特征标记下的蓄热式连续加热炉炉管安全VR检测技术

VR Detection Technology for the Safety of Regenerative Continuous Heating Furnace Tubes under Feature Labeling

蓄热式连续加热炉炉管开展安全检测时,若采集的炉管损伤图像质量差,会直接降低后续的安全检测精度。为了提升安全检测精度,提出特征标记下的蓄热式连续加热炉炉管安全VR检测技术。该方法首先使用红外热成像技术获取炉管红外图像,并利用双层小波系数对图像展开去噪处理;完成图像去噪后,利用VR技术对去噪图像实施目标重现;结合差分卷积模板,对炉管图像的温度特征实施标记处理,完成温度特征标签化;最后,将温度标签作为模型参数,基于人工神经网络建立安全检测模型,完成蓄热式连续加热炉炉管的安全状态检测。实验结果表明,使用该方法开展炉管安全检测时,炉管红外图像处理效果较好,安全检测精度和效率均较高。

When conducting safety inspections on the furnace tubes of a regenerative continuous heating furnace,if the quality of the collected furnace tube damage images is poor,it will directly reduce the accuracy of subsequent safety inspections.In order to improve the accuracy of safety detection,a feature marked VR detection technology for the safety of regenerative continuous heating furnace tubes is proposed.This method first uses infrared thermal imaging technology to obtain infrared images of furnace tubes,and uses double-layer wavelet coefficients to denoise the images;After completing image denoising,use VR technology to reproduce the target of the denoised image;By combining differ-ential convolutional templates,label the temperature features of the furnace tube image and complete temperature feature labeling;Finally,u-sing temperature labels as model parameters,a safety detection model is established based on artificial neural networks to complete the safety status detection of the regenerative continuous heating furnace tubes.The experimental results show that when using this method for furnace tube safety detection,the infrared image processing effect of the furnace tube is good,and the safety detection accuracy and efficiency are both high.