A voxel-based machine-learning digital-oven-twin for precise cooking

In an increasingly competitive market, oven manufacturers are developing a variety of combinations of convection and radiation processes within ovens that attempt to precisely control temperatures around a target object. Industrial applications include materials drying, annealing, sintering, viral decontamination, etc. In the culinary world, in addition to classical baking and roasting, newer processes such as air-frying have become extremely popular as a healthy alternative to the classical saturated-fat-based frying methods. In this work, a machine-learning framework is developed that rapidly combines oven radiative heating surfaces and convective flows in order to induce a desired temperature profile across a time series of voxel-frames around a target object. In order to achieve this, a digital-twin based thermo-fluid radiative model is developed, utilizing the Navier-Stokes equations and the first law of thermodynamics in combination with a spatio-temporal voxel rendering of the system. The resulting coupled model equations are rapidly solved with a voxel-tailored, temporally-adaptive, iterative solution scheme. This voxel-framework is then combined with a genomic-based machine-learning algorithm in order to ascertain system settings to precisely “cook” the object. Numerical examples are provided to illustrate the framework.