A Distributed Online Pricing Strategy for Demand Response Programs

Abstract

We study a demand response problem from utility (also referred to as operator)'s perspective with realistic settings, in which the utility faces uncertainty and limited communication. Specifically, the utility does not know the cost function of consumers and cannot have multiple rounds of information exchange with consumers. We formulate an optimization problem for the utility to minimize its operational cost considering time-varying demand response targets and responses of consumers. We develop a joint online learning and pricing algorithm. In each time slot, the utility sends out a price signal to all consumers and estimates the cost functions of consumers based on their noisy responses. We measure the performance of our algorithm using regret analysis and show that our online algorithm achieves logarithmic regret with respect to the operating horizon. In addition, our algorithm employs linear regression to estimate the aggregate response of consumers, making it easy to implement in practice. Simulation experiments validate the theoretic results and show that the performance gap between our algorithm and the offline optimality decays quickly.

Serverless Computing: A Survey of Opportunities, Challenges and Applications

“…Finding the optimal pricing strategy with multiple providers and customer constraints is a challenging issue that must be addressed in research studies. A similar notion have extensively discussed in the area of smart grids where supply and demand are considered in extracting online pricing models [59].…”

Online Learning for Network Constrained Demand Response Pricing in Distribution Systems

“…t , which is the error inflicted by the uncontrollable disturbance ǫ in Eq. (8). This error is the same in the oracular and non-oracular cases and, therefore, the learning error in the three non-oracular cases can be recovered as ∆…”

“…the difference between the decision of the oblivious model and the oracle (perfect foresight) model, allows assessment of the performance of the learning process. We define two regret metrics similar to [8]. First, the expected regret defines the difference between the objective values of the oblivious and oracle models:…”

“…These methods learn the price sensitivity of each DR participants by inferring it from the historical price signals and observed DR responses, and use the inferred value to generate a more accurate price signal. Li et al [8] use an iterative regression algorithm to learn price sensitivities of individual DR participants that can be used by profit-seeking DR aggregators to optimize the total DR capacity offered to the utility. This algorithm is shown to achieve logarithmically progressing regret, i.e.…”

Is Machine Learning in Power Systems Vulnerable?

“…The modern power systems, with deeper penetration of renewable generation and higher level of demand-side participation, are faced with increasing degree of complexities and uncertainties [1], [2]. Reliable operation of the grid in this context calls for improved techniques in system modeling, assessment and decision making [3], [4], [5]. On the one hand, smart meters and advanced sensing technologies have made the collection of fine-grained electricity data, both historical and streaming, available to system operators [6].…”