A decentralised bi-level control approach to wind power regulation via thermostatically controlled loads

This paper studies the wind power regulation problem by controlling a population of thermostatically controlled loads (TCLs) in smart grids. A decentralised bi-level control approach is proposed, where the first-level controller achieves load following based on wind power forecast, while the second-level controller deals with the forecast error effectively using the frequency deviation information. Specifically, to realise the fair dispatch of wind power in a users’ comfort sense, the wind power following problem is formulated as a quadratic optimisation problem, which is solved by the first-level controller. The second-level adopts a local proportional controller to handle the frequency deviation caused by the forecast errors of wind power. The proposed algorithm converges fast and is decentralised in the sense that the TCLs conduct local computation and keep the parameters’ privacy from the aggregator. Simulations are given to show the performance of the proposed approach.