Methods and systems for classifying optically detected power quality disturbances
Inventors
Sternberg, Oren • Rockway, John D. • Allen, Jeffery C.
Assignees
Publication Number
US-10436855-B2
Publication Date
2019-10-08
Expiration Date
2037-08-10
Interested in licensing this patent?
MTEC can help explore whether this patent might be available for licensing for your application.
Abstract
An optically detected power quality disturbance caused by a remote load is classified as belonging to a class of known classes of power quality disturbances. Features associated with different power quality disturbances that belong to a plurality of different known classes of power quality disturbances are learned. Cross-covariance is applied to the optically detected power quality disturbance and the different power quality disturbances that belong to the different known classes of power quality disturbances to recognize features of the optically detected power quality disturbance that at least partially match the learned features. The class of power quality disturbances among the plurality of classes of different known power quality disturbances to which the optically detected power quality disturbance belongs is determined based on the recognized features.
Core Innovation
The invention provides a method and system for classifying optically detected power quality disturbances caused by a remote load as belonging to a known class of power quality disturbances. It learns features associated with different power quality disturbances across multiple known classes, applies cross-covariance to the optically detected disturbance and known disturbances to recognize matching features, and determines the class based on these recognized features.
The invention addresses the problem that traditional power quality disturbance detection relies on invasive electronic techniques requiring insertion of devices into power infrastructures. Although passive optical sensing techniques have been developed to extract vibration signatures, no existing systems remotely detect or classify power quality disturbances based on optical data.
Claims Coverage
The patent includes three independent claims covering a method and a device for classifying optically detected power quality disturbances, focusing on feature learning, cross-covariance application, and classification.
Learning features associated with power quality disturbances
Learning features associated with different power quality disturbances that belong to multiple known classes as a basis for classification.
Applying cross-covariance for feature recognition
Applying cross-covariance to an optically detected power quality disturbance and the known disturbances to recognize features that at least partially match the learned features, enabling remote detection from a load causing the disturbance.
Determining class based on recognized features
Determining the class of power quality disturbances among known classes to which the optically detected disturbance belongs based on recognized features from cross-covariance analysis.
Using intensity and frequency responses
Incorporating intensity responses and frequency responses in learned features and mapping responses of unknown disturbances to known classes for classification.
Handling multiple simultaneous disturbances
Recognizing features of multiple simultaneous optically detected disturbances and determining their respective classes based on learned features.
Device implementation for classification
A device comprising a processor and memory with instructions to perform the learning, cross-covariance application, and classification steps associated with optically detected power quality disturbances.
Optical analysis methodology
Analyzing video data of a light source to isolate indicative pixels and extract intensity and frequency responses for classification.
The claims collectively cover a comprehensive approach to remotely detect, analyze, and classify power quality disturbances optically by learning disturbance features, applying cross-covariance for recognition, and determining disturbance class using intensity and frequency response analysis.
Stated Advantages
Enables non-invasive, remote optical detection and classification of power quality disturbances without modifying existing power infrastructure.
Allows classification of disturbances caused by particular loads or modes of operation, facilitating load utilization tracking.
Supports machine learning integration for feature learning and improved classification accuracy.
Documented Applications
Tracking utilization of various loads through classification of power quality disturbances associated with those loads.
Application in Internet of Things (IoT) environments for monitoring power disturbances remotely via optical sensing.
Interested in licensing this patent?