MAtheuristic Algorithm Based On Neighborhood Structure to Solve The Reconfiguration Problem of Active Distribution Systems

Abstract

The problem of reconfiguration for active distribution systems is formulated as a stochastic mixed-integer second-order conic programming (MISOCP) model that simultaneously considers the minimization of energy power losses and CO2 emissions. The solution of the model determines the optimal radial topology, the operation of switchable capacitor banks, and the operation of dispatchable and non - dispatchable distributed generators. A stochastic scenario-based model is considered to handle uncertainties in load behavior, solar irradiation, and energy prices. The optimal solution of this model can be reached with a commercial solver; however, this is not computationally efficient. To tackle this issue a novel methodology which explores the efficiency of classical optimization techniques and heuristic based on neighborhood structures, referred as matheuristic algorithm is proposed. In this algorithm. the neighborhood search is carried out using the solution of reduced MISOCP models that are obtained from the original formulation of the problem. Numerical experiments are performed using several systems to compare the performance of the proposed matheuristic against the direct solution by the commercial solver CPLEX. Results demonstrate the superiority of the proposed methodology solving the problem for large-scale systems.