基于数字图像处理的温度检测技术研究

【作者】 李汉舟；

【导师】 潘泉； 张洪才；

【作者基本信息】 西北工业大学 ， 控制理论与控制工程， 2002， 硕士

【摘要】 高温物体的温度场实时测量一直是理论界和工业界探索和研究的一个重要课题。现代工业生产急需发展一种实时性强、精度较高的温度场测量技术，这对提高产品质量、优化燃烧过程、节约能源、保护环境以及工业生产安全都有重要意义。 应用彩色CCD摄像机获取炉内火焰图像，进而计算出工业炉内温度场分布情况是一项全新的技术。本文详细推导了在高温辐射体具有实物面的情况下，物体温度和CCD图像灰度之间的关系；讨论了该方法的测量误差及减少误差的措施。 我们利用自己研制的WHT-01黑体炉标定了一套测温系统。根据标定结果，本文用最小二乘和神经网络两种方法分别对黑体炉和普通煤炉的图像进行了温度计算；通过比较热电偶输出的温度值和计算的结果，提出了一种改进的神经网络方法，该方法更适用于工程实际，其测温精度较高。实验表明我们的图像预处理和温度计算方法基本可行。 最后，针对目前工业炉燃烧控制现状，本文将基于数字图像处理的温度检测技术与模糊控制技术相结合，初步探讨了将其应用于燃烧控制的可行性。更多还原

【Abstract】 Temperature field measurement is an important subject in both practice and theory. Modern industry expects a real time and high precision temperature distribution test method, which would be positively significant in improving the quality of products, optimizing burning process, saving energy resource, protecting environment and production safety.It is a new technology to use color CCD camera to acquire flame images and calculate the temperature distribution. The expression between the temperature of the objects that have real surface and the luminance of pixel in the image produced by CCD camera is deduced in this paper. The measurement errors are analyzed in this paper.A black body furnace WHT-01 has been developed in order to validate mis new temperature field measurement method. After the temperature measurement system is scaled by the black body furnace, we use least square method and neural network to calculate the temperature respectively according to the images taken from both the black body furnace and the common coal furnace. Through comparison the calculation data with the result of thermocouple, an improved method of neural network is proposed. This method is more practical and precise. It is shown that the new temperature measurement and our algorithms are effective.At last, this paper tries to combine temperature field measurement based on digital image processing with fuzzy control to design a new burning process controller.更多还原