[Tracing the Surface Water Pollution in a Chemical Park Based on the Fusion of Spectral and Chromatographic Characteristic Data].

• DOI: 10.13227/j.hjkx.202401073
• Publication Date: 2025-01-08
• Authors: Yuan Zhao; Xinyu Yin; Xiaofeng Liu; Meng Jin; Ya-Qiong Lan; Rui Liu
• Keywords: Tracing

Identification of the pollution source surface water in a chemical park was difficult because many industrial enterprises with complex wastewater components and similar characteristics are located there. Therefore, national-level Jiaxing City studied, samples from ten batches seven key were collected analyzed using 3D excitation emission matrix spectrometry （EEMS） gas chromatography-mass （GC-MS）. Parallel factor analysis used to extract common EEMS spectra different drainage same enterprise construct an characteristic data matrix. Furthermore, specific substances high detection rates or that could effectively distinguish other screened out GC-MS Pollution identification attempted models based on matrices. The results showed regardless whether being original matrix, accuracy BP neural network model not high, only 71.43%, 76.19%, respectively, which slightly higher than support vector machine （76.19%, 57.14%）. However, when fusion used, performance significantly improved. accuracy, macro precision, recall, harmonic mean for 95.24%, 96.43%, 95.10%, while better, all four indicators close 100%. study provides effective method identifying sources parks.