- 學位論文
運用人工智慧於動壓頸軸承之優化設計
Optimizable Design of Hydrodynamic Journal Bearings by Artificial Intelligence
若您是本文的作者，可授權文章由華藝線上圖書館中協助推廣。
摘要
本文使用類神經網路，分析國際標準ISO 7902動壓頸軸承設計參數數據和潤滑油黏溫數據，並將四種軸承設計參數如長徑比 、比摩擦係數 、摩擦係數 與潤滑油流量比 及使用不同潤滑油黏溫關係數據輸入MATLAB類神經網路模型訓練，得到可輸出數據趨勢之模型，應用於設計軸承實務。並說明軟體設定介面與參數。 故本文使用基於軸承設計參數數據建立的類神經網路模型，能使用約總數據量的1/3建立類神經網路，且誤差 ，並運用開頭使用並說明流程的Webplotdigitizer[15]網路開源免費網頁軟體，能快速準確的將圖表取得數據，此一方法亦能運用於僅有實驗圖表，無完整數據之研究項目，能加快得到所求數值的NN模型，應用於如計算問題或比對其他標準上。
並列摘要
This study uses neural networks to analyze design parameter data of hydrodynamic journal bearings based on international standard ISO 7902 and lubricant viscosity-temperature data. Four bearing design parameters—length-to-diameter ratio, specific friction coefficient, friction coefficient, and lubricant flow ratio—along with different lubricant viscosity-temperature relationships were input into a MATLAB neural network model for training. The resulting model, which can output data trends, is applied to practical bearing design. The software interface and parameters are also described. In this paper employs a neural network model based on bearing design parameter data, demonstrating that the neural network can be established using approximately one-third of the total data, with minimal error. Additionally, the Webplotdigitizer [15], a free and open-source web software, is utilized at the beginning to extract data from graphs quickly and accurately. This method is particularly useful for research projects where only experimental graphs are available without complete datasets, allowing for faster acquisition of the required values for neural network models. It can be applied to computational problems or used for comparison with other standards.
參考文獻
[1] ISO 7902-1,Hydrodynamic plain journal bearings under steady-state conditions - Circular cylindrical bearings - Part 1: Calculation procedure,2020
[2] ISO 7902-2,Hydrodynamic plain journal bearings under steady-state conditions - Circular cylindrical bearings - Part 2: Functions used in the calculation procedure,2020
[3] ISO 7902-3, Hydrodynamic plain journal bearings under steady-state conditions - Circular cylindrical bearings - Part 3: Permissible operational parameters,2020
[4] Ståhl, J., & Jacobson, B. O. (2001). Design functions for hydrodynamic bearings. Proceedings of the Institution of Mechanical Engineers, Part J: Journal of Engineering Tribology, 215(5), 405-416.
[5] Hirani, H., & Suh, N. P. (2005). Journal bearing design using multiobjective genetic algorithm and axiomatic design approaches. Tribology International, 38(5), 481-491.