**4. Conclusions**

In this study, a flexible SERS substrate of β-CD@AuNP/PTFE combined with a lightweight network was designed to achieve the in situ detection and identification of various PAH residues on fruit and vegetable surfaces. AuNPs were modified with β-CD to enhance adsorption of PAHs. The flexible β-CD@AuNP/PTFE substrate was prepared by assembling β-CD@AuNPs on a PTFE film coated with a perfluorinated solution, contributing to the generation of a large number of hot spots and realising convenient in situ detection. The concentrations of BaP, Pyr, and Nap residues on fruit and vegetable surfaces can still be detected at 0.25, 0.5, and 0.25 μg/cm2, and all the RSD values were less than 10%. Subsequently, SqueezeNet, MobileNet, and ShuffleNet networks were used to establish recognition models for various PAH residues on fruit and vegetable surfaces. ShuffleNet obtained the best recognition results with *ACCT* = 100%, *ACCV* = 96.61%, and *ACCP* = 97.63%. These results demonstrated that the proposed method could achieve simple, sensitive, stable, and intelligent in situ detection and identification of various PAH residues on fruit and vegetable surfaces. This method offers great potential for the practical application of rapid, non-destructive analysis of surface contaminant residues in the food industry. However, owing to the large variety and low content of residual PAHs, it is necessary to optimise the SERS substrate to achieve highly sensitive PAH detection in complex matrices. Meanwhile, the SERS spectra of additional types of PAHs should be collected to further improve the recognition performance of the model.

**Supplementary Materials:** The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/foods12163096/s1. Figure S1: SEM image (A) and Size (B) of the prepared β-CD@AuNPs; Figure S2: The preparation process of β-CD@AuNPs/PTFE; Figure S3: Structure of SqueezeNet; Figure S4: Structure of MobileNet; Figure S5: Structure of ShuffleNet; Figure S6: SEM image of (a) PTFE and (b) β-CD@AuNPs/PTFE; Figure S7: The field distribution of four AuNPs with A particle size of 24 nm with different gaps of (A) 8 nm, (B) 6 nm, (C) 5 nm, (D) 4 nm and (E) 3 nm; (F) Gap distribution of β-CD@AuNPs/PTFE; Table S1: Parameter setting of different models; Table S2: Raman shifts of BaP, Nap and Pyr and the corresponding vibration modes.

**Author Contributions:** M.Q.: Methodology, Validation, Investigation, Writing—original draft, Writing—review and editing. L.T.: Software, Methodology, Data curation. J.W.: Visualisation, Investigation. Q.X.: Investigation, Resources. S.Z.: Formal analysis, Supervision, Funding acquisition. S.W.: Conceptualisation, Methodology, Supervision, Project administration, Funding acquisition. All authors have read and agreed to the published version of the manuscript.

**Funding:** This study was supported by National Key R&D Program of China (No. 2022YFD2401305), Key Research and Development Program of Anhui Province (Nos. 202104a06020025), and National Natural Science Foundation of China (Nos. 31971789, 32001421).

**Data Availability Statement:** All related data and methods are presented in this paper. Additional inquiries should be addressed to the corresponding author.

**Conflicts of Interest:** The authors declare that they have no known competing financial interest or personal relationships that could have appeared to influence the work reported in this paper.
