Application of Distributed Constraint Optimization in Power Network Problems

capable of operating independently in the absence of a main grid. MG could be considered as one of the most promising key technologies in ensuring the adequacy of future green energy demands, as the distributed energy resources in a MG are primarily renewables. Customer-driven micro-grid (CDMG) is a relatively new paradigm in which utility-compatible generation sources are installed by the customers in their homes or facilities. CDMG is a step towards placing some control of energy delivery in the hands of the customer, which inherently makes the management of CDMG a distributed control problem. Moreover, distributed control promises a more secure, robust, cheap, and environmentally friendly power supply. CDMGs can be evolved in distribution system of the utility itself if proper control and communication are made available. The management and control of MGs create many challenging problems to power system researchers, such as active and reactive power control, islanding, voltage and frequency control, etc. So far, power researchers have been trying to solve these issues independently as separate MG problems. However, these problems are actually related to one another. For example, the choice of which island (a disjoint sub-grid in the MG) to create in the islanding problem affects the amount of available power in the island (in the power control problem). NMSU power researchers have recently developed a comprehensive optimization formulation that solves a majority of these problems together as a whole, thus eliminating the need to solve them separately. Moreover, global optimization is needed in a MG to minimize the total power consumption of MGs and reduce as much as possible the distribution losses, which one is not able to do by solving the problems separately. We are proposing to solve this formulation by modeling it as a Distributed Constraint Optimization Problem (DCOP). DCOP is a recently proposed problem solving paradigm, introduced by distributed artificial intelligence researchers. In this presentation, we will provide a short introduction to DCOPs along with our accomplishments and future challenges.