International Journal of Computer Applications (0975 – 8887) Volume 41– No.16, March 2012 29 Performance Comparison of Invasive Weed Optimization and Particle Swarm Optimization Algorithm for the tuning of Power System Stabilizer in Multi-machine Power System Ashik Ahmed Assistant Professor EEE Department Islamic University of Technology, Gazipur, Bangladesh B.M. Ruhul Amin Lecturer EEE Department Bangladesh University Of Business and technology, Bangladesh ABSTRACT In this paper, two evolutionary algorithms- Invasive Weed Optimization (IWO) based power system stabilizer (PSS) and particle swarm optimization (PSO) based power system stabilizer is designed for multi-machine power system to compare their tuning performances. IWO is a derivative-free real parameter optimization technique that mimics the ecological behavior of colonizing weeds. PSO is also a derivative-free and flexible optimizer which is powered by the behavior of organism, such as bird flocking. Eigen-value based objective function is considered for the tuning of PSSs to enhance system damping of electromechanical mode. The performance of proposed IWO-based PSS and PSO-based PSS is tested and demonstrated under different disturbances for a four machine example power system. The Eigen value analysis and non-linear time domain simulation results shows that both IWO-based PSS and PSO-based design can successfully damp out the oscillations and thus improve the stability of the system. However, the abilities like faster convergence and greater shifting of critical modes to the left of s-plane keeps the choice of IWO based design in front of PSO based design for the system under consideration. General Terms Power system stability, Multi-machine system, Optimization. Keywords PSS design, Invasive Weed Optimization, Particle Swarm Optimization, Dynamic stability. 1. INTRODUCTION Newer approaches are being tried by the power system engineers to maximize power transfer among different areas in a stable manner. Usually the transmission networks are overdesigned to run the operation of the overall system safely. Constraints like thermal limit, rotor angle stability limit and voltage stability limit influences the consideration of safety limit [1]. Presence of low frequency oscillation among the interconnected group of generators creates serious threat to the proper operation of the overall system. Usually plant mode and interconnected mode of oscillations with frequency ranges of 0.7-2.0 Hz and 0.1-0.8 Hz respectively are observed [2]. Substantial efforts have been made to realize the impact of Power System Stabilizer (PSS) in damping low frequency oscillation and thus improve the small signal stability of power system [3]. The effectiveness of PSSs from both cost and operational point of view has ensured wide use of it by the utilities. Different modern control theory based approaches have been applied to PSS design problem. These contain optimal control, fuzzy & neuro-fuzzy control, variable structure control and adaptive control [4-7]. It is shown in [8] that appropriate selection of conventional PSS parameters result in satisfactory performance under system disturbances. Sequential and simultaneous tuning based approaches are used in [9]. In sequential design of PSS only one electromechanical mode is considered for damping among many available modes. It is shown that the optimal parameters designed for a certain mode in this fashion can create adverse effect on some other modes of the system and the overall performance may not be optimal. The PSS tuning problem is formulated as non-linear non differentiable optimization problem in [10] which is found to be very hard to solve using traditionally differentiable optimization algorithms. Several random exploration techniques like Genetic Algorithm (GA), Simulated Annealing (SA), Tabu Search (TS), and Evolutionary Programming (EP) has been successfully used [11-15] to optimize the PSS parameters. These techniques have gained acceptance because of their ability and effectiveness of searching an optimal solution in a problem space. A frequency domain based approach [16] is found to be more appropriate where the PSS design problem is formulated as a multi-objective optimization problem and GA is used to optimize them. Even though GA is found to be very satisfactory in searching global or near global optimal result of the problem, the long run-time constraint limits the use of this. Lately a derivative- free, meta-heuristic optimization algorithm named as Invasive Weed Optimization (IWO) is proposed [17] which imitate the ecological behavior of the colonizing weeds. Successfully utilization of IWO has been found since then in different optimization problems like tuning of a robust controller [17], optimal positioning of piezoelectric actuators [18], designing an E-shaped MIMO antenna [20], designing the encoding sequences for DNA computing [21], developing a recommender system [22] and studying electricity market dynamics [23]. PSO was introduced by Kennedy and Eberhart in 1995 [24]. In contrast to the traditional evolutionary algorithms, PSO keeps track of the information regarding both position and velocity of the particle [25]. Several upgraded PSO algorithms are implemented and applied in [26-27]. In this paper the optimizing capability of IWO and PSO is compared for designing the PSSs parameters for obtaining the