基于GA⁃BP神经网络的颗粒阻尼器减振特性预测

Prediction of Vibration Damping Characteristics with Particle Dampers Based on GA⁃BP Neural Network

提出了一种基于遗传算法改进的BP神经网络(GA⁃BP)的颗粒阻尼效应预测模型.首先通过悬臂梁阻尼检测实验建立数据集,然后对建立的数据集进行训练非线性复杂模型,用于描述颗粒阻尼器的阻尼效应.为了进一步验证所提模型的有效性,通过CA⁃YD⁃1181压电传感器采集相关数据进行二次验证.结果表明,与传统的BP神经网络预测模型相比,遗传算法优化后的模型能够通过不同参数的变化对颗粒阻尼器减振效果进行精准预测,收敛速度提高了近36.8%.该模型具有良好的拟合效果,能准确、合理地预测阻尼特性,并调整颗粒阻尼器的相关参数.

This paper proposes a prediction model for the particle damping effect based on an improved BP neural network using genetic algorithm(GA⁃BP).Firstly,a dataset is established through the cantilever damping test.Then,a nonlinear complex model is trained to describe the damping effect of the particle damper.In order to further verify the effectiveness of the proposed model,the CA⁃YD⁃1181 piezoelectric sensor is used to collect relevant data for secondary verifica⁃tion.The results show that compared with the traditional BP neural network prediction model,the optimized GA⁃BP model can accurately predict the vibration reduction effect of the particle damper under different parameter variations and with the convergence rate improved by nearly 36.8%.The model has a good fitting effect,can predict the damping characteristics accurately and reasonably,and adjust the relevant parameters of the particle damper.