Real-Time Detection and Classification of Power Quality Disturbances.

Area of Science:

• Electrical Engineering
• Power Systems Analysis
• Signal Processing

Background:

• Power quality disturbances impact the reliability and efficiency of power delivery systems.
• Real-time detection and classification are crucial for mitigating these disturbances.
• Existing methods may lack the speed or accuracy required for dynamic power grids.

Purpose of the Study:

• To develop a novel method for real-time detection and classification of power quality disturbances.
• To improve the speed and accuracy of disturbance identification in power systems.
• To leverage multivariate data analysis for enhanced detection capabilities.

Main Methods:

• A sequential and multivariate disturbance detection method is proposed.
• The detector employs a non-parametric, supervised approach learning from clean and disturbed signals.
• The method is extended to a multi-hypothesis setting for classification, using specific training data for each disturbance type.

Main Results:

• The multivariate approach enables cooperative analysis of data from multiple meters, accelerating detection.
• The multi-hypothesis extension allows for rapid and accurate classification of various disturbance events.
• The proposed method demonstrates effective real-time detection and classification of power disturbances.

Conclusions:

• The developed sequential, multivariate method offers a significant advancement in real-time power quality disturbance analysis.
• Cooperative analysis of multi-meter data enhances detection efficiency.
• The multi-hypothesis classification framework provides accurate and timely identification of power system disturbances.