火焰燃烧的特征量提取及稳定性识别

Feature Extraction and Stability Recognition of Flame Combustion

为了有效提取火焰燃烧特征量,进而对火焰燃烧稳定性进行分析,文章通过理论分析和实验总结,首先提取了火焰强度方差及闪烁频率作为特征量,之后重点运用高阶统计量对火焰信号进行分析,提取了对火焰稳定性有较大影响的相位作为另一个新的特征量。最后考虑到各个特征量之间的相关性及量纲的影响,运用马氏距离作为判别准则,对火焰燃烧稳定性进行了有效的识别。结果表明,提取的3个特征量对火焰燃烧稳定性都有重要影响,而且运用马氏距离能够综合这3个特征量,使得从总体样本中选出的训练样本识别率达到93%,识别效果理想。

To extract the features of flame combustion and recognize the flame stability effectively, theoretical analysis and experimental summary were taken. Firstly, the paper extracted intensity variance and flicker frequency of flame as a feature. Then, appling higher - order statistics analysis to the flame signal, the phase that had greater impact on flame stability was extracted as another one. Finally, we used the Mahalanobis distance as a criterion to recognize the flame stability effectively. Results show that the features ex- tracted have great impact on flame stability, and the Mahal- anobis distance can synthesize these features, therefore make the recognition rate of the training samples chosen from the total samples reach to 93%.