REAL-TIME ESTIMATION OF ROCKET MOTOR PERFORMANCE USING MULTI-SENSOR FUSION AND AN ADAPTIVE KALMAN FILTER

Ground and flight testing of solid-propellant rocket motors requires accurate, low-latency knowledge of chamber pressure, thrust, and specific impulse (Isp), yet no single instrumentation channel provides all three with adequate bandwidth and noise immunity. This paper presents a real-time multi-sensor fusion architecture that combines a chamber-pressure transducer, a thrust load cell, and an axial accelerometer-derived thrust channel through a discrete Kalman filter with innovation-based adaptive process noise (IAE). The adaptive mechanism inflates the process-noise covariance for a short, bounded interval whenever the normalized innovation exceeds a threshold, allowing the filter to track the fast ignition and tail-off transients of a solid-motor burn without sacrificing steady-state noise rejection during the quasi-steady burn phase. The approach is validated against a physically grounded internal-ballistics truth model of an AP/HTPB/aluminized composite motor (Saint-Robert burn law, chamber-filling dynamics, choked-nozzle thrust with correct exit-pressure correction) corrupted with realistic transducer, load-cell, and accelerometer noise. The fused estimator reduces chamber-pressure RMS error by 22.2% and thrust RMS error by 39.0% relative to the raw sensor channels, and reconstructs real-time specific impulse to within 0.02% of the true burn-averaged value during the quasi-steady phase. The results indicate that adaptive multi-sensor fusion is a practical, low-cost path to real-time motor-performance telemetry suitable for static test stands, abort-decision logic, and small-scale flight avionics.