This paper proposes a cooperative demand response scheme for load management in smart grid. The cooperative demand response scheme is formulated as a constrained optimization problem that generates a Pareto-optimal response strategy profile for consumers. Comparing with the noncooperative response strategy (i.e., Nash equilibrium) obtained from the oneshot demand management game, the Pareto-optimal response strategy reduces the electricity costs to the consumers. We further develop an incentive-compatible trigger-and-punishment mechanism to avoid the noncooperative behaviors of the selfish consumers. Furthermore, the cooperative demand response scheme is applied to load management of industrial refrigerated warehouses. To implement the cooperative demand response scheme in large-scale system, we divide the refrigerated warehouses into different clusters and implement the cooperative demand response scheme inside each cluster. Numerical results demonstrate that the cooperative demand response scheme can reduce the electricity costs, drop the electricity prices, and curtail the total energy consumption comparing with the noncooperative demand response scheme.

This paper provides a study of the optimal scheduling of building operation to minimize its energy cost under building operation uncertainties. Opposed to the usual way that describes thermal comfort using a static range of air temperature, the optimization of a tradeoff between energy cost and thermal comfort predicted mean vote (PMV) index is addressed in this paper. In order to integrate the calculation of the PMV index with the optimization procedure, we develop a sufficiently accurate approximation of the original PMV model which is computationally efficient. We develop a model-based periodic event-triggered mechanism (ETM) to handle the uncertainties in the building operation. Upon the triggering of predefined events, the ETM determines whether the optimal strategy should be recalculated. In this way, the communication and computational resources required can be significantly reduced. Numerical results show that the ETM method is robust with respect to the uncertainties in prediction errors and results in a reduction of more than 60% in computation without perceivable degradation in system performance as compared to a typical closed-loop model predictive control.