Multisynchrosqueezing transform aided transfer learning based approach for diagnosis of single and mixed power quality events

This paper presents a technique for recognizing the single stage and multiple PQ (Power Quality) events using an algorithm based on ST (Stockwell's-Transform) and ANN (Artificial Neural Network) based classifier and a rule based decision tree. The ST which combines elements of WT (Wavelet Transform) and STFT (Short-Time Fourier Transform) is used for the analysis of various single stage and multiple power quality events. Single stage PQ events such as sag, swell, interruption, harmonics, transients, notch, spike, flicker and multiple power quality events which include the harmonic disturbances with sag, swell, flicker and interruption are analyzed using the proposed algorithm. A data base of these events is generated in MATLAB as per IEEE-1159 standard. Significant features of various PQ events are extracted using the S-transform and are used as an input to this hybrid classifier. The results are presented for the effective recognition of the PQ events with the proposed algorithm.

This paper presents a new method for detection and classification of various Power Quality (PQ) events in power distribution system by using Cross Wavelet Transform (CWT)-Support Vector Machine (SVM). The proposed approach collects the numerous power quality events data through the simulation of MATLAB for training and testing. First the data is used for training the SVM with mathematical model and control parameters of various classifiers. In second stage it is validated with trained data. As per IEEE-519-1992, 14 classes of single and multiple power quality events are also tested with untrained data in simplified manner. The proposed intelligence recognition system gives the results of confusion matrix and accuracy for all events. The results confirm that the CWT-SVM technique efficiently detect and classify the various power quality events.

In this paper a technique for detection of multiple power quality (PQ) events is illustrated. An algorithm based on wavelet transform and Random Forest based classifier is proposed in this paper. The developed technique is implemented on 11 different power quality events consisting of single stage power quality events such as sag, swell, flicker, interruption and multi stage power quality events such as harmonics combined with sag, swell, flicker, interruption. PQ events are simulated in MATLAB using standard IEEE-1159 standard. Significant features of PQ events are extracted using wavelet transform and used to train random forest based classifier. The efficiency of Random Forest Based classifier is compared with other widely used machine learning algorithms such as K-Nearest Neighbour (KNN) and Support Vector Machine (SVM). From confusion matrix of different algorithms it is concluded that Random Forest shows superior classification accuracy as compared to SVM and KNN.

Grid edge power quality (PQ) issues are receiving growing attention due to the increasing complexity, decentralization, and digital connectivity of modern energy systems. Ensuring stable and high-quality power at the grid edge is essential to prevent equipment damage, enhance energy efficiency, and maintain the reliable operation of bidirectional systems such as microgrids, electric vehicles (EVs), and smart homes—all of which require fast and accurate detection and classification of PQ disturbances. This work proposes an unsupervised classification method for detecting mixed PQ events at the grid edge, leveraging feature-engineered discrete wavelet transform (DWT) applied to real-field PQ meter data. A five-level DWT decomposition is used to extract rich statistical features from real-time voltage and current waveforms, which are subsequently clustered using a domain knowledge-driven K-Means approach. The classification results demonstrate that the proposed method effectively distinguishes between normal and abnormal conditions, as well as differentiates multiple simultaneous PQ disturbances—improving event characterization without the need for synthesized labeled datasets.

This work presents a framework to accomplish the power quality disturbances classification by employing a novel hybrid methodology that combines the Gramian Angular Difference Field (GADF) approach with a deep neural network. The Gramian Angular Difference Field (GADF) method is applied to convert Power quality signals to two dimensional image and then, convolutional neural network (CNN) is applied to learn high level features for classification purpose. The mixed and single synthetic power quality (PQ) disturbances are taken into account for classification purpose. A unit is constructed with two dimensional convolutional, pooling, and batch-normalization layers to extract the meaningful features of the power quality disturbances. The GADF-based convolutional neural network (GADF-CNN) for classifying PQ disturbance signals is validated subsequently. The results shows that the proposed classification model has high efficiency and reliability.

Diagnosing utility grid disturbances in photovoltaic integrated DC microgrid using adaptive multiscale morphology with DFA analysis

The increasing penetration of non-linear and power-electronic-based loads in industrial distribution systems has led to a growing prevalence of power quality (PQ) disturbances such as voltage sags, harmonics, transients, and mixed events, which adversely affect equipment reliability and operational efficiency. Conventional PQ assessment techniques based on time-domain indices and Fourier analysis are limited in their ability to accurately characterize non-stationary and transient disturbances commonly observed in industrial environments. This study presents an advanced PQ assessment framework that integrates wavelet-based signal processing with machine learning (ML) classification to enable automated, high-resolution disturbance analysis. Multi-level wavelet decomposition is employed to extract discriminative time–frequency features, including energy distribution, statistical measures, and entropy, which effectively capture the intrinsic characteristics of diverse PQ events. These features are subsequently used to train and evaluate supervised ML classifiers, including support vector machines, random forest models, artificial neural networks, and convolutional neural networks. The proposed framework is validated using representative industrial distribution system data under varying operating conditions, including noisy and mixed PQ scenarios. Comparative results demonstrate that the wavelet–ML approach significantly outperforms traditional RMS-, FFT-, and STFT-based methods in terms of classification accuracy and robustness. The findings highlight the suitability of the proposed framework for real-time industrial PQ monitoring, predictive maintenance, and intelligent decision support, contributing to enhanced reliability and resilience of modern industrial power systems.

Based on intrinsic mode functions (IMFs), standard energy difference of each IMF obtained by EEMD and probabilistic neural network (PNN), a new method is proposed to the recognition of power quality transient disturbances. In this method, ensemble empirical mode decomposition (EEMD) is used to decompose the non-stationary power quality disturbances into a number of IMFs. Then the standard energy differences of each IMF are used as feature vectors. At last, power quality disturbances are identified and classified with PNN. The experimental results show that the proposed method can effectively realize feature extraction and classification of single and mixed power quality disturbances.

This paper presents a variational mode decomposition (VMD) and decision tree based detection and classification method of single and mixed power quality (PQ) disturbances in grid-connected distributed generation system. Applicability of VMD technique is investigated for discrimination of stationary PQ disturbances (such as harmonics, interharmonics, and flicker), non-stationary events (e.g., transients) and noise. Studies indicate usefulness of VMD for accurate estimation of phasor quantities such as amplitude, phase angle, and frequency and other describing features. Features namely, mode central frequencies, relative mode energy ratios, zero crossings, and instantaneous amplitude (IA) are extracted for classification of single and mixed PQ disturbances using a decision tree algorithm. A set of synthetic test signals, disturbance signals obtained from real events as well as signals generated from real time digital simulator platform are used to test effectiveness of proposed method, under various system operating scenarios and noise levels. Proposed method is found to be capable of accurate detection, estimation, localization, and classification of all kinds of PQ disturbances in both noisy and noise-free cases.

This work develops a new approach to recognize multiple disturbances for a power quality (PQ) event in power systems. It is usual that several different types of disturbances simultaneously exist in a PQ event. However, most of the existing methods treat a PQ event as a single type of PQ disturbance. The performance of these methods might be limited and impracticable for application in the real power systems. This work proposes a novel approach integrated the wavelet transform and dynamic structural neural network (DSNN) to identify disturbance waveforms. The proposed neural network has the capability of adapting to multiple disturbances for a PQ event. In the proposed approach, the disturbance waveforms are extracted by the wavelet transform and then fed to the DSNN for identifying the types of disturbances. The distinctive features of the proposed method are that it can estimate the amplitude of the considering event, recognize transient and steady state disturbances which are simultaneous existed in a PQ event

The paper deals with the application of Volterra bound Interval type −2 fuzzy logic techniques in power quality assessment. This work proposes a new layout for detection, localization and classification of various types of power quality events. The proposed method exploits Volterra series for the extraction of relevant features, which are used to recognize different PQ events by Interval type-2 fuzzy logic based classifier. Numerous single as well as multiple powers signal disturbances have been simulated to testify the efficiency of the proposed technique. This time–frequency analysis results in the clear visual detection, localization, and classification of the different power quality events. The simulation results signify that the proposed scheme has a higher recognition rate while classifying single and multiple power quality events unlike other methods. Finally, the proposed method is compared with SVM, feed forward neural network and type −1 Fuzzy logic system based classifier to show the efficacy of the proposed technique in classifying the Power quality events.

—Since decade, many time-frequency analysis methods with combination of classifiers have been studied in literature for recognition of power quality (PQ) events. In these studies, feature extraction and selection have a vital role to enhance PQ classification accuracy and to reduce computational complexity for PQ recognition. This paper presents a new method to detect and classify PQ disturbances based on modified Stockwell Transform (ST) for extraction of features and Hybrid Grey Wolf Optimization (HGWO) for feature selection along with K Nearest Neighbor (KNN) classifier. Simulation of the proposed method using MATLAB is carried out through a wide range of eighteen synthetic PQ events to validate the effectiveness of the selected features. Further, an experiment is extended for six classes of real PQ events acquired from self excited induction generator (SEIG) system in a laboratory experimental setup. Proposed method is also employed on those real time data to study the classification accuracy performance. In these experiments, an impressive accuracy of 99.94% and 99.3% for synthetic and real time PQ event data, respectively are reported. Hence, it is observed from result analysis, this proposed method can be utilized for recognition of PQ events in real time power system.

Power quality (PQ) events are referred to any abnormal deviation from the standard sinusoidal behavior of power signals within a power system. PQ events are usually studied by tracking the behavior of voltage signals over observation points of the system. IEEE Standards have defined standard categories for PQ events based on their time behavior. Each class of these events may have different level of importance from different contributors' perspective (utilities, system operators, or costumers). Due to increasing the usage of sensitive technological loads such as transportation, banking systems, and databases on one hand in addition to the uncertainty injected to the system from aggregation of renewables on the other hand, the fast and reliable PQ events classification is an important monitoring task in the future smart grid. In this paper, combining the theory of sparse recovery with a new high-dimensional convex hull approximation framework we have developed a fast, reliable, and adaptive PQ events classification methodology named "compressive-informative sparse representation-based" PQ events classifier. Unlike usual classification approaches, the proposed classifier does not need any training procedure while due to its linear mathematical formulation acts inherently fast. Moreover, it can be easily adapted to recognize the challenging combined PQ events in addition to any permanent change in the behavior of PQ patterns.

The proposed work aims at the accurate detection and classification of various single and multiple power quality (PQ) disturbances. To this end, a modified optimal fast discrete Stockwell transform (ST) with random forests (RF) classifier-based PQ detection framework has been proposed in this paper. In modified ST, a single signal-dependent window is introduced, with optimally selected window parameters via energy concentration maximization based constraint optimization. As a result of which accurate time-frequency localization of various PQ events is achieved, with sharper energy concentration. In classification stage, the proposed PQ framework utilizes the RF-based classifier, which follows the bagging approach by random selection of features and data points, at each node, to train the classifier. Decision stumps are used as weak classifiers, and using a simple majority voting of these decision stumps, RF builds a strong classifier. The RF gives less variance and less bias estimation due to injection of randomness into the training phase, and its performance is found to be reasonably immune to input parameter selection. As a result of this, the classification results of the proposed PQ framework are found to be very accurate and quite insensitive to the presence of noise in the data. Various test cases are presented in this paper to clearly demonstrate the superiority of the proposed scheme. The proposed approach has also been tested on real field data and very promising results have been obtained.

A novel method of classifying Power quality (PQ) events using Wavelet Packet Transform (WPT) and Extreme Learning Machines (ELM) has been proposed. In recent times, the power quality has been a major research concern due to changing regulations, liberalized distribution market and increased use of power electronic based equipment. The first step of any remedial action requires proper identification of PQ events. One of the major challenge of this event identification is to extract significant features from the limited measurements, which can subsequently be used for the classification. Therefore, in the present study Wavelet Packet Transform (WPT) has been used to obtain several mathematical features. These features can segregate both single and simultaneous PQ event occurrences. Further to improve the classification performance, the ELM based classifier has been used. This classifier significantly reduces the training time by many-fold. The performance of the proposed approach has been compared with ANN based classifier considering over 1000 PQ signals from various PQ events. The results of the simulation demonstrate that the proposed approach can achieve over 99% classification accuracy.