This paper presents a method to design power system stabilizer(PSS) based on parametric identification modeling and the robust gain scheduling control theory. Operating conditions of large power system vary with network configurations and loading conditions in a complex manner. The proposed gain scheduling control is a successful design methodology for a time varying and/or a nonlinear system. The parameter dependency of the generator output as a scheduling parameter is discussed, which directly reflect fluctuations of operating conditions. Another issue in applying the advanced PSS to each generator in a large power system is how to design the each controller under the consideration with interconnection. In this paper, from view point of the decentralized control, the generator model to be controlled is obtained by using the partial parameter identification, which is represented by single-machine and infinite-bus system. The basic idea is that whole network system except designing generator is unknown. The Improvement of the transient performance is demonstrated for a variety of operating conditions and disturbances by performing some numerical simulation studies.