Abstract
Dynamic security assessment of a large power system operating over a wide range of conditions requires an intensive computation for evaluating the system's transient stability against a large number of contingencies. In this study, we investigate the application of multilabel learning for improving training and prediction time, along with the prediction accuracy, of neural networks for online transient stability assessment of power systems. We introduce a new multilabel learning method, which uses a contingency clustering step to learn similar contingencies together in the same multilabel multilayer perceptron. Experimental results on two different power systems demonstrate improved accuracy, as well as significant reduction in both training and testing time.
Highlights
Dynamic security assessment (DSA) of large interconnected power systems is a challenging task, often requiring an intensive computation power for both off-line and online studies
Conclusion and future work For the online transient stability assessment (TSA) using pattern recognition methods, both prediction success and the time needed to train and test the predictors can become limiting factors, especially when the task gets complicated by a large number of contingencies
We developed a method that uses multilabel artificial neural networks (MLMLP) for predicting concurrently the system security against multiple contingencies and demonstrated its success in the TSA of two test systems
Summary
Dynamic security assessment (DSA) of large interconnected power systems is a challenging task, often requiring an intensive computation power for both off-line and online studies. We focus on the application of a multilayer perceptron (MLP) model, a specific subclass of ANNs, which enables fast online prediction of the security status of a large power system This approach can effectively be used to predict the security of the system either operating at a single OP for the assessment of the current status, or a large number of possible OPs for an assessment to be used in developing preventive control actions. The novel contributions of this paper are as follows: (a) the use of multilabel MLPs for the TSA problem, (b) comparison of both training and testing time and accuracy performance of MLMLP to an MLP for each label, (c) instead of a single MLMLP for all labels or a single MLP for each label, using multiple MLMLPs on adaptively determined clusters of correlated labels.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
