基于神经网络的间隙机构运动副磨损预测

应用BP神经网络建立了磨损率与接触应力、滑动速度和材料硬度之间的非线性关系模型,并对该网络模型进行了验证和测试,结果表明,训练良好的神经网络模型能够准确反映样本所蕴含的内在磨损规律,且具有较好的预测效果。基于非线性弹簧阻尼模型和修正的Coulomb摩擦力模型对含间隙曲柄滑块机构进行数值仿真分析,获得间隙机构运动副的接触应力和相对滑动速度,利用训练好的神经网络磨损模型对轴套的磨损进行迭代磨损预测分析,发现随着曲柄转数的增加,轴套表面一些特定位置处的磨损越来越严重,最终导致轴套表面出现非均匀磨损现象,其原因是间隙机构运转过程在一些特定位置处产生了较大接触应力和碰撞力。

Prediction of wear in mechanism with revolute joint clearances based on neural network

A nonlinear wear model was presented for studying and quantifying the wear phenomenon in revolute clearance joints.The model of relationship between wear rate and contact pressure,sliding velo-city and material hardness was established based on BP neural network.The test results have indicated that the model can accurately reflect the inherent wear law of the samples,and has a higher prediction precision.In the process,a simple slider-crank mechanism with a clearance in revolute joints was utilized.The evaluation of the contact forces developed was based on the elastic-damping contact force model and modified Coulomb friction model.The contact pressure and relative sliding velocity of the revolute clea-rance joint were obtained by numerical simulations.Then,the trained BP neural network model was employed to predict the wear of clearance joint.Through repeated iterative prediction,it can be found that the wear depth occurred in the joint surface is non-uniform,owing to the fact that the large contact force and impact force between the bushing and pin occurs frequently in some special range of crank angle.