Setpoint Tracking with Partially Observed Loads

TL;DR

This paper develops online convex optimization algorithms for setpoint tracking in uncertain load systems, providing regret bounds and evaluating performance on thermostats and electric vehicles under various feedback scenarios.

Contribution

It introduces algorithms for load tracking with partial and probabilistic feedback, extending OCO methods to more realistic feedback models with proven regret guarantees.

Findings

Algorithms achieve sublinear regret in all feedback settings.

Numerical results demonstrate effective load tracking with thermostats and electric vehicles.

Performance varies with feedback type, showing robustness of the proposed methods.

Abstract

We use online convex optimization (OCO) for setpoint tracking with uncertain, flexible loads. We consider full feedback from the loads, bandit feedback, and two intermediate types of feedback: partial bandit where a subset of the loads are individually observed and the rest are observed in aggregate, and Bernoulli feedback where in each round the aggregator receives either full or bandit feedback according to a known probability. We give sublinear regret bounds in all cases. We numerically evaluate our algorithms on examples with thermostatically controlled loads and electric vehicles.