21 November 2019 in IEEE Access
Photovoltaic (PV) output is susceptible to meteorological factors, resulting in intermittency and randomness of power generation. Accurate prediction of PV power output can not only reduce the im-pact of PV power generation on the grid but also provide a reference for grid dispatching. Therefore, this paper proposes an LSTM-attention-embedding model based on Bayesian optimization to predict the day-ahead PV power output. The statistical features at multiple time scales, combined features, time features and wind speed categorical features are explored for PV related meteorological factors. A deep learning model is constructed based on an LSTM block and an embedding block with the connection of a merge layer. The LSTM block is used to memorize and attend the historical information, and the embedding block is used to encode the categorical features. Then, an output block is used to output the prediction results, and a residual connection is also included in the model to mitigate the gradient transfer. Bayesian optimization is used to select the optimal combined features. The effectiveness of the proposed model is verified on two actual PV power plants in one area of China. The comparative experimental results show that the performance of the proposed model has been significantly improved compared to LSTM neural networks, BPNN, SVR model and persistence model.
29 November 2023 in IEEE Transactions on Industrial Informatics
Electricity theft users with zero electricity usage (UZEU) should be specifically concerned in electricity theft detection (ETD) research. The challenges are: they provide no effective information on electricity usage behaviors, and they are easily confused with vacant house users. This has caused the majority of the existing detection methods relying on single electricity usage to fail to identify UZEU accurately. Hence, this article first analyzes the underlying correlation between water and electricity (W&E) usage collected by the smart meter. This analysis then lends the theoretical basis to propose a new ETD method by comprehensively using the multisource information. More precisely, the proposed method utilizes the mutual information coefficient (MIC) to construct a correlation model between W&E usage and in turn the wavelet clustering algorithm to cluster the MIC of the power distribution users. Thereafter, the resulting weak correlations indicate the suspected users as the electricity theft UZEU in case of zero electricity usage. Finally, the proposed method is validated by numerical experiments in the real world and illustrated to be more accurate than existing methods in detecting UZEU.
13 June 2022 in IEEE Transactions on Power Delivery
Operating Acoustic Signal (OAS) emitted by electric equipment reveals its state. However, environmental noise (EN) could result in misleading result. Distribution transformers are deployed in noisy environment and its OAS is mixed with substantial noise. OAS of distribution transformer in normal and typical anomaly state behave spectro-temporal self-similarity (STSS), which is notable different from known ENs. A STSS based approach is proposed to identify whether OAS is corrupted by EN. 60s OAS of transformer are divided into overlapping segments. The features of these segments are grouped to determine whether they belong to the same cluster. Once they belong to the same cluster, it can be used for condition monitoring. Otherwise, it is corrupted by EN. Numerical simulation of onsite recorded OAS indicates that the proposed approach can identify OAS corrupted by EN precisely. Difficulty of removing EN from OAS could be avoided in a much easier alternative way to single out valid OAS not corrupted by EN instead of removing EN from OAS. The proposed approach facilitates OAS based condition monitoring of distribution transformers in noisy environment.
6 January 2025 in IEEE Access
The construction of smart cities contributes to promoting residents’ life convenience and sustainable energy development. Despite these advancements, the challenge of fully analyzing and understanding residents’ energy usage behaviors leads to inefficient energy use and potential economic losses. Current resident anomaly detection technologies rely on single-source energy data, lacking detailed behavior pattern analysis. Hence, this paper proposes a method to detect abnormal residential water and electricity usage by incorporating multi-source information. Specifically, the correlation between water and electricity usage of residential customers is analyzed based on real metering data and the use of the Copula distribution function, followed by the integration of two innovative data mining techniques to form an anomaly detection framework. The distance correlation coefficient algorithm is used to measure the relevance of users’ water and electricity usage data. Then, the Density-Based Spatial Clustering of Applications with Noise (DBSCAN) algorithm is utilized to cluster the distance correlation coefficient for users and detect abnormal users whose distance correlation coefficient curves deviate from the normal user clusters. This multi-source approach avoids single-source bias by improving the data accuracy over one-dimensional methods. Experiments are implemented in a real low-voltage transformer area to prove the validity of the proposed method.
9 December 2025 in IET Smart Energy Systems
In September 15, 2011, an unseasonal heat wave triggered an energy shortfall in Korea, leading to rolling blackouts: approximately 4000 MW of load was shed and 7.5 million users lost power. Using historical temperature records and power‐system operation data, we show that the generation deficit primarily stemmed from maintenance scheduling based on traditional seasonal patterns, which coincided with a heat‐driven peak‐load surge under a warming climate. We further analyse future climate projections from CMIP6 (Coupled Model Intercomparison Project Phase 6) and find that the likelihood of September heat waves in Korea will increase notably over the next three decades. It is prudent to estimate the risk of unseasonal heat wave when determining the maintenance scheme.
1 January 2025 in IET Generation, Transmission & Distribution
Wiring errors, caused by improper connections between neutral lines and protective earth (PE) lines due to negligence by electrical technicians, are a prevalent type of earth fault in low‐voltage distribution systems (LVDS). These errors can cause the load current of affected users to flow back to the transformer's neutral point via the PE line as residual current, leading to nuisance tripping of residual current devices (RCDs). To maintain normal power supply, users may disable RCDs, which compromises safety and can result in severe hazards such as electric shocks and electrical fires. This paper proposes a method to locate users with wiring errors by leveraging abundant metering data within LVDS and utilising leakage fault analysis devices. We construct a linear model of residual current considering multiple error scenarios. Based on this model, a multiple linear regression (MLR) approach is developed to identify and locate anomalous users by analysing the correlation between their load currents and the residual current of the LVDS. Experimental results under various scenarios validate the performance of the proposed method.