Data-driven analysis of magnetorheological fluid seal degradation for performance prediction

To quantitatively performance degradation in magnetorheological fluid (MRF) seals, this paper proposes a data-driven method, iKansformer, to predict sealing pressure changes. First, the porous media sealing theory of MRF is introduced for understanding sealing mechanisms, and the microscopic mechanisms underlying MRF seal degradation are examined. Subsequently, the effects of degradation caused by rotational speed, initial pressure, and wear on MRF seals are analyzed and experimentally validated through extensive testing. Moreover, leveraging MRF seal data characteristics, the iKansformer is developed for accurate performance prediction under varying conditions. Finally, long-term airtightness predictions and experimental validations are performed under specified conditions. The results demonstrate that the iKansformer maintains high reliability in predicting sealing performance over a 2880-minute period.