【摘要】 提出了一种基于传感信号二维时频谱特征与注意力机制深度学习模型相结合的智能多类别光纤振动模式检测方案。首先，通过广义Warblet变换，将一维时序振动传感信号转换为二维时频谱信号，以获取振动传感信号的具体时频分布特征。其次，构建了一种卷积神经网络与双向门控循环单元相结合的注意力机制多模式精准识别分类深度学习模型，以实现振动输入时频特征的精准识别与分类检测。最后，通过周界安防应用场景进行了实际性能验证测试。结果表明，所提方案可对常见的6种振动传感事件进行精准识别与分类检测，且平均识别准确度高达99.2%。更多还原

【Abstract】 We propose an intelligent multi-class fiber optic vibration pattern detection scheme. This scheme is based on the two-dimensional time-frequency spectrum characteristics of sensing signals and an attention mechanism deep learning model. First, by utilizing a generalized Warblet transform-based method, the one-dimensional temporal disturbance sensing signal is converted into a two-dimensional time-frequency spectrum, thereby obtaining the specific time-frequency distribution characteristics of the vibration signal. Second, a deep learning model, which integrates convolutional neural network and Bidirectional gated recurrent unit, is constructed. This model leverages an attention mechanism to achieve accurate recognition and classification of input time-frequency vibration features, enabling precise recognition and classification of multiple vibration modes. Finally, practical application tests are conducted in perimeter security scenarios. The results demonstrate that the proposed scheme can accurately identify six common vibration sensing events, with an average recognition accuracy of 99.2%.更多还原