This paper is mainly concerned with the health estimation of a gas turbine using a symbolic linearization model approach. Health parameters will change with the degradation of gas turbine performance. Monitoring and evaluating these health parameters can assist in the development of predictive control techniques and maintenance schedules. Currently, various health parameter estimation methods have been studied extensively, but there have been less related studies on how to obtain statespace models. In this paper, a symbolic linearization model method is presented to overcome the shortcoming of high time consumption suffered by existing methods. In this method, each component model of the dynamic nonlinear gas turbine model is decomposed into several sub-modules, each of which contains a simple nonlinear equation. By means of symbolic computation, a linear model of the components is derived by linearizing these sub-modules, and then the generalized linear state-space model of the gas turbine is derived from the relationship among the components. In the generalized linear state-space model, the Jacobian matrices are functions of the parameters under a steady-state operating condition. Therefore, it is easy to obtain a linear model that represents the dynamics of the gas turbine under a given operating condition. To estimate the health parameters of a gas turbine, a piecewise linear model is developed using the proposed approach, and this model is verified in a simulation environment. The results show that the developed piecewise linear model can capture the behavior of a gas turbine quite closely. Then, a linearized Kalman filter is designed for estimating the health parameters under steady-state and transient conditions. The results show that the generalized linear model established using the presented method can be used to accurately estimate the health parameters of a gas turbine.
MODEL-BASED PREDICTIVE CONTROL OF A PRE-DENITRIFICATION PLANT: A LINEAR STATE-SPACE MODEL APPROACH
