MINIMAX OPTIMIZATION FOR POWER SYSTEM CONTROL: A MULTIPLE OBJECTIVE APPROACH (RELAYS, STABILIZIER, COORDINATION)

In this work, the classical minimax optimization problem formulation is extended for the consideration of multiple perturbations by introducing multiple objectives. This extended minimax technique permits the use of multiple objective optimization concepts and algorithms to solve the problem of tuning control systems where the performance index as well as the system structure depend on uncertain parameters or perturbations. The controller structure is assumed to be fixed and specified a priori.; The proposed approach is tested with two study cases in the power system field: the coordination problem of protective devices, giving special emphasis to the case of directional overcurrent relays, and the problem of tuning power system stabilizers. Both problems are solved for the particular examples using a multiple objective weighting technique.; The coordination problem of directional overcurrent relays is stated as an optimization problem and solved using a partitioning scheme based in the type of the settings. For each subproblem, the optimal time dial settings are determined using the simplex method of linear programming and the optimal pickup currents are determined using the generalized reduced gradient method. The sum of the operation times is to be minimized as a global objective. This is accomplished by coordinating the subproblem solutions using a Gauss-Seidel type algorithm which shows good convergence characteristics for the particular examples. Alternate partitionings of the problem, based on the power system configuration and fault locations are also used. A restriction coordination procedure is used to iterate the subproblem solutions, showing rapid and reliable convergence. This contribution opens the field of optimization theory to those engineers working in power system relaying, aiming to reduce and eventually eliminate the need of interactive algorithms for coordinating power system protective devices.; The method developed in this thesis is also used to study the problem of power system stabilizer tuning. The approach is to use parameter optimization algorithms based in functional evaluations. In particular, the algorithm proposed by Hooke and Jeeves is tested, showing good performance for a 241 bus system with 34 equivalent generators.; The proposed technique is proved to be a new simple and practical tool, which applies to a broad spectrum of systems.