Trajectory sensitivity and genetic algorithm based-method for load identification

Load identification is an important issue in power system representations to ensure that simulations will reproduce the dynamic response of a system during a disturbance. For a load model to be accurate, its parameter must be appropriately estimated by a parameter fitness algorithm. The success of the estimation depends mainly on the availability of a good initial parameter guess. If it is not available, the estimation process takes plenty of time to converge or to diverge. This paper proposes a hybrid algorithm based on trajectory sensitivity and generic algorithm. The advantages of the fitness algorithms of Trajectory Sensitivity and Generic Algorithm are combined so as to provide a robust algorithm regarding the initial parameter guess that guarantees the convergence even in the case of unavailability of a good initial parameter set. The combined algorithm was tested in one hundred simulations, in which the initial parameter guesses were randomly generated between limits (parameter uncertainties) for the assessment of the robustness of the algorithm. The results show that in 99 cases, the proposed methodology converged to the true values in a short time.