航空学报 > 2023, Vol. 44 Issue (S2): 729309-729309   doi: 10.7527/S1000-6893.2023.29309

基于多光谱和复阻抗的复合结冰探测技术

刘进一1,2, 熊建军3, 桂康1,2,3(), 葛俊锋1,2,3, 叶林1,2   

  1. 1.华中科技大学 人工智能与自动化学院,武汉 430074
    2.多谱信息智能处理技术全国重点实验室,武汉 430074
    3.中国空气动力研究与发展中心 结冰与防除冰重点实验室,绵阳 621000
  • 收稿日期:2023-07-11 修回日期:2023-07-16 接受日期:2023-07-17 出版日期:2023-12-25 发布日期:2023-07-21
  • 通讯作者: 桂康 E-mail:guik@hust.edu.cn
  • 基金资助:
    国家自然科学基金(62101196);中国博士后科学基金(2021M691138)

Integrated ice detection technology based on multispectral and complex impedance principles

Jinyi LIU1,2, Jianjun XIONG3, Kang GUI1,2,3(), Junfeng GE1,2,3, Lin YE1,2   

  1. 1.School of Artificial Intelligence and Automation,Huazhong University of Science and Technology,Wuhan 430074,China
    2.National Key Laboratory of Multispectral Information Intelligent Processing Technology,Wuhan 430074,China
    3.Key Laboratory of Icing and Anti/De-icing,China Aerodynamics Research and Development Center,Mianyang 621000,China
  • Received:2023-07-11 Revised:2023-07-16 Accepted:2023-07-17 Online:2023-12-25 Published:2023-07-21
  • Contact: Kang GUI E-mail:guik@hust.edu.cn
  • Supported by:
    National Natural Science Foundation of China(62101196);China Postdoctoral Science Foundation(2021M691138)

摘要:

针对基于光强调制的光纤结冰传感器容易受污染物干扰的问题,提出了一种基于多光谱和复阻抗原理的复合式结冰探测技术,并通过实验对该方法进行了评估。根据待识别污染物的类型,多光谱和复阻抗传感器分别采集了具有区分性的4个近红外波段以及4个频率点数据。通过对每种不同类型表面状态提取21维多光谱特征和15维复阻抗特征,结合主成分分析及支撑向量机算法,建立了对明冰、积水表面状态的识别方法。实验结果表明,该方法可以排除防冻液、沙尘等5种常见的污染物干扰,在3 010条数据中实现了99.6%的状态识别准确率。对于明冰、积水的情况,进一步分析了液态水结冰过程的探测结果以及两种传感器的冰厚测量特性。实验结果表明940 nm波段的电压信号以及频率点为1.5 kHz的复阻抗等效电容值,能够表征结冰的相变过程并与冰厚具有明显相关性。

关键词: 多光谱传感器, 复阻抗传感器, 光纤结冰传感器, 复合式结冰探测, 污染物识别

Abstract:

Optical fiber ice sensors based on light intensity modulation are susceptible to contaminant interference. To solve this problem, an integrated ice detection technique is proposed based on multispectral and impedance principles, and the technique is evaluated by experiments. Depending on the types of contaminants to be identified, the data of the multi-spectral and impedance sensor in four distinguishable near-infrared bands and four frequency points are collected respectively. By extracting 21-dimensional multispectral features and 15-dimensional complex impedance features for each different type of surface state, a method for recognizing the surface state of glazed ice and water is established by using principal component analysis and support vector machine algorithms. The experimental results show that the method can exclude the interference of five common pollutants, such as antifreeze and sand, and achieve 99.6% state recognition accuracy in 3 010 data. For the case of glazed ice and water, the detection results of the liquid water icing process and the characteristics of the two sensors for measuring ice thickness are further analyzed. The experimental results also show that the voltage signal in the 940 nm band and the impedance equivalent capacitance value at the frequency point of 1.5 kHz can characterize the phase change process of icing and correlate significantly with ice thickness.

Key words: multispectral sensor, complex impedance sensor, optical fiber ice sensor, integrated ice detection, pollutant identification

中图分类号: