A near-optimal maintenance policy for automated DR devices

摘要

Demand side participation is now widely recognized as being extremelycritical for satisfying the growing electricity demand in the US. The primarymechanism for demand management in the US is demand response (DR) programs thatattempt to reduce or shift demand by giving incentives to participatingcustomers via price discounts or rebate payments. Utilities that offer DRprograms rely on automated DR devices (ADRs) to automate the response to DRsignals. The ADRs are faulty; but the working state of the ADR is not directlyobservable --one can, however, attempt to infer it from the power consumptionduring DR events. The utility loses revenue when a malfunctioning ADR does notrespond to a DR signal; however, sending a maintenance crew to check and resetthe ADR also incurs costs. In this paper, we show that the problem ofmaintaining a pool of ADRs using a limited number of maintenance crews can beformulated as a restless bandit problem, and that one can compute anear-optimal policy for this problem using Whittle indices. We show that theWhittle indices can be efficiently computed using a variational Bayes procedureeven when the load-shed magnitude is noisy and when there is a random mismatchbetween the clocks at the utility and at the meter. The results of ournumerical experiments suggest that the Whittle-index based approximate policyis within 3.95% of the optimal solution for all reasonably low values of thesignal-to-noise ratio in the meter readings.