A Deep Learning Approach for System Identification of an Electric Oven Using Long Short-Term Memory Recurrent Neural Network

This paper introduces a novel deep-learning approach for system identification of a Multiple-Input SingleOutput (MISO) household electric oven in temperature control applications. The study collects input-output data from the oven under various operating conditions and compares traditional methods like transfer function, state space, and process model estimations with a Long Short-Term Memory Recurrent Neural Network (LSTM-RNN) approach. A custom LSTM architecture was developed to overcome challenges in modeling the oven system. The models were evaluated using multiple validation datasets, with experimental results showing that the LSTMRNN technique more accurately captures the complex dynamics of the oven temperature compared to traditional methods. This research highlights the potential of integrating classical and deep learning techniques for MISO system identification in household appliance temperature control.