基于图像对抗卷积自编码的燃烧稳定性定量监测

Quantitative Monitoring of Combustion Stability Based on Image Adversarial Convolutional Autoencoder

燃烧稳定性的准确监测对于优化燃烧状态具有重要意义。传统燃烧稳定性监测方法不仅高度依赖先验专家知识,而且难以实现定量评估。为了克服这些局限性,该文提出一种新的燃烧稳定性定量评价方法。该方法建立一种对抗卷积自编码(adversarial convolutional autoencoder,ACAE)提取火焰图像的深层特征,并利用一种定量评价指标进行特征分析。其中,ACAE采用一种新型对抗机制来提高训练效率,从而增强特征学习能力;定量评价指标的数值区间为[0,1],且当评价指标低于0.5时,燃烧状态稳定。通过乙烯燃烧实验测试了稳定性监测方法的可行性。结果表明,由ACAE提取的深层图像特征能够用于定量估计燃烧稳定性。此外,该监测方法具有较强的泛化性能,能够准确识别训练数据集范围以外的火焰图像。

Accurate monitoring of combustion stability is of great significance in optimizing the combustion state.Traditional combustion stability monitoring methods are not only highly dependent on prior expert knowledge,but also difficult to achieve quantitative evaluation.To overcome these limitations,a novel quantitative assessment method for combustion stability is proposed in this study.In this method,an adversarial convolutional autoencoder(ACAE)is established to extract deep features of the flame images,and a quantitative assessment index is applied for feature analysis.Especially,the ACAE adopts a novel adversarial mechanism to improve the training efficiency and thereby enhance the feature learning ability.The numerical interval of the quantitative assessment index is[0,1],and when the assessment index is lower than 0.5,the combustion state is stable.The feasibility of the stability monitoring method is verified by ethylene combustion experiments,and the testing results confirm that the deep image features extracted by the ACAE can be used to quantitatively estimate the combustion stability.Furthermore,the proposed monitoring method has a strong generalization performance that can accurately identify flame images beyond the scope of the training dataset.