Frost thickness estimation in a domestic refrigerator using acoustic signals and artificial intelligence

This paper proposes a novel method to estimate the amount of accumulated frost using acoustic signals and artificial intelligence. The objective of this method is to estimate the amount of accumulated frost on the surface of the evaporator inside a domestic refrigerator. This methodology generates a set of acoustic signals with different frequency values. These signals interact inside the evaporator cavity. Each additional signal is absorbed and reflected depending on the amount of accumulated frost. This method uses an intelligent model and a combination of an active speaker and a simple microphone to estimate the accumulation of frost on the evaporator surface. Additionally, is proposed the use of sound pressure levels to reduce the data dimensionality. These sound pressure level values create the training and the validation dataset. In this work, is proposed the use of two intelligent models to analyze the datasets. These models are based on artificial neural networks and probabilistic neural networks. Several tests were performed in a domestic refrigerator following the typical usage habits to collect the data. Each test ran for 24 h to measure sound pressure levels at different frequencies and periods. The main contribution of this work is a low-cost method to estimate the amount of frost accumulated in the evaporator surface. One of the main advantages of the proposed method is that it can be easily incorporated into a domestic refrigerator. The results indicate that this method is capable of estimating with accuracy four different levels of frost accumulation on the evaporator surface of a domestic refrigerator.