This study selected five typical types of chemical industry volatile organic compounds (VOCs) emission characteristics in China for analysis. The results from 70 source samples showed that alkanes were the dominant VOCs category from synthetic material industry sources, petrochemical industry sources, and coating industry sources (accounting for 43%, 63%, and 68%, respectively); olefins were the main VOCs category from the daily supplies chemical industry (46%); and halogenated hydrocarbons were the dominate VOCs category from specialty chemicals industry account source emissions (43%). Additionally, the machine learning method was applied in this study to analyze the marker components of the above industries. The results showed that decane and tetrahydrofuran were the source markers of the synthetic material industry; n-butanol and toluene were the markers of the daily supplies industry source; 1,2,3-trimethylbenzene and 1,3,5-trimethylbenzene were the markers of the petrochemical industry source; propylene and 3-methyl pentane were the source markers of the coating industry; and P-Xylene and cumene were the markers of the specialty chemicals industry source. The maximum incremental reactivity method (MIR) was used to estimate the ozone formation potential (OFP) of different VOCs-sources. The calculation results showed that when considering per unit TVOCs concentration emissions, the contribution to the ozone generation potential was in the order of the daily supplies chemical industry, specialty chemical industry, petrochemical industry, synthetic material industry, and coating industry. Therefore, we suggest that more attention should be paid to the key active species emitted by various industry sources rather than only the total amount of VOCs emissions in future ozone prevention and control efforts.
Keywords: chemical industry soureces; machine learning; ozone formation potential （OFP）; ozone（O3）; volatile organic compounds（VOCs）.