Industrial load identification based on sparse representation classification

Abstract

In this paper, sparse representation classification (SRC) model based on discriminant dictionary is proposed to train loads of different categories by K-SVD algorithm. Firstly, The absolute value of sparse coefficient and the category of atoms are used to determine the load category. Secondly, the identification of load types is improved in the traditional SRC method based on dictionary learning. Thirdly, the actual identify effect is optimized. Finally, an example is given to illustrate the recognition process. The discriminant dictionary-based SRC method is compared with the traditional SRC method. It proved the feasibility and effectiveness of the method.