1 April 2023 in Production and Operations Management
5 March 2025 in IEEE Transactions on Instrumentation and Measurement
Topological reconfiguration caused by distributed energy consumption changes the propagation path of fault features rendering localization more difficult. To address this issue, a two-stage graph spatiotemporal attention network (Ts-GSAN) fault location method is proposed for distributed energy systems. Firstly, a low-observable node selection mechanism is proposed to realize global awareness of the system based on the least amount of data, which accounts the propagation mechanism of fault unbalanced current. Secondly, a graph spatiotemporal attention mechanism is proposed to solve the problem of feature propagation path change due to topology reconfiguration, which captures the feature attenuation during fault propagation. Further, a two-stage graph neural network method is proposed for fault localization, which solves the problem of interference between various fault types through a localization-correction mechanism. Ts-GSAN realizes the global fault awareness of the system based on a small number of devices, which can ensure the generalizability to different topologies and the localization accuracy. Finally, experimental validation is carried out through IEEE 123 node distributed energy systems. The results show that the fault localization accuracy reaches more than 98% with high effectiveness and feasibility.
23 January 2024 in IET Control Theory & Applications
With the popularity of smart meters, the frequent information exchange between smart grids and consumers leads to easy leakage of consumers' electricity consumption data. These leaked electricity consumption data are obtained by some malicious attackers and used to infer consumers' behavioural patterns by non‐intrusive load monitoring (NILM), which seriously threatens consumers' privacy. Therefore, the multi‐agent Hidden Markov energy management model is proposed in this paper to safeguard the privacy of consumers. First, a weighted Bayesian risk model is proposed, which combines privacy leakage risks and energy storage system (ESS) losses in a microgrid with multiple agents. Next, a three‐loop model for lithium batteries is constructed to quantify the capacity degradation and cost issues of the ESS. Finally, the multi‐objective optimization problem is resolved by integrating the Bayesian risk model with a hidden Markov model to simulate attackers. The proposed multi‐agent Markov decision process method is validated on Electricity Consumption and Occupancy (ECO) dataset, and control strategies are evaluated based on different weights in the Bayesian risk model. The results demonstrate that by incorporating the multi‐agent approach and energy storage system capacity degradation into the privacy protection strategy, the lifespan of the energy storage system can be significantly increased.
4 April 2022 in Frontiers in Energy Research
Based on long air gap discharge test data and lightning return stroke observation data, an improved electrogeometric model (EGM) considering terrain conditions is established and verified to analyze the lightning shielding performance of UHV transmission lines. The striking distance formula is modified as rs = 0.13 (I2 + 40I)0.814. In this paper, the lightning shielding failure rate of three-phase conductors of EHV and UHV transmission lines calculated by the improved EGM model is consistent with the lightning observation data of actual transmission lines in Japan and the scaled lightning discharge simulation experimental results of UHV transmission lines in plains and mountainous areas of China, which verifies the applicability of the improved EGM model to evaluate the lightning shielding performance of large-scale UHV transmission lines. The improved EGM model is applied to evaluate the influence of tower type and slope steepness on the shielding failure tripping rate of UHV transmission lines. The shielding failure tripping rate of the SZ322 tower is higher than that of the SZT1 tower, and the shielding failure tripping rate of UHV transmission lines is greatly affected by slope gradient and increases with the increase of slope gradient.
18 September 2024 in IEEE Transactions on Instrumentation and Measurement
Fault detection is an essential aspect of power generation in wind turbines (WTs). However, existing fault detection methods are developed specifically for identifying a single type of fault and rely on a sufficient amount of labeled data. These approaches tend to be less accurate when used to detect many types of faults with limited data. To solve those problems, this paper proposes a self-supervised fault detection method based on a time-frequency feature fusion module (TF-FFM). First, a periodicity-based hybrid data augmentation is presented in order to expand the number and diversity of fault samples. Second, time-frequency feature fusion module can fully extract the features of fault in both time and frequency domains. Third, a fault detection method based on self-supervised learning is proposed during the training process to reduce the cost of data collection and labelling. Meanwhile, the loss function is optimised based on the energy conservation theorem in the time-frequency domain. This optimization leads to an advanced accuracy of the fault detection method for WTs by establishing a power consistency relationship. Finally, this paper evaluates the effectiveness of the proposed method through a comparative analysis with various fault diagnosis techniques, feature visualization, and ablation experiments. The accuracy of the proposed method achieves 95% in the context of multiple faults detection, which is 3% higher than the results of existing methods.
19 March 2025 in IEEE Transactions on Consumer Electronics
Applying high-precision 3D reconstruction for driver facial expression recognition can effectively tackle difficulties such as facial occlusion that are commonly encountered when using 2D image-based detection methods. However, challenges such as weak inter-correlation among multi-scale features as well as cost volumes, and insufficient attention to critical features in deep learning-based multi-view stereo (MVS) methods still persist. Therefore, a cross-scale attention-based MVS (CSA-MVSNet) model is proposed to address these challenges. Specifically, a channel-spatial cross-scale attention strategy is designed to enhance the interdependence of features between encoder and decoder. In this strategy, channel cross-scale attention is employed to address the convolutional neural network’s limitations in handling long-range dependencies, while spatial cross-scale attention enhances the correlation among multi-scale features. Meanwhile, a feature enhancement attention module is proposed to facilitate information exchange among sub-features. Furthermore, a multi-scale cost volume fusion method is presented to propagate low-scale cost volume information to higher scales progressively, ensuring the effective transfer of information across cost volumes from fine to coarse levels. Extensive experiments on DTU dataset demonstrate that the proposed model achieved improved accuracy and completeness in reconstruction compared to existing methods, validating its effectiveness. Moreover, the model offers theoretical potential for achieving high-accuracy driver facial expression recognition.