Gas Turbine Model Parameter Classification During Abnormal Operation Using Support Vector Machine For Maintenance

Abstract

When there is a contingency in a major transmission system, it is crucial to locate and detect abnormal parameters using an accurately modeled gas turbine (GT) generator. In this paper, a new method was proposed to model, a GT generator system. Firstly, MATLAB/Simulink was used to rebuild GT-based model which was validated using a model in PSS/E. Secondly, an artificial intelligence (AI) based approach, namely Support Vector Machine (SVM) was used for problem recognition in GT power plant. The results showed that, under steady-state, the electrical power of the rebuilt GT model was the same as the model in PSS/E. The advantage of having the MATLAB / Simulink GT model was the ability to visually observe the output from each block which was not possible in PSS/E. In addition, the average training accuracy was over 90% for the detection of GT governor parameter during abnormal behavior. As an implication, the developed model should be considered to apply in real-world grids for operation engineers to detect abnormal governor parameters in GT in the next stage. In turn, it assists in the restoration of GT to its operating condition and minimizes troubleshooting time.