基于差分多光谱成像原理的铸坯表面高温场连续测量传感器

The surface temperature of billets is very signigicant to optimize secondary cooling system and to improve billets quality. An on-line accurate measurement for it is a yet unsolved difficult problem in metallurgical detections. The key problems are: the uncertainty of surface emissivity is difficult to realize true temperature measurement; it is difficult to achieve equal-accuracy measurement for various temperature ranges in whole temperature field due to the effect of oxide coatings and as large as 400 ℃ temperature span. This project proposes an idea of the new imaging temperature sensor based on CCD with features of differential multi-spectrum, and makes its prototype design. By optimizing the wavelength set effectively, the measuring error resulted from the factor of emissivity could be overcome and the absolute accuracy of sensor could be greatly improved; by photochopper set the thermal images of different spectrums and dark signals are sequentially acquired on a single-color CCD. In order to improve the realtime of measurements for the sensor, multi-pipeline is used for the main processor DSP; To remove the disturbances resulted from the random oxide coatings of billets and the influence of CCD dark voltage, a spatial digital filtering method is adopted. In this method, firstly the temperature extremums about each row of pixels are extracted from raw temperature field, and they are regarded as the information from a single pixel sensor in time domain. Then through discrete Fourier transform and low-pass filter, the random noises supression is realized.

铸坯表面温度对优化二冷制度和提高铸坯质量具有重要意义，对其在线准确测量是至今尚未解决的冶金检测难题。关键问题是：表面发射率的不确定性难以实现真温测量；表面氧化皮影响和高达400℃的温度梯度难以实现整个温度场不同温度段的等精度稳定测量。 本项目提出基于CCD的差分多光谱新型图像测温传感器设想并进行原型机设计。通过优化波长组有效克服发射率产生的测温误差，提高传感器的绝对测量精度； 采用斩光结构在单色CCD上顺次采集多光谱热图像和暗电压信息，并通过DSP及多级流水线结构保证温度场测量的实时性；为消除连铸坯表面随机分布的氧化皮和CCD像素暗电压的扰动，采用空间数字滤波法，从原始温度场中提取行像素传感器阵列测得的极值分布，经列重组后将其视作单一像素的时域信息，再经离散傅氏变换和低通滤波实现噪声抑制。

铸坯表面温度的准确稳定测量对优化二冷制度、判断液相穴深度和优化拉坯速度具有重要意义。本项目通过研制高温场图像测温传感器重点解决了多流铸坯的表面温度场的稳定测量问题，研究内容包括原型机的设计、高温场图像测温模型的建立、空间滤波算法和传感器标定及噪声分析。经过一年的探索，原型机在南钢进行了试应用，测试结果显示传感器性能基本正常，测量稳定度可控制在±6℃之内，动态范围可达300℃，证明了项目整体方案的可行性。同时，项目建立了较为完善的CCD高温测量理论，促进了冶金行业中表面测温技术的发展，为后续多光谱图像测温及测温装置的实用化奠定了坚实的理论和技术基础。项目资助获得省科技进步二等奖一项；撰写SCI论文一篇（修改中）；申请国家发明专利一项（已受理）；培养硕士研究生5名，青年教师1名。项目总经费16万元，支出12.2万元，各项支出基本与预算相符。剩余经费3.8万元，计划用于后续研发。