Key Technologies Of On-board Health Diagnosis For Civil Turbofan Engine

Engine health management (EHM) is a prerequisite to achieve condition based maintenancefor aircraft engines. EHM can solve the contradiction between engine security and economicaffordability. With continuing advances in avionics, research focus of EHM is developingfrom off-board to on-board. The on-line health diagnosis, as the key technology of on-boardEHM, has been widely researched by both domestic and foreign scholars. In this paper, thekey technologies of on-board health diagnosis for civil turbofan engine are researchedaccording to the requirements of developing the EHM system for the domestic civil turbofanengine. This research is funded by the programs named “Civil Turbofan Engine On-boardHealth Diagnostic Techniques” and has important theoretical significance and applicationvalue.Firstly, the component-level modeling method based on GasTurb/MATLAB is researched.According to the overall performance demand of the domestic civil turbofan engine, the highbypass ratio turbofan engine nonlinear model and associated linear modeling tools aredeveloped for the EHM research. The original program and developed program arerespectively simulated with the same inputs. Comparative results verify the correctness of themodel.Secondly, it is pointed out that continuing developments in avionics are enabling themigration of portions of the conventional ground-based diagnostic functionality on-board. Anintegrated on-board diagnostic architecture composed by the on-board adaptive model andon-line fault diagnostic system is presented for the civil turbofan engine on-line healthdiagnosis. The simulation results show that, the benefits of using the on-line real-time datainclude the real-time continuous monitoring of engine health, the estimation of engineperformance parameters, the early diagnosis of fault conditions and a greater probability ofdetecting intermittent faults.Next, based on hybrid Kalman filter (HKF)，an algorithm of a bank of HKF (BHKF) and afusion algorithm of HKF and back-propagation neural network (HKF-BPNN) are proposedfor on-line diagnosis. BHKF and HKF-BPNN inherit the system architecture of HKF whichallows the reference health baseline of BHKF and HKF-BPNN to be updated to the healthcondition of degraded engines. BHKF and HKF-BPNN extend the application range of HKF to not only fault detection but also fault isolation. The performance of the two algorithms isevaluated in a simulation for diagnosing the four common faults of the civil turbofan engine.The simulation results show that the two algorithms can detect and isolate the four commonfaults at both the healthy and degraded conditions.Subsequently, a public testing approach based on Monte Carlo simulation is proposed for thefault diagnosis algorithms of the civil turbofan engine. The quantitative performanceevaluation indicators including the rates of fault detection, fault isolation and false alarm canbe obtained to evaluate and compare the fault diagnosis algorithms. BHKF and HKF-BPNNdiagnosis algorithms are tested using the approach. The testing results show that thediagnostic capability of HKF-BPNN is better than BHKF.Finally, a hardware platform for real-time simulation of civil turbofan engine health diagnosisis constructed using rapid prototype method. The overall structure and each part of thesimulation platform are described respectively. The development process of hardware andsoftware is showed. The real-time simulations of the integrated on-board diagnosticarchitecture and its algorithms are carried out. Simulation results prove that the integratedon-board diagnostic architecture can run in real time and work correctly.