Investigation of power quality disturbances by using 2D discrete orthonormal S-transform, machine learning and multi-objective evolutionary algorithms

Karasu, Seçkin; Saraç, Zehra

doi:10.1016/j.swevo.2018.11.002

Cited by 35 publications

(17 citation statements)

References 34 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Labani et al [57] considered the relevance of the text features to the target class and the correlation between the features as two objectives in text FS, and proposed a multi-objective algorithm, namely MORDC. Karasu and Saraç [58] used NSGA-II to find the optimal solutions for two different fitness functions, i.e., NF and classification accuracy. In [59], NSGA-II are used to obtain a set of Pareto-optimal solutions in different pattern recognition domains, the number of used features and the classification error are set as two objectives, see [60].…”

Section: Multi-objective Gas For Fsmentioning

confidence: 99%

A self-adaptive multi-objective feature selection approach for classification problems

Xue

Zhu

Neri

2021

ICA

View full text Add to dashboard Cite

In classification tasks, feature selection (FS) can reduce the data dimensionality and may also improve classification accuracy, both of which are commonly treated as the two objectives in FS problems. Many meta-heuristic algorithms have been applied to solve the FS problems and they perform satisfactorily when the problem is relatively simple. However, once the dimensionality of the datasets grows, their performance drops dramatically. This paper proposes a self-adaptive multi-objective genetic algorithm (SaMOGA) for FS, which is designed to maintain a high performance even when the dimensionality of the datasets grows. The main concept of SaMOGA lies in the dynamic selection of five different crossover operators in different evolution process by applying a self-adaptive mechanism. Meanwhile, a search stagnation detection mechanism is also proposed to prevent premature convergence. In the experiments, we compare SaMOGA with five multi-objective FS algorithms on sixteen datasets. According to the experimental results, SaMOGA yields a set of well converged and well distributed solutions on most data sets, indicating that SaMOGA can guarantee classification performance while removing many features, and the advantage over its counterparts is more obvious when the dimensionality of datasets grows.

show abstract

Section: Multi-objective Gas For Fsmentioning

confidence: 99%

A self-adaptive multi-objective feature selection approach for classification problems

Xue

Zhu

Neri

2021

ICA

View full text Add to dashboard Cite

show abstract

“…Similar to the WT, it can offer an improved time-frequency interpretation of an input signal. [19][20][21] The stationary modulating sinusoids in regards to the time-axis along with the scalable and variable Gaussian window are some of the exceptional properties of ST. The mathematical representation of ST can be expressed in Mishra et al 22 HHT Energy, entropy, skewness, minimum, and maximum of amplitude curve from the first IMF, standard deviation, skewness, and energy of phase curve.…”

Section: S-transformmentioning

confidence: 99%

“…The ST can be defined as an extension of either “phase correction” of WT or movable window STFT. Similar to the WT, it can offer an improved time‐frequency interpretation of an input signal 19‐21 . The stationary modulating sinusoids in regards to the time‐axis along with the scalable and variable Gaussian window are some of the exceptional properties of ST.…”

Section: Techniques For Vscs Detection and Classificationmentioning

confidence: 99%

Advanced signal processing and machine learning techniques for voltage sag causes detection in an electric power system

Mishra

Panigrahi

2019

Int Trans Electr Energ Syst

View full text Add to dashboard Cite

Summary In this manuscript, voltage sag causes (VSCs) detection in an electric power system is studied using S‐transform (ST) and variational mode decomposition (VMD) techniques. The advantages of these approaches are compared in detail against earlier implemented wavelet transform (WT), empirical mode decomposition (EMD), and Hilbert‐transform (HT) techniques, as a novel contribution to previous studies. Voltage sag can trigger periods of downtime, considerable damage of product and moreover, it can attribute to malfunctions, instabilities, and decrease the lifespan of the connected loads. Therefore, accurate detection of VSCs can avoid the loss and problems instigated by voltage sag in an electric power system. In this work, initially, the extracted disturbance voltage signals from the relaying point are passed through the aforementioned signal processing algorithms to extract unique features. Then, these extracted features are used as inputs to the extreme learning machine (ELM) classifier for recognition of VSCs. Additionally, to augment the precision of the output result, the performance of ELM classifier is compared with the artificial neural network (ANN), K‐nearest neighbors (KNN), and support vector machine (SVM) classifier. Three types of VSCs are used in this analysis such as (a) sag due to incipient fault, (b) sag due to induction motor stating, and (c) sag due to transformer energization. The simulation results show good performance and feasibility of the stated methods for VSCs recognition even under in cooperation of noisy condition.

show abstract

“…One‐dimensional signal processing and two‐dimensional signal processing can be used for signal processing in pattern recognition. Although one‐dimensional (1D) signal processing methods were widely used to analyze the PQDs, but two‐dimensional (2D) signal processing methods due to generating more feature groups than 1D signal processing and distinctive features can be better 3 . In recent studies regarding PQDs recognition, only 1D methods are used 4‐12 Simultaneous study of the current and voltage signals using 2D signal processing methods can help to better identify some PQDs, which is discussed in this article by using two‐dimensional discrete wavelet transform (2D‐DWT).…”

Section: Introductionmentioning

confidence: 99%

A novel flicker detection method for vertical axis wind turbine using two‐dimensional discrete wavelet transform

Doustmohammadi

Foroud

2020

Int Trans Electr Energ Syst

View full text Add to dashboard Cite

Background Pattern recognition applications have increased dramatically over the past four decades. In power system, pattern recognition can be used in fields such as power quality. One of the important power quality aspects is flicker, which is caused by torque fluctuations in wind turbines. Aims In this paper, the various mechanical faults including single and double faults that cause flicker in vertical axis wind turbines (VAWTs) are modeled; then flicker caused by single and double faults are analyzed. Then, a novel method for pattern recognition is presented which results in more accuracy in flicker detection. Materials and Methods In this paper, features of voltage and current signals are extracted by the 2D‐DWT. The 2D‐DWT is capable of producing valuable features than the discrete wavelet transform (DWT) at a lower level of decomposition. The 2D‐DWT is even capable of compressing data that can be suitable for smart grids, which high volume data is one of their problems. In order to reduce the redundant data, dimension of features vector, and computational burden, a new method is presented for feature selection. The backbone of the presented feature selector is the used classifier in the corresponding pattern system; so it is compatible with the classifier. Results Results show the high accuracy of the proposed method in identifying the factors that cause flicker in VAWTs. Conclusion In this paper, it was shown that the features extracted from the current and voltage signal by the 2D‐DWT were superior to the features extracted from the only voltage signal by the DWT. Also, the advantage of the proposed feature selection over the Relieff method was shown.

show abstract

Investigation of power quality disturbances by using 2D discrete orthonormal S-transform, machine learning and multi-objective evolutionary algorithms

Cited by 35 publications

References 34 publications

A self-adaptive multi-objective feature selection approach for classification problems

A self-adaptive multi-objective feature selection approach for classification problems

Advanced signal processing and machine learning techniques for voltage sag causes detection in an electric power system

A novel flicker detection method for vertical axis wind turbine using two‐dimensional discrete wavelet transform

Contact Info

Product

Resources

About