Gear noise is an important factor that affects both mechanical system performance and user experience. It
is greatly affected by design parameters and manufacturing errors. Thus, it requires a precise analysis based on
gear working surface deviation measurements. In this study, we verified a new approach using Legendre
Superposition method to implement mathematical modeling of gear working surface deviation and evaluated its
effectiveness for gear noise prediction compared to existing method using linear interpolation and uniformly spaced
measurements. Gear noise prediction was performed by Loaded Tooth Contact Analysis (LTCA). Transmission
errors (TE) and TE harmonics were analyzed. Results showed that the Legendre Superposition method achieved
higher accuracy in TE prediction while maintaining similar performance in TE harmonic analysis. This result
suggests that the proposed method can improve precision in modeling gear profile by reflecting real gear geometry
more effectively. This study concludes that the Legendre Superposition method has the potential to significantly
improve gear noise prediction accuracy.