• Energy Research

With the liberalization of the retail market, new parties such as load aggregators are participating in the demand response (DR). Aggregated baseline load (ABL) estimation provides a basis for aggregators to quantify the total responsiveness. This paper aims to improve the ABL estimation accuracy by using Gaussian mixture model (GMM). Modeling the distribution of consumption patterns by Gaussian distributions, GMM first divides the customers into several groups. Then, support vector regression (SVR) is utilized to estimate the baseline load over each group. And the estimated loads are summed up to form the final result. We make comprehensive comparisons in the case study. The results prove that the proposed method can improve the ABL estimation accuracy. And it is better than similar day, exponential moving average, and other regression model-based estimation methods.

Demand response (DR) is an important technique to explore the demand-side flexibility. The wide deployment of smart meters makes it possible to quantify the baseline load. As an intermediate agent, demand response aggregator needs to obtain the aggregated baseline load (ABL) for the DR event. Previous studies about the household level estimation focus on the estimation method. However, for ABL estimation, customer division is an important issue. A major limitation is the mismatch between the objectives of segmentation and estimation. Therefore, this paper proposes a new closed-loop method for estimating the ABL, which utilizes the contextual bandit with policy gradient to link the segmentation with the estimation. As such, the ABL estimation accuracy can guide the segmentation to divide the customers. The segmentation and estimation optimize collaboratively to improve the ABL estimation accuracy. An ensemble method for combining network’s weights during the training process is proposed. Moreover, a pre-and post-event adjustment method is developed to further improve the estimation accuracy. Comprehensive comparisons demonstrate the proposed method can achieve the best estimation performance with regard to the MAPE and RMSE. It improves the estimation accuracy by 7% in terms of MAPE, and 11% in terms of RMSE.

Energy forecasting is an essential task in power system operations. Operators usually issue forecasts and leverage them to schedule energy dispatch ahead of time. However, forecast models are typically developed in a way that overlooks the operational value of the forecasts. To bridge the gap, we design a value-oriented point forecasting approach for sequential energy dispatch problems with renewable energy sources. At the training phase, we align the loss function with the overall operation cost function, thereby achieving reduced operation costs. The forecast model parameter estimation is formulated as a bilevel program. Under mild assumptions, we convert the upper-level objective into an equivalent form using the dual solutions obtained from the lower-level operation problems. Additionally, a novel iterative solution strategy is proposed for the newly formulated bilevel program. Under such an iterative scheme, we show that the upper-level objective is locally linear regarding the forecast model output, and can act as the loss function. Numerical experiments demonstrate that, compared to commonly used statistical quality-oriented point forecasting methods, forecasts obtained by the proposed approach result in lower operation costs. Meanwhile, the proposed approach is more computationally efficient than traditional two-stage stochastic programs. Accepted in IEEE Transactions on Smart Grid

Prediction intervals (PIs) offer an effective tool for quantifying uncertainty of loads in distribution systems. The traditional central PIs cannot adapt well to skewed distributions, and their offline training fashion is vulnerable to the unforeseen change in future load patterns. Therefore, we propose an optimal PI estimation approach, which is online and adaptive to different data distributions by adaptively determining symmetric or asymmetric probability proportion pairs for quantiles of PIs’ bounds. It relies on the online learning ability of reinforcement learning (RL) to integrate the two online tasks, i.e., the adaptive selection of probability proportion pairs and quantile predictions, both of which are modeled by neural networks. As such, the quality of quantiles-formed PI can guide the selection process of optimal probability proportion pairs, which forms a closed loop to improve PIs’ quality. Furthermore, to improve the learning efficiency of quantile forecasts, a prioritized experience replay (PER) strategy is proposed for online quantile regression processes. Case studies on both load and net load demonstrate that the proposed method can better adapt to data distribution compared with online central PIs method. Compared with offline-trained methods, it obtains PIs with better quality and is more robust against concept drift.

Electric vehicle (EV) charging couples the operation of power and traffic networks. Specifically, the power network determines the charging price at various locations, while EVs on the traffic network optimize the charging power given the price, acting as price-takers. We model such decision-making processes by a bilevel program, with the power network at the upper-level and the traffic network at the lower-level. However, since the two networks are managed by separate entities and the charging expense term, calculated as the product of charging price and charging demand, is nonlinear. Solving the bilevel program is nontrivial. To overcome these challenges, we derive the charging demand function using multiparametric programming theory. This function establishes a piecewise linear relationship between the charging price and the optimal charging power, enabling the power network operator to manage EV charging power independently while accounting for the coupling between the two networks. With the derived function, we are also able to replace the nonlinear charging expense term with a piecewise quadratic one, thus guaranteeing solution optimality. Our numerical studies demonstrate that different traffic demands can have an impact on charging patterns and the power network can effectively incentivize charging at low-price nodes through price setting. submitted to IEEE Transactions on Smart Grid

Prediction interval (PI) is an effective tool to quantify uncertainty and usually serves as an input to downstream robust optimization. Traditional approaches focus on improving the quality of PI in the view of statistical scores and assume the improvement in quality will lead to a higher value in the power systems operation. However, such an assumption cannot always hold in practice. In this paper, we propose a value-oriented PI forecasting approach, which aims at reducing operational costs in downstream operations. For that, it is required to issue PIs with the guidance of operational costs in robust optimization, which is addressed within the contextual bandit framework here. Concretely, the agent is used to select the optimal quantile proportion, while the environment reveals the costs in operations as rewards to the agent. As such, the agent can learn the policy of quantile proportion selection for minimizing the operational cost. The numerical study regarding a two-timescale operation of a virtual power plant verifies the superiority of the proposed approach in terms of operational value. And it is especially evident in the context of extensive penetration of wind power. the revision to IEEE Transactions on Smart Grid