EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Design of a Lightweight Edge-AI System for Predictive Maintenance on ESP32-S3

Gaurav Kumar, Maris Terauds, Amal Ajayakumar Raji, Janis Semenako, Vladimirs Smolaninovs, Pauls Eriks Sics, Arun Kumar Malayidinja Poikayil Thankappan

Applied Sciences · 25.05.2026 · 16 (11) · 5287

DOI: 10.3390/app16115287 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Applied Sciences, online erschienen am 25.05.2026. Verfasst von Gaurav Kumar und 6 weiteren Autor:innen. Die Arbeit wurde bisher 4-mal zitiert (Crossref) und nennt 35 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Latvian Council of Science.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

While predictive maintenance increasingly relies on artificial intelligence, strict dependence on cloud computing introduces network latency and demands continuous connectivity, creating critical bottlenecks for time-sensitive industrial applications. To overcome this, we introduce a novel hybrid edge-cloud architecture, which allows deploying an ultra-low-power microcontroller (ESP32-S3) without dedicated AI acceleration hardware to perform complete, operational, predictive maintenance on ultra-constrained embedded hardware. The edge model is optimized to be very small to ensure that increasing model complexity does not cause inference latency to exceed 100 ms or make real-time operation infeasible. We created a very compact INT8-quantized neural network to perform the simultaneous classification of faults and estimation of Time-to-Failure (TTF) with a deterministic mean inference time of 42.3 ms. It dynamically estimates prediction confidence, processes high-confidence predictions locally, and offloads uncertain predictions to a higher-capacity cloud model, and recovers 97.3% of the cloud accuracy gain at 92% of the cloud latency budget. An asymmetric loss function penalizes over-prediction of the remaining useful life, and thus it provides conservative and safe warnings of fault. Operators’ interpretability is improved with Shapley Additive exPlanations (SHAP) and natural-language recommendations. Network outages of up to 50% have not influenced the safety-critical fault recall (above 0.924), so graceful degradation is reached when the network is used in real time in industrial applications. The edge-first with adaptive cloud fallback approach is demonstrated to be technically feasible for a full predictive maintenance workflow—including inference, confidence fusion, and explainability on a low-cost commercial microcontroller.

Förderung

  • Latvian Council of Science

Zitieren

APA

Kumar, G., Terauds, M., Raji, A. A., Semenako, J., Smolaninovs, V., Sics, P. E., & Thankappan, A. K. M. P. (2026). Design of a Lightweight Edge-AI System for Predictive Maintenance on ESP32-S3. Applied Sciences, 16(11), 5287. https://doi.org/10.3390/app16115287

Vancouver

Kumar G, Terauds M, Raji AA, Semenako J, Smolaninovs V, Sics PE, et al. Design of a Lightweight Edge-AI System for Predictive Maintenance on ESP32-S3. Applied Sciences. 2026;16(11):5287. doi:10.3390/app16115287

BibTeX
@article{Kumar2026,
  author = {Kumar, Gaurav and Terauds, Maris and Raji, Amal Ajayakumar and Semenako, Janis and Smolaninovs, Vladimirs and Sics, Pauls Eriks and Thankappan, Arun Kumar Malayidinja Poikayil},
  title = {Design of a Lightweight Edge-AI System for Predictive Maintenance on ESP32-S3},
  journal = {Applied Sciences},
  year = {2026},
  volume = {16},
  number = {11},
  pages = {5287},
  doi = {10.3390/app16115287}
}
Kostenlos starten →