Study Of Vibration Reduction Structure Damping Characteristics With Particle Damper Based On SVR

Particle damping technology has better robustness than other vibration damping technologies, such as viscoelastic damping. Damping characteristics of particle damping structure is highly nonlinear changed with particle parameters, structure parameters and damper parameters. Influence law of particle parameters, structure parameters and damper parameters on vibration reduction structure is the mainly research direction which most of the scholars at home and abroad have focused on. At present so many scholars have adopt the experimental method to measure a large number of damping characteristics data for the sake of studying its damping effect, the advantage of this way is that the data was high reliability but the disadvantage is time-consuming and inefficient. Therefore, building a way which could predict the vibration structure damping characteristics is of great importance.First of all, in order to explore the prediction method of vibration damping characteristics with particle damper, damping characteristics of particle damping structure were predicted based on structural risk minimization of support vector regression machine in this paper. The research of this part includes four sections: 1.Established a experimental system of cantilever beam structure with particle damper, studied the effect of various parameters on the vibration reduction performance of cantilever beam structure with particle damper; 2.Influence of values in advance on the accuracy of forecast results was investigated; 3.Performance in searching the best SVR structural parameters was compared among GA, CV and PSO; 4.Damping characteristics of particle damping structure were predicted by using SVR and have been verified by welding experiments. The results show: the influence of values(average, root mean square value, previous label value) in advance on the accuracy of forecast results was little, the three approaches can meet the need of the SVR model;Three kinds of optimization algorithm will be able to search the best parameters for prediction model and the precision of established forecast model was almost unanimous, the average relative error between the actual and estimated values was 15.1%; particle density, filling rate, amplitude and fixed distance were determined as the main influencing factors, particle diameter was determined as the secondary factor through forecasting analysis.Second, in order to make the prediction method more generality, particle damping plate structural damping characteristics were predicted by using SVR on the basis of structural parameters which were optimized through GA in this paper. The research of this part includes three sections: 1.Established a experimental system of plate structure with particle damper, studied the effect of various parameters on the vibration reduction performance of plate structure with particle damper; 2.Influences of kernel function types, terminated algebra and population size on the accuracy of prediction model were investigated. 3. Damping characteristics of plate structure with particle damper was predicted by using SVR and GA, have been verified by welding experiments. The results show: the average relative error between the actual and estimated values was 12.9% on the basis of choosing a gaussian radial basis kernel function, suitable terminated algebra and population size; particle density and filling rate were determined as the main influencing factors, hole cavity arrangement was determined as the secondary factor through forecasting analysis.At the end of the paper,the particle damping structure damping characteristics prediction software was designed and developed by using the MATLAB, has pushed parameters optimization algorithm and SVR in the application of particle damping vibration reduction structure, further was accelerated the application of the particle damping vibration technology in engineering practice.