ABSTRACT
GTs have been used widely in industrial plants all over the world. They are the main source of power generation in places such as offshore plants and oil fields, which are far away from urban areas. The key role of GTs in the developing industry has motivated researchers to explore new methodologies in order to be able to predict the dynamic behavior of these complex systems as accurately as possible. A variety of analytical and experimental techniques has been developed so far to approach an optimal model of GTs. Fortunately, black-box system identification techniques and specifically ANN-based approaches can effectively assist researchers who work in this field. The study in this area can be categorized into IPGT, aero, and low-power GT models [227,228]. ANN is one of the techniques that has played a significant role in system identification and modeling of industrial systems. This is due to its capability to capture dynamics of the systems without any prior knowledge about their complicated dynamical equations. Because of sophisticated and nonlinear dynamic behavior of GTs, significant attention still needs to be paid to the dynamics of these systems to unfold unknowns behind undesirable events during GT operation. As it can be seen, each research activity in the field of modeling of GTs investigated the issue from a specific perspective and had its own limitation. According to the methodology used in this chapter, various backpropagation training functions, different number of neurons, and a variety of transfer functions were employed to train the network in order to explore an accurate ANN model using an MLP structure. To increase the level of generalization for the model, the data sets were partitioned randomly for training, validation, and test purposes.
