一种基于CEEMDAN与Teager能量算子的气泡信号自动检测研究

Research on Automatic Detection of Bubble Signals Based on CEEMDAN and Teager Energy Operator

摘要: 锁骨下静脉的气泡超声多普勒信号是评估潜水后潜水员罹患减压病风险的重要指标。传统的方法是通过人耳监听来判断，这种方法非常依赖监听者的经验和主观判断，难以满足当前潜水作业的实际需求。本研究针对锁骨下静脉气泡检测问题，提出一种基于CEEMDAN分解与Teager能量算子的超声气泡信号自动检测方法。该算法首先对原始信号进行标准化与带通滤波以增强目标频段特征，随后使用CEEMDAN分解提取本征模态函数(IMF)并重构信号，再通过Teager算子计算瞬时能量变化并设定动态阈值实现气泡信号的检测。通过对大量实验数据的验证，系统整体检测精确率达到80.28%，召回率为84.18%，综合性能指标(F1分数)达到82.18%，显示出良好的检测效果。结果表明本方法在气泡样本中表现出良好的检测能力，能为气泡自动分级提供重要技术支持。

Abstract: The bubble ultrasonic Doppler signals in the subclavian vein are an important indicator for assessing the risk of decompression sickness in divers after diving. The traditional method involves listening with the human ear to make a judgment, which is highly dependent on the listener’s experience and subjective judgment and is difficult to meet the actual needs of current diving operations. This study addresses the problem of subclavian vein bubble detection and proposes an automatic ultrasonic bubble signal detection method based on CEEMDAN decomposition and Teager energy operator. The algorithm first standardizes and applies band-pass filtering to the original signal to enhance the characteristics of the target frequency band. Then, it uses CEEMDAN decomposition to extract the intrinsic mode functions (IMF) and reconstruct the signal, and then calculates the instantaneous energy change through the Teager operator and sets a dynamic threshold to detect the bubble signal. Through the verification of a large number of experimental data, the overall detection accuracy of the system reaches 80.28%, the recall rate is 84.18%, and the comprehensive performance index (F1 score) reaches 82.18%, demonstrating good detection performance. The results show that this method exhibits good detection ability in bubble samples and can provide important technical support for automatic bubble classification.