SVM颗粒运动形态及其能量耗散的预测分析  

Prediction of Particle Motion Modes and Energy Dissipation Based on SVM

作　　者：徐志凯 尹忠俊[1] 韩天[1] XU Zhikai;YIN Zhongjun;HAN Tian(School of Mechanical Engineering,University of Science and Technology Beijing,Beijing 100083,China)

机构地区：[1]北京科技大学机械工程学院,北京100083

出　　处：《实验室研究与探索》2020年第1期59-64,82,共7页Research and Exploration In Laboratory

摘　　要：基于离散单元法建立了圆柱形颗粒阻尼器的仿真模型,研究了阻尼器内颗粒在不同激振条件下的运动形态及其能量耗散大小。为了获取两者之间的定量关系,应用基于网格搜索法(GS)的支持向量机(SVM)建立了颗粒运动形态的分类预测模型及其能量损耗的回归预测模型,对颗粒运动形态的分布及其能量损耗的大小进行了预测,并通过仿真进行了验证。结果表明:基于GS方法优化的SVM能够建立一个预测准确度很高、推广泛化能力很强的分类和回归预测模型,该预测模型不仅能够很好地揭示颗粒系统在不同运动形态下的能量耗散的变化规律,而且还能在较大的激振条件范围内确定系统能量耗散最大值及对应的运动形态。The discrete element method theory(DEM)was applied to establish a cylindrical particle damper simulation model.The variation of the motion modes and energy dissipation of the particles in the damper under different excitation conditions were studied.In order to obtain the quantitative relationship between the particle motion modes,energy dissipation and excitation conditions,support vector machine(SVM)based on grid search method(GS)was used to establish a classification prediction model for particle motion modes and a regression prediction model for its energy loss.On the basis of GS,we optimized the SVM parameters,predicted the distribution of particle motion modes and its energy loss,and then verified the forecasting data by simulation.The results show that the SVM based on GS can establish a prediction model with high accuracy and strong generalization.The prediction model established by SVM reveal the law of the energy dissipation of the particle system can work well,and is accessible to obtain the maximum energy dissipation and its corresponding motion mode within the range of excitation conditions studied.

关 键 词：颗粒运动形态 能量耗散 支持向量机 网格搜索法 

分 类 号：O328[理学—一般力学与力学基础] TB34[理学—力学]