Data-driven Based Monitoring Of Gear Health Condition

The process of gear transmission mainly occurs in the gearbox.Gearbox is the core component of turbofan engine.It has high transmission efficiency,compact structure,high reliability and long service life and can achieve high torque output in limited space.Therefore,the reduction gearbox has been widely used in aeroengine.As the power source of the turbofan,the gearbox has high operating strength and bad operating environment,which lead to a high failure rate.The reliability of gearbox directly affects the performance and safety of the aircraft.So,it is of great practical significance to study the fault monitoring of gearbox of turbofan engine.This paper takes the gearbox as the research object and systematically studies the fault monitoring method based on the batch process from division and synchronization of the stage of gear transmission process.Firstly,aiming at the problem of unequal length phase in the process of gear transmission,a phase division method based on sliding window is proposed.That is,the unidentified batch samples are projected to the feature space of the adjacent window data to judge whether the data feature has obvious variation.In this way,the phase division point can be determined.According to the characteristics of uneven phase over batches,this paper divides phase recognition process into two steps: stable period recognition and transition period recognition.By adjusting the window adaptively,uneven phases can be accurately identifiedSecondly,aiming at the problem that the batch trajectories are not synchronous in the process of gear transmission.The synchronization model and the corresponding fault diagnosis model based on dynamic time warping(DTW)are proposed.Two of them are focused on,one is the selection of target length in synchronization.For this issue,the optimal synchronization length can be calculated by the statistical distribution,which makes the length of phase model closer to online condition.Another is the estimation of missing data in synchronization.Considering the process characteristics and the statistical distribution of variables,a principal component analysis(PCA)model based on sliding window is proposed to estimate the missing data accurately.Finally,the phase division and synchronization method is tested on the gearbox experiment platform to verify the applicability of the methods in the gear transmission process.Compared with the traditional methods,the experimental results show that the method proposed in this paper has good performance in gearbox fault monitoring.