Research On Modal Parameters Identification For Vibration Mechanical Structure

Vibrating machines is a type of equipment that makes use of vibrating to achieve various processes. A good investigation of its dynamic properties is needed for its inspection, assessment and health monitoring. Modal parameters are the key structure dynamic property parameters for such study as faults diagnosis, et al. Combining the present research status and perspectives of vibrating. machines, the theory and methodology of the modal parameter identification are studied in this dissertation, of which the main achievements are as follows:(1) Modal parameters of the non-proportional viscous damping vibration system were identified using complex modal theory. In order to eliminate the amplitude unstable situation by resonance using, the polyurethane springs were installed in the main mass to avoid the instability of near resonance vibrating machines effectively. The stiffness of the springs was designed at the same time.(2) Modal parameters of blocking forming machine were simulated by 3-D finite element method (FEM). The FEM model of blocking forming machine was created with the elasticity of polyurethane springs equivalent to the bottom of the kickstand. The vertical direction (Y direction) translational degree is retained. The result of modal analysis shows that only the 5th order natural frequency (working-frequency) can give birth to resonance and the mode shape at this stage is Y direction vibration, which will benefit forming process.(3) Experimental modal analysis for block forming machine was performed. First the methods to eliminate the errors in the testing process were analyzed. The pulse exciting modal experiment was carried on to verify the FEM analysis modal parameters, whereas the consistency of simulation and experimental result prove the accurate and reliable of the experiments. Finally the structure of block forming machine has been improved to solve such problems as housing abrasion, bad block forming quality, working noise et al.(4) The modal parameter identification method combining wavelet transform (WT) and Bootstrap theory was proposed. Through comparison with other mother wavelet functions, Morlet Wavelet was selected due to its better capacity in time-frequency analysis than others. Considering the effects of modal separation, end-effect and the effect of wavelet parameters choose on the accuracy of identification results, the complex Molet wavelet parameters were optimized by Shannon Entropy. There is no adequate information for appropriate and estimate the identification accuracy, due to the effects of noise, environment change and measurement errors. Thus, a method that combines WT and Bootstrap Theory was proposed to solve this problem. Simulation and experiment results show that:①the method does better in estimating the modal frequency and damping ratio;②the method is not sensitive to noise and doesn’t need large sample;③this method has a higher accuracy than WT in the process of identifying damping ratio.(5) The vibration prediction technology of artificial neural network (ANN) was studied. Considering the structure parameters diversity and the number of hidden layer uncertainty, the ANN models were optimized by signal under 2 kind of vibrating conditions i.e.Condition 1, vibration only; Condition 2, vibration with press. Simulation results show that condition 2 is superior to condition 1. ANN technology was applied to predict vibration system. Comparing the results at exciting frequencies of 35Hz and 25Hz shows that 4-7-1 model can forecast the vibration system perfectly.