为了精确获得光纤布拉格光栅(FBG)传感系统的中心波长, 提出了一种新的峰值检测算法, 即三点寻峰法。研究了该算法的流程, 在此基础上确定了影响峰值检测精度的三要素, 即窗口大小、稀疏数据个数和波长间隔。考虑窗口大小的合理选取是提高寻峰精度的关键, 本文采用求导法来确定光谱处理窗口大小。然后, 设置采样间隔对窗口内光谱进行重采样, 获得稀疏光谱数据并进行多项式拟合。最后, 根据拟合结果选取合适的波长间隔, 从而确定三点坐标并对FBG反射谱进行峰值检测。仿真实验结果表明：当光谱处理窗口大小为0.4 nm、重采样间隔为0.18 nm时获得了3个光谱数据; 其波长间隔取0.22 nm时, 所提方法的峰值误差为0.4 pm, 好于其他寻峰算法, 满足精确解调FBG传感系统的要求。

To obtain the center wavelength of a Fiber Bragg Grating(FBG) sensing system, an improved peak detecting method, three-point peak-detection, was proposed. After researching on the process of algorithm, three elements(window sizes, sparse data and wavelength intervals)that impact the accuracy of peak-detection were determined. As selecting window sizes reasonably would enhance the accuracy of peak-detection, a derivation method was used to determine the size of the spectral processing window. Then, by setting sampling intervals, spectral window data were re-sampled to obtain sparse spectrum data and perform a polynomial fitting for the sparse data. Finally, the wavelength interval was chosen to determine the coordinates of three points and to search the peak of FBG wavelength. The results of experiment and simulation indicate that the number of spectrum data is three, when the size of processing window is chosen to be 0.4 nm and the re-sampling interval is 0.18 nm. Moreover, when the wavelength interval is chosen to be 0.22 nm, the peak error of proposed method is 0.4 pm, better than that of other methods. It will meet the requirements of FBG sensing systems for demodulating precisely.