Wei Dong

Temperature fluctuations in cold chain logistics accelerate protein degradation, lipid oxidation, and color deterioration of Litopenaeus vannamei, reducing product value and compromising food safety. To address this issue, this paper proposes a dynamic quality prediction method based on the Tad-Transformer neural network. First, storage experiments were conducted under six cold chain temperature conditions to collect multi-dimensional physicochemical and texture data. Core quality indicators were selected through temperature sensitivity analysis, and a time-series dataset was constructed. Second, an improved K-means++ clustering algorithm incorporating min-max constraints was applied for quality grading. Finally, the Tad-Transformer model was employed to predict quality indicators and temporal grade evolution. Comparative validation with Transformer, Informer and FEDformer on the self-constructed dataset demonstrates that, for the most challenging high-quality samples, the proposed model achieves both precision and recall exceeding 89%, representing improvements of 4.17–10.77% and 2.28–8.37%, respectively, over the comparison models. This method provides technical support for quality grading control and early warning of abnormal risks in cold chain logistics, offering a scientific reference for dynamic quality monitoring and intelligent evaluation of aquatic products.

To assess and predict the food safety risk of benzopyrene (BaP) in edible oils in China, this study collected national sampling data of edible oils from 20 Chinese provinces and their prefectures in 2019, and constructed a risk assessment model of BaP in edible oils with consumption data. Initially, the k-means algorithm was used for risk classification; then the data were pre-processed and trained to predict the data using the Long Short-Term Memory (LSTM) and the eXtreme Gradient Boosting (XGBoost) models, respectively, and finally, the two models were combined using the inverse error method. To test the effectiveness of the prediction model, this study experimentally validated the model according to five evaluation metrics: root mean square error (RMSE), mean absolute error (MAE), precision, recall, and F1 score. The variable-weight combined LSTM-XGBoost prediction model proposed in this paper achieved a precision of 94.62%, and the F1 score value reached 95.16%, which is significantly better than other neural network models; the results demonstrate that the prediction model has certain stability and feasibility. Overall, the combined model used in this study not only improves the accuracy but also enhances the practicality, real-time capabilities, and expandability of the model.

At present,China's food sampling inspection work has the problems of large workload and high cost. In this paper, we try to extract some valuable association rules from the national sampling data by analyzing them. Based on the data of spices major category in the 2019 national sampling database, this paper applies CARMA algorithm to mine association rules for the data studied in terms of food categories (sub-categories), sampling provinces, testing sites, contaminant categories, and relative risk levels of contaminants, and eight relatively valuable and effective strong association rules are obtained after the experiment, and some of them are interpreted. The results show that through association rule mining of relevant sampling data, the connection existing between some key and non-key testing objects and some unqualified items can be determined, which in turn can provide some reference for the allocation of resources for sampling work.

The supervision of security risk level of carbofuran pesticide residues can guarantee the food quality and security of residents effectively. In order to predict the potential key risk vegetables and regions, this paper constructs a security risk assessment model, combined with the k-means++ algorithm, to establish the risk security level. Then the evaluation index value of the security risk model is predicted to determine the security risk level based on the deep learning model. The model consists of a convolutional neural network (CNN) and a long short-term memory network (LSTM) optimized by an arithmetic optimization algorithm (AOA), namely, CNN-AOA-LSTM. In this paper, a comparative experiment is conducted on a small sample data set of independently constructed security risk assessment indicators. Experimental results show that the accuracy of the CNN-AOA-LSTM prediction model based on attention mechanism is 6.12% to 18.99% higher than several commonly used deep neural network models (gated recurrent unit, LSTM, and recurrent neural networks). The prediction model proposed in this paper provides scientific reference to establish the priority order of supervision, and provides forward-looking supervision for the government.