Multi-stage algorithm of estimation and fault diagnose for closed-loop MIMO gas turbine engine systems

Aircraft engines have characteristics with complex structure and prone to fault, in addition, conventional modeling methods based on mechanism analysis has troubles of complicated process and low operational efficiency. To solve these problems, an algorithm was proposed for identification and fault diagnose based on data, which could be used in closed-loop MIMO aircraft engine systems. Using the algorithm, external input and output data of running closed-loop system measured online by sensors and data acquisition system were needed without understanding complexity construction and operation process. By choosing appropriate auxiliary parameters, auxiliary models were established. Then time delay was calculated by determining the order of the auxiliary model, parameters of forward channels and feedback channels were estimated by multi-stage least square algorithm. An illustrative example of biaxial gas turbine engine system was given for simulation. At the same time, the algorithm can be used for diagnosing gas path fault when the forgetting factor was changing by convergence of recognition result, the fault values can be got at last. Finally, two kinds of important characteristics were introduced, about importance and principium of selecting forgetting factor and excluding disturb of noises by distinguish it by the result of fault.