Controlled islanding of power networks based on graph reduction and spectral clustering

Abstract

Intentional controlled islanding aims to split the power system into self-sustainable islands after a severe disturbance, but prior the uncontrolled network separation. Given its nature (i.e. last resort for blackout prevention), this emergency control technique must be adopted as quickly as possible. This paper proposes a computationally efficient method based on graph reduction and spectral clustering. The paper contributes by describing important details of the graph reduction process in the context of controlled islanding and by the formalisation of this process. Furthermore, it demonstrates how to adopt embedded graphs to enhance the Multiway Spectral Clustering graph partitioning. Finally, it is shown how to explicitly incorporate important cannot-link constrains between coherent generator groups into the islanding problem. The proposed method is detailed using the IEEE 39-bus test case. To evaluate the algorithm performance, the method is applied to realistically-sized PEGASE test networks. 'case39'

Slow Coherency Identification and Power System Dynamic Model Reduction by Using Orthogonal Structure of Electromechanical Eigenvectors

“…As discussed in Section V-B, it is known that the NPCC 48-machine test system can be well decomposed into 9 areas based on its 9 slowest electromechanical modes. Our previous case study in Section V-B comes to the same conclusion, but it additionally identifies alternative area structures consisting of 1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,33,34,35 , 36 2 32, 37, 38, 39, 40, 41, 42 3 43, 44, 45, 46, 47, 48 3 and 6 areas (see Figure 2). The present case study illustrates the validity of decomposing the NPCC system into 6 areas by using our grouping algorithm in Section IV.…”

“…Another use case of coherency-based reduced models is motivated by the aversion of system operators to share the full models of their control areas with third parties for confidentiality reasons. Besides applications to power system model reduction, coherency identification techniques are instrumental for the design of certain system integrity protection schemes (SIPS) such as intentional controlled islanding (ICI) [4]- [6], as well as the design of wide-area monitoring and control systems (WAMCS) [7], [8].…”

Generator grouping cutset determination based on tree construction and constrained spectral clustering

“…Once the graph nodes were embedded geometrically, it is possible to construct an embedded graph having the same sets of nodes and edges as G , but with its edge weights defined by the distances between the graph nodes in geometric graph embedding [9, 10]. The weighted adjacency matrix corresponding to graph is denoted as .…”

“…Then the remaining task is to produce a contiguous partitioning of the reduced graph that separates the k merged nodes from each other, which was detailed e.g. in [10] (in general, this problem is known as k‐way cut in the literature). The partitioning resulting from spectral clustering can often be noticeably improved by using various post‐processing algorithms [11], with the two relevant refinement algorithms suitable for both active power balance improvement and increase in islands’ electrical cohesion detailed in [12].…”

Adaptive Day-Ahead Prediction of Resilient Power Distribution Network Partitions

“…The graph laplacian for the spectral clustering is tabulated from the admittance matrix of the power grid. The graph partitioning method based on graph reduction and spectral clustering for power distribution networks is discussed in [9]. The authors in [10] have integrated a self organizing map algorithm with spectral clustering with a goal of tackling problems like optimal number of clusters, multi-objective partitioning, and big-data.…”