旋涡泵叶轮和流道水力尺寸智能算法研究  

Study on the intelligent approach for the calculation of hydraulic dimension for impeller and passage of vortex pump

作　　者：王军[1] 吴凤和[1] 陈恩平[1] 

机构地区：[1]燕山大学机械工程学院,河北秦皇岛066004

出　　处：《机械设计》2013年第8期1-6,共6页Journal of Machine Design

摘　　要：旋涡泵内部流动理论不完善,其叶轮和流道水力尺寸计算通常采用相似计算法和经验统计计算法。文中针对这两种方法在应用中存在的诸多局限性,提出了基于人工神经网络的智能算法。该方法既不以寻求相似的模型泵并进行换算为基础,也不以统计方法获取经验数据,而是以现有旋涡泵产品数据为基础,应用神经网络构建旋涡泵性能参数与水力尺寸之间的关系模型,从而实现泵的性能参数与水力尺寸的非线性映射,并利用这一关系模型进行新产品设计。文中介绍了这一方法的基本原理和关键技术,其中包括神经网络的结构设计、模型选择、样本对设计和网络训练等。解决了由于旋涡泵叶轮和流道结构型式繁多、设计中缺乏足够的经验数据支持及缺乏统一的经验统计方法的问题,提供了使不同企业能够有效地利用已有产品开发新产品的统一方法,该方法不仅支持知识重用,而且具有通用性,对提高旋涡泵的设计质量和缩短设计周期具有重要意义。As the internal flow theory for vortex pump is not mature,the similarity and empirical approach are often employed for the calculation of the impeller and passageway dimension.This paper introduces a intelligent approach based on artificial neural network that can overcome the limitation of the similarity and empirical approach.In this new approach,the relational model between the pump parameters and its dimension is established with artificial neural network based on the existing products data,and the nonlinear mapping from the parameters to dimension is realized with this model.The principle technology of this new approach includes network structure design,model selection,training samples design,and network training process.This new approach provides a new way for new products design based on the existing products,and also can be used for different pump types.This new approach not only utilizes data from previous products,but also attains a high level of versatility and can be applied to a variety of pumps.

关 键 词：旋涡泵 人工神经网络 水力尺寸 知识重用 人工智能 

分 类 号：TH314[机械工程—机械制造及自动化] TP301[自动化与计算机技术—计算机系统结构]