Abstract: To address the reconstruction inaccuracies in traditional modal transformation analysis (MTA) for large steel structures under elastic constraints—arising from the omission of rigid-body modal information, an elastic-constrained modal transformation analysis (EMTA) method is proposed. By incorporating finite measurement point displacement data to construct a composite error function, the generalized modal coordinates—encompassing both rigid-body and elastic modes—are optimized using the least-squares method. These coordinates are then combined with modal shape superposition to reconstruct the full-field displacement of the structure. Numerical simulations and experiments demonstrate that under three distinct loading conditions, MTA yields surface displacement fitting accuracies below 0.5, whereas EMTA consistently achieves accuracies exceeding 0.9. The minimum mean displacement error for MTA is 36.56%, with a maximum mean error of 56.32%. In contrast, the EMTA method achieves a minimum mean displacement error of 8.01% and a maximum mean error of 11.42%, with the maximum absolute error not exceeding 0.25 mm. Consequently, the EMTA method demonstrates significantly superior reconstruction accuracy and stability compared to traditional MTA, providing effective technical support for displacement monitoring and control in large-scale steel structures.