【作者】 周颖慧；

【导师】 郑德忠；

【作者基本信息】 燕山大学 ， 精密仪器及机械， 2010， 博士

【摘要】 温度是高温热处理工业中的重要热工参数。实时准确的温度场监测是提高产品质量、预防工业事故、节约能源、减少污染的有效手段。传统的高温测量技术多采用热电偶等接触式传感器,但其侵入式单点测量的方式已不能满足目前工业测量的需要。基于CCD图像传感器的非接触测温技术是集电子技术、传热学、图像处理技术于一体的高新技术,以其响应迅速、非接触测量、适用范围广等优点成为高温检测领域中的研究热点之一。目前直接利用CCD测温的方法还处于实验室研究和工厂试验阶段。测温精度低、动态测温范围窄,实时处理能力弱、便携性差是制约CCD测温技术应用的主要因素。论文分析了CCD的基本工作原理与辐射测温方法,首先搭建了线阵CCD测温系统。采用高灵敏度黑白线阵CCD实现辐射体单点温度或均匀温度场测量。通过在CCD前加载滤光片获取单波长下的灰度输出,运用比色法获取辐射温度。结合实验讨论了波长选择对测温精度的影响,针对系统在不同波长组合下的温度输出具有多传感器关联特性,提出了自适应加权实时数据融合算法,在改善测量结果的基础上提高了系统的实时性。研究了面阵CCD数字成像原理与图像测温的基本方法,建立了基于彩色面阵CCD的测温系统,用于全温度场监测。在分析传统测温系统的优势与不足的基础上,设计了基于面阵CCD的单镜头双色测温系统,采用一个分光滤光光路实现了单CCD对三色图像的同步获取,降低了系统误差。分析了系统中辐射体温度与辐射图像像点灰度之间的映射关系,利用比色原理推导出该系统的温度计算公式。根据人造黑体原理,设计了满足工程需要的实验室用黑体炉,用于标定公式中的系统修正因子并进行测温实验研究。为提高系统测温精度且便于工程应用,对系统潜在的误差源进行了分析并提出了相应的解决方案。针对测温系统中CCD光谱响应带宽引起的误差,提出了基于CCD光谱响应曲线寻迹的数字滤光方法,利用曲线寻迹思想和数字离散化技术获得单波长下的灰度输出,以算法模拟滤光功能,使系统无需任何外置滤光光路即可实现辐射测温,减小了系统误差。此外,受周围环境和测温设备本身的影响,获取的辐射图像含有大量的未知噪声。针对该问题,提出了基于改进GCV准则的小波图像去噪方法。利用小波分析技术和改进的广义交叉验证算法对摄取后的含噪图像进行降噪处理,降低噪声信号对测温精度的影响。研究了CCD测温系统的动态测温范围。针对CCD感光特性导致系统可测温区较窄的问题,提出了彩色图像灰度拟合算法。在黑体炉上建立灰度拟合关系式,估算饱和温区的虚拟灰度值,以增加可用灰度比来扩展测温区间。为提高拟合精度,通过分析灰度比曲线对算法进行了校正。为便于测温系统的工程应用,设计开发了基于FPGA(Field Programmable Gate Array)的便携式实时测温系统和基于PC的全温度场实时监控系统。前者采用一块FPGA实现了数据采集、温度计算和显示,适用于对实时性要求较高的现场测温,为开发便携式测温仪打下了基础;后者可进行图像采集、算法处理、灰度分析和温度场显示等,适用于工业热处理过程的后台监控与后续分析,为全面监测热处理过程,改善燃烧状况、节能减排提供了技术支持。更多还原

【Abstract】 Temperature is one of the important parameters in the high temperature heat treatment field. The real-time and accurate measurement of temperature field is an effective mean for improving product quality, preventing industrial accident, conserving energy and reducing environment pollution. The conventional contact sensor such as thermocouples would not meet the demands of industry development because of the drawback of single-point and intrusive measurement. The uncontact temperature measurement using a CCD image sensor is a new technique that integrates electronics, heat transfer theory and image processing technique, and attracts the attention of the researchers on account of its obvious advantages like quick response, uncontact measurement and wide available range. The method now is still in research and experimental stage. The main factors limiting the technology application include low measurement accuracy, narrow measurable temperature scope, weak real-time processing ability and poor portability.In the paper, the basic principle of CCD is analyzed and a prototype system of temperature measurement using a linear CCD sensor is presented firstly. The system is able to get the radiation image gray-level at a single wavelength by loading the filter in front of the CCD and then calcuates point temperature or uniform temperature field of radiation object by applying the two-color theory. Then the method of wavelength selection and its effect on the measured results are discussed combined with the experimental. In view of the multi-sensor characteristic of temperature output at varity wavelength combinations, an algorithm of adaptive weighted real-time data fusion is presented. It improves the accuracy of temperature measurement and efficiency of system operation.Combining the digital imaging processing technique and two-color temperature measurement method, a novel instrumentation system is built for whole temperature field monitoring using a color matrix CCD camera. On the basis of analyzing the advantages and the disadvantages of the conventional methods, a two-color temperature measurement system with a single camera is designed. The system can capture the three monochrome images synchronically by using a splitting/filtering optical path, which reduces the system error. The relationship is deduced between the radiation temperature and pixel gray-level, and then a temperature expression is given based on the two-color theory. To meet measurement needs, an industrial black-body furnace is made for system calibration and experimental research.To improve the accuracy of temperature measurement and easy to be used in engineering application, the system’s potential error sources are analyzed and the corresponding solution are given. Aiming at the error aroused by CCD spectral response bandwidth, a digital filtering method is proposed to get gray-level output at single wavelength by applying the thought of CCD spectral responsed curve tracing and the digital image discretization technique. The system simulates the filter effect with an algorithm to accomplish the temperature measurement using CCD camera without any external filtering optical path. Besides, the radiation image captured by CCD camera involves much unknown noise from bug dust, environment light and optical device. For this problem, an algorithm of multiple wavelet images de-noising based on improved Generalized Cross Validation (GCV) law is adopted to purify the radiation image and reduce measuring error.The issue on the dynamic temperature measurement scope is discussed. The CCD sensor’s photo-behavior results in measurable temperature range too narrow to show the whole temperature field accurately. For solving the problem, a gray-level fitting algorithm is presented to establish an expression about RG gray-level to estimate the virtual gray-level on the saturation pixel. Increased gray-level ratios will extend the range of temperature measurement. To decrease the fitting error, the algorithm mentioned above is corrected through analyzing the curve of gray-level ratio.Two application tools have been developed for convinent to apply the temperature measurement system into industrial field. One is a FPGA(Field Programmable Gate Array)-based portable real-time temperature measurement system. The system, using a FPGA chip to implement data acquisition, temperature calculation and showing, is suitable for getting real-time results of simple point temperature or mean value of temperature zone. The other one is a PC-based monitoring system of temperature field, which is applied to monitor and analyze combustion temperature field with the powerful image processing function including image acquisition, algorithm processing, gray-level analysis and temperature field showing, etc. The two tools provide technique support for the over-all monitoring of heat-treatment processing, improving combustion efficiency, saving energy source and reducing pollution.更多还原