摘要
燃烧器火焰状态的有效监测对燃煤电厂的安全生产和节能减排都具有重要作用。现有的火检系统主要以检测火焰是否存在为主要任务,不能对火焰状态进行准确判断。因此以实际条件下获取到的火焰图像为对象,以支持向量机作为基础分类器,提出了一种融合火焰图像颜色和轮廓特征且使用改进粒子群优化算法进行参数寻优的火焰状态识别方法。实验结果表明,在数据量充足的前提下,所提出的方法对3种不同负荷状态下的燃烧器火焰图像分类准确率达到99%以上,且改进后的参数寻优算法较之传统算法有更好的性能,同时在低数据集容量情况下表现更好。
The effective monitoring of burner flame status in coal-fired power plants is crucial to production safety,energy-saving and emission reduction.Existing fire detection systems mainly focus on detecting the presence of flame as the main task,and cannot make accurate judgments on the state of the flame.Therefore,a method for determining the flame states with different load conditions is proposed.The support vector machines are used as a basic classifier to combine the flame color and contour characteristics,which are derived from the flame images obtained in an actual power plant.Furthermore,an improved particle swarm optimization algorithm is employed for parameter optimization.Experimental results demonstrate that with sufficient data,the proposed method achieves a classification accuracy of more than 99%on flame images with three load conditions.In addition,the improved parameter optimization algorithm outperforms traditional methods,and the proposed method exhibits better performance with low dataset capacity.
作者
钱相臣
马赟
徐伟程
付伟
QIAN Xiangchen;MA Yun;XU Weicheng;FU Wei(School of Control and Computer Engineering,North China Electric Power University,Beijing 102211,China)
出处
《计量学报》
北大核心
2025年第11期1568-1573,共6页
Acta Metrologica Sinica
基金
国家自然科学基金(51827808)。
关键词
火焰状态监测
燃烧诊断
数字图像处理
支持向量机
粒子群优化算法
flame states monitoring
combustion diagnosis
digital image processing
support vector machine
particle swarm optimization algorithm
