Microgrids, as emerging means for localized management, supervision, and control of energy production and consumption are changing the traditional centralized grid topology, making it more distributed and autonomous. However, the intermittent nature of renewable energy producers as major source of energy production in a microgrid, along with the fluctuation in energy demand, can destabilize it if not dealt with properly. Hence, there is a need for a comprehensive Energy Management System (EMS) for microgrids which can efficiently handle the loads and generation uncertainties. There exist several challenges in developing such an EMS for microgrid operation under uncertainties. 

One of the main challenges is providing the active participation of consumers for Demand-Side Management (DSM) in power imbalance situations. DSM is the main EMS tool to manipulate the consumption pattern on the customer side in case of unforeseen events caused by uncertainties in the microgrid. Most existing research uses utility functions to represent user’s priorities for certain appliances for their Demand-Response (DR) programs. However, there is no standard way to model the utility functions of various appliances belonging to different users. Researchers also tried to model the utility functions of the appliances based on their type or power consumption level, which in neither case represent the realistic consumption priorities of individual users. 

The other challenge is the implementation of the decision-making module in EMS. In most of the state-of-the-art EMS, the decision-making and optimization process are solely done by a central entity based on the detailed patterns of users’ consumption, generation data, weather forecast, etc. Besides the privacy concerns and massive computational burden which are the common drawbacks of centralized control schemes, the users’ inconvenience due to load curtailment is usually neglected. In other words, there should be a structure that allows the active participation of users in the demand-response programs considering their realistic consumption preferences.

However, the participation of the users in the decision-making process indeed is another source of uncertainty and complexity of EMS that remarkably cause high computational cost if a close to optimal operation is required. Therefore, novel metaheuristic optimization approaches along with the proper uncertainty modeling technique to deal with that complexity is another challenge in the development of microgrid EMS. Here, the state-of-the-art suffers from major simplifications and assumptions about generation and demand models which neglect the effects of the uncertainties in microgrid operation.

The focus of this thesis is on the development of an EMS for microgrids that effectively handles the loads and generation uncertainties, on the one hand, by optimizing the microgrid operation and, on the other hand, by deploying proper Demand-side Management approaches. The incorporation of a realistic consumption model and the provision of novel algorithms to support the active participation of the users in the decision-making processes are the main contributions of this research. Moreover, the optimization challenges that imposed to the new microgrid EMS are addressed and the negative effects of uncertainties on microgrid operation are compromised by taking novel measurements and optimization techniques. 

To the best of our knowledge, this research is the first that provides and implements a comprehensive multi-agent-based EMS, incorporating a realistic consumption preference model with an interactive decision-making schemes and specific metaheuristic optimization approaches for a comprehensive and effective microgrid operation under uncertainty.