首页 >

基于梯度下降的水滴收集率计算方法

任靖豪1,王强1,李维浩1,刘宇2,易贤1   

  1. 1. 中国空气动力研究与发展中心
    2.
  • 收稿日期:2021-09-14 修回日期:2021-12-15 出版日期:2021-12-24 发布日期:2021-12-24
  • 通讯作者: 易贤
  • 基金资助:
    国家科技重大专项;国家自然科学基金项目

A Prediction Algorithm of Collection Efficiency Based on Gradient Descent Method

  • Received:2021-09-14 Revised:2021-12-15 Online:2021-12-24 Published:2021-12-24
  • Contact: YI Xian

摘要: 拉格朗日方法在计算三维复杂外形的水滴撞击特性时,一般需要通过插值获得物面网格上的水滴收集率,而这一过程容易造成计算结果失真。本文建立了一种基于梯度下降的水滴收集率计算方法,通过确定撞击物面网格角点的水滴轨迹实现物面网格中心处的水滴收集率计算,避免插值运算。该方法的计算思路如下:1)根据撞击点到目标点的距离,采用梯度下降算法自适应地调控水滴的初始释放位置,使撞击点落在壁面网格角点附近的给定阈值范围内;2)采用邻点检索算法,快速确定迎风面内所有撞击网格角点的水滴轨迹;3)基于流管定义,计算出网格中心处的水滴收集率。采用该方法对典型三维外形的水滴收集率进行了计算,并与相关文献结果进行对比。结果显示计算结果与文献数据吻合良好,并有效克服了现有技术中因插值导致的结果失真问题,能够为飞机结冰特性及防除冰技术研究提供参考。

关键词: 飞机结冰, 水滴收集率, 拉格朗日方法, 方形邻域, 梯度下降法

Abstract: The collection efficiency transfer from impinging point to grid nodes is indispensable when determining droplet impingement property by the traditional Lagrange method. But, the general interpolation technique may bring errors and make the results disagree with the reality in the three-dimensional case. To voids the interpolation, this paper proposed a new strategy that gets the trajectory of impinging near the grid's node by the gradient descent method and calculates the collection efficiency on the cell center. The improved algorithm uses a deviation function iteration by gradients descent method to reducing the distance between an impinging point and the grid’s node. An neighbor search algorithm is then used to obtain all of the trajectories that impacting on the grid’s node on the windward region. And the collection efficiency on the center of the surface control element could be acquired by tube defined model. In this paper, some representative cases valuated the improved method. The testing report showed that the calculation results using the proposed method are consistent with the experimental data and better than the interpolation results by the traditional algorithm.

Key words: aircraft icing, collection efficiency, Lagrangian method, directional neighborhood, gradient descent algorithm

中图分类号: