不可压缩大流场多体运动建模与数值仿真
文献类型：学位论文
| 作者 | 于军 |
| 学位类别 | 博士 |
| 答辩日期 | 2005-12-15 |
| 授予单位 | 中国科学院沈阳自动化研究所 |
| 授予地点 | 中国科学院沈阳自动化研究所 |
| 导师 | 聂义勇 |
| 关键词 | 不可压缩流场 多体运动 计算流体动力学 重叠网格法 神经网络控制 |
| 其他题名 | Model-building and Numerical Simulation for Multi-body Moving in Infinite Incompressible Fluid |
| 学位专业 | 机械电子工程 |
| 中文摘要 | 不可压缩流体区域中多个（主动）物体的运动是许多工程问题的理论基础，因此吸引了不同领域研究人员的浓厚兴趣。经过科研人员的努力，该问题的研究工作已经取得了一些突破性的成果。但是此运动包含复杂的流体和刚体耦合运动，所以仍有大量问题需深入研究，如流体和刚体运动机理、流 体和刚体运动方程的求解方法、耦合运动的解耦方法以及物体的运动控制方法等等，为此本文进行了如下工作。 首先，本文在参考水下物体运动模型的基础上，提出一种不可压缩流体中多体运动模型。相比于已有水下物体运动模型，该模型不仅准确计算物体受到流体阻力，而且能够建立起流体阻力随时间变化的关系。 其次，流体动力学方程组求解的准确程度直接关系该模型的计算精度，本文在分析流体动力方程组对流项误差产生原因的基础上，提出了一种改进离散方法，提高了物体运动模型的计算精度。由于流体中物体自由运动，导致离散网格随时间改变，这增加了求解模型的难度。在比较了多种可变网格技术的基础上，我们引入了重叠网格技术，从而实现了流体中多体运动模型的求解。 第三，本文提出的物体运动模型包含复杂的非线性偏微分映射关系，因此应用此模型设计运动控制系统时，需要建立一种适用于该模型的控制方法。本文利用神经网络对非线性函数的逼近能力，并通过采用差商近似代替微分的方法，建立了系统输入输出之间近似映射关系，在此基础上设计了一种控制方法。应用该控制方法，对运动轨迹跟踪问题进行了仿真试验，结果表明该控制方法的可行性和有效性。 |
| 索取号 | O23/Y74/2005 |
| 英文摘要 | It is theoretical base of many engineering problems that (initiative) multi-body moving in incompressible fluid, so it attracts people from various disciplines. And plenty of breakthroughs are received through the work of researchers. Because it includes complicated model of interaction between fluid and rigid body, there are many problems which are required to research. For example, the mechanism of interaction between fluid and rigid body, algorithms for fluid and rigid body movement, a method to resolve the interaction, and a control algorithm to rigid body movement, and so on. So the following is work done in this paper: Firstly, referring to mathematical model of body moving underwater, a new model of multi-body moving in incompressible fluid is proposed. Comparison to old model, not only resistance is exactly computed in this model, but also it is easily built the map between fluid resistance and time. Secondly, the precision of fluid dynamic equations influences on precision of this model. According to analysis of error to convection terms of these equations, an improved numerical method is proposed, which can get precise root of this model. Because body can arbitrarily move in fluid, the grid is varying with time which adds difficulty to computation. Overlapping grid is introduced to realize the simulation of multi-body moving in fluid by comparison to lot of variable grid technology. Thirdly, the proposed model includes complicated nonlinear partial difference relation, so a control algorithm is needed to be built for this model when movement control system is designed. In this paper, a controller is gained based on the approximate relation of input and output. This relation is built on utilizing approximating ability to nonlinear function of neural network, and approximating differential by difference. In the end, this controller is applied to trajectory tracing problem simulation, and result demonstrates its feasibility and effectively. |
| 语种 | 中文 |
| 公开日期 | 2012-08-29 |
| 产权排序 | 1 |
| 分类号 | O23 |
| 源URL | [http://ir.sia.ac.cn/handle/173321/9453]  |
| 专题 | 沈阳自动化研究所_工业信息学研究室_工业控制系统研究室 |
| 推荐引用方式 GB/T 7714 | 于军. 不可压缩大流场多体运动建模与数值仿真[D]. 中国科学院沈阳自动化研究所. 中国科学院沈阳自动化研究所. 2005. |
入库方式： OAI收割
来源：沈阳自动化研究所
浏览0
下载0
收藏0
其他版本
除非特别说明，本系统中所有内容都受版权保护，并保留所有权利。