Advanced topology optimization in cyber-physical distribution systems for multi-security and operational flexibility enhancement

Abstract

The distribution grids evolve from the passive network with having the goal of supplying reliably and efficiently the end users, gradually to active networks with integrating distributed energy resources (DERs). With the extensive use of operational technologies (OT) and information and communication technologies (ICT) networks, the transition to cyber-physical distribution systems enables the complete observability enhancement of measurements and smartization of control components. Under this background, an effective distribution network reconfiguration (DNR) scheme plays a key role in smart energy management of today’s active distribution networks (ADNs) for substantial cost reductions and operational flexibility enhancements subject to system observability and privacy concerns of different stakeholders. Firstly, we propose a disjunctive convex hull relaxation (DCHR) to tackle with the classical DNR problem. This classic DNR problem is a mixed integer second order conic programming (MISOCP) problem which is non-convex and nonlinear. However, our proposed DCHR approach can perfectly addresses this problem and it is proven to be a tighter relaxation than the existing relaxation techniques for DNR problems, such as the Big-M and McCormick linearization methods. Secondly, the system observability enabled by distribution-level PMUs becomes increasingly crucial for cyber-physical security enhancement. We formulate this system observability as a disjunctive relaxed connected dominating set problem for reconfigurable ADNs with the least defense cost in theory. For the benefits of system observability, an observability defense-constrained DNR model has been proposed. Thirdly, the topology switch for the loss minimization may expose the private load change information of an agent, e.g., transition from a light load to a heavy load, in interconnected ADNs managed by multiple agents. To address this issue, this paper proposes a differentially private distribution network reconfiguration (DP-DNR) mechanism based on a consensus alternating direction method of multipliers (C-ADMM) algorithm. This can tackle privacy leakage challenges on the agent’s and customer’s levels. To suppress private load change leakage as an agent’s concern, this DP-DNR mechanism provides a mixture output of realistically optimal topology switch status and corresponding obfuscated-but-feasible load flows, part of which may have reverse load flow directions. On the customer’s level, the C-ADMM-based decentralized DP-DNR approach can seek the optimal topology switch without customer’s load datasets of agents, whilst exchanged communication signals in C-ADMM algorithm are also synthetic based on the proposed DP-DNR mechanism. Lastly, a distribution-level topology optimization contributes to the flexibility enhancement of a look-ahead rolling economic dispatch of wind-thermal-bundled power system (WTBPS), which offsets the insufficient ramping margins of retrofitted coal-fired units. Since WTBPS connects to high voltage distribution networks (HVDNs), graph characterization of typical HVDNs is summarized, and then the simplified voltage-constrained load transfer strategy via topological structures can be developed. This proposed look-ahead economic dispatch model is cast as a MISOCP problem. For this established MISOCP-based model, it is highly desirable to combine the Multi-cut Benders Decomposition (MBD) and Generalized Benders Decomposition (GBD) as the devised Multi-cut GBD (MGBD) to tackle this MISOCP problem, which can enhance overall computational efficiency and be suitable for online rolling economic dispatch.