Bi-level particle swarm optimization and evolutionary algorithm approaches for residential demand response with different user profiles

The deregulation of electricity retail markets requires the development of new modeling approaches for the optimal setting of dynamic tariffs, in which consumers' responses according to their flexibility to schedule demand are considered. Retailers and consumers have conflicting goals: the former aim to maximize profits and the latter aim to reduce electricity bills. Also, there is a hierarchical relation between them, as retailers (upper-level decision makers) determine the pricing strategy and consumers (lower-level decision makers) react by scheduling their loads according to price signals and comfort requirements. This is a bi-level optimization problem. In this paper, typical residential loads are considered and three scenarios of feasible windows of appliance operation are established. Two new population-based approaches, an evolutionary algorithm and a particle swarm optimization algorithm, are developed to solve the bi-level problem. The results obtained are then compared with a hybrid algorithm that solves the lower-level problem exactly.