Gas turbine combustion profile modelling for predictive maintenance using an artificial neural network

Abstract. Dry Low Emission (DLE) gas turbine has been developed as a solution to encounter the harmful high NOx emission from conventional turbine. However, it is prone create Lean Blowout (LBO) error that causes frequent shutdown due its stringent condition needs be operate inside desired operating can monitored through temperature, and CO concentration. This paper develops an Artificial Neural Network – Multilayer Perceptron (ANN-MLP) predictive maintenance model using actual DLE data predict trips exhaust classification of warning stages on LBO error. 94.12% R2 for regression 100% accuracy Python obtained four months period data. proposed ANN-MLP manage suitable time real which help reduce cost lost unscheduled shutdown.