AUTHOR=Zhang Ruoyuan , Wang Yuan , Song Yang 

TITLE=Nonintrusive Load Monitoring Method Based on Color Encoding and Improved Twin Support Vector Machine

JOURNAL=Frontiers in Energy Research

VOLUME=Volume 10 - 2022

YEAR=2022

URL=https://www.frontiersin.org/journals/energy-research/articles/10.3389/fenrg.2022.906458

DOI=10.3389/fenrg.2022.906458

ISSN=2296-598X

ABSTRACT=Non-Intrusive Load Mmonitoring (NILM) is a method to realize automatic perception of users' electrical behavior by external analysis tools and means without relying on users' internal devices. This paper proposes a power load recognition method based on color image coding and Improved Twin Support Vector Machine (ITWSVM). Firstly, based on the traditional voltage-current grey trajectory method, bilinear interpolation technique is used to solve the problem of pixel discontinuity effectively. Considering the complementarity of features, the numerical features are embedded into the gray V-I trajectory by constructing three channels, namely current (), voltage () and phase (), so the color V-I image with rich electrical features is obtained. Secondly, two-dimension Gabor wavelet is used to extract the texture features of the image, and the dimension is reduced by means of Local Linear Embedding (LLE), reduce the amount of computation. Thirdly, the parameters of TWSVM are used as the location information of Artificial Fish Swarm Algorithm (AFSA), and the optimal parameters of TWSVM are obtained by AFSA, which improves the classification performance of TWSVM. Finally, the ITWSM is used to identify the features. The test results show that the accuracy of the proposed method is up to 86.73%. The results fully verify the effectiveness of the above methods.