Research On Health Assessment And Fault Prognosis For Aircraft Electromechanical System

In recent years, in order to reduce maintenance manpower, increase sortie rates, achieve condition based maintenance and autonomous security, the technique has developed from traditional status monitoring and fault diagnosis technology to health assessment and the failure prognosis. Aircraft health assessment and fault prognosis achieve fault monitoring, diagnosis, prediction, condition assessment and the capabilities of integrated decision-making. These can also reduce aircraft maintenance costs, improve aircraft readiness, mission success rate as well as security and availability. Aircraft electromechanical systems is an important system in aircraft structure whose failure accounts for a large proportion of the entire aircraft failure, it not only affects the operation performance of the aircraft, but also greatly reduces the reliability of aircraft operations, causes great threat for aircraft safety. Failure of aircraft electromechanical systems is often characterized by concealment, randomness, diversity, complexity and uncertainty. The use of a single intelligent fault diagnosis technology often has the disadvantage that diagnostic accuracy is not high, the ability of generalization is not strong, training efficiency is low. New ideas and methods to solve practical problems on these projects are needed. So the research of aircraft electromechanical system health assessment method and fault prognosis has a great significance.The HMM basic theory and algorithms are studied and researched in this thesis, on the one hand, the practical function of three classical HMM algorithm in fault prediction is explored, and the MATLAB program is compiled, the feasibility of HMM theory in Rolling Bearing Fault Diagnosis is verified. On the other hand, for the failure prognosis, using the genetic algorithm to train the optimal HMM models to predict failure and using fault evaluation index to evaluate this method, the experimental results demonstrate the effectiveness of fault identification and prediction proposed in this paper.The theoretical basis of particle filter algorithm is studied, the method of particle filter is introduced, the application of particle filter in the signal noise reduction is researched. The rolling bearing common faults, failure mechanisms, signal characteristics and common fault diagnosis methods are deeply studied. Damage and vibration signal acquisition is done on the rotor test bench. Rolling fault identification method based on the improved particle filter is proposed. The simulation for rolling failure is made by using the method which is mentioned above. The results prove the accuracy and effectiveness of the method.The basic principles of EEMD, AR model and its application are respectively introduced, the diagnosis method of analysis of the bearing inner ring fault based on EEMD and AR spectrum is proposed. Experimental verification is made. The analysis results of experimental signals of rolling inner fault indicate that: Compared with the traditional spectral analysis method, the analysis method based on EEMD AR and AR spectral has obvious advantages. It greatly improves the signal to noise ratio, can identify rolling bearing faults more effectively.