基于BP网络的齿根弯曲应力计算方法  

Calculation Method of Gear Bending Stress Based on BP Neural Network

作　　者：杨小安[1] 张伟社[1] 高勇[1] 成建联[1] 

机构地区：[1]长安大学理学院,西安710064

出　　处：《机械设计与研究》2008年第6期52-55,共4页Machine Design And Research

基　　金：陕西省科研基金资助项目(2007611)

摘　　要：为实现齿轮传动寿命的精确计算,需要探讨精确的轮齿齿根弯曲应力计算方法。文中将渐开线直齿圆柱齿轮齿根应力的影响因素分为5因素5水平,用正交试验表L25(45)组建了24组因素组合。依据渐开线齿轮范成加工原理及UG、ANSYS三维建模功能,建立的有限元齿根弯曲应力精确计算模型,并进行了24组数据计算。将其计算结果作为BP网络训练样本,得到了载荷作用于齿顶及单齿啮合区最高点时,渐开线直齿圆柱齿轮齿根弯曲最大应力计算的BP网络模型。用其计算结果与ANSYS计算结果对比,最大相对误差不超过2.4%。因此,该BP网络模型可直接用于齿根弯曲应力值的计算,特别是对于高精度齿轮传动齿根弯曲应力的计算。In order to achieve the accurate calculation of the life time of the gear drive, it is necessary to probe into the accurate calculation method of the bending stress of the gear tooth root. The article divided the stress influencing factors of involutes spur gear tooth root into 5 levels of 5 factors, with orthogonal table L25 （45 ） setting up a group of 24 combinations of factors. Based on the involutes gear generating method and UG,ANSYS three-dimensional modeling capabilities, models of accurate calculation of the finite element tooth root bending stress are established with 24 groups of data. The results of about calculation are used as BP network training samples, which lead to the BP network model of the largest stress calculation of involutes spur gear tooth root bending when the load has been exerted on the roof and the highest point of the single tooth meshing. The greatest error is no more than 2.4% compared with the ANSYS results. Therefore, the BP network model can be used directly in the tooth root bending stress calculation, especially for gear tooth root bending stress calculations of high-precision.

关 键 词：BP网络 齿根应力 正交试验 ANSYS分析 UG建模 

分 类 号：TH114.413[机械工程—机械设计及理论]