In this paper, a performance improvement scheme for an autonomous AC microgrid in the presence of dispatchable and renewable units, and Plug-in Hybrid Electric vehicles has been proposed. In a standalone mode of operation, a droop-control strategy is implemented at the primary control level for a communication-free accurate power-sharing among multiple dispatchable generators. For a network that is neither highly reactive nor highly resistive, active and reactive powers are coupled to both voltage magnitudes and phase angles. A mixed-droop feature is incorporated at the primary control level for autonomous power-sharing among the units for such networks. The mixed-droop feature is considered in this paper. Further, stochastic models of load demands, renewable generation, and plug-in hybrid electric vehicle charging loads are considered. Objectives are to reduce the cost of operation and minimize the emission of greenhouse gases from generation activities. The bi-objective problem is solved in the fuzzy domain. Performance improvement is achieved in two steps, first, by choosing optimal static droop constants using the particle swarm optimization, and then a further performance improvement is achieved by network reconfiguration. The proposed algorithm is tested on a thirty-three bus autonomous microgrid and gives satisfactory results.