文章详情

摘要

文章通过添加伯努利方程和尼古拉兹公式物理约束条件构建了混合神经网络模型，探索了涡流管冷端温度变化规律并进行了相应预测。网络采用多层前馈模型和Levenberg-Marquardt学习算法，选择双曲正切函数作为传递函数。此外，利用决定系数（R2）、均方根误差（RMSE）来确定所开发模型的统计有效性，并分析了模型不确定性和鲁棒性。混合模型的指标R2为0.993 6，RMSE为0.339 2，在不确定性和鲁棒性方面也具有良好的表现。这些数据表明文章构建的模型成功地预测了涡流管冷端温度的变化，并且具有良好的精度。

关键词

涡流管；预测模型；混合神经网络；温度性能

参考文献

1. [1] SINGH J，GILL S S，DOGRA M，et al. Effect of Ranque-Hilsch Vortex Tube Cooling to Enhance the Surface-Topography and Tool-Wear in Sustainable Turning of Al-5.6Zn-2.5Mg-1.6Cu-0.23Cr-T6 Aerospace Alloy [J/OL].Materials，2022，15：5681[2024-10-03].https://doi.org/10.3390/ma15165681.
2. [2] PARKER M J，STRAATMAN A G. Experimental Study on the Impact of Pressure Ratio on Temperature Drop in a Ranque-Hilsch Vortex Tube [J/OL].Applied Thermal Engineering，2021，189（7）：116653[2024-10-23].https://doi.org/10.1016/j.applthermaleng.2021.116653.
3. [3] ATTALLA M，AHMED H，AHMED M S，et al. Experimental Investigation of the Effect of Nozzle Numbers on Ranque-Hilsch Vortex Tube Performance [J].Experimental Heat Transfer，2017，30（1-3）：253-265.
4. [4] AVCI M. The Effects of Nozzle Aspect Ratio and Nozzle Number on the Performance of the Ranque-Hilsch Vortex Tube [J].Applied Thermal Engineering，2013，50（1）：302-308.
5. [5] RAFIEE S E，SADEGHIAZAD M M. Experimental Analysis on Impact of Navigator's Angle on Velocimetry and Thermal Capability of RH-Vortex Tube [J/OL].Applied Thermal Engineering，2020，169：114907[2024-09-28].https://doi.org/10.1016/j.applthermaleng.2020.114907.
6. [6] POURARIA H，KIA S M，PARK W G，et al. Modeling the Cooling Performance of Vortex Tube Using a Genetic Algorithm-Based Artificial Neural Network [J].Thermal Science，2016，20（1）：53-65.
7. [7] KAYA H，KIRMACI V，ES H A. Performance Modeling of Parallel-Connected Ranque-Hilsch Vortex Tubes Using a Generalizable and Robust ANN [J].Heat Transfer，2020，51（15）：1399-1415.
8. [8] KAYA，HÜSEYIN，GULER E，et al. Prediction of Temperature Departure of a Nitrogen-Driven Vortex Tube with Linear, Knn, SVM, and RF Regression Models [J].Neural Computing and Applications，2022，35：6281-6291.
9. [9] LI H W，HAO S，NI Y Q，et al. Hybrid Structural Analysis Integrating Physical Model and Continuous-Time State-Space Neural Network Model [J].Computer-aided civil and infrastructure engineering，2025，40（2）：166-180.
10. [10] ZHENG X W，ZHOU S，LV H，et al. Hybrid Physics-BP Neural Network-Based Strength Degradation Model of Corroded Reinforcements Under the Simulated Colliery Environment [J].Structures，2023，50：524-537.
11. [11] XIAO M J，YU T C，ZHANG Y S，et al. Physics-Informed Neural Networks for the Reynolds-Averaged Navier-Stokes Modeling of Rayleigh–Taylor Turbulent Mixing [J/OL].Computers & Fluids，2023，266：106025[2024-09-17].https://doi.org/10.1016/j.compfluid.2023.106025.
12. [12] WEN T，CHEN C，ZHENG G，et al. Using a Neural Network-Physics-Based Hybrid Model to Predict Soil Reaction Fronts [J/OL].Computers & Geosciences，2022，167（C）[2024-09-26].https://doi.org/10.1016/j.cageo.2022.105200.

基金项目

国家自然科学基金（52066006）；景德镇市科技局项目（2019GYZD008-13）

作者简介

李申申（1997—），男，汉族，河南周口人，硕士在读，研究方向：智能检测与控制。