Search

In order to reduce the dependence on accumulated experience that is difficult to replicate for the human dispatchers to make power generation scheduling more automatically, quickly, and intelligently, higher requirements are placed on the level of the auxiliary decision-making system. In this paper, a short-term scheduling problem was treated as a regression task that concentrates on how to learn a reliable statistic model that concludes the intrinsic logic of the dispatching policy from extensive historical dispatching experience. In this way, since Kullback-Leibler distance can better measure the distance between two distributions, we designed a novel GAN-based learning method for the scheduling task. Also, we proposed a feasible framework that combines the stage of learning, decision-making, and deployment to support the practical implementation of the proposed algorithm. In the experiment, a real case that takes short-term scheduling tasks on a regional large-scale power system is considered in our experiments. As a result, the comparison of several methods further shows the superiority of the proposed GAN-based method under our implementations. [ABSTRACT FROM AUTHOR]

Demand response (DR) currently plays a significant role in the operation of the electric grid. As a result, quantification of DR flexibility is an important aspect in the utilization of various DR resources. Generally, the evaluation of DR flexibility at bulk supply points (BSPs) is a challenging problem, especially without the monitoring of downstream customers’ load profiles in some areas. To solve this problem, we develop a two-layer DR flexibility estimation framework. In the top layer, a top-down optimization approach is proposed to disaggregate the BSP load into different building categories based on a suite of prototype building (PB) load profiles. In the bottom layer, simplified DR estimation models are deployed to quantify the theoretical DR flexibility of each PB type. Key advantages of this framework include: 1) quantifying DR flexibility at BSPs without relying on smart meter data or detailed customer surveys and 2) providing day-ahead, hour-ahead, and near real-time prediction of DR resources based on weather forecasts and other data. Case studies demonstrate the effectiveness of load disaggregation and DR flexibility quantification at a BSP. The prediction is compared with detailed physical models, and the mean relative errors for upper/lower DR capacity at the BSP are 1.5% and 3.1%, respectively. [ABSTRACT FROM AUTHOR]

Aggregate thermostatically controlled loads (TCLs) are good candidates for providing load following services in power systems. This paper is concerned with the modeling, evaluation, and control problems of a population of heterogeneous TCLs. Specifically, the heterogeneous population is divided into multiple homogeneous clusters and each cluster, i.e., TCL aggregator, is modeled by an approximated three-input single-output state space model. Here, the aggregators serve as a bridge connecting the load utility and the terminal TCLs, which have their own decision makers and are responsible for aggregate estimation and command issuing. And aggregate evaluation is carried out for the aggregator so as to provide the aggregate regulation capacities and ramping rates, which is useful for setting of the reference power trajectory. Based on the established control model, we furthermore propose a hierarchical centralized control algorithm for a bus load utility to regulate all TCLs inside it so as to provide load following service, while not affecting the customers’ comfort levels. Finally, simulation results with respect to a common bus load are provided to demonstrate the effectiveness of the proposed aggregate modeling and the centralized load following strategy. [ABSTRACT FROM AUTHOR]

In response to the computational challenges produced by the integrated dispatch of generation and load (IDGL), this paper proposes a novel and efficient decomposition method. The IDGL is formulated using the mixed-integer quadratic constrained programming (MIQCP) method. To efficiently solve this complex optimization problem, the nodal equivalent load shifting bidding curve (NELSBC) is proposed to represent the aggregated response characteristics of customers at a node. The IDGL is subsequently decomposed into a two-level optimization problem. At the upper level, grid operators optimize load shifting schedules based on the NELSBC of each node. Transmission losses are explicitly incorporated into the model to coordinate them with generating costs and load shifting costs. At the bottom level, customer load adjustments are optimized at individual nodes given the nodal load shifting requirement imposed by the grid operators. The key advantage of the proposed method is that the load shifting among different nodes can be coordinated via NELSBCs without iterations. The proposed decomposition method significantly improves the efficiency of the IDGL. Parallel computing techniques are utilized to accelerate the computations. Using numerical studies of IEEE 30-bus, 118-bus, and practically sized 300-bus systems, this study demonstrates that accurate and efficient IDGL scheduling results, which consider the nonlinear impact of transmission losses, can be achieved. [ABSTRACT FROM PUBLISHER]