摘要
由于动载系数的计算公式比较复杂,机械设计手册中提供动载系数线图供设计人员近似计算,但通过人工查取线图数据的方法会给设计带来误差,并且不利于实现设计过程的CAD。因此,提出了应用径向基函数神经网络来确定动载系数的计算方法,映射了引起动载系数与其影响因素(齿轮节线速度、齿轮制造精度)的非线性关系,并与其他文献中提出的应用BP神经网络的方法进行对比。结果表明:径向基函数神经网络可通过较少的训练次数达到较高的精度,大大超过了BP神经网络的收敛速度和训练效率。该方法可广泛应用于工程设计中计算齿轮动载系数及其他以线图表示的参数。
The diagram of gear dynamic factor was given in the mechanical design manual for gear design due to its complicate and implieit expression, whieh may induee a larger error beeause of the reeognition aeeuraey of engineers. Radial Basis Function （RBF） neural network,which was widely used to map the non-linear relation between input and output variables,was introduced to determine the relation between dynamic factor and some factors sueh as piteh line veloeity and quality grade of gears. In addi- tion, a comparison was made between our RBF method with BP one proposed by another study. The results show that RBF neural network are more effeetive and has a faster eonvergenee rate than BP neural network. This method could be widely used to ealeu- late the dynamic faetor and other parameters which are shown diagrammatically.
出处
《现代制造工程》
CSCD
北大核心
2013年第1期76-78,120,共4页
Modern Manufacturing Engineering
