Abstract:In order to prevent the major economic losses caused by the failure of the gas turbine combustion chamber in the power plant during the working period,a fault early warning method based on Grey Relation Analysis-Deep Autoregressive Model (GRA-DeepAR) is proposed.Firstly,the normal operation data of the gas turbine are collected for GRA,and the characteristic parameters highly related to the turbine exhaust temperature are extracted; Then,DeepAR is used to predict the turbine exhaust temperature,and an early warning threshold is set to establish a deep autoregressive fault early warning model; Finally,the operation of the combustion chamber is judged indirectly according to whether the absolute value of the residual exceeds the warning line.Taking the operation data of Ansaldo gas turbine in a power plant as an example,the results show that the method can identify the abnormal conditions of the combustion chamber in advance,and send an early warning signal to remind the staff to deal with them,providing practical reference for the early warning of gas turbine combustion chamber failure.