A general method for trace substance detection based on cross-correlation theory

In order to improve the limit of detection (LOD) of trace substances, a general signal analysis and processing method based on cross-correlation theory is proposed, including correlation detection methods and multi-channel signal modified methods. In this paper, both methods were validated on Raman and GC-MS detection, trace substance in insulating oil (dibenzyl disulfide) is introduced as an example, and verification test of multi-substances detection is also conducted. More generally, the cross-correlation theory based method can be applied to all detection based on spectral peak areas. After applying this method, the LOD can be considerably improved. The traditional detection method based on linear fitting between characteristic peak area and trace substance concentration is implemented as a benchmark. The LOD of Raman detection in traditional way is 338.51 ppm, The LOD of GC-MS detection in traditional way is 317.26 ppb. The correlation detection results show that the LOD of Raman spectroscopy and GC-MS are respectively increased by 1.90/2.76 times compared with traditional way. And when the multi-channel signal modified method is introduced, the qualitative LOD of Raman detection reaches 9.8 ppm (number of channel is 100), the qualitative LOD of GC-MS detection reaches 39.1 ppb (number of channel is 10). Comparative study with conventional noise reduction methods have also been carried out, and the results show that correlation detection and multi-channel signal modified method are superior to them.