实验采集多菌灵农药的固体和液体拉曼光谱信号, 对固体的原始拉曼光谱信号进行小波去噪预处理, 利用正交试验方法筛选小波去噪参数的最优组合。 结果表明, 采用db2小波基函数、 分解层数为2、 阈值方案选择为rigrsure、 重调方式为sln时, 去噪效果最好, 信噪比为62.483。 根据不同官能团的振动模式, 对去噪后的拉曼光谱分3个波数段（1 400～2 000, 700～1 400, 200～700 cm-1）进行谱峰归属和分析, 得到了多菌灵农药分子在不同波数范围内的特征振动模式, 其中, 在619, 725, 964, 1 022, 1 265, 1 274和1 478 cm-1处的拉曼信号较强, 可作为固体多菌灵农药的特征峰。 从多菌灵农药的液体拉曼光谱中, 找到了629, 727, 1 001, 1 219, 1 258和1 365 cm-1特征峰, 这些特征峰跟固体多菌灵农药的特征峰基本吻合。 研究结果可为拉曼光谱分析技术在食品及农产品中农药残留的快速筛选提供判别依据。

Raman signal of solid and liquid carbendazim pesticide was collected by laser Raman spectrometer. The acquired Raman spectrum signal of solid carbendazim was preprocessed by wavelet analysis method, and the optimal combination of wavelet denoising parameter was selected through mixed orthogonal test. The results showed that the best effect was got with signal to noise ratio (SNR) being 62.483 when db2 wavelet function was used, decomposition level was 2, the threshold option scheme was ‘rigrsure’ and reset mode was ‘sln’. According to the vibration mode of different functional groups, the de-noised Raman bands could be divided into 3 areas: 1 400～2 000, 700～1 400 and 200～700 cm-1. And the de-noised Raman bands were assigned with and analyzed. The characteristic vibrational modes were gained in different ranges of wavenumbers. Strong Raman signals were observed in the Raman spectrum at 619, 725, 964, 1 022, 1 265, 1 274 and 1 478 cm-1, respectively. These characteristic vibrational modes are characteristic Raman peaks of solid carbendazim pesticide. Find characteristic Raman peaks at 629, 727, 1 001, 1 219, 1 258 and 1 365 cm-1 in Raman spectrum signal of liquid carbendazim. These characteristic peaks were basically tallies with the solid carbendazim. The results can provide basis for the rapid screening of pesticide residue in food and agricultural products based on Raman spectrum.