弹道
阻尼器
润滑
物理
控制理论(社会学)
振幅
机械
灵敏度(控制系统)
理论(学习稳定性)
粒子群优化
人工神经网络
体积热力学
粒子(生态学)
时域
作者
Zhongliang Xie,Yihao Sun,Ming Jin Yang,Jiabin Tian,Bin Zhao,Xiaohui Zhang,Boyao Ding
出处
期刊:Physics of Fluids
[American Institute of Physics]
日期:2025-09-01
卷期号:37 (9)
被引量:1
摘要
The influence of non-circular whirl trajectory parameters on the dynamic characteristics of squeeze film damper has not been well understood. In this study, a fluid domain model was established to analyze the influence of whirl trajectory parameters on lubrication characteristics under different structure and working condition parameters, revealing the underlying evolution mechanism. Four machine learning models were established to enable rapid evaluation of dynamic behavior. The results show that non-circular whirl trajectories cause lubrication characteristics to vary sinusoidally with the whirl angle. The whirl trajectory has the most pronounced effect on the maximum pressure, the average value, and amplitude of the steam volume fraction. With increase in the whirl trajectory parameters, the stability of the dynamic characteristics decreases. Under most operating conditions, when the non-circular whirl trajectory parameter is 2.5, the amplitude increase in dynamic characteristics fluctuation is more than 10 times compared with the circular whirl trajectory. The sensitivity of width to damping is the highest for conventional parameters. Among the four machine learning models, the particle swarm optimization-backpropagation neural network algorithm has the best prediction performance, with R2-values exceeding 0.93. The research results provide a new approach for rapid evaluation of dynamic coefficients in squeeze film dampers under non-circular whirl conditions.
科研通智能强力驱动
Strongly Powered by AbleSci AI