Walnut oil is a kind of nutty vegetable oil with rich nutrition and high prices. Using cheaper oil mixed with walnut oil is one of the primary means of adulteration. To achieve a rapid and efficient quantitative analysis method and detection technology of walnut oil adulteration, this paper proposes to predict the mixed concentration of walnut oil and sunflower oil using the UV LED fluorescence spectrum combined with a convolutional neural network algorithm. Firstly, a series of mixed samples of walnut oil and sunflower oil were prepared, and the fluorescence spectra of the mixed oil samples were excited by UV LED. The noise information in the fluorescence spectra was removed by using the EMD-PSO optimization threshold algorithm. The superposition spectral peaks of the fluorescence spectra were calculated and analyzed theoretically, and the fluorescence spectra database of the mixed oil samples was established. Then, a convolutional neural network model was constructed based on the fluorescence spectrum database to predict the concentration of walnut oil in mixed oil samples. The experimental results show that the detection technology proposed in this paper not only characterizes the difference in the fluorescence spectra of the two vegetable oils but also has good accuracy and stability in predicting the mixed concentration based on a convolutional neural network. The R² and ME predicted by the model for the test set are 0.9853 and 0.0783, respectively. In summary, this study provides a new approach for rapid and non-destructive detection of the adulteration concentration of walnut oil and sunflower oil, which is expected to be widely applied in the food and oil industry.