| 摘 要: 为了克服传统当量比测量模型存在的局限性,提出了一种基于机器学习的多特征输入方法。该方法摒弃了以往单一依赖CH*与C2*比值的方式,而是综合利用C2*swan band的多个化学发光波段。为了验证这种多输入方法的有效性,采用支持向量机(SVM)和多层感知机(MLP)两种不同的机器学习模型进行实验验证。实验结果表明,多特征输入在这两种模型上展现出优异的回归预测性能,回归系数均达到了0.99,证明了所提方法在精确测量当量比方面的可靠性和有效性。 |
| 关键词: 当量比;多特征输入;支持向量机;多层感知机 |
|
中图分类号: TP391.41
文献标识码: A
|
|
| Measurement of Methane Premixed Flame Equivalence Ratio Based on Machine Learning |
|
ZHANG Yan, YANG Jiansheng
|
(The Electrical Engineering College, Guizhou University, Guiyang 550025, China)
357994559@qq.com; 1158705802@qq.com
|
| Abstract: In order to overcome the limitations of the traditional equivalence ratio measurement model, a multifeature input method based on machine learning is proposed. Instead of relying on a single CH to C2 ratio, this method integrates multiple chemiluminescence bands of the C2 swan band. To validate the effectiveness of this multifeature method, two distinct machine learning models, Support Vector Machine (SVM) and Multilayer Perceptron (MLP), are employed for experimental verification. The experimental results demonstrate that the multi-feature input exhibits excellent regression prediction performance on both models, with regression coefficients reaching 0.99, thereby proving the reliability and effectiveness of the proposed method in accurately measuring the equivalence ratio. |
| Keywords: equivalence ratio; multi-feature input; Support Vector Machine (SVM); Multilayer Perceptron (MLP) |
