DISTRIBUTED CONTROL AND OPTIMIZATION TECHNIQUES FOR DEMAND-SIDE MANAGEMENT UNDER UNCERTAINTY IN RENEWABLE POWER GENERATION

Uncertainty in the renewable energy source raises the difficulty to achieve an accurate energy scheduling within the electricity distribution network. In this work, we consider a smart grid model where the users behave selfishly, seeking to reduce their total expenses in the presence of uncertainty wind speed forecast. We implemented a demand-side management model, where the active users can own both private generation and storage devices. These prosumers participate in modifying their energy scheduling, considering their local preference and the global grid constraint. The resulting problem is a non-cooperative Nash game, whose equilibrium is reached using a rolling-horizon semi-decentralised DSM algorithm. For every time slot, we defined the energy price that the users pay to the grid as a function of the current aggregate load and wind power production. Besides the rolling-horizon approach, to handle the inherent uncertainty in wind power production, an approximated stochastic optimisation framework is employed. We generate several wind power production scenarios from a defined probability density function, defining an approximate stochastic cost function. The overall method dynamically, at every time step, retrieve the most recent forecast and together with a weather instability index generate the probability density function for the wind power production; moreover, historical wind speed data are used to make the distribution reliable with high uncertainty weather conditions. Simulation results show that the proposed methodology, compared to the deterministic counterpart, exhibit a lower expense with a substantial benefit over competitors in diverse scenarios.