Neurocomputing 70 (2007) 2902–2912 Design and comparison of adaptive power system stabilizers based on neural fuzzy networks and genetic algorithms Jesu´s Fraile-Ardanuy à , P.J. Zufiria SISDAC Group (Grupo de Sistemas Dina´micos, Aprendizaje y Control), Polithecnic University of Madrid, Madrid, Spain Available online 21 May 2007 Abstract This paper presents two different power system stabilizers (PSSs) which are designed making use of neural fuzzy network and genetic algorithms (GAs). In both cases, GAs tune a conventional PSS on different operating conditions and then, the relationship between these points and the PSS parameters is learned by the ANFIS. ANFIS will select the PSS parameters based on machine loading conditions. The first stabilizer is adjusted minimizing an objective function based on ITAE index, while second stabilizer is adjusted minimizing an objective function based on pole-placement technique. The proposed stabilizers have been tested by performing simulations of the overall nonlinear system. Preliminary experimental results are shown. r 2007 Elsevier B.V. All rights reserved. Keywords: Fuzzy logic; Genetic algorithms; Power system stabilizer; Neural networks; ANFIS 1. Introduction Stability of power systems is one of the most important aspects in electric system operation. This arises from the fact that the power system must maintain frequency and voltage levels, under any disturbance, like a sudden increase in the load, loss of one generator or switching out of a transmission line, during a fault. The stability of the system determines whether this system can settle down to a new operating point after the transients disappear [35,43]. The use of high performance excitation systems is essential for maintaining steady state and transient stability of modern synchronous generators and provides fast control of the terminal voltage. However, these fast acting exciters, with high gains, can contribute to oscillatory instability in the power system. This type of instability is characterized by low frequency oscillations which can persist or even grow in magnitude [12,32]. Several real examples have been recorded and studied [7,52]. Power system stabilizers (PSSs) are used to generate supplemen- tary control signals for the excitation system in order to damp these oscillations. The conventional PSS (CPSS) was first proposed in the 1960s [12]. These devices are designed using the theory of phase compensation in the frequency domain and are introduced as a lead–lag compensator [37]. Other authors have presented PI—PID PSSs with the characteristic feature of being relatively simple for practical implementa- tion [29,30]. The parameters of CPSS are determined based on the linearized model of the power system. Usually, the operating condition where synchronous generator operates most of the time is chosen to adjust the CPSS and, its parameters, such as gains and time constants, are fixed to ensure its optimal performance at this specific operating point. In daily operation of a power system, the operating condition may change as a result of load variation and major disturbances, making the dynamic behavior of the power system to become different, at different operating points. These effects of the machine loading, on the synchronous-generator dynamics can be evaluated by means of an analysis of the linearized model’s eigenvalues [17]. Fixed-parameters CPSS can guarantee stability over a wide range of system loading conditions [1–3,20,36], but it is clear that, if the parameters of the stabilizer are kept fixed, it cannot maintain the same quality of performance whenever the operating point changes. ARTICLE IN PRESS www.elsevier.com/locate/neucom 0925-2312/$ - see front matter r 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.neucom.2006.06.014 à Corresponding author. Tel.: +34 913365354; fax: +34 913366764. E-mail address: jefar@ieee.org (J. Fraile-Ardanuy).