Based on a self-developed partitioned plantar perturbation force plate (PP-PFP), this paper proposes a phase-partitioning method for analyzing the dynamic process of human postural control following support-surface perturbations. Using acceleration and force data from the built-in sensors of the PP-PFP, three key time points were defined based on a threshold method: the onset of perturbation, the onset of postural adjustment, and the end of postural adjustment. Accordingly, the perturbed postural response process was divided into four phases: balanced standing (BS), imbalanced standing (IBS), compensatory standing (CS), and rebalanced standing (RBS). The center of pressure (CoP) variables calculated include: root mean square distance (RDIST), mean velocity (MVELO), 95% confidence ellipse area (AREA-CE), and the ratio of trajectory length to the straight-line distance between the start and end points (length ratio, LR). Paired-sample t-tests were conducted on the CoP variables for the BS vs. RBS phases and the IBS vs. CS phases, respectively. The results showed significant differences in LR, MVELO, and RDIST between the IBS and CS phases (p<0.01), and significant differences in AREA-CE and RDIST between the BS and RBS phases (p<0.01). Combined with the clear transitional features observed in the CoP trajectory plots across the four phases, these findings preliminarily validate the effectiveness of the proposed PP-PFP and its phase-partitioning method, providing a refined analytical framework for in-depth investigation of the dynamic regulatory mechanisms of postural control.