针对目前布里渊光时域分析(BOTDA)分布式光纤传感系统存在的实时性较差的问题,为了缩短测量时间,提出一种基于互相关卷积与高阶矩质心计算相结合的布里渊散射谱特征提取方法。首先将布里渊散射谱沿光纤的扫频数据与理想Lorentz曲线作互相关卷积,然后利用卷积结果峰值附近的理想Lorentz线型特征进行高阶矩质心提取,并将提取结果作为布里渊频移(BFS)的估计值;其次,搭建1.5km的瑞利BOTDA温度传感系统对所提算法进行可行性验证。结果表明:不同于常用的Lorentz拟合(LCF),所提算法避免了复杂的迭代求解所造成的测量时间延长,具有良好的实时性与测量精度,选取恰当的数据点数与阶数可将误差控制在小于1 MHz;当进行长距离高分辨率的动态测量而不得不加大扫频间隔以减少测量时间时,所提算法的测量误差远小于基于莱文伯-马奈特(LM)算法的LCF的测量误差。 In order to reduce the measurement time, aiming at the shortcoming of BOTDA distributed optical fiber sensing system in current real time, a cloth based on cross-correlation convolution and high-order centroid calculation is proposed The Method of Feature Extraction of Brillouin Scattering Spectrum. Firstly, the Brillouin scattering spectrum is cross-correlated with the ideal Lorentz curve along the fiber sweep data, and then the high-order centroid extraction is performed using the ideal Lorentz linear feature near the peak of the convolution result. (BFS). Secondly, a 1.5 km Rayleigh BOTDA temperature sensing system is set up to verify the proposed algorithm. The results show that, unlike the commonly used Lorentz fitting (LCF), the proposed algorithm avoids the prolonged measurement time caused by complex iterative solution, has good real-time performance and measurement accuracy. Choosing the appropriate data points and orders can reduce the error The control method is less than 1 MHz. The measurement error of the proposed algorithm is far less than that of the LM algorithm based on the Levenberg-Marquardt (LM) algorithm when the long-range and high-resolution dynamic measurements have to be performed to increase the sweeping interval to reduce the measurement time. The LCF measurement error.