Ensemble learning-based data-driven analytics for power system stability

• Main Author: Shu, Zhengge
• Other Authors: Xu Yan
• Format: Thesis-Master by Coursework
• Language: English
• Published: Nanyang Technological University 2023
• Subjects: Engineering::Electrical and electronic engineering::Electric power
• Online Access: https://hdl.handle.net/10356/170081
• Institution: Nanyang Technological University

With the continuous development of society, electricity has become a necessity for human production and livelihood. In order to meet the increasing demand for electricity, modern power systems are developing towards ultra-high voltage, large capacity, and long-distance transmission. In addition, with the integration of a high proportion of renewable energy, the power grid structure has become increasingly complex. Studying the stability of the power system, extracting data on the stability situation of the power system through data-driven methods for analysis and stability analysis, is of great significance for improving the voltage stability of the power system and preventing large-scale power outages in the power system. The main contributions of this dissertation are as follows: 1.Pre-processing of transient stability data in power systems and dimensionality reduction based on principal component analysis. Principal component analysis is used to reduce the dimensionality of the datasets, as well as selecting the principal components containing the majority of information as inputs to the power system stability analysis model, in order to reduce information redundancy and reduce the difficulty of prediction. 2.Construction of a data-driven model for transient stability assessment of power systems based on DT, SVM, and ANN. On the basis of simulated power system model data, different data-driven methods are used for training to test the accuracy of the model. In response to the issue of imbalanced transient stability data in power systems, a Bagging ensemble power system stability assessment model based on under-sampling algorithm and voting method is constructed. Multiple datasets are generated through under-sampling method, and the parameters of the prediction model were selected. At the same time, compared with the single learner, the evaluation indicators showed that the Bagging ensemble power system transient stability evaluation model is able to solve the problem of imbalanced data in power systems. Compared with the single learner, Bagging ensemble models have higher sensitivity and accuracy in addressing power system instability.