Research On Fault Diagnosis Method For Transmission Shaft Of Helicopter Transmission System

As a common aircraft,helicopters are widely used in national defense and the national economy due to their advantages such as hovering in the air and vertical takeoff and landing.The transmission system is an important component of helicopters,and as one of the three important components of helicopters,its safety margin is the smallest.Therefore,it is crucial to study the reliability of the transmission system.Based on this research background,this article mainly explores the dynamic characteristics of misalignment faults in the transmission shaft shaft system.Starting from this,neural networks are used to identify structural damage features based on vibration response signals,and the reliability of the method is verified through ground experiments.Firstly,this article establishes a simplified dynamic model of the misaligned transmission shaft,which is transformed into an eccentric rotating thin-walled cylindrical shell model.Based on the Flugge nonlinear shell theory,considering the influence of centrifugal force,Coriolis force caused by rotation and initial stress after eccentricity,the vibration differential equations of eccentric rotating thin-walled cylindrical shells are derived by Hamilton variational principle.Since the traditional assumed mode method cannot solve the traveling wave vibration problem of rotating cylindrical shells,a method of expanding assumed mode functions is proposed for the first time to solve this problem,and the influence of Coriolis force generated by rotation on the structural state matrix is clarified.Simultaneously applying finite element method to analyze the vibration characteristics of rotating cylindrical shells under different boundary conditions.By comparing the results with literature,the correctness of the established model,solution method,and calculation program was verified.This paper investigates the effects of aspect ratio,thickness to diameter ratio,rotational speed,and boundary conditions on the natural frequency of eccentric rotating thin-walled cylindrical shells.The influence of eccentricity on the natural frequency and mode shape is analyzed by plotting the structural modal shapes under different eccentricity states.On this basis,by analyzing the eigenvectors of cylindrical shells under eccentric conditions,it is clarified that the localized modes generated by eccentricity are formed by the superposition of uniform modes with different specific gravity.By conducting block wise research on the structure,the causes of modal deformation distribution were explained.Based on the phenomenon of modal localization,the vibration response of eccentric cylindrical shells was studied as one of the basis for fault diagnosis.Finally,a set of transmission shaft fault diagnosis program based on BP-LM neural network was established.The time-domain response of a point on the transmission shaft shell under six different faults was obtained through simulation,and the input samples were obtained through preprocessing and feature extraction of the original signal.Introduce the BP-LM neural network algorithm to train input samples and obtain a fault diagnosis model.Afterwards,ground tests were conducted to obtain measured time-domain response data of three common faults at a point on the simplified transmission shaft shell.By constructing a neural network training model,the identification of measured faults was achieved,and the correctness of the proposed method in this paper was verified.