Parameter Optimization of Droop Controllers for Microgrids in Islanded Mode by the SQP Method with Gradient Sampling

Abstract

For enhancing the stability of the microgrid operation, this paper proposes an optimization model considering the small-signal stability constraint. Due to the nonsmooth property of the spectral abscissa function, the droop controller parameters’ optimization is a nonsmooth optimization problem. The Sequential Quadratic Programming with Gradient Sampling (SQP-GS) is implemented to optimize the droop controller parameters for solving the nonsmooth problem. The SQP-GS method can guarantee the solution of the optimization problem globally and efficiently converges to stationary points with probability of one. In the current iteration, the gradient of the nonsmooth function can be evaluated on a set of randomly generated nearby points by computing closed-form sensitivities. A test on the microgrid system shows that the optimality and the efficiency of the SQP-GS are better than those of the heuristic algorithms.