Neural networks for modeling and identification of the dough rising process inside an industrial proofing chamber

The main problem we meet with when we want to preview and control the quality of the yeast leavened bakery foods is the lack of operative models able to relate the dough rising process with the environmental conditions inside the proofing chamber. In this work we propose a methodology which relates easily measurable temperatures inside the proofing chamber with the height of the rising dough inside a closed mould. The proposed models are identified by means of neural networks and linear ARX systems. The identification has been carried out using of measurements carried out both on the industrial plant and in a small laboratory climatic chamber. The good fit of the results shows that the proposed architecture is well suited for the considered plant.