Recent years have witnessed the prominent advancements of deep learning (DL) in the arsenal of prognostics and health management. However, the prognostic uncertainty problem extensively existed in industrial devices is not addressed by most DL approaches. This article formulates a novel Bayesian Deep Learning (BDL) framework to characterize the prognostic uncertainties. A distinguished advantage of the framework is its capacity of capturing the comprehensive effects of two critical uncertainties: 1) epistemic uncertainty, accounting for the uncertainty in the model, and 2) aleatoric uncertainty, representing the impact of random disturbance, such as measurement errors. The former arises from the variability of the model weights, and the latter is characterized by selected lifetime distributions. We integrate both uncertainties by defining BDL as priors of lifetime parameters. A sequential Bayesian boosting algorithm is executed to improve the estimation accuracy and compress the credible intervals. The superior prediction performance of our framework is validated by a real-world dataset collected from hydraulic mechanisms of circuit breakers.

In the presence of H ₂ O and O ₂ , the dissociation products of SF ₆ will decompose to form several main stable byproducts (i.e. SOF ₂ , SOF ₄ and SO ₂ F ₂ ) in an electrical discharge. These byproducts are chemically active and have been shown experimentally to be associated with discharge faults. However, the relationships between the discharges and types of decomposition components are still not clear, mainly due to the fact that the complex chemical processes during SF ₆ discharges are not fully understood. In order to comprehensively investigate the decomposition of SF ₆ , an approach combining density functional theory (DFT) and transition state theory (TST) was used to study the pathways of SF ₆ decomposition in mixtures with H ₂ O and O ₂ that involve electrically-neutral species. The complex chemical reactions were analyzed, and the corresponding rate constants were predicted. The structural optimizations, vibrational frequency calculations and zero-point energy calculations of the species involved in each chemical reaction considered were carried out using the DFT-B3LYP method. Single-point energies were calculated using the CCSD(T) method. Based on the energy information obtained, the rate constants were predicted by TST, over a large temperature range, from 300 to 12 000 K.

Partial discharge (PD) is not only an important symptom for monitoring the imperfections in the insulation system of a gas-insulated switchgear (GIS), but also the factor that accelerates the degradation. At present, monitoring ultra-high-frequency (UHF) signals induced by PDs is regarded as one of the most effective approaches for assessing the insulation severity and classifying the PDs. Therefore, in this paper, a deep learning-based PD classification algorithm is proposed and realized with a multi-column convolutional neural network (CNN) that incorporates UHF spectra of multiple resolutions. First, three subnetworks, as characterized by their specified designed temporal filters, frequency filters, and texture filters, are organized and then intergraded by a fully-connected neural network. Then, a long short-term memory (LSTM) network is utilized for fusing the embedded multi-sensor information. Furthermore, to alleviate the risk of overfitting, a transfer learning approach inspired by manifold learning is also present for model training. To demonstrate, 13 modes of defects considering both the defect types and their relative positions were well designed for a simulated GIS tank. A detailed analysis of the performance reveals the clear superiority of the proposed method, compared to18 typical baselines. Several advanced visualization techniques are also implemented to explore the possible qualitative interpretations of the learned features. Finally, a unified framework based on matrix projection is discussed to provide a possible explanation for the effectiveness of the architecture.

Active current injection circuit breakers (ACICBs), which utilize the mechanical switch and a parallel-connected pre-charged capacitor, have broad application prospects in future multi-terminal DC grids. However, with the increase of system voltage, the pre-charged capacitor leads to the great cost and size being spent on the charging circuit and auxiliary commutating appliances. To address this shortcoming, an innovative DC circuit breaker (DCCB) based on a magnetic induction current commutation module (MICCM) is proposed. The influence of the main component parameters on the commutation performance is investigated by calculation, in which the self, mutual, and stray inductances are involved. Besides, the validity of calculation is experimentally proved. Then, by applying the MICCM to the ACICB, the breaking characteristics of the ACICB are analyzed in detail. Finally, a 10-kV DCCB prototype is developed, verifying the feasibility of current commutation concept and the validity of simulated interruption results. Our studies show that a 10-kA bidirectional fault current can be interrupted successfully with a commutation duration less than 250 's.

Image online monitoring technology has been widely used in transmission lines inspection, but the intelligent and efficient foreign object detection still has a gap with the ideal. In this paper, we propose a deep learning method to detect invading foreign objects for power transmission line inspection. Specifically, we design our network based on the regression strategy with oriented bounding boxes to accurately predict spatial location and orientation angle of foreign objects, as well as their categories in cluttered backgrounds. Moreover, an easy yet effective Scale Histogram Matching method is proposed to be applied to the publicly available dataset, allowing useful patterns to be exploited to detect tiny foreign objects during the pretraining procedure and boosting detection performance even with limited annotated samples. Besides, we construct an image dataset that contains common foreign objects in transmission line scenarios to evaluate proposed methods, on which experiment results show our full model achieves accuracy with 88.1% mean Average Precision (mAP). Additionally, the efficient and compact network structure allows our network to run in real-time, which provides possibilities for practical use.

SF ₆ decomposition products could reflect the running status and inner faults of power equipment, and it's expected to realize a timely warning. In this work, six faults including spark and corona discharge were simulated, and SF ₆ decomposition products with various types and contents were obtained as well. Different from previous investigations employing precision instruments, such as gas chromatography and infrared spectroscopy, a micro sensor array loaded with three gas-sensitive nanomaterials was used to discriminate fault characteristic gases, performing obvious advantages in small size, high integration, and rapid detection. Gas chromatography-mass spectrometry (GCMS) indicated that seven analytes had significant differences in types and contents. Meanwhile, the as-prepared micro gas sensor array also outputted significantly various signals for seven analytes, which provided a basis for gas identification. With the assistance of stacked denoising autoencoder (SDAE)-based discrimination algorithms, the recognition model between the response signals of the array and the discharge faults in power equipment could be established. In comparison with KNN (66.67 %), decision tree (70.47 %), and BPNN (73.33 %), SVM has achieved the highest average accuracy of 75.23 %. Totally, this work provides a promising novel method for rapid on-site inspection of SF ₆ -insulated power equipment.