摘要:
为确保航空薄壁锥齿轮工作可靠,运行平稳,其工作转速范围内不应当有振动应力很大的危险共振存在。因此需要调整轮体结构参数,使其在工作转速范围内避开行波共振转速,同时尽可能不使质量增加。本文中用正交试验法安排结构参数取值,建立了参数化轮体截面模型,用AN-SYS参数化设计语言APDL计算了各样本的动频。建立双隐层BP神经网络,经过训练使其映射结构参数与动频间的非线性关系,采用混合遗传算法得到一组结构参数,使新的轮体结构避开共振,且轮体重量减轻。
Abstract:
In order to ensure the security and low noise of the aeronautic thin-spoke bevel gear, there should not belarge stress aroused by resonance existing in the range of operating speed. Thus it is necessary to modulate thestructural parameters of the gear spoke to assure that the travelling-wave resonance do not arise in the working conditions, at the same time, the mass should not be added. This study prepared the structural parameters by Orthogonal Experimental Design, and established the axial section model through parameterization, and computed the dynamic frequency by APDL (Ansys Programing Design Language), then founded the double latent BP ArtificialNeural Network and trained it to make it reflect the nonlinear relationship between the dynamic frequency and thestructural parameters. Finally, a set of parameters which avoided the travelling-wave resonance was calculated successfully, simultaneously, the mass of the gear was also decreased.
点击查看大图
下载:
