Classification of Power Quality Events Using Optimal Time-Frequency Representations—Part 2: Application

Abstract

Abstract-Better software and hardware for automatic classification of power quality (PQ) disturbances are desired for both utilities and commercial customers. Existing automatic recognition methods need improvement in terms of their capability, reliability, and accuracy. This paper presents the theoretical foundation of a new method for classifying voltage and current waveform events that are related to a variety of PQ problems. The method is composed of two sequential processes: feature extraction and classification. The proposed feature extraction tool, time-frequency ambiguity plane with kernel techniques, is new to the power engineering field. The essence of the feature exaction is to project a PQ signal onto a lowdimension time-frequency representation (TFR), which is deliberately designed for maximizing the separability between classes. The technique of designing an optimized TFR from timefrequency ambiguity plane is for the first time applied to the PQ classification problem. A distinct TFR is designed for each class. I. INTRODUCTIONHE proliferation of highly sensitive computerized equipment places increasingly more stringent demands on the quality of electric power supplied to the customer [2]. Today, power quality (PQ) has become a very interesting cross-disciplinary topic, coupling power engineering and power electronics with digital signal processing, software engineering, networking, and VLSI.Voltage disturbances are the most frequent cause of a broad This work is supported by the Advanced Power Technologies (APT) Center at the University of Washington. The APT Center is supported by ALSTOM ESCA, LG Industrial Systems, RTE France, and Mitsubishi Electric Corp. This work is also partially supported by the National Science Foundation Career Award Grant #0093716 and the American Public Power Association DEED program.Both authors are with the SEAL (Sensors, Energy, and Automation Laboratory), Department of Electrical Engineering, Box 352500, University of Washington, Seattle, WA98195. (E-mails: mwang@ee.washington.edu; mamishev@ee.washington.edu) range of disruption in industrial and commercial power supply systems. These disturbances, often referred to as power quality problems, significantly affect many industries. Major causes of PQ-related revenue losses are interrupted manufacturing processes and computer network downtime. The examples abound in semiconductor industry, chemical industry, automobile industry, paper manufacturing, and ecommerce. A report by CEIDS (Consortium for Electric Infrastructure to Support a Digital Society) shows that the U.S. economy is losing between $104 billion and $164 billion a year due to outages and another $15 billion to $24 billion due to PQ phenomena [3].The conventional methods currently used by utilities for power quality monitoring are primarily based on visual inspection of voltage and current waveforms. Highly automated monitoring software and hardware is needed in order to provide adequate coverage of the entire system, understand the causes of these disturba...

Classification of Power Quality Events Using Optimal Time-Frequency Representations—Part 1: Theory

“…The proposed algorithm has been successfully tested with 860 real world power quality events. A recognition rate of 98.0% has been achieved in a classification experiment that covers five classes; a DSP-based hardware system, which is capable of processing a five-cycle (83.3 ms) PQ waveform within 11.2 ms, has verified the realtime computing capability of this algorithm [1]. …”

“…These five types of PQ events were used to validate the proposed new classification algorithm, in the Application Part of this paper [1]. The case study tested a total of 860 realworld waveforms.…”

Classification of power quality events – A review

“…This final winner determines the class of the test data. In [115], authors proved SVM has upper-hand over optimal time-frequency representation [116,117] and low complexity event classification method [118]. In [119], authors designed a weighted SVM and trained by 5-dimension feature space points for making decision regarding the simulated without noise disturbances.…”

Classification of Induction Machine Faults using Time Frequency Representation and Particle Swarm Optimization

“…In order to avoid unnecessary computation to separate classes, we have proposed the principle of the remaining classes [11]. The discrimination between different classes is made by separating the class i from all the remaining classes {i+1,…,N}.…”

“…In fact, such a representation may conflict with the goal of classification, generating a TFR that maximizes the separability of TFRs from different classes. It may be advantageous to design TFRs that specifically highlight differences between classes [11][12][13][14].…”