Research On Analysis Method Of Dynamic Characteristics And Reliability For The Double Swing Head Transmission System Of CNC

Due to China's transition from a manufacturing country to a manufacturing powerhouse in recent years,CNC equipment has become increasingly sophisticated,composite and intelligent.And a double swing head system that is one of the core functional components of high-end CNC equipment plays an important role in ensuring the machining accuracy of CNC equipment.Its dynamic characteristics and operational reliability directly affect the machining accuracy and machining efficiency of CNC equipments.Therefore,this paper studied a double swing head system and carried on the dynamic analysis and operational reliability assessment for it.The main research contents are as follows:(1)First,the single-roller enveloping hourglass worm gear of the key transmission component of a double swing head transmission system has been studied.The effect of the load on the single-roller enveloping hourglass worm gear is studied by analyzing the meshing relationship between the space structure and the transmission process.And the worm stiffness is further refined into the worm tooth bending stiffness,shear stiffness,radial compression stiffness and worm foundation stiffness.Similarly,the worm wheel stiffness is divided into worm wheel tooth bending stiffness,shear stiffness and worm wheel foundation stiffness.Then,based on the energy method,the analytical expressions of each stiffness are obtained,and the contact stiffness is used as a medium to integrate the each stiffness into a tooth pair mesh stiffness.Considering that the contact ratio of single roller enveloping hourglass worm gear is not integer,the number of mesh teeth changes periodically with the transmission process,and the time-varying mesh stiffness during the transmission process is further obtained.Second,based on the stiffness analysis result,the dynamic differential equation of the single roller enveloping hourglass worm gear is established by applying Newton's law.And the Newmark-? numerical analysis method is used to solve the differential equation and obtain its dynamic responses in the transmission process.(2)In order to solve the fault pattern recognition problem of the weak components of a double swing head transmission system,a fault pattern recognition method based on multiclassifier ensemble and parallel learning is proposed in this paper.The easily acquired running data is used as the data source,and deep features are extracted from different dimensions by the parallel learning channel,which composed of the stacked auto-encoder(SAE)with different activation functions.Then the features are evaluated based on the characteristics of fault classification,and the features corresponding to each fault mode are selected and formed into feature subsets.Afterwards,multiple classifier models are constructed by using feature subsets,and the output of multi-classifiers is determined by majority voting method.The results are integrated to enable effective identification of weak component failure modes.The bearing fault example shows that this method can effectively extract the deep features of the fault pattern in the operational data,and can fully utilize the independence and complementarity between multiple classifiers to improve the diagnostic accuracy.(3)Aiming to tackle the performance degradation of the weak components of a double swing head transmission system,a 2-D convolutional neural network(CNN)model is proposed for the operational reliability assessment and the remaining useful life prediction.This method uses the data conversion model to realize the data type conversion of one-dimensional data to two-dimensional data,and the operational reliability assessment of weak components is completed based on the 2-D CNN model with improved loss function and network structure.Based on the reliability evaluation results,a double exponential model is used to establish the remaining useful life prediction model.The result shows that the method can effectively extract the deep features of the performance degradation in the operational data and complete the performance assessment and prediction quickly and effectively.