|
| Prediction of Surface Temperature Change Trend of Electric Heating Anti-icing and De-icing Based on Machine Learning |
| Received:March 25, 2021 Revised:May 08, 2021 |
| View Full Text View/Add Comment Download reader |
| DOI：10.7643/issn.1672-9242.2021.08.006 |
| KeyWord:machine learning icing wind tunnel electric heating anti-icing and de-icing K-nearest neighbor regression algorithm local linear weighted regression algorithm |
| Author | Institution |
| RAN Lin |
Key Laboratory of Icing and Deicing, China Aerodynamics Research and Development Center, Mianyang , China |
| XIONG Jian-jun |
Key Laboratory of Icing and Deicing, China Aerodynamics Research and Development Center, Mianyang , China |
| ZHAO Zhao |
Key Laboratory of Icing and Deicing, China Aerodynamics Research and Development Center, Mianyang , China |
| ZUO Cheng-lin |
Key Laboratory of Icing and Deicing, China Aerodynamics Research and Development Center, Mianyang , China |
| YI Xian |
Key Laboratory of Icing and Deicing, China Aerodynamics Research and Development Center, Mianyang , China |
|
| Hits: |
| Download times: |
| Abstract: |
| For the application of aircraft component cycle control law electric heating anti-icing, this paper proposes to predict the surface temperature change trend of electric heating anti-icing surface based on machine learning. The large-scale icing wind tunnel and its electric heating anti-icing and de-icing control system is used to complete the test, obtain valid test data, and divide the test data into several samples with the on and off cycles as the division unit. According to the heat exchange of the electric heating anti-icing de-icing process, the characteristic parameters of the sample are constructed. The supervised learning method is used to predict and calculate the temperature change rate and convert it to temperature through the K-nearest neighbor regression algorithm and the local linear weighted regression algorithm. The temperature change obtained is the linear correlation of the measured data reaches a high similarity result of more than 80%, which indicates that the use of machine learning can quickly predict the change trend of the surface temperature of electric heating anti-icing components, and different regression algorithms have different prediction effects for different temperature measurement points of the model. |
| Close |
|
|
|