Detecting anomalies in power consumption of electrical systems

摘要

A method for detecting power consumption anomalies. The method obtains two different time power series and compares them to generate a dissimilarity matrix. A dimensionality reduction equation is applied to the dissimilarity matrix to obtain a low dimensional embedding of the power data. A probability score of each of the time series being anomalous is then calculated.

主权项

1. A method of detecting power consumption anomalies in an electrical system comprising:
determining a power time series from monitoring at least one power meter electrically connected to the electrical system; determining a first power time period and creating a first power time curve for power data from the power time series; determining a second power time period and creating a second power time curve for power data from the power time series, wherein the first and second power time periods are different; comparing power data in the first power time curve for the first power time period with power data in the second power time curve for the second power time period to create a clustering of power data in the compared power time periods; generating a dissimilarity matrix from the clustering of power data in the compared power time periods; applying a dimensionality reduction algorithm to obtain a low dimensional embedding of the dissimilarity matrix; and using the low dimensional embedding of the dissimilarity matrix to perform a density estimation on the clustering of power data in the compared power time periods in order to compute a probability score of power consumption in each of the first power time period and the second power time period being anomalous.