Gated recurrent unit-based data-driven model for predicting pot temperature in induction cooking systems

Abstract

On an induction cooker, accurate pot temperature monitoring can cut down on energy use and prevent overheating. It can be challenging to create a mathematical model for predicting pot temperature, though, because of the different materials and sizes of the pots and the intricate electromagnetic coupling between the induction coil and the pot. To solve this problem, this paper offered a data-driven model that forecasts pot temperature by the inverter module's measured temperature. We used thermocouples in contact with the bottom surfaces of the pots and inverters to record time-series data on the temperatures of the devices to make accurate predictions. This information was gathered under a variety of circumstances, including various pot sizes and materials, various power and temperature settings, and various water levels. We used a gated recurrent unit-based autoencoder to reduce noise in the inverter temperature that was brought on by the cooling fan operation. The data-driven model was then trained using a variety of machine learning and deep learning techniques, including support vector machines, decision trees, linear regression, artificial neural networks, and gated recurrent units. Finally, we compared the predicted and measured pot temperatures to validate the data-driven models that had been learned. According to the statistical findings, there was a 3 °C or so median temperature difference between predicted and actual temperatures.