Comparative study between subspace method and prediction error method for identification of gas turbine power plant

This paper aims to introduce generalized statespace models of power plant gas turbine system. Once verified, those models can be used in system real-time monitoring and control of power plants. The models are of generalized statespace structure and their parameters are optimized from normal operating records first using subspace identification method. Then those results are compared with the identified responses using prediction error method. Simulation results have shown corresponding advantages of simplicity and following the main dynamic variations of the system for the proposed models with some essence differences between the two methods. Also, residual analysis has been run to confirm the models validity.