根据磁流变减振器的非线性特性,提出磁流变减振器广义回归神经网络(GRNN)模型辨识方法,利用台架试验获取的力学特性数据,建立磁流变减振器广义回归神经网络正、逆模型,并与反向传播神经网络(BPNN)模型进行比较.结果表明:通过合理选取网络变量并优化光滑因子,GRNN模型能准确预测磁流变减振器的阻尼力和控制电流,其正、逆模型辨识精度优于BPNN模型.此外,GRNN还具有结构简单、快速收敛等特点,为磁流变减振器的准确建模与控制提供了重要手段.%According to the nonlinear characteristics of magneto-rheological (MR) damper,a model identification method based on generalized regression neural network (GRNN) is proposed.By using the mechanical characteristics data obtained in bench test,both forward and inverse GRNN-based models for MR damper are built and compared with models based on back propagation neural network (BPNN).The results show that by reasonably selecting network variables and optimizing smooth factors,the GRNN-based model can accurately predict the damping force and the control current of MR damper with model identification accuracy higher than BPNN-based models.In addition,GRNN is also simple in structure and fast in convergence,providing an important means of accurate modeling and control for MR damper.
展开▼
