New framework enables in-sensor AI for bearing fault diagnosis

Researchers have developed BearingNAS, a Hardware-Aware Neural Architecture Search (HW-NAS) framework that enables intelligent fault diagnosis directly on sensor hardware. This framework is designed to operate within extremely limited resource budgets, such as 4-8 kiB of RAM and 16-32 kiB of Flash, and avoids the need for expensive GPUs by using a lightweight search strategy. The system was tested on the Case Western Reserve University bearing benchmark and achieved a diagnostic accuracy of 99.50% on STMicroelectronics' LSM6DSO16IS Intelligent Sensor Processing Unit, demonstrating the feasibility of in-sensor machine learning for cost-effective fault detection. AI

IMPACT Enables low-cost, real-time AI-driven fault diagnosis directly on edge devices.

RANK_REASON Academic paper detailing a new framework and methodology. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.AI →

• Andrea Mattia Garavagno
• BearingNAS
• Case Western Reserve University
• hardware-aware neural architecture search
• LSM6DSO16IS Intelligent Sensor Processing Unit
• STMicroelectronics

• paper
• infra

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →