Scholarlink

<p>In recent years, due to military and environmental reasons, attention to the issue of aerodynamic noise has increased a lot in the world. One of the most crucial aspects of engineering structural design is noise prediction. The special attention paid to the design and construction of modern equipment with minimum noise is the origin of extensive research in noise control methods. In the present study, a model drawing on artificial neural networks (ANNs) that forecasts the sound pressure level released by airfoils was provided. For this purpose, different layers were used to define the architecture of the recommended scheme, such as the dropout layer, which had a fraction of 0.2. Also, the Adam algorithm was deployed to boost the learning rate and sigmoid function for activation. The accuracy of the suggested model was compared with five popular traditional schemes, including decision tree (DT), support vector regression (SVR), linear support vector regression (linear SVR) and adaptive boosting (AdaBoost), using a case study based on the dataset for NACA 0012 airfoils. The findings have displayed that, when compared to other schemes, the suggested model had the best evaluation index values and the lowest mistake rate. Therefore, the defined ANN-based model is suggested to project the sound pressure level in this study.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Prediction of airfoil sound pressure level using ANN-based model

  • 期刊论文

  • 发表日期：

    2026年3月11日

查看全文
  • Yetong Wang,
  • Bing Zheng,
  • Kongduo Xing

摘要

In recent years, due to military and environmental reasons, attention to the issue of aerodynamic noise has increased a lot in the world. One of the most crucial aspects of engineering structural design is noise prediction. The special attention paid to the design and construction of modern equipment with minimum noise is the origin of extensive research in noise control methods. In the present study, a model drawing on artificial neural networks (ANNs) that forecasts the sound pressure level released by airfoils was provided. For this purpose, different layers were used to define the architecture of the recommended scheme, such as the dropout layer, which had a fraction of 0.2. Also, the Adam algorithm was deployed to boost the learning rate and sigmoid function for activation. The accuracy of the suggested model was compared with five popular traditional schemes, including decision tree (DT), support vector regression (SVR), linear support vector regression (linear SVR) and adaptive boosting (AdaBoost), using a case study based on the dataset for NACA 0012 airfoils. The findings have displayed that, when compared to other schemes, the suggested model had the best evaluation index values and the lowest mistake rate. Therefore, the defined ANN-based model is suggested to project the sound pressure level in this study.