Prediction model for aerodynamic coefficients of iced quad bundle conductors based on machine learning method

Area of Science:

• Aerospace Engineering
• Mechanical Engineering
• Electrical Engineering

Background:

• Iced conductors are susceptible to aerodynamic instability, specifically galloping.
• Accurate prediction of aerodynamic forces (lift, drag, torsional moment) is crucial for conductor safety.
• Existing models may not fully capture the complex interactions of ice, wind, and conductor geometry.

Purpose of the Study:

• To numerically simulate and investigate the aerodynamic coefficients of iced quad bundle conductors.
• To develop an efficient prediction model for these coefficients using machine learning.
• To validate the prediction model's effectiveness in analyzing galloping behavior.

Main Methods:

• Numerical simulation of lift, drag, and torsional moment coefficients under various ice and wind conditions.
• Utilizing Latin hypercube sampling to create comprehensive datasets.
• Employing the extra-trees algorithm to train and test a prediction model.

Main Results:

• A validated prediction model for aerodynamic coefficients of iced quad bundle conductors was successfully developed.
• The model accurately predicts lift, drag, and torsional moment coefficients based on input parameters.
• The model's aerodynamic coefficients were used to effectively analyze galloping features.

Conclusions:

• The developed prediction model offers an efficient tool for the rapid investigation of iced conductor aerodynamics.
• This model is vital for the prediction and early warning system of conductor galloping.
• The study highlights the importance of considering various ice and wind parameters in conductor design and safety.