Abstract: At present, there are relatively mature methods for identifying the dynamic properties of ancient buildings based on the assumption of stationary excitation, but there are fewer methods for identifying the true modes of structures under the influence of non-stationary excitation caused by anthropogenic activities. A Ming Dynasty hall-type wooden building, the Great Compassion Hall of Chongshan Temple is taken as an example, proposes the Forced Vibration Dense Array Identification Network (FDIN) of ancient buildings, and synthesizes the COV-SSI and the FDD, proposes the Modal Parameter Deviation Quantification Screening Parameters (MDSP) to remove the spurious system modes, and obtains the exact modes of the Great Compassion Hall of Chongshan Temple, as well as modal frequencies of each order, Damping ratio and vibration mode. The result shows that: FDIN is able to capture the local vibration characteristics of complex ancient building structures and improve the correlation between the responses of the measurement points and the overall structural vibration characteristics; MDSP combines the subjectivity and objectivity of the modal recognition methods in time and frequency domains, sifting out a large number of spurious modes and retaining the real modes of the Great Compassion Hall of Chongshan Temple; Under environmental excitation, the Great Compassion Hall of Chongshan Temple exhibits third-order modes, which are bi-directional translations and north-south torsion, respectively; The stiffness of the Great Compassion Hall of Chongshan Temple is unevenly distributed in the horizontal and vertical directions, as evidenced by the low stiffness of the inner columns, the south main door without the constraint of the adobe wall, and the double-step beams on the roof.