Application of random matrix model in multiple abnormal sources detection and location based on PMU monitoring data in distribution network

Abstract

With the conversion of the global power economy and energy structure, access to a large amount of renewable energy has led to a decrease in power system inertia. The slight abnormal disturbance in the distribution network may have a significant impact on social and economic development. Aim at enhancing power stability and system resiliency; this study focuses on the detection and location of multiple abnormal sources in the distribution network. Most traditional methods use models relying on precise line parameters, subject to poor adaptability to the distribution network with a large number of nodes, and rapidly changing topology. Therefore, this study proposes a novel random matrix model, driven by monitoring data from phasor measurement units distributed on the overhead transmission lines. In this model, linear shrinkage (LS) theory, and Marchenko–Pastur law are combined for noise reduction to ensure the dynamic character and anti‐noise ability. Moreover, data dimensions and sample points may be at the same level in an extensive scale network. The LS and standard condition number rule (SCN) are used for estimating the number of abnormal sources. Finally, the effectiveness of this paper's model is verified in PSCAD. The results indicate that the method has specific dynamic performance and anti‐noise ability.

A Disturbance Localization Method for Power System Based on Group Sparse Representation and Entropy Weight Method

“…Within power systems, it is well-established that disturbances typically exhibit a sparse spatial distribution owing to their localized nature [12,13]. The task of localizing disturbances can be effectively accomplished through the utilization of compression-aware sparse reconstruction techniques.…”

Convolutional autoencoder anomaly detection and classification based on distribution PMU measurements

“…In addition, the robustness of proposed method to various factors is analysed in this study. Figure 13 depicts the test feeder [20] along with 5 DER units and 7 DPMUs. Various operating conditions, including different loading and configuration, have been considered in the simulation.…”

“…In [13], an ensemblebased unsupervised approach is proposed to identify outliers based on the anomaly score detected by three different base detectors (Chebyshev-based, DBSCAN-based, and Regressionbased detectors). Other techniques such as dynamic time warping classifier [14], multi-class SVM [15,16], Bayesian Network [17], and random matrix theory [18][19][20] are also used to address the anomaly detection problem.…”