With the continuous increase in renewable energy penetration within power systems, the operational frequency of substation on-load tap changers (OLTCs) has exhibited a significant increase, leading to significantly elevated risks of mechanical wear and electrical failures. The OLTC condition monitoring platform developed through digital twin technology enables early fault warning and health state assessment via real-time multi-physical field data fusion and operational state prediction models, thereby providing critical support for proactive grid defense systems. High-precision prediction models constitute the fundamental cornerstone ensuring monitoring reliability. Among various physical parameter monitoring methodologies, contact action timing detection stands out as the most direct and effective approach for OLTC condition assessment. This paper systematically investigates the electromagnetic transient characteristics during contact switching processes in commonly deployed VKMD-type vacuum OLTCs. Building upon these findings, we innovatively establish a transient current waveform-based analytical model for contact timing determination, subsequently developing a data-drive online monitoring solution that integrates physical models with real-time operational data. Comparative experimental validation demonstrates that the maximum prediction error of this solution across all contact switching phases remains merely 2%. The successful implementation of this proposed methodology not only overcomes technical bottlenecks in vacuum OLTC online condition monitoring but also provides key technological foundations for constructing digital twin monitoring systems for transformer OLTCs.