Power quality events are usually associated with more than one disturbance and their recognition is typically based on multi-label learning. In this study, we propose a new method for recognizing multiple disturbances (MPQDs) based on variational mode decomposition (VMD) and a random discriminative projection extreme learning machine for multi-label learning (RDPEML). First, VMD is employed to decompose the MPQDs into several intrinsic mode functions and the standard energy differences of each mode are extracted as features that form the input vectors of the classifier. Second, a novel multi-label classifier called RDPEML is constructed by combining a random discriminative projection multiclass extreme learning machine (ELM) and a thresholding learning method-based kernel ELM. In order to obtain better classification performance, a tenfold cross-validation embedded particle swarm optimization approach is utilized to search for the optimal values of the structural parameters. Finally, a test study was conducted using Matlab synthetic signals and real signals sampled from a three-phase standard source under different noise conditions. Compared with several recent state-of-the-art multi-label learning algorithms, RDPEML achieved better classification performance with superior computational speed.

In this paper, a new classification approach for detection and classification of complex power quality disturbances (PQDs) using a three-level multiply connected Bayesian network is proposed. First, the model consisting of features evidence layer, disturbances state layer, and circumstance evidence layer is established, which represent the features extracted from the sample signal, the state of each single label of PQDs and the circumstance factors that may affect the PQDs, respectively. Second, the parameters of three-level multiply connected Bayesian network (TLBN) are studied from statistical data and Monte Carlo simulations. Finally, the classification is determined by computing the posterior marginal probabilities of each event given observed evidences. The new method not only utilizes the existing features extracting methods, but also takes the historical data, and other surrounding factors into account. Simulation results and real-life PQ signal tests show that the performances of TLBN classification of complex disturbances are better than the other approaches in existing literatures.

Measurement accuracy is an important performance indicator for high-voltage direct current (HVDC) voltage dividers. The temperature rise effect for a HVDC voltage divider’s internal resistance has an adverse effect on measurement accuracy. In this paper, by building a solid model of a DC voltage divider, the internal temperature rise characteristic and error caused by the temperature rise in a resistance voltage divider were theoretically simulated. We found that with the increase in height and working time, the internal temperature of the voltage divider increased. The results also showed that the lowest temperature was near the lower flange and the highest temperature was near the upper flange in the middle of the voltage divider. The error caused by the temperature rise increased first and then decreased gradually with divider height, increasing with its working time. The measurement error caused by the internal temperature difference in steady state reached a maximum of 158.4 ppm. This study provides a theoretical basis to determine the structure and accuracy improvement for a resistive voltage divider, which is helpful for the selection of components and the optimization of the heat dissipation structure.

With the increasingly penetration of nonlinear loads in the power system, power quality (PQ) has become a significant issue for the power utilities and end users. In order to improve the PQ, the PQ detection is essential. In this paper, a new method for detecting the PQ disturbances via empirical wavelet transform (EWT) and Hilbert (HT) is proposed. Firstly, EWT is applied to the signal for obtaining different modes. Then the instantaneous amplitude and frequency of each mode are calculated by using the HT. By applying it to two stationary signals and two non-stationary signals, the efficiency of the proposed method is evaluated. With no frequency aliasing like the S transform (ST), the proposed method presents more accurate results than the ST.

This paper proposes a highly accurate and fast power quality disturbances (PQDs) classification using dictionary learning sparse decomposition (DLSD). Firstly, an over-complete dictionary is constructed by combining an identity matrix with a learning dictionary trained by K-SVD algorithm. Secondly, the features and the fuzzy primary classifications of PQDs are obtained by calculating the sparse decomposition coefficients based on the learning dictionary. For being adaptive to sparsity and reducing computational complexity, a fast adaptive matching pursuit (FAMP) using sparsity adaptive algorithm and regularized atom selection is proposed. Then, a decision tree is adopted to accomplish accurate classification by using the estimated features and the pre-classification results. Finally, the proposed approach is tested by PQDs from simulations, IEEE PES database and actual measurements. Moreover, several testing signals, which contain strong noise and frequency deviation, are introduced to further validate DLSD. The results demonstrate that DLSD has a good improvement on computational complexity and classification accuracy when dealing with PQDs classification.