| 法布里-珀罗游标光谱信号的深度学习解调 |
| Deep Learning-Based Demodulation of Fabry-Pérot Vernier Spectral Signals |
| 投稿时间：2025-02-17 修订日期：2025-04-28 |
| HTML 查看/发表评论 下载PDF阅读器 |
| 中文摘要: |
| 为提升法布里-珀罗(Fabry-Perot, F-P)传感器游标光谱信号解调的准确性，提出基于深度学习的光谱数据直接解调方法。首先对光谱数据进行预处理，将复杂的游标光谱信息转化为卷积神经网络(Convolutional Neural Network, CNN)可以处理的数据格式，然后采用深度学习模型对预处理后的完整光谱数据进行训练和测试，并利用卷积神经网络对光谱数据进行特征提取和分类，最终实现待测信号的准确解调。使用灵敏度为112.5 nm / MPa的双腔法布里-珀罗传感器采集光谱数据并开展信号解调实验，结果表明：CNN模型对未知光谱进行10折交叉验证的平均准确率为92.49%，均方根误差RRMSE（Root Mean Square Error, RMSE)为0.0392 MPa，相对误差的平均值为3.31%；卷积神经网络-长短期记忆(Convolutional Neural Network-Long Short Term Memory, CNN-LSTM)模型对未知光谱进行10折交叉验证的平均准确率为96.98%，RRMSE为0.0390 MPa，相对误差的平均值为3.28%。基于CNN-LSTM模型的方法仅通过解调256个采样点的数据就实现了较高准确度，具有便捷高效的优点，为推动光谱信号解调技术发展提供了重要借鉴。 |
| 英文摘要: |
| To enhance the demodulation accuracy of Vernier spectral signals in F-P sensors, this study proposes a deep learn-ing-based direct demodulation method for spectral data. The methodology involves preprocessing spectral data to con-vert complex Vernier spectral information into formats compatible with CNN, followed by training and testing deep learning models on the processed full-spectrum data. The CNN architecture was employed for feature extraction and classification of spectral data, enabling accurate demodulation of target signals. Experimental validation utilized spectral data collected from a dual-cavity F-P sensor with 112.5 nm/MPa sensitivity. Results demonstrate that the CNN model achieved 92.49% average accuracy with 10-fold cross-validation, accompanied by 0.0392 MPa RRMSE and 3.31% mean relative error. The hybrid CNN-LSTM model exhibited superior performance with 96.98% average accuracy, 0.0390 MPa RRMSE, and 3.28% mean relative error. Notably, the CNN-LSTM approach attained high precision using only 256 sampled data points, demonstrating remarkable efficiency. This methodology provides an effective technical pathway for advancing spectral signal demodulation technology, offering significant reference value for developing intelligent optical sensing systems. |
|
| 中文关键词: 光纤传感器 法布里-珀罗干涉仪 光谱解调 深度学习 游标效应 |
| 英文关键词:optical engineering Fabry-Perot sensor spectrum demodulation deep learning vernier sensitivity enhancement |
| 基金项目:国家自然科学基金项目（面上项目，重点项目，重大项目）,其他 |
| DOI： |
| 关闭 |
