Adaptive isolation of fault areas in distribution networks based on multi-scale entropy fractal dimension composite features

With the large-scale integration of distributed power sources, the power grid often experiences random load fluctuations. When a fault occurs in the distribution network, electrical signals such as fault current and voltage exhibit highly nonlinear characteristics, making it difficult to accurately extract fault area feature information, which in turn affects the positioning accuracy of the fault area and cannot achieve precise isolation. Therefore, a multi-scale entropy fractal dimension composite feature based adaptive isolation of distribution network fault areas is proposed. By constructing a multi-scale entropy fractal dimension composite model, fault features of the distribution network are extracted, a composite feature vector is constructed, and a subset of fault features is obtained through principal component analysis. Combined with the perception ability of feeder terminals, a fault state matrix is constructed, and Pearson correlation coefficient is introduced to determine the fault category. The fault location is determined through fault traveling wave analysis, and the equipment fault correlation degree is calculated based on the distribution network topology to clarify the isolation range. An artificial neural network is introduced to design the basis for starting the isolation scheme, and isolation operations are implemented according to the prescribed process. Through experimental verification, this method adaptively isolates the fault area of the power grid, and the similarity between the electrical information features and the fault features in the distribution network is less than 3%. This indicates that this method can accurately adaptively isolate the fault area of the distribution network and avoid the propagation of fault features.