Abstract: To address the limitations of the traditional stabilization diagram methods, which rely heavily on artificial experience, suffer from the high sensitivity of clustering algorithms to parameter selection, and face difficulties in modal order determination, this study proposes an automatic modal parameter identification and order determination method based on the DBSCAN clustering algorithm and the FCNN neural network. The proposed method employs four modal filtering criteria—damping ratio threshold, detection of complex conjugate modes, MPC, and stability point density—to eliminate spurious modes. An innovative modal similarity measurement method based on spatial cosine distance is proposed, which, combined with the DBSCAN clustering algorithm, enables the accurate separation of similar modes in the stabilization diagram. Furthermore, an FCNN neural network classification model is established, using frequency and mode shape as input features and modal order as the output, to train a classifier for automatic modal order determination. The method is validated using wind tunnel tests of a monopole tower and field measurement data. The results demonstrate that the automatic modal parameter identification method can effectively eliminate spurious modes and achieve reliable modal identification. The proposed automatic modal order determination method achieves a 100% accuracy in experimental modal classification and 98.9% accuracy in field measurements. Additionally, the results confirm the existence of orthogonal modal coupling phenomena in monopole tower structures, which can be accurately identified and processed using the proposed method.