Parallel intelligent search for loss minimization in distribution systems

Abstract

The problem of obtaining a network configuration of minimum energy losses for electric power distribution systems is addressed. It can be regarded as a generalization of the minimum spanning tree problem, where edge costs vary as the configuration changes. A solution is found with a recursive two-step procedure: the constraint of radial operation is relaxed in the first step, leading to an optimistic solution (a lower bound); information from this approximate solution is used in the second step to approach a feasible optimal solution. Nonlinear network flow optimization techniques team with search strategies from the field of artificial intelligence to cope with computation intractability. Parallel processing speeds the search of optimal solutions. A case study sheds light on the possibilities and limitations of the procedure.