摘要
在陶瓷产品生产过程中,不同烧制阶段陶瓷梭式窑烧结带温度发生相应的变化,其对应的火焰图像也随着变化。本文针对陶瓷梭式窑烧结带温度检测提出一种基于改进BP神经网络的火焰图像识别方法。首先对获取的火焰图像利用改进的小波阈值算法去除图像中的噪音进行预处理,其次基于改进的BP神经网络对得到的火焰图像三个分量值R、G、B和测得的火焰温度进行数据拟合,最后测试已训练的神经网络识别火焰图像的效果。实验结果表明,改进后的BP神经网络收敛速度更快、训练时间更短、误差更小,能够更好地检测陶瓷梭式窑火焰图像温度。
In the production process of ceramic products,the sintering zone temperature of ceramic shuttle kiln changes corr espondingly at different firing stages,so does the corresponding flame image.In this paper,an improved BP neural network-based flame image recognition method is proposed for the temperature detection in ceramic shuttle kiln.Firstly,the improved wavelet threshold algorithm is used to remove the noise in flame image,that is,image preprocessing.Secondly,based on the improved BP neural network,the data of the three components R,G and B obtained from the flame images and the measured flame temperature data are fitted.Finally,the recognition effect of the trained neural network on flame images is tested.The experimental results show that the improved BP neural network has faster convergence speed,shorter training time and smaller error,and can detect the flame image temperature of the ceramic shuttle kiln better.
作者
朱永红
夏力
王俊祥
ZHU Yonghong;XIA Li;WANG Junxiang(School of Mechanical&Electronic Engineering,Jingdezhen Ceramic Institute,Jingdezhen 333403,Jiangxi,China)
出处
《中国陶瓷工业》
CAS
2020年第1期16-20,共5页
China Ceramic Industry
基金
国家自然科学基金(61563022,61762054)
江西省重大自然科学基金(20152ACB20009)。
关键词
BP神经网络
动态学习效率函数
激活函数
火焰图像识别
BP neural network
dynamic learning efficiency function
activation function
flame image recognition
