Heat transfer analysis using ANN with experimental data of 180° turn channels with rib turbulators
-
摘要: 在对带有不同肋间距和肋高的直肋变截面U型通道的换热性能进行了实验研究的基础上,分单元处理了数据,并采用经过LM(Levenberg-Marquardt)算法和贝叶斯正则化方法改进的基于BP(Back-Propagation)算法的前馈神经网络对数据结果进行了建模,实现了对带直肋的变截面U型通道换热性能的预测,实践证明,人工神经网络ANN(Artificial Neural Network)方法可以用于航空发动机涡轮叶片内通道换热性能的预测,并且其预测精度明显高于非线性拟合.Abstract: The heat transfer character inside the 180° turn channel with variable cross-section and different rib turbulators was studied by experiment. The channel is the middle part of a gas turbine blade. The data set evaluated by the experiment was processed by dividing the channel into several sections. A feed-forward neural network based on back-propagation algorithm (BP) was evolved with levenberg-marquardt(LM) algorithm and Bayesian regularization, and a model was established for the result referred before using BP net. It is validated by being compared with curve regression that it is available to use artificial neural networks (ANN) to predict heat transfer character of channel inside a gas turbine blade. And it-s much more efficient than curve regression.
-
Key words:
- air engines /
- turbine /
- convective heat transfer /
- neural networks /
- prediction
-
[1] 周开利,康耀红.神经网络模型及其MATLAB程序设计[M].北京:清华大学出版社,2005 Zhou Kaili,Kang Yaohong. Model of neural net work and design with Matlab[M]. Beijing:Publishing Company of Tsinghua University,2005(in Chinese) [2] 刘湘云,丁水汀,陶智,等.横肋变截面U通道内换热特性的实验研究[J].工程热物理学报,2004,25(5):858-860 Liu Xiangyun, Ding Shuiting, Tao Zhi, et al. Experimental study of heat transfer characteristics in the U shaped channels with variable cross-section and rib [J]. Journal of Engineering Thermophysics, 2004, 25(5):858-860(in Chinese) [3] Hagen M T, Menhaj M B. Training feed-forward networks with the Levenberg-Marquardt algorithm IEEE Trains Neural Networks,1994,5(6):989-993 [4] Foresee F D, Hagan M T. Gauss-Newton approximation to Bayesian regularization Proceedings of the 1997 International Joint Conference on Neural Networks,1997:1930-1935 -
点击查看大图
计量
- 文章访问数: 2546
- HTML全文浏览量: 176
- PDF下载量: 943
- 被引次数: 0
下载:
