基于图像的高炉风口燃烧充分性检测方法研究

摘要

研究高炉燃烧效率检测问题,提高检测的准确性.针对高炉风口中火焰受到不定风向影响,火苗像素出现随机性方向波动时,建立的像素模型发生变形,造成图像失真.传统的图像的燃烧充分性检测方法过于依赖充分的像素模型,造成对锅炉内可燃物质燃烧充分性检测准确度不高.为了解决上述问题,提出了一种基于神经网络的火焰燃烧充分性检测方法,选取与火焰燃烧充分性直接相关的多个特征作为神经网络的输入向量,通过对样本的多次迭代训练去除风向突变引起的微小脉动的影响.仿真结果表明,改进方法能够有效避免风向突变对像素波动造成的影响,提高了检测的准确性,取得了满意的结果.%The blast furnace burning efficiency detection problems were stued to improve the accuracy of the test. The paper proposed a flame burning sufficiency detection method based on neural network. The characteristics which directly related to fire burning sufficiency were selected as the input vector of the neural network. The the neural network was trained to remove the influence of small pulse caused by wind mutation. Experiments show that the method can effectively avoid the influences of wind mutations on pixels, and satisfactory results were obtained.