Real-Time Data Assimilation in Digital Twins via Particle Filters for Discrete Event Simulation

Integrating real-time data into digital twins using Discrete Event Simulation (DES) faces a fundamental challenge: the asynchronous, event-driven updates of DES conflict with the fixed-interval arrival of sensor data from physical systems. To address this, we propose particle filters with a time-windowed smoothing technique, estimating system states by aggregating DES events within a configurable time window around each measurement, aligning irregular simulation updates with periodic data streams. We validate the method through synthetic experiments emulating a manufacturing production line. Results demonstrate a reduction in buffer-level RMSE compared to traditional methods (direct injection, moving average), while remaining computationally practical. Statistical validation via Monte Carlo trials confirms significant improvements ($p< 0.001$). Resolving this temporal mismatch enhances digital twin reliability in dynamic environments like industrial automation and predictive maintenance.