2019/03/15 by Sanjeev Raja, Raja, Sanjeev, Ernest Fokoué +1
Engineering · #62F25 #62F40 #62H30 #62J12 #Energy Load and Power Forecasting #FOS: Computer and information sciences #G.3 #I.2.6 #I.5.3 #Machine Learning (cs.LG) #Machine Learning (stat.ML) #Power System Reliability and Maintenance #Smart Grid and Power Systems
paper · pdf · doi:10.48550/arxiv.1903.06700
openalex publication_date 2019/03/15 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
In the monitoring of a complex electric grid, it is of paramount importance\nto provide operators with early warnings of anomalies detected on the network,\nalong with a precise classification and diagnosis of the specific fault type.\nIn this paper, we propose a novel multi-stage early warning system prototype\nfor electric grid fault detection, classification, subgroup discovery, and\nvisualization. In the first stage, a computationally efficient anomaly\ndetection method based on quartiles detects the presence of a fault in real\ntime. In the second stage, the fault is classified into one of nine pre-defined\ndisaster scenarios. The time series data are first mapped to highly\ndiscriminative features by applying dimensionality reduction based on temporal\nautocorrelation. The features are then mapped through one of three\nclassification techniques: support vector machine, random forest, and\nartificial neural network. Finally in the third stage, intra-class clustering\nbased on dynamic time warping is used to characterize the fault with further\ngranularity. Results on the Bonneville Power Administration electric grid data\nshow that i) the proposed anomaly detector is both fast and accurate; ii)\ndimensionality reduction leads to dramatic improvement in classification\naccuracy and speed; iii) the random forest method offers the most accurate,\nconsistent, and robust fault classification; and iv) time series within a given\nclass naturally separate into five distinct clusters which correspond closely\nto the geographical distribution of electric grid buses.\n