The accuracy of the load model has great effects on power system stability analysis and control. Based on our practice in China on modeling load from field measurements, this paper systematically develops a measurement-based composite load model. Principles guiding the load modeling practice are discussed based on detailed analysis on stochastic characteristics of the modeling procedure. The structure of the measurement-based composite load model is presented. A multicurve identification technique is described to derive parameters. The generalization capability of this built load model is also investigated in this paper. Two cases are studied to illustrate the accuracy of the developed load model on describing the load dynamic characteristics in the actual power system.

This paper proposes a robust optimization approach for optimal operation of microgrids. The uncertain output variation of renewable energy sources (RESs) is addressed by collaboratively scheduling of energy storage (ES) and direct load control (DLC) through a two-stage complementary framework: an hour-ahead charging/discharging of ES and a quarter-hour-ahead activation of DLC. The objective is to maximize the total profit of the microgrid considering operation and maintenance costs of ES units, wind turbines and photovoltaics, and transaction with main grid and customer loads. Assuming the power output of RES randomly varies within a bounded uncertainty set, the problem is modeled to a two-stage robust optimization model and solved by a column-and-constraint generation algorithm. Compared with conventional operation methods, the ES and DLC are coordinated in different time-scales, and RES uncertainties are fully addressed during operation decision-making, ensuring the solutions to be optimal and robust for any realization of uncertainty. The proposed methodology is verified on the IEEE 33-bus distribution system through a wide range of different tests.

To address the uncertain renewable energy in the day-ahead op-timal dispatch of energy and reserve, a multi-stage stochastic programming model is established in this paper to characterize the uncertainties over the multiple stages by a scenario tree. Most importantly, to deal with the “Curse of Dimensionality” of sto-chastic programming, stochastic dual dynamic programming (SDDP) is employed, which decomposes the original problem into several sub-problems according to the stages. Specifically, the SDDP algorithm performs forward pass and backward pass repeatedly until the convergence criterion is satisfied. At each iteration, the original problem is approximated by creating a lin-ear piecewise function. The results on the IEEE 118-bus system and real-life provincial power grid show the effectiveness of the proposed model and method.

A smart power grid is an integration of the advanced measurement, communication, computer, and control techniques. Among all the state-of-the-art technologies in building a smart power grid, the phasor measurement unit (PMU) is an important and promising one. Nowadays in China, most ultrahigh- and high-voltage buses are all equipped with PMUs. This paper reports an application of PMU data on locating the disturbance source for the low-frequency oscillation that has happened several times in the interconnected big power grids in China recently. The posterior analysis by the Electric Power System Research Institute of China (CEPRI) has confirmed that these low-frequency oscillations are caused by the resonance phenomena. It is urgent to find the disturbance source in case it happens again. However, there have been no systematic and effective methods to locate the disturbance in a very large power system. The trial and error method has been implemented via the digital simulations, but in vain. To overcome this problem, a novel method is developed in this paper to locate the disturbance source with the help of PMUs. The method takes advantage of the PMU measurements to reduce the searching area for locating the disturbance source. Case studies show the efficiency of the proposed method.

Appliance-level load models are expected to be crucial to future smart grid applications. Unlike direct appliance monitoring approaches, it is more flexible and convenient to mine smart meter data to generate load models at device level nonintrusively and generalise to all households with smart meter ownership. This paper proposes a comprehensive and extensible framework to solve the load disaggregation problem for residential households. Our approach examines both the modelling of home appliances as hidden Markov models and the solving of non-intrusive load monitoring based on segmented integer quadratic constraint programming to disaggregate a household power profile into the appliance level. Structure of our approach to be implemented with current smart meter infrastructure is given and simulations are performed based on public datasets. All data are down-sampled to the rate that is consistent with the Australia smart meter infrastructure minimum functionality. The results demonstrate that our approach is able to work with existing smart meters to generate device level load model for other smart grid research and applications.

System splitting problem, also known as controlled system separation problem, is to determine the proper splitting points for splitting the entire power network into islands when island operation of system is unavoidable. By "proper" we mean that the splitting strategies should guarantee both the power balance and satisfaction to capacity constraints of transmission lines and other facilities in each island. The system splitting problem is very hard because the strategy space is huge for even middle-scale power networks. This paper proposes a two-phase method to search for proper splitting strategies in real-time. The method narrows down the strategy space using highly efficient OBDD-based algorithm in the first phase, then finds proper splitting strategies using power-flow analysis in the reduced strategy space in the second phase. Simulation with symbolic model checking tool SMV indicates that this method is very promising.