Artificial neural network-based shelf life prediction approach in the food storage process: A review

Ce Shi,Zhiyao Zhao,Zhixin Jia,Mengyuan Hou,Xinting Yang,Xiaoguo Ying,Zengtao Ji

DOI: https://doi.org/10.1080/10408398.2023.2245899

Abstract:The prediction of food shelf life has become a vital tool for distributors and consumers, enabling them to determine storage and optimal edible time, thus avoiding unexpected food waste. Artificial neural network (ANN) have emerged as an effective, fast and accurate method for modeling, simulating and predicting shelf life in food. ANNs are capable of tackling nonlinear, complex and ill-defined problems between the variables without prior knowledge. ANN model exhibited excellent fit performance evidenced by low root mean squared error and high correlation coefficient. The low relative error between actual values and predicted values from the ANN model demonstrates its high accuracy. This paper describes the modeling of ANN in food quality prediction, encompassing commonly used ANN architectures, ANN simulation techniques, and criteria for evaluating ANN model performance. The review focuses on the application of ANN for modeling nonlinear food quality during storage, including dairy, meat, aquatic, fruits, and vegetables products. The future prospects of ANN development mainly focus on optimal models and learning algorithm selection, multiple model fusion, self-learning and self-correcting shelf-life prediction model development, and the potential utilization of deep learning techniques.