19 December 2022 in IEEE Transactions on Sustainable Energy
Power system resilience has become a critical topic in recent years because of the increasing trend of extreme events and the growing integration of intermittent renewable energy sources. To enhance grid resilience against high-impact, low-frequency events, two questions should be answered: how to quantify the resilience of a given grid and how to incorporate the quantification into power system planning, operation, and restoration. This paper develops a new set of quantitative metrics with clear physical interpretation to comprehensively evaluate power system resilience. Using microgrids as an example, an event-based corrective scheduling (ECS) model and an online model predictive control (OMPC) model are developed to integrate the proposed quantitative resilience metrics into power system optimization models for resilience enhancement. The ECS model employs extreme event data to investigate the optimal restoration solution and to help microgrid operators prepare to respond to similar events. The OMPC model provides online decision-making support for operators to handle ongoing outages in the most resilient fashion. The effectiveness and superiority of the proposed quantitative resilience metrics and the resilience enhancement models are demonstrated through simulations and comparative studies on an IEEE test feeder and a real distribution feeder in Southern California.
2 January 2023 in IEEE Transactions on Smart Grid
This letter proposes a multi-task deep reinforcement learning (DRL) approach for distribution system voltage regulation considering topology changes via PV smart inverter control. The key idea is to encode the topology as an additional state for the DRL and leverage the multi-task learning scheme for joint learning of all task control policies. Unlike other DRL-based methods, our approach is robust to different topologies. Comparison results on the modified IEEE 123-node system demonstrate the enhanced robustness of the proposed method.
20 July 2018 in IEEE Transactions on Smart Grid
Distribution system state estimation (DSSE) is one of the vital components in the next-generation distribution management system (DMS), which allows the operators to monitor the entire system’s operating conditions. Due to the lack of real-time measurements, DSSE has to process measurements whose quality varies significantly across different sources, which causes convergence issue to the Gauss-Newton solver. In this paper, a semidefinite programming (SDP) framework is developed to reformulate the DSSE problem into a rank-constrained SDP problem. One challenge of this technique is the nonconvex rank-one constraint, which is generally relaxed. However, the relaxed SDP-DSSE problem cannot guarantee a rank-one solution and hence lose optimality. Therefore, we propose two solution approaches, namely the rank reduction approach and the convex iteration approach, to obtain rank-one solutions for the SDP-DSSE problem. The proposed model and the effectiveness of the proposed solution approaches are numerically demonstrated on the IEEE 13-, 34-bus, and 123-bus distribution systems.
18 April 2019 in IEEE Transactions on Smart Grid
The proliferation of electric vehicles (EVs) offers additional opportunities for improving the ambient air quality and reducing the associated health impacts. However, such opportunities could become unattainable if the electricity for charging EVs is mainly supplied by coal-fired units. This paper aims to enhance the environmental benefits of EVs by differentiating the marginal impacts of coal-fired unit emissions at various times and locations on ambient air pollutant concentration (AAPC) and human health conditions. The proposed cost-effective approach will assign EV battery charging loads and other electricity demands to generation units with low health impacts. The proposed air pollution dispersion model maps the emission into spatial AAPC increments (DAAPCs), and concentration-response function maps DAAPCs into health impacts. As such, we estimate differentiating health impacts of emissions at various times and locations without executing large-scale atmospheric models. We also integrate differentiating emission regulations with the joint optimization of security-constrained unit commitment (SCUC) and EV battery swapping-charging system (BSCS). BSCS refers to the optimal scheduling of an aggregated number of EV batteries which are centrally charged and then dispatched via delivery trucks to battery swapping stations (BSSs) to supply local EVs. BSCS determines optimal schedules for battery charging, swapping, and truck routing. A Lagrangian decomposition method decouples the formulated large-scale MILP model into SCUC and BSCS subproblems. Case studies demonstrate the effectiveness of the proposed approach for managing ambient air quality and human health conditions.
25 July 2019 in IEEE Transactions on Power Systems
This paper discusses several methods to improve commercial optimization solver performance on day-ahead security constrained unit commitment through warm start and lazy constraint settings. Data analytics is performed to greatly improve the quality of the initial commitment solution and lazy constraint setting. A distributed optimization framework is proposed to take advantage of the diversity from prevalent solvers (GUROBI and CPLEX) and different warm start strategies. A systematic distribution profile based benchmarking method is also proposed.
5 May 2021 in IEEE Systems Journal
High penetrations of distributed photovoltaics (PV) could cause adverse grid impacts, such as voltage violations. The recent development in inverter technologies provides the opportunity to develop control systems to realize effective PV governance, and thus, to improve dynamic PV hosting capacity for distribution grids. In this article, a novel distributed energy resource management system (DERMS) solution is proposed by adopting the real-time optimal power flow approach for coordinated control of the distributed PV inverters. The proposed approach eliminates the dependence on load knowledge via measurement feedback correction, and it can be implemented in real time. One challenge is that the technique is sensitive to the data availability and integrity of voltage measurements. Therefore, a decentralized DERMS approach is developed by leveraging the concepts of state estimation. The framework and the effectiveness of the solution approach are numerically demonstrated on a real distribution feeder in Southern California.