基于变分模态分解与非对称双S函数拟合的光纤SPR传感器畸变信号解调算法研究

Research on distortion signal demodulation algorithm for optical fiber SPR sensor based on variational mode decomposition and asymmetric double S function fitting

摘要:

在光纤表面等离子体共振（Surface Plasmon Resonance，SPR）传感器表面修饰功能化膜层能够有效提升传感器灵敏度，但同时会导致传感器的透射光谱畸变，系统信噪比降低。因此本文通过引入变分模态分解（Variational mode decomposition，VMD）滤波算法提升系统的信噪比，引入非对称双S函数法拟合共振波长，改善信号的畸变问题。本文理论上阐述了VMD算法以及非对称双S函数应用于光纤SPR生物传感系统的理论基础。实验上搭建了光纤SPR生物传感系统，并对实验数据采用多种降噪以及拟合算法进行了处理。结果显示，在提升信噪比方面，与经验模态分解（Empirical Mode Decomposition，EMD）算法相比，VMD滤波能够有效提升系统的信噪比（信噪比从53.72 dB提升到61.57 dB）。在解决信号畸变方面，与最小值法、质心法相比，非对称双S函数法拟合度更高。进一步地，使用所提出算法对光纤SPR传感系统监测C反应蛋白的透射光谱数据进行了解调，结果表明，传感器的最低探测浓度相比于EMD算法由1.119 ug/ml下降到0.429 ug/ml。这一研究结果对于提升光纤SPR生物传感系统的商业应用价值具有重要的意义。

Abstract:

The functionalization of the surface of fiber-optic surface plasmon resonance (SPR) sensors can effectively enhance the sensor's sensitivity, but it can also cause distortion in the sensor's transmission spectrum and a decrease in the system's signal-to-noise ratio. Therefore, this study introduces the variational mode decomposition (VMD) filtering algorithm to improve the system's signal-to-noise ratio and the asymmetric double-S function method to fit the resonance wavelength and address the signal distortion issue. This paper theoretically explains the basis of applying the VMD algorithm and the asymmetric double-S function to fiber-optic SPR biosensing systems. In the experiment, a fiber-optic SPR biosensing system was constructed, and various denoising and fitting algorithms were applied to the experimental data. The results show that compared to the empirical mode decomposition (EMD) algorithm, the VMD filtering effectively improves the system's signal-to-noise ratio (increased from 53.72dB to 61.57dB). In solving signal distortion, the fitting accuracy of the asymmetric double-S function method is higher compared to the minimum value method and centroid method. Furthermore, the proposed algorithm was used to demodulate the transmission spectrum data of the fiber-optic SPR sensor for monitoring C-reactive protein, and the results demonstrate that the lowest detection concentration of the sensor decreased from 1.119 ug/ml (using EMD algorithm) to 0.429 ug/ml. These research findings are of significant importance for enhancing the commercial value of fiber-optic SPR biosensing systems.