The prediction model of automotive aerodynamic drag coefficient can make a quick decision on the influence of characteristic parameters on aerodynamic drag coefficient and reduce the calculation times in automotive aerodynamic development. In order to establish a high-precision prediction model to meet the requirements of automobile aerodynamics design, Hammersley sampling is used to select sample points, and the prediction model of wind drag coefficient based on extreme learning machine (ELM) is established. Then, the moth flame algorithm (MFO) is used to optimize the model to improve its generalization ability. Finally, the prediction model is applied to aerodynamics design of a passenger car. The results show that the optimized prediction model is lower than the original prediction model of extreme learning machine, and the root mean square error is only 0.593%, which is 43.57% less than the original prediction error. In the aerodynamic design of a passenger car, the resistance value obtained by using MFO-ELM model is lower than that obtained by conventional optimization, and the drag reduction rate is 7.04%, and the number of simulation calculation is reduced by 20.8%, which meets the design requirements.
Prediction model of automobile wind drag coefficient based on moth optimization algorithm and ELM
01.08.2022
452158 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
Real world drag coefficient - is it wind averaged drag
Tema Archiv | 2014
|Accurate Fuel Economy Prediction via a Realistic Wind Averaged Drag Coefficient
SAE Technical Papers | 2017
|Accurate Fuel Economy Prediction via a Realistic Wind Averaged Drag Coefficient
British Library Conference Proceedings | 2017
|