A Data-Driven Approach for Online Inter-Area Oscillatory Stability Assessment of Power Systems Based on Random Bits Forest Considering Feature Redundancy

To utilize the rapidly refreshed operating data of power systems fully and effectively, an integrated scheme for inter-area oscillatory stability assessment (OSA) is proposed in this paper using a compositive feature selection unit and random bits forest (RBF) algorithm. This scheme consists of offline, update, and online stages, and it can provide fast and accurate estimation of the oscillatory stability margin (OSM) by using the real-time system operating data. In this scheme, a compositive feature selection unit is specially designed to realize efficient feature selection, which can significantly reduce the data dimensionality, effectively alleviate feature redundancy, and provide accurate correlation information to system operators. Then, the feature set consisting of the selected pivotal features is used for the RBF training to build the mapping relationships between the OSM and the system operating variables. Moreover, to enhance the robustness of the scheme in the face of variable operating conditions, an update stage is developed. The effectiveness of the integrated scheme is verified on the IEEE 39-bus system and a larger 1648-bus system. Tests of estimation accuracy, data processing speed, and the impact of missing data and noise data on this scheme are implemented. Comparisons with other methods reveal the superiority of the integrated scheme. In addition, the robustness of the scheme to variations in system topology, distribution among generators and loads, and peak and minimum load is studied.