The Gear Life Prediction Based On The Motion Vibration Signals

The gear as the key motion transmission of the machine, plays a critical role inthe stable operation. The life prediction of the operating gear is significant for thedetermination of the equipment overhaul period and the accidents preventing. Thetraditional life prediction method based on fatigue damage accumulation hypothesiswhich need to know the external loads and fatigue life curve, serving for gear designcan not used to estimate the gear life reduction because of the gear teeth broken. Thelife ending of gear due to the gear cracks, teeth broken and other faults, the cracks andbroken of teeth often beginning a subtle change from the gear micro-cracks andmaterial morphology, with some evidence tend to be reflected in the meshingvibration signals.In this study, we predict the life of the gear based on the fault diagnosis,take theresidual error signals as the analysis object, which represents the departures of TSAremoved the mesh frequency and its harmonics and usually shows evidence of faultsearlier and more clearly than TSA.And the researchs are as follows:1. The fault indicators based on the AR model and the Two-Sample K-S test.Design the AR model-based filte to calculate the residual error signals, get theprediction error signals of residual error signals. Two-sample K-S test can be used totest the prediction error signals of the normal gear with the prediction error signal ofthe needed gear. The test statistics D can be used to find the early crack and the faultindicators.2. The gear life prediction based on the gear fault indicators.Using theGray-Markov model as the prediction model. GM (1,1) model predict the overalltrend of the fault progress. Markov model correct the prediction errors.3. Design a gear fault diagnosis test-bed with power loop back and built gearvibration signal acquisition system based on LABVIEW, Collecting vibration dataused to subsequent analysis by multi-channel interval in the whole life cycle of thegear test. The full life cycle test is the gear from the normal state to complete failure.The result shows that, the fault indicators based on the AR model and theTwo-Sample K-S test can find the gear fault early and well reflect the gear faultevolving.The life prediction model based on the gray-markov can well forecast gearfault trends, facilitate remaining life prediction, the prediction method has certain engineering significance for the remaining life of gear online prediction.