Performance Comparison of No-preference and Weighted Sum Objective Methods in Multi-Objective Optimization of AVR-PSS Tuning in Multi-machine Power System

Simultaneous optimization of controllers in power systems is a challenging research due to the inherent nonlinearity of such a system. Multi-objective optimization is a useful tool for tuning excitation controllers and minimizing oscillations that are described through definition of transient and small-signal stability in power systems. In this paper, a Two-Area-Four-Machine (TAFM) power system model is tested on multiple short circuit and load disturbances. A multi-objective performance analysis is investigated by observing the system's behaviour in different cases involving the no-preference method and a priori method called weighted sum objective. The analysis is done through observation of two different objective functions. First objective function includes the sum of the integral of time-weighted absolute errors of rotor speed differences, generator voltage, and tie-line power transfer. Second objective function observes time domain elements: overshoot, undershoot, and settling time of machines' rotor speeds. Results are compared for two methods combined with four different algorithms to provide better insight into the computational performance of each algorithm and objective search method. Algorithms used for controllers' parametrization include two novel algorithms: multi-objective ant lion optimizer (MOALO) and salp swarm algorithm (MOSSA), and two classic algorithms: multi-objective particle swarm optimization with velocity relaxation (MOVRPSO) and simulated annealing (MOSA).