Impact of Sesbania Gum Addition on the Quality of Salami Based on Backpropagation-Artificial Neural Network Analysis
This study explored the impact of adding sesbania gum on the quality of salami using backpropagation-artificial neural network (BP-ANN) analysis. Four treatment groups were designed: blank control (CK), inoculation of a mixed culture (CG), addition of sesbania gum (SE), and sesbania gum addition com...
Saved in:
| Main Author: | |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
China Food Publishing Company
2025-07-01
|
| Series: | Shipin Kexue |
| Subjects: | |
| Online Access: | https://www.spkx.net.cn/fileup/1002-6630/PDF/2025-46-13-005.pdf |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | This study explored the impact of adding sesbania gum on the quality of salami using backpropagation-artificial neural network (BP-ANN) analysis. Four treatment groups were designed: blank control (CK), inoculation of a mixed culture (CG), addition of sesbania gum (SE), and sesbania gum addition combined with mixed culture inoculation (SE-CG). The quality of salami was evaluated in terms of its pH, water activity (aw), color difference, texture, sensory evaluation, and electronic nose analysis. It was demonstrated that the combined treatment rapidly decreased the pH and aw of the product, thereby contributing to the formation of the final quality of salami. Compared with the CK and CG groups, the SE-CG group exhibited significantly improved a* value (4.64 ± 0.38) and hardness ((60.95 ± 1.48) N). Furthermore, the electronic nose analysis revealed that the SE-CG treatment significantly increased the contents of sulfur-containing compounds, alcohols, and aromatic compounds in the product. The developed BP-ANN model had good classification accuracy and predictive ability with a 96% accuracy. Additionally, the Shapley additive explanations (SHAP) method was employed to interpret the BP-ANN model, highlighting the significance of various quality indicators in the prediction. Notably, the signal of electronic nose sensor S12, hardness, and chewiness were identified as the most important features for the model prediction. |
|---|---|
| ISSN: | 1002-6630 |