Probabilistic neural networks method for dividing the blade damage image of aero-engine is used.The paper selects 80 pixel RGB values of the image as input samples of a network.The effective weights and threshold are achieved after training, and expected segmentation results can be realized.The results show that the probabilistic neural networks can better the image segmentation, compared with the traditional image segmentation and other neural networks.
ZHANG Wei-liang, LI Nan, LI Ang, SHI Hong
. Segmentation of blade damage image of aero-engine based on probabilistic neural networks[J]. Journal of Shenyang Aerospace University, 2013
, 30(2)
: 22
-26
.
DOI: 10.3969/j.issn.2095-1248.2013.02.006
[1]McAulay A D.Computerized model demonstrating magnetic submarine localization[J].IEEEtrans, 1977, 13(3):246-254.
[2]孙文雅, 黄民, 李天剑, 等.基于BP神经网络管道裂纹图像分割[J].计算机测量与控制, 2012, 20(5):1363-1365.
[3]叶志峰, 孙健国.基于概率神经网络的发动机故障诊断[J].航空学报, 2002, 23(2):155-158.
[4]罗云林, 赵梅.数字图像处理在发动机孔探技术中的应用[J].中国民航学院学报, 2005(23):54-56.
[5]周开利, 康耀洪.神经网络模型及其MATLAB仿真程序设计[M].北京:清华大学出版社, 2005.
[6]黎群辉, 张航.基于改进概率神经网络的交通标志图像识别方法[J].系统与工程, 2006, 24(4):97-101.
[7]徐小力, 徐洪安, 王少红.旋转机械的遗传算法优化神经网络预测模型[J].机械工程学报, 2003, 39(2):140-144.
[8]朱虹.数字图像处理基础[M].北京:北京科学出版社, 2005.
[9]刘小勇, 樊思齐.采用BP网络辨识航空发动机数学模型[J].航空动力学报, 1996, 11(2):153-156.
[10]丁凯峰, 樊思齐.基于RBF网络的航空发动机辨识模型[J].航空动力学报, 2000, 15(2):205-208.
