A temperature control method based on a physical neural network, a device, and an application thereof

Abstract

The present invention provides a temperature control method based on a physical neural network, a device, and an application thereof. By employing a predicted temperature control curve generated by a temperature prediction model, the predictive control model determines the thermal control parameters of the thermal control elements required to achieve the target temperature in a predictive state. This approach enables real-time adaptive temperature regulation and overcomes the limitations of traditional thermal control models, which are susceptible to prediction errors caused by external unstable factors. The invention is widely applicable to various fields, including household appliances, automotive and motorcycle temperature control management systems, heating systems, semiconductor thermal annealing equipment, and industrial furnaces.

Images