The research on the quantification of peony seed oil adulteration is still focused on the adulteration of a single cheap vegetable oil, so it is necessary to consider more complex adulteration. This paper uses portable near-infrared Raman spectroscopy technology combined with particle swarm optimization generalized regression neural network algorithm (PSO-GRNN) to help interpret the complex adulteration behavior of peony seed oil. Firstly, the mixed oil of cheap sunflower seed oil and corn oil was used as the adulteration object. The adulteration concentration of peony seed oil was prepared from low to high, and the concentration gradient was relatively uniform. Secondly, a portable near-infrared Raman spectrometer was used to collect the Raman spectrum signals of all oil samples. The Raman spectra were manually reduced and weighted based on spectral analysis. Finally, a quantitative analysis model based on PSO-GRNN algorithm was established. The results show that the complex adulteration behavior of peony seed oil can be interpreted based on portable near-infrared Raman spectroscopy technology combined with the PSO-GRNN algorithm. This scheme can predict the concentration of peony seed oil and effectively evaluate the concentration of various cheap vegetable oils. The test set of the model R²>0.94, RMSE<0.036. This paper deeply studied the possible complex adulteration of peony seed oil and proposed the corresponding solutions. This method is significant for the market supervision and quality detection of peony seed oil.