摘要： 针对电站锅炉火焰监视、温度场测量和燃烧诊断中存在的问题，提出了一种以DSP为核心的嵌入式图像火焰监测系统，在对彩色图像法测量温度场的原理论述的基础上，对图像采集和处理各部分的电路和系统软件进行了分析和说明。通过优化设计，使系统的数据处理能力和实时性得到大幅的提升。经过200MW机组的试验表明，系统能在1帧时间内完成一幅图像的处理与特征量的提取，在1s内完成锅炉燃烧状况的诊断，火焰监测和燃烧诊断实时性得到了保证。温度场测量最大偏差不会超过50℃，相对误差小于5%，能满足系统对测量精度的要求。

关键词: 炉膛火焰, 温度场, 数字图像处理, 燃烧诊断, DSP

Abstract: This paper proposes embedded system for picture flame detection with the core of DSP, aiming at the problems on flame monitoring, temperature field measure and combustion diagnosis. On the basis of temperature field measure principle using color image method, all circuits and system soft of image acquisition and processing are analyzed and illuminated. Though optimum design, system ability of data processing and real-time is heightened heavily. Experimentation of 200MW unit shows that system can accomplish image processing, extract statistic feature within 1 frame time for one picture, and accomplish diagnosis for boiler burning conditions within 1 second. So the real-time of flame detection and burning diagnosis is assured. The maximum error of temperature distribution measure is at range of 50℃ and relative error is less than 5%. Therefore, these can meet the requirement of accuracy measurement.

Key words: Furnace flame, Temperature field, Digital image processing, Burning diagnosis, DSP