本文研究是以国内现有RV减速器为参考对新型号进行正向开发,要实现系列化设计,一种方法是全新结构的设计,其在结构设计优化过程中对现有复杂CAD及CAE模型信息利用率不足,由大量的结构设计方案带来的大量分析试验任务难以实现;另一种方式是对RV减速器的改良设计,这种设计方法中包括对大量旧型号模型的几何参数信息的重新组织。为使设计更加高效,需要简化的减速器参数化模型、可在组件库中重复调用的样机模型,建立减速器零部件的通用优化方法。 目前RV减速器结构设计主要通过试验及数值模拟两种手段,尚未有采用参数推荐方法进行产品开发的相关报导,相比人工调整参数,参数推荐算法可以减少试错成本,但要获得可用的结构参数,训练样本及算法的学习能力是关键。为了结合工程需求及减少设计工作量,本章将从建立小样本推荐算法、算法调优及评估算法的实现效果等方面进行研究。
The research in this paper is the positive development of new models with reference to the existing domestic RV reducers. To realize the series design, one way is the design of a brand new structure, which has insufficient utilization of the existing complex CAD and CAE model information in the process of structural design optimization, and it is difficult to realize a large number of analysis test tasks brought by a large number of structural design solutions; the other way is the improved design of RV reducers, which The other way is the improved design of RV gearboxes, which includes the reorganization of the geometric parameter information of a large number of old models. To make the design more efficient, a simplified parametric model of the gearbox, a prototype model that can be repeatedly called in the component library, and a general optimization method for gearbox components are needed. At present, RV reducer structure design is mainly through two means: experimental and numerical simulation, and there is no report on product development using parametric recommendation method. Compared with manual adjustment of parameters, parametric recommendation algorithm can reduce the trial and error cost, but the training samples and the learning ability of the algorithm are key to obtain available structural parameters. In order to combine the engineering requirements and reduce the design workload, this chapter will investigate the establishment of small-sample recommendation algorithms, algorithm tuning and evaluation of the algorithm implementation effect.