Wear State Recognition And Residual Life Prediction Of Gearbox Based On Oil And Vibration Analysis

As a key component in the transmission system of chemical equipment,gear boxes are widely used in industrial production fields such as light industry and petrochemical,metallurgy and mining,chemical energy,fire protection,and ocean engineering.The stable and reliable operation of chemical equipment is closely related to the operating status of this component.However,due to long-term harsh working conditions such as overload and high pressure,gear boxes are extremely prone to failure,leading to major safety accidents and huge losses.Therefore,identifying the wear failure status and predicting the remaining life of the gearbox can not only avoid safety accidents,ensure its reliable operation,and master its operating conditions,but also provide a basis for predictive maintenance in the later stage and reduce the risk of equipment failure.Based on data driven,this paper conducts research on the operating status of gear boxes from both oil analysis and vibration monitoring.The main research contents are as follows:(1)The occurrence of early failures in gear boxes is often accompanied by changes in wear status,but the changes and causes of wear status are often difficult to effectively identify and judge.Therefore,this paper proposes a gear box wear state recognition model based on the limit value of oil element concentration.The model is based on understanding the principles of gearbox fault and wear identification,and uses the limit value principle to diagnose the operating state of gearbox wear based on oil detection data.Finally,based on the actual detection results such as ferrography analysis and contamination level,the wear status of the gearbox is comprehensively compared and verified,and the causes and locations of wear are refined.The results show that the model can effectively judge the operating state of gearbox wear based on oil information combined with actual detection results.(2)In order to solve the problems such as the difficulty in effectively extracting and identifying early fault features of gear boxes,the nonlinear and unstable characteristics of gear faults,and the limitations in parameter selection of variational mode decomposition methods,this paper proposes a gear fault identification method based on WOA-VMD combined with RBFNN.This method first uses WOA algorithm to synchronously optimize the number of VMD decomposition K and penalty factors α.The improved VMD is used to decompose the original vibration signal and generate multiple intrinsic mode components(IMF).Then,the kurtosis criterion is applied to select the IMF with significant impact components,and the energy entropy is calculated to extract the fault feature vector,which constitutes a feature vector matrix.Then,the RBFNN model is input for fault identification.Finally,experimental fault data are used to verify the effectiveness of the optimized VMD-RBFNN method for gear fault identification.The results show that the diagnostic accuracy rate is 95.83%,further verifying the effectiveness of the method.(3)Due to the increasingly large-scale and complex nature of gear boxes,their degradation trends have become difficult to grasp and predict.At the same time,the performance degradation characteristics of residual life prediction methods are difficult to establish and the prediction accuracy is low.To solve these problems,this paper establishes a residual life prediction model based on multiple characteristic parameters and LSTM.The model utilizes VMD to decompose vibration data to obtain modal component features,and combines the feature information that can characterize the bearing degradation process to construct a multi feature parameter set.Then,input these parameters into the LSTM model for training,and set the output sample as the remaining life ratio for prediction.Finally,the model is validated and analyzed using bearing data sets,and a comparative analysis of BPNN and SVM models is conducted.The results show that the model proposed in this paper can basically fit the bearing degradation curve,and the overall prediction effect is good,and can basically reflect the actual residual life trend of the bearing.