Embedded-AI Based Fault Detection in Domestic Dishwashers: A Tale of On-Device Time Series Deep Classification

In this study, we address the challenge of detecting faults in the rotation of the spray arm in non-connected domestic dishwashers using highly noisy accelerometer data from a commercial MEMS sensor. We propose a Deep Neural Network model optimized to run in an entry-level, single-core microcontroller unit (MCU) aimed at discriminating between blocked and free states of the arm component. The proposed Embedded-AI classifier, also referred to as the 1DCNN-MLP model, is verified to be suitable for the selected hardware by a Processor-in-the-loop profiler. It is intended to execute multiple times during the washing cycle to adjust appliance control and notify the customer of necessary interventions, thereby preventing cycle repetition and saving electricity, water, and time. Copyright (c) 2025 The Authors. This is an open access article under the CC BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0/)