No abstract

Detecting, measuring, and classifying partial discharges (PDs) are important tasks for assessing the condition of insulation systems used in different electrical equipment. Owing to the implementation of the phase-resolved PD (PRPD) as a sequence input, an existing method that processes sequential data, e.g., the recurrent neural network, using a long short-term memory (LSTM) has been applied for fault classification. However, the model performance is not further improved because of the lack of supporting parallel computation and the inability to recognize the relevance of all inputs. To overcome these two drawbacks, we propose a novel deep-learning model in this study based on a self-attention mechanism to classify the PD patterns in a gas-insulated switchgear (GIS). The proposed model uses a self-attention block that offers the advantages of simultaneous computation and selective focusing on parts of the PRPD signals and a classification block to finally classify faults in the GIS. Moreover, the combination of LSTM and self-attention is considered for comparison purposes. The experimental results show that the proposed method achieves performance superiority compared with the previous neural networks, whereas the model complexity is significantly reduced.

In recent years, deep learning has been successfully used in order to classify partial discharges (PDs) for assessing the condition of insulation systems in different electrical equipment. However, fault diagnosis using deep learning is still challenging, as it requires a large amount of training data, which is difficult and expensive to obtain in the real world. This paper proposes a novel one-shot learning method for fault diagnosis using a small dataset of phase-resolved PDs (PRPDs) in a gas-insulated switchgear (GIS). The proposed method is based on a Siamese network framework, which employs a distance metric function for predicting sample pairs from the same PRPD class or different PRPD classes. Experimental results over the small PRPD dataset that was obtained from an ultra-high-frequency sensor in the GIS show that the proposed method achieves outstanding performance for PRPD fault diagnosis as compared with the previous methods.

Nowadays, modern technologies in power systems have been attracting more attention, and households can supply a portion of or all of their electricity based on on-site generation at their location. This can be challenging for utilities in terms of monitoring and recording the data because the households' facilities can generate or consume the energy without passing it through a meter, increasing the complexity of a distribution network. The speed of transferring data to utilities is another important concern. There is a necessity to send the smart meter (SM) data of each house to a distribution management system (DMS) for more analysis in the shortest possible time. This paper presents a novel deep learning framework collaborating with sequence-to-sequence (seq2seq), long short-term memory (LSTM), and stacked autoencoders (SAEs) to forecast residential load profiles considering the photovoltaic (PV), battery energy storage system (BESS), and electric vehicle (EV) loads with more capability based on pre-defined patterns. Experimental results show that the proposed method achieves outstanding performance in the forecasting process of residential load profiles in comparison with other algorithms. Also, a smart distribution transformer can help utilities to receive the data instantly via wireless communication, which can reduce the transfer duration to every minute and make the prediction and monitoring more manageable considering the different combinations of distributed energy resources (DERs) in residential locations.

Deep neural networks (DNNs) are widely used for fault classification using partial discharges (PDs) to evaluate various electrical apparatuses and achieve high classification accuracy pertaining to trained PD faults. However, there is a risk of false alarm in the case of untrained PD faults because it is difficult for DNNs to predict data that were not included in the training process. In this paper, we research classification problems of unknown classes using PDs in gas-insulated switchgears (GISs) and propose a deep ensemble model to obtain the confidence of output probability and determine thresholds to detect unknown fault classes. The proposed model was verified by real-world phase-resolved PD (PRPD) experiments using online ultra-high frequency (UHF) PD measurement systems. The experimental results show that the proposed model achieves better unknown detection performance for the untrained PD faults and retains the classification performance for the trained PD faults.INDEX TERMS Fault diagnosis, convolutional neural network (CNN), ensemble model, partial discharges (PDs), gas-insulated switchgear (GIS).