Summary
Multi-constrained pipes conveying high-speed fluid, such as aircraft hydraulic control lines, operate in harsh vibration environments. In addition, one or more natural frequencies of the pipe system may fall within the frequency band of environmental excitation. As a result, resonance fatigue can occur, leading to system failure or even catastrophic accidents. Therefore, much attention has been paid to the design of pipe systems with low vibration levels. In this work, a typical hydraulically controlled slender pipe is taken as the object. The dynamic model of the pipe system is set up to obtain the vibration characteristics. By comparing the natural frequency of the pipe with the frequency band of the environmental excitation, a data set of the absolute safety length and the absolute resonance length of the pipe is obtained. Based on the genetic programming (GP) algorithms extensively trained from the data set, the GP model is proposed to accurately predict the absolute safety length and absolute resonance length of pipes. Therefore, mathematical expressions are obtained for the variation of the absolute safety length and absolute resonance length with the location, stiffness, and number of retaining clips. The proposed GP model, as a critical component of the design strategy, effectively bridges the data set with the prediction results, thereby completing the design strategy for low vibration level pipes conveying fluid. In summary, this work combines machine learning methods to suppress pipe system vibration through inverse dynamics design, which provides insights for the design of low vibration pipe systems.
