE 24-bus The proposed LRCN GCN models achieve an 97.34% 98.15% Furthermore, zero generation injection bus optimal PMU placement (ZGIB-OPP) has proved be able maximize system observability, which subsequently improves performance all models.