Search

In this paper, a long short-term memory (LSTM)-based method with a multi-input tensor approach is used for the classification of events that affect the power quality (PQ) in power systems with distributed generation sources. The considered events are line faults (one line, two lines, and three lines faulted), islanding events, sudden load variations, and generation tripping. The proposed LSTM-based method was trained and tested using the signals produced by the events simulated in a study system with distributed generation sources via PSCAD®. Then, noise with different levels was added to the testing set for a thorough assessment, and the results were compared with other well-known methods such as convolutional and simple recurrent neuronal networks. The LSTM-based method with multi-input proved to be effective for event classification, achieving remarkable classification performance even in noisy conditions. [ABSTRACT FROM AUTHOR]

Identification of coherent generators and the determination of the stability system condition in large interconnected power system is one of the key steps to carry out different control system strategies to avoid a partial or complete blackout of a power system. However, the oscillatory trends, the larger amount data available and the non‐linear dynamic behaviour of the frequency measurements often mislead the appropriate knowledge of the actual coherent groups, making wide‐area coherency monitoring a challenging task. This paper presents a novel online unsupervised data mining technique to identify coherent groups, to detect the power system disturbance event and determine status stability condition of the system. The innovative part of the proposed approach resides on combining traditional plain algorithms such as singular value decomposition (SVD) and K‐means for clustering together with new concept based on clustering slopes. The proposed combination provides an added value to other applications relying on similar algorithms available in the literature. To validate the effectiveness of the proposed method, two case studies are presented, where data is extracted from the large and comprehensive initial dynamic model of ENTSO‐E and the results compared to other alternative methods available in the literature. [ABSTRACT FROM AUTHOR]

In this paper, a deep learning approach for power quality monitoring in systems with distributed generation sources is presented. The proposed method focuses in the multi-scale analysis of multi-component signals for power quality disturbances classification. The proposed methodology combines a signal processing stage using variational mode decomposition (VMD) to obtain the times scales of multi-component signals, and a deep learning stage using a simple feedforward neural network (FFNN) to classify the disturbances. The simple proposed architecture allows minimum training time of the classification model. In addition, the proposed method is able to classify different disturbance combinations based on a reduced training-set. The proposed VMD-FFNN method is tested using synthetic and simulated signals, and it is compared with other well-known methods based on convolutional and recurrent deep neuronal networks. Finally, the proposed method is assessed using lab measurements in order to shown its performance in a real-world environment. • A multi-component signal analysis with a minimum training class number. • A fast-training process for the managing of different system configurations. • The detection of the disturbance's onset time, through an observation window analysis. [ABSTRACT FROM AUTHOR]