Optimal Thermostatically Controlled Loads Coordination Under Stochastic Renewable Energy

This paper investigates the multi-time horizon coordination of thermostatically controlled loads (TCLs) in a neighborhood area, where each TCL provides demand response (DR) services under unknown distributions of renewable energy (RE). We design a distributed scheme based on a designed consensus real-time price such that each TCL user minimizes his electricity payments when satisfying the comfort requirement. During the process of achieving the optimal coordination of TCLs, on one hand, TCLs face spatial coupling constraints generated by the feeder capacity limits in the neighbourhood network, which are decoupled by the so-called Lagrangian multiplier. On the other hand, unknown information such as expectations or moments of renewable generation restricts the development of gradient-based algorithm. An accelerated distributed zeroth order (ADZO) algorithm with heterogeneous smoothing parameters is devised, which updates by the batch oracles instead of the gradient, to achieve the adjustment of TCLs in a distributed way. Numerical simulations demonstrate the effectiveness of the proposed method.