Diagnosis and remedy of fault in a gas turbine with neural network and adaptive control

Meisam FATAHİ, Hamed KHODADADİ
1.668 370

Abstract


Abstract. This article is based on a design method of (neural network) in two different modes with the development and application of artificial neural network technique to diagnose (identifying and modeling) using an levenberg-marquardt algorithm and 4 conmon fault applied to gas turbine and then an exact detailed model of the procurement  process and its control units provided and by comparing the output with the same variables in the nerrous pulsing estimates set the modeling and simulation of extraction and the difference they can be identified as residual .And in the next step using the remedy instrument (control instrument maker ADRS) to reach on optimal response in a sensitive , complex and nonlinear gas turbines in order to protect it against the fault progress of gas turbine that leads to adverse events and occurrence of break down that be great deal . the result of simulation of optimum  performance shows the proper function of proposed controller


Keywords


Gas turbine, fault diagnosis , levenberg-marquardt algorithm, compensation ADRC

Full Text:

PDF