Embedded Files
Wind Tunnel Wall Interference Correction based on Data and Artificial Neural Network
Wind Tunnel Wall Interference Correction based on Data and Artificial Neural Network
Fig 1. Wind tunnel test
Wind tunnel test is an important part in the development of aircrafts, such as analyzing performance of aircraft and generating aerodynamic data. In case of the wind tunnel test, unlike the free-air state, errors due to the wind tunnel wall effect are included in the data because it simulates flights in the closed wall state. These errors cause a difference between CFD data and wind tunnel data and make it difficult to predict the performance of the aircraft.
Wind tunnel test is an important part in the development of aircrafts, such as analyzing performance of aircraft and generating aerodynamic data. In case of the wind tunnel test, unlike the free-air state, errors due to the wind tunnel wall effect are included in the data because it simulates flights in the closed wall state. These errors cause a difference between CFD data and wind tunnel data and make it difficult to predict the performance of the aircraft.
Fig 2. Prior methods used to correct wind tunnel wall interference
Wind tunnel wall interference correction methods include empirical method, numerical method and wall boundary measurement method. Classical method which is one of the numerical methods are mainly used in academia and on the ground. However, in the case of existing methods, there is a limit to correcting wall interference by accurately considering the effects of parameters such as wind tunnel type, test wall type and flow conditions. Therefore, in this study, after analyzing the sensitivity of parameters affecting the wall interference through data mining, wall interference was corrected by constructing the deep neural network which uses major parameters as input parameters.
Wind tunnel wall interference correction methods include empirical method, numerical method and wall boundary measurement method. Classical method which is one of the numerical methods are mainly used in academia and on the ground. However, in the case of existing methods, there is a limit to correcting wall interference by accurately considering the effects of parameters such as wind tunnel type, test wall type and flow conditions. Therefore, in this study, after analyzing the sensitivity of parameters affecting the wall interference through data mining, wall interference was corrected by constructing the deep neural network which uses major parameters as input parameters.
Fig 3. Research overview
Final goal of this research : Confirmation of the possibility of a data-based simple and rapid wind tunnel wall interference correction method that exceeds the limitations of existing methods.
Final goal of this research : Confirmation of the possibility of a data-based simple and rapid wind tunnel wall interference correction method that exceeds the limitations of existing methods.
Fig 4. Self-organizing map
Fig 5. Structure of deep neural network
Page updated
Report abuse