K-nearest neighbor and naïve Bayes based diagnostic analytic of harmonic source identification

This paper proposes a comparison of machine learning (ML) algorithm known as the k-nearest neighbor (KNN) and naïve Bayes (NB) in identifying and diagnosing the harmonic sources in the power system. A single-point measurement is applied in this proposed method, and using the S-transform the measurement signals are analyzed and extracted into voltage and current parameters. The voltage and current features that estimated from time-frequency representation (TFR) of S-transform analysis are used as the input for MLs. Four significant cases of harmonic source location are considered, whereas harmonic voltage (HV) and harmonic current (HC) source type-load are used in the diagnosing process. To identify the best ML, the performance measurement of the proposed method including the accuracy, precision, specificity, sensitivity, and F-measure are calculated. The sufficiency of the proposed methodology is tested and verified on IEEE 4-bust test feeder and each ML algorithm is executed for 10 times due to prevent any overfitting result.

A Method for Responsibility Division of Multi-Harmonic Sources Based on Canonical Correlation Analysis

The variable background harmonic data and incomplete phasor information make multi-harmonic source responsibility division in three-phase symmetrical power system a significant challenge. In this paper, a background harmonic data selection method based on canonical correlation analysis is proposed to deal with multi-harmonic source responsibility division without phasor information. Firstly, the canonical correlation coefficient between harmonic voltage and harmonic current is used to characterize the fluctuations of background harmonic voltage. Then, the sliding window method is adopted to select the harmonic voltage and harmonic current with small fluctuations. Next, the canonical correlation results for selected data are used to calculate the harmonic responsibility index via the linear regression method. The harmonic responsibility index in the form of percentage represents the harmonic responsibility division. Finally, several experimental results demonstrate that the proposed method has a high accuracy in calculating the harmonic responsibility division, particularly when the user side contains fluctuations of unknown harmonic sources.

Linear discriminate analysis and k-nearest neighbor based diagnostic analytic of harmonic source identification

The diagnostic analytic of harmonic source is crucial research due to identify and diagnose the harmonic source in the power system. This paper presents a comparison of machine learning (ML) algorithm known as linear discriminate analysis (LDA) and k-nearest neighbor (KNN) in identifying and diagnosing the harmonic sources. Voltage and current features that estimated from time-frequency representation (TFR) of S-transform analysis are used as the input for ML. Several unique cases of harmonic source location are considered, whereas harmonic voltage (HV) and harmonic current (HC) source type-load are used in the diagnosing process. To identify the best ML, each ML algorithm is executed 10 times due to prevent any overfitting result and the performance criteria are measured consist of the accuracy, precision, geometric mean, specificity, sensitivity, and F measure are calculated.