With consumers' increasing attention to the ingredients of skincare products and their preference for natural and additive-free products, the issue of adding preservatives to skincare products has become one of the hot topics in analytical chemistry. This article proposes a method based on particle swarm optimization and backpropagation neural network algorithm combined with surface enhanced Raman spectroscopy (SERS) to achieve quantitative prediction of preservatives such as benzyl alcohol and phenoxyethanol in skincare products for the rapid and non-destructive quantitative analysis of preservatives. We collected and analyzed the Raman spectra of purified preservatives such as benzyl alcohol and phenoxyethanol in skincare products and predicted their vibrational spectra using density functional theory. The enhancement effects of different nanoparticles on preservatives such as benzyl alcohol and phenoxyethanol were compared in this innovative research. The Raman spectra of different concentrations of preservatives in skincare products were detected, and quantitative analysis models for preservatives such as benzyl alcohol and phenoxyethanol were constructed using the novel particle swarm optimization backpropagation neural network algorithm models. The predictive ability of the models was evaluated by the coefficient of determination and root mean square error. Among them, the PSO-BP model had the best prediction results for the test set, with R² of 0.9518 and RMSE of 7.669×10^-6. In summary, the particle swarm optimization backpropagation neural network algorithm combined with surface enhanced Raman spectroscopy proposed in this article has good performance in quantifying skincare preservatives. This study can provide strong technical reference and support for relevant law enforcement or quality supervision departments.