A non‐intrusive load state identification method considering non‐local spatiotemporal feature

Zhang, Zhenyu; Li, Yong; Duan, Jinxia; Duan, Yilong; Guo, Yixiu; Rehtanz, Christian

doi:10.1049/gtd2.12330

Cited by 8 publications

(2 citation statements)

References 22 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…To be specific, the Multi-Layer Perceptron neural network (MLP) is a classic tool to capture nonlinear features, allows feature refitting and reduces the loss of information with fully connected hidden layers [32], but is less capable of high dimensional tasks [33]. Besides, CNN (Convolutional Neural Network) is adept at revealing high-dimensional temporal features, utilizing continuous convolution operations to extract crucial traits [34] rapidly and alleviate the influence of noise. Moreover, the addition of a Gated Recurrent Unit (GRU) can further enhance time-dependent feature tracking.…”

Section: Load Disaggregation Model Based On the Multi-channel Structurementioning

confidence: 99%

Simulation‐data‐driven load disaggregation based on multi‐channel neural network for industrial and commercial users

Yang

Jiang

Sun

et al. 2023

IET Generation Trans & Dist

View full text Add to dashboard Cite

The immeasurable energy-saving potential of industrial parks and commercial buildings with unprecedented growth in scale and quantity can be well exploited by load disaggregation, a promising technique for demand-side refinement management. However, traditional load disaggregation studies mainly focus on residential users and neglect the collection and analysis of industrial and commercial loads. This study employs a simulated response model of versatile load categories with voltage-dependent load functions to overcome the unavailability of industrial and commercial datasets. Specially, a weighting factors allocation method considering probability statistics of power consumption distribution characteristics is proposed to construct a typical daily multi-type load aggregation model. Furthermore, an intelligent load disaggregation strategy is developed based on a multi-channel neural network to constructively improve multi-feature tracking and convergence efficiency. Finally, the simulation results of the detailed study case confirm the tractability and effectiveness of the proposed approach to carry out load identification among multi-type users, with an accuracy of 94.9% and errors of individual categories controlled between 3.32% and 10.686%. Moreover, additional interference tests demonstrate that the developed model has superiority over common load disaggregation structures, with better tolerance to exceptional voltage fluctuation and noise.

show abstract

Section: Load Disaggregation Model Based On the Multi-channel Structurementioning

confidence: 99%

Simulation‐data‐driven load disaggregation based on multi‐channel neural network for industrial and commercial users

Yang

Jiang

Sun

et al. 2023

IET Generation Trans & Dist

View full text Add to dashboard Cite

show abstract

“…. , x n }, where n is the dimension of the feature vector, and feature extraction can be done by calculating the mutual information between X and Y and sorting and filtering them [22]. In the training stage, we use a seq2seq classification model f θ () with parameters θ to train the length-L X[t : t+L] and Y [t : t + L], with the objective function:…”

Section: Problem Definitionmentioning

confidence: 99%

Achieving Sustained Improvement in Identification Accuracy with a Semi-Supervised Learning Approach for NILM Systems

Zhang¹,

Li²,

Guo³

et al. 2023

Preprint

View full text Add to dashboard Cite

<p>Non-Intrusive Load Monitoring (NILM) systems are widely used for energy management, but their accuracy can be limited by the removal of monitoring equipment after the initial training phase. This removal results in a lack of ongoing training data and impairs the ability to improve recognition accuracy. To address this challenge, this research proposes a novel semi-supervised learning approach for NILM systems. The approach is based on a previously proposed temporal convolutional network-conditional random field (TCN-CRF) model. It uses optimal interval filtering and unanimous voting to control the quality of pseudo-labels. The approach identifies high-precision segments in the sequence-to-sequence algorithm as semi-supervised learning targets and generates pseudo-labels, allowing for continuous self-supervised learning and sustained improvement in recognition accuracy. The results show that the proposed method can effectively combine unlabeled data to improve model accuracy, and achieve up to 14% highest improvement in algorithm accuracy over fully self-supervised methods in small sample training.</p>

show abstract

Technical and practical application of integrated energy technology in smart community

Wang,

Li,

Cao

et al. 2024

Electric Power Systems Research

View full text Add to dashboard Cite

A non‐intrusive load state identification method considering non‐local spatiotemporal feature

Cited by 8 publications

References 22 publications

Simulation‐data‐driven load disaggregation based on multi‐channel neural network for industrial and commercial users

Simulation‐data‐driven load disaggregation based on multi‐channel neural network for industrial and commercial users

Achieving Sustained Improvement in Identification Accuracy with a Semi-Supervised Learning Approach for NILM Systems

Technical and practical application of integrated energy technology in smart community

Contact Info

Product

Resources

About