Research on loom on-board condition monitoring technology based on fuzzy rough set and improved DSmT theory¹.

This article focuses on the development of an onboard condition monitoring system for high-speed rapier looms, addressing limitations in existing methods such as shallow information mining and low recognition rates. It proposes a novel approach combining fuzzy rough set theory—enhanced by α-information entropy for attribute reduction—with an improved Dezert-Smarandache Theory (DSmT) fusion decision method that incorporates uncertainty and importance weighting to enhance decision reliability. The system architecture includes multi-level data acquisition, processing, integration, reasoning, cloud fusion, and application, implemented on a hardware platform using the STM32F407ZET6 microcontroller and AD7730/AD7190 ADCs for tension and vibration signal sampling. Experimental results demonstrate that the fuzzy rough set combined with improved DSmT fusion achieves higher spindle wear recognition rates compared to traditional rough set and fuzzy rough set methods, supporting more accurate and reliable loom condition assessment critical for intelligent weaving machine management and fault diagnosis.