| Travelling wave ultrasonic motor(USM) works by the inverse piezoelectric effect of piezoelectric materials. Micro vibrations in ultrasonic frequency are inspired in stator, then the rotor is driven by frictional effect between stator and rotor, finally the loads are driven. USM is a typical strong nonlinear systems and friction material is the key components. Wear of friction material not only lead to decrease of the friction material thickness, but also reduce the preload and changing the contact state of interface between rotor and stator. These changes have significant influence on the vibration characteristics of stator. The coupling of these factors makes it difficult to predict the performance life of USM accurately which greatly limits the application of USM. The establishment of an accurate life prediction model of USM has important practical significance.Preload has great influence on the running of USM. In order to adjust preload, an open type preload controlled USM performance testing device was developed. No-close load type was adopted, the actual operate conditions were not changed and the preload could be adjusted expediently and accurately. Using this test device, the wear process and wear mechanisms in different wear stages of the contact interface were researched. Based on the results, the accelerated life test method was proposed and then the method was validated by experiment. Using this method, plenty of time cost could be saved which provided experiment basic to the following research.Using the accelerated life test method, the mechanical and electrical load characteristics of TWUSMs in the whole life of friction material were researched. The effects of preload, load torque and friction material thickness on the USM characteristics were abtained. The energy transfer process was analyzed based on the results, then the coupling effect between mechanical characteristics and electrical load characteristics were explored and the relativity between electric signals and life of feiction material was proved. These findings provided theory history data for the ultrasonic motor remaining useful performance life prediction model.The variations of stator vibrations with preload and load torque were researched by theory and experiment. The effects of preload and load torque on resonance frequency and vibration velocity were obtained. The contact state of stator and rotor was measured by improved electrical contact method and the effects of preload and load torque on contact state were analyzed. Then the effects of preload and load torque on the stator drive effect were explored. Based on the analysis results, preload-amplitude coefficient and load torque-amplitude coefficient were proposed to establish the visco-elastic contact model considering preload and load torque effect. The out performance of USM could be predicted accurately using the proposed contact model.An ultrasonic motor remaining useful performance life prediction model based on data-driven and combined visco-elastic contact model was established. In the model, the electrical signal was as input vector and artificial neural networks and support vector machines were adopted. First, the structure of ANN hidden layer was optimized and the results shown the optimal structures were double hidden layer 20-4 and triple hidden layer 20-7-3. The radial basis function was used as kernel function of SVM, and the parameters(C, γ) of kernel function were optimized. The optimum value was(11.9, 0.012). Compared to SVM, ANN had better precision when the thickness of friction material was small. Second, put the predicted thickness of friction material into the visco-elastic contact model, the output performance of USM could be calculated. Last, according to working request, the failure thickness could be obtained and remaining useful performance life percentage could be calculated, then the maintenance policy could be made based on the results. The remaining useful performance life of USM could be predicted using the proposed model. |
PDF Full Text Request