Research On Method Of Civil Aircraft System Fault Diagnosis And Prediction Based On Status Monitoring Data

With the rapid development of China’s civil aviation industry,the number of civil aircraft is increasing,and the safety of civil aviation is becoming more and more serious.The “safety gateway forward” and “continuous safety” have become the development goals of the current air transportation system.Operational efficiency and safety have become major research problems and are urgent needed in the civil aviation field.Modern civil aircraft is usually equipped with an aircraft condition monitoring system.The operating parameters of each subsystem and component are collected through the on-board data bus.The status,performance,operating environment and load of the recording system are monitored online and transmitted to the ground for real-time analysis or stored on the airborne device(such as QAR,Quick Access Recorder),for offline analysis after navigation.Mining flight data and integrating expert experience,traditional reliability data,inspection data and system/component mathematical models for aircraft complex system fault diagnosis,prediction technology research and application,and achieving data-driven operation and maintenance decision support have become the current needs of China’s aviation industry.Therefore,from the application needs of civil aviation engineering in China,this paper carries out related researches on the new maintenance mode,fault diagnosis and prediction key technologies of civil aircraft complex systems based on flight data.In this paper,based on the full analysis of the application of flight data in the field of aviation application development and related key technology methods,combined with the domestic flight data in the domestic civil aircraft maintenance engineering practice fault diagnosis and prediction problems.The QAR data is used in the system of abnormality detection,residual life prediction modeling,fault isolation and planned maintenance,and the predictive maintenance mode based on continuous monitoring is studied in depth which provides a feasible technical method reference for fault diagnosis and prediction,and can also provide technical reserves for the development of domestic civil aircraft PHM system.The research work of the thesis is as follows:(1)Research on classification method of flight data work conditionsDue to the influence of many factors,such as multi-flight profile of aircraft,different operation condition of related system and combination working of various components of the system.The flight data of civil aircraft complex system are characterized by variability and complexity,and the flight data are often incomplete and multi-working conditions,which is difficult to meet the requirements of fault diagnosis and prediction of aircraft system directly.In order to extract the high quality and reliable data of anomaly detection,fault diagnosis,RUL prediction and prediction maintenance from flight data,the filling of flight data missing value based on multiple interpolation technology and the classification method of multi-working condition flight data based on decision tree c5.0 classification algorithm are studied according to the characteristics of flight data.(2)Research on fault detection method of the civil aircraft complex systemThere are multiple flight phases in a flight cycle of a civil aircraft.Complex systems often have different working conditions in different flight phases.The time series of system performance monitoring data in different flight cycles and flight phases in QAR data are different between different flight envelopes and noises.There will be timeline scaling,discontinuity,amplitude shifting,and scaling,and most monitoring data of the system performance in QAR data is normal system performance data.Aiming at solving these problems,the DTW clustering,Exponentially Weighted Moving Average(EWMA)control chart method and the fault detection method of multiple linear regression model are proposed to realize fault detection of the civil aircraft complex system suitable for QAR data.(3)Fault Diagnosis of Civil Aircraft System Based on Bayesian NetworkThe flight data truly reflects the working conditions and system performance of different complex systems under different flight conditions,providing an objective basis for maintenance support,assessment of flight training quality and investigation and analysis of flight accidents.When the function of the aircraft system/component is abnormal or failed during flight and the ground is normal,and the crew is unable to provide sufficient system/component abnormality or failure information,only based on the troubleshooting manual,maintenance manual and expert experience.It is difficult to analyze and isolate faults.Aiming at this situation,the fault isolation research of civil aircraft system based on the combination of troubleshooting manual and maintenance manual and flight data is studied.The fault diagnosis process of civil aircraft system based on knowledge and flight data is designed.The Bayesian network is proposed.The fusion of civil aircraft troubleshooting manual,maintenance manual,expert experience and other methods of fault isolation of flight data,the feasibility of the method is verified by the case of bleed air system.(4)Research on modeling method of health state and residual use life predictionThe degradation data selection and health index determination method for the remaining service life prediction of the complex system of the civil aircraft based on the flight data are studied on the basis of the above-mentioned classification and the abnormality detection of the flight data.The paper studies the modeling method of the state space and the joint estimation method based on the state and parameters of the Bayesian framework for the uncertainty representation and management of the RUL prediction of the civil aircraft system.Taking the performance degradation of the pneumatic system of the civil aircraft as an example,the degradation data is selected by using the EWMA control chart abnormality detection limit,and the HI of the degradation of the temperature control performance of the pneumatic system is determined by using the mixed measurement method of the health index,and then the performance degradation state space model is established.The effectiveness of this method is verified by the joint estimation of the state and parameters of the Bayesian framework to the RUL prediction and the reliability evaluation of the three sets of different degradation time flight data.(5)Preventive maintenance strategy based on continuous monitoringThe use of flight data for performance degradation prediction and reliability estimation of civil aircraft systems presents an opportunity for existing maintenance planning(MRB)-based civil aircraft maintenance optimization,and analyzes the new logic of MSG-3 maintenance tasks integrated into PHM.The PHM task is candidated under the process,but the in-depth study of the PHM task engineering analysis process and method is lacking,which leads to the difficulty of predictive maintenance in practical application.Therefore,the maintenance task engineering analysis process and method suitable for predictive maintenance need to be established.Aiming at this problem,this paper puts forward the basic idea of preventive maintenance mode based on continuous monitoring,analyzes the new logic and process of MSG-3 maintenance task analysis integrated into PHM,designs the process of predictive maintenance,and studies the method of application of PHM fault diagnosis and prediction results to civil aircraft maintenance.Then these effects are verified by the bleed air system.The new maintenance mode can effectively reduce unplanned maintenance events,improve aircraft utilization,and has better economics.