Digital twins for monitoring and predicting the cooking of food products: A case study for a French crêpe

The food industry is shifting toward automated and customized processes, leading to the emergence of smart cooking devices that improve cooking outcomes. However, these devices can be invasive, costly, and only applicable to certain foods. To address these issues, a noninvasive digital twin that monitors food during cooking using a common frying pan with a temperature sensor and a weighing scale is proposed. A case study for a French crêpe is presented, in which we developed a digital twin using a neural network trained on over 400,000 simulation data points. The results show that the digital twin can accurately estimate the properties of the crêpe during cooking in real time with a mean absolute percentage error of less than 5% and predict when it will be cooked according to user criteria. The approach offers significant benefits over existing smart cooking devices, as it can be applied to a wide range of cooking processes. The proposed technology enables food process automation and has potential applications in both home and professional kitchens.