Parallel Optimization of Generation Capacity Structure in a Local Grid under Military Threats

We investigated the specific features of the problem of optimizing the structure of generation capacities in a Local Grid operating under military threats. We formulated this problem as a large-scale mixed-integer linear programming (MILP) problem. We proposed a method of decomposing this problem into interrelated subproblems: a first-stage master subproblem and a set of second-stage subproblems. For the independent solution of the second-stage subproblems, we used the SCIP solver, and for solving the master subproblem, we applied one of two algorithms: the Ant Colony Optimization (ACO) algorithm implemented in the MIDACO solver or the Evolutionary Centers Algorithm (ECA). The proposed decomposition method for optimizing the structure of generation capacities in a Local Grid was implemented in a parallel computing environment, where we conducted test experiments to compare the solution speeds of the ACO-SCIP and ECA-SCIP algorithms. The presented computational experiment results demonstrate the advantages of the solver implementing the proposed decomposition method with ECA-SCIP over the SCIP solver and the ACO-SCIP solver.