多孔质静压轴承静态特性预测研究

闫如忠; 石俊伟; 马晓建; 安星宇

doi:10.13433/j.cnki.1003-8728.20220214

多孔质静压轴承静态特性预测研究

doi: 10.13433/j.cnki.1003-8728.20220214

闫如忠^{1, 2,},
石俊伟^{1, 2},
马晓建^{1, 2},
安星宇^{1, 2}

1.
东华大学机械工程学院，上海　201620
2.
纺织装备教育部工程研究中心，上海　201620

详细信息

作者简介:
闫如忠，副教授，硕士生导师，博士，yanrz@dhu.edu.cn

中图分类号: TH133.3；TP183
计量
- 文章访问数: 22
- HTML全文浏览量: 6
- PDF下载量: 0
- 被引次数: 0
出版历程
- 收稿日期: 2021-12-09
- 刊出日期: 2024-03-25

Study on Static Characteristics Prediction of Porous Hydrostatic Bearing

YAN Ruzhong^{1, 2
,},
SHI Junwei^{1, 2},
MA Xiaojian^{1, 2},
AN Xingyu^{1, 2}

1.
College of Mechanical Engineering, Donghua University, Shanghai 201620, China
2.
Engineering Research Center of Advanced Textile Machinery Ministry of Education, Shanghai 201620, China

摘要

摘要: 在多孔质静压轴承设计中，轴承设计参数是影响其静动态特性的关键因素之一，通常情况下，要得到合适的轴承设计参数，需要多次重复建模和仿真，且由于轴承结构复杂，建模难度大，仿真时间长，严重影响了轴承的设计效率。本文构建了一种基于遗传算法（Genetic algorithm，GA）优化反向传播（Back propagation, BP）神经网络的轴承静态特性预测模型，采用拉丁超立方抽样方法在轴承参数设计空间内采样，并进行Fluent流体仿真，将仿真数据用于GA-BP神经网络模型的训练与测试，实现了对设计空间内任意设计参数下的多孔质静压轴承静态特性的预测。研究结果表明，训练出的GA-BP神经网络模型能够准确预测多孔质静压轴承的静态特性，预测精度在95%以上，对多孔质静压轴承的快速设计和参数优化具有重要意义。
- 多孔质静压轴承 /
- 静态特性 /
- GA-BP神经网络 /
- 拉丁超立方抽样 /
- 预测
Abstract: In the design of porous hydrostatic bearings, bearing's design parameter is one of the key factors affecting its static and dynamic characteristics. Normally, several repetitions of modeling and simulation are required to obtain suitable bearing design parameters, and due to the complex bearing structure makes modeling difficult and simulation time long, which seriously affects the efficiency of bearing design. In this paper, a bearing static characteristic prediction model based on genetic algorithm (GA) optimized BP (Back Propagation) neural network is constructed, Latin hypercube sampling method is used to sample the bearing parameter design space, and perform Fluent fluid simulation. The data is used for the training and testing of the GA-BP neural network model to realize the prediction of static characteristics of porous hydrostatic bearing under any design parameters in the design space. The research results show that the trained GA-BP neural network model can accurately predict the static characteristics of porous hydrostatic bearings with a prediction accuracy of over 95%, which is great significance for the rapid design and parameter optimization of porous hydrostatic bearings.
- porous hydrostatic bearing /
- static characteristic /
- GA-BP neural network /
- latin hypercube sampling /
- prediction

HTML全文

图 1 静压止推轴承多孔质结构

Figure 1. Porous structure of hydrostatic thrust bearing

下载: 全尺寸图片幻灯片

图 2 多孔质轴承承载原理

Figure 2. Load-bearing principle of porous bearing

下载: 全尺寸图片幻灯片

图 3 BP神经元模型

Figure 3. Model of BP neuron

下载: 全尺寸图片幻灯片

图 4 GA-BP神经网络训练流程

Figure 4. Training processes of GA-BP neural network

下载: 全尺寸图片幻灯片

图 5 拉丁超立方抽样样本点分布

Figure 5. Sample point distribution of Latin hypercube sampling

下载: 全尺寸图片幻灯片

图 6 BP神经网络结构

Figure 6. Structure of BP neural network

下载: 全尺寸图片幻灯片

图 7 GA-BP神经网络模型训练结果

Figure 7. Training results of GA-BP neural network model

下载: 全尺寸图片幻灯片

图 8 静态承载力预测结果与预测误差百分比

Figure 8. Prediction results and prediction error percentage of static bearing capacity

下载: 全尺寸图片幻灯片

图 9 静刚度预测结果与预测误差百分比

Figure 9. Prediction results and prediction error percentage of static stiffness

下载: 全尺寸图片幻灯片

图 10 耗气量预测结果与预测误差百分比

Figure 10. Prediction results and prediction error percentage of gas consumption

下载: 全尺寸图片幻灯片

表 1 多孔质静压止推轴承设计参数

Table 1. Design parameters of porous hydrostatic thrust bearing

参数名称	最小值	最大值
多孔质长度L/mm	20.0	80.0
多孔质宽度W/mm	10.0	50.0
多孔质厚度H/mm	3.0	15.0
气膜厚度h/μm	5	20
供气压力P_s/bar	5.0	8.0
多孔质渗透率α/m²	1.0×10⁻¹³	1.0×10⁻¹³

下载: 导出CSV

表 2 部分样本数据

Table 2. Partial sample data

样本编号	多孔质长度L/mm	多孔质宽度W/mm	多孔质厚度H/mm	气膜厚度h/μm	供气压力P_s/bar	渗透率α/10⁻¹³m²	静态承载力W_b/N	耗气量Q/（L·min⁻¹）	静刚度K/（N·μm⁻¹）
1	23.2	43.3	11.1	18	5.9	1.0	122.5	13.3	11.5
2	39.0	19.5	14.7	14	7.0	1.0	105.8	9.1	12.0
3	51.4	23.3	4.7	15	6.4	1.0	398.7	24.6	22.5
4	43.2	35.3	8.4	20	6.3	1.0	310.3	24.4	20.6
5	31.0	38.3	12.4	10	6.1	1.0	324.4	10.0	28.2
6	68.8	31.9	3.3	16	7.7	1.0	1174.8	49.1	36.5
$\vdots $	$\vdots $	$\vdots $	$\vdots $	$\vdots $	$\vdots $	$\vdots $	$\vdots $	$\vdots $	$\vdots $
200	78.4	16.7	3.4	7	5.5	1.0	569.7	13.9	27.1

下载: 导出CSV

表 3 BP神经网络参数设置

Table 3. BP neural network parameters settings

参数	取值/方法
输入层神经元数	5
隐含层神经元数	11
输出层神经元数	3
激活函数	Tansig、Purelin
训练函数	Trainlm
最大训练次数	150
学习速率	0.01
训练精度	1.0 × 10⁻⁴

下载: 导出CSV

表 4 GA-BP神经网络模型预测精度

Table 4. Prediction accuracy of GA-BP neural network model

静态特性	E_MR/%	R²	预测精度/%
静态承载力	3.32	0.991	96.68
静刚度	2.99	0.996	97.01
耗气量	4.65	0.988	95.35

下载: 导出CSV

参考文献(15)

[1]	伍奕桦. 多孔质可倾瓦轴承支承转子系统动力学分析及其实验研究[D]. 长沙: 湖南大学, 2019. WU Y H. Theoretical analysis and experimental investigation on rotordynamic performance of a rigid rotor supported on porous tiliting pad bearings[D]. Changsha: Hunan University, 2019. (in Chinese)
[2]	段柳颖, 张晓峰, 刘玚, 等. 多孔质气体静压径向轴承倾斜状态特性研究[J]. 润滑与密封, 2021, 46(6): 10-14. DUAN L Y, ZHANG X F, LIU Y, et al. Inclination performance study of porous aerostatic journal bearing[J]. Lubrication Engineering, 2021, 46(6): 10-14. (in Chinese)
[3]	QIN P, SHEN Y, ZHU J, et al. Dynamic analysis of hydrodynamic bearing–rotor system based on neural network[J]. International Journal of Engineering Science, 2005, 43(5-6): 520-531. doi: 10.1016/j.ijengsci.2003.10.008
[4]	ZHU X X, LUO X Z, ZHAO J H, et al. Research on deep feature learning and condition recognition method for bearing vibration[J]. Applied Acoustics, 2020, 168: 107435. doi: 10.1016/j.apacoust.2020.107435
[5]	TANG J, HUANG X D, FANG C G. BP neural network in prediction of the constant-current hydrostatic bearing static stiffness[J]. Advanced Materials Research, 2011, 199-200: 271-274. doi: 10.4028/www.scientific.net/AMR.199-200.271
[6]	李寅平, 陈国定, 李盼, 等. 基于GA-LM-BP算法的圆柱轴承动特性预测[J]. 机械制造, 2013, 51(1): 24-27. doi: 10.3969/j.issn.1000-4998.2013.01.009 LI Y P, CHEN G D, LI P, et al. Prediction of dynamic characteristics of cylindrical bearing based on GA-LM-BP algorithm[J]. Machinery, 2013, 51(1): 24-27. (in Chinese) doi: 10.3969/j.issn.1000-4998.2013.01.009
[7]	石莹, 王学智, 刘金营, 等. 基于Fluent-BP神经网络的液体动静压轴承热特性分析[J]. 润滑与密封, 2016, 41(1): 37-42. doi: 10.3969/j.issn.0254-0150.2016.01.007 SHI Y, WANG X Z, LIU J Y, et al. Thermal characteristics analysis of liquid hybrid bearing based on Fluent-BP neural networks[J]. Lubrication Engineering, 2016, 41(1): 37-42. (in Chinese) doi: 10.3969/j.issn.0254-0150.2016.01.007
[8]	林桂强, 毕超, 江波, 等. 基于BP神经网络的套筒式串联推力轴承组均载性能及承载能力细化研究[J]. 工程塑料应用, 2012, 40(7): 50-53. doi: 10.3969/j.issn.1001-3539.2012.07.012 LIN G Q, BI C, JIANG B, et al. Detailed research on load uniformity and load-bearing capacity of sleeve type tandem thrust bearings based on BP neural network[J]. Engineering Plastics Application, 2012, 40(7): 50-53. (in Chinese) doi: 10.3969/j.issn.1001-3539.2012.07.012
[9]	顾延东. 多孔质气体径向轴承静动特性研究及优化设计方法[D]. 镇江: 江苏大学, 2019. GU Y D. Investigation on the static and dynamic characteristics and optimization design of aerostatic radial bearing with porous restrictor[D]. Zhenjiang: Jiangsu University, 2019. (in Chinese)
[10]	顾延东, MARTIN B, ARTUR S, 等. 多孔质气体静压轴承研究现状及发展趋势[J]. 排灌机械工程学报, 2021, 39(8): 818-825. GU Y D, MARTIN B, ARTUR S, et al. Aerostatic bearing with porous restrictor: research status and future perspectives[J]. Journal of Drainage and Irrigation Machinery Engineering, 2021, 39(8): 818-825. (in Chinese)
[11]	雷明远. 基于遗传算法与BP神经网络的RV减速器结构优化设计[D]. 兰州: 兰州理工大学, 2021. LEI M Y. Structure optimization design of RV reducer based on genetic algorithm and BP neural network[D]. Lanzhou: Lanzhou University of Technology, 2021. (in Chinese)
[12]	马高, 刘康. 基于BP神经网络CFRP约束混凝土抗压强度预测[J]. 湖南大学学报(自然科学版), 2021, 48(9): 88-97. doi: 10.16339/j.cnki.hdxbzkb.2021.09.010 MA G, LIU K. Prediction of compressive strength of CFRP-confined concrete columns based on BP neural network[J]. Journal of Hunan University (Natural Sciences), 2021, 48(9): 88-97. (in Chinese) doi: 10.16339/j.cnki.hdxbzkb.2021.09.010
[13]	胡新荣, 刘嘉文, 刘军平, 等. 基于改进GA-BP-MC神经网络的人体三维尺寸预测[J]. 计算机工程与科学, 2021, 43(8): 1443-1453. doi: 10.3969/j.issn.1007-130X.2021.08.013 HU X R, LIU J W, LIU J P, et al. 3D human dimension prediction based on improved GA-BP-MC neural network[J]. Computer Engineering and Science, 2021, 43(8): 1443-1453. (in Chinese) doi: 10.3969/j.issn.1007-130X.2021.08.013
[14]	李静, 姜峰, 牛彩云. 实验设计和神经网络法对柴油机性能的优化研究[J]. 机械设计与制造, 2020(6): 100-104. doi: 10.3969/j.issn.1001-3997.2020.06.025 LI J, JIANG F, NIU C Y. Research on performance optimization of diesel engine based on DOE and ANN method[J]. Machinery Design & Manufacture, 2020(6): 100-104. (in Chinese) doi: 10.3969/j.issn.1001-3997.2020.06.025
[15]	CHEN H, WANG Y, ZUO M S, et al. A new prediction model of CO₂ diffusion coefficient in crude oil under reservoir conditions based on BP neural network[J]. Energy, 2022, 239: 122286. doi: 10.1016/j.energy.2021.122286