Flexible ferroelectric materials demonstrate considerable potential for wearable electronics and bio-inspired devices, yet their mechano-electric coupling mechanisms under dynamic bending conditions remain incompletely understood. This study systematically investigates the effects of bending deformation on domain structures and macroscopic ferroelectric responses in (SrTiO₃)₁₀/(PbTiO₃)₁₀/(SrTiO₃)₁₀ flexible ferroelectric trilayer films using phase-field simulations. By constructing computational models for upward-concave (U-shaped) and downward-concave (N-shaped) bending configurations, we analyze strain distribution under varying curvature radii and its regulation mechanisms on polarization patterns. Results reveal distinct strain gradients across bending modes: U-shaped bending induces compressive strain in the upper layer and tensile strain in the lower layer, generating a negative out-of-plane strain gradient. Conversely, N-shaped bending reverses this strain distribution. Such inhomogeneous strains drive significant polarization reconfiguration within the PTO layer. At moderate curvature (large R), the system retains stable vortex-antivortex pairs. Reducing bending radius (smaller R) promotes divergent topological transitions—U-shaped bending facilitates vortex pair transformation into zigzag-like domains, while N-shaped bending drives vortex-to-out-of-plane c-domain evolution. Notably, bending-induced strain gradients impose transverse flexoelectric fields that markedly alter trilayer hysteresis loops. U-shaped bending introduces a negative flexoelectric field, shifting loops rightward with reduced maximum polarization (P_max). In contrast, N-shaped bending generates a positive field, enhancing P_max via leftward loop shifting. Polarization switching analysis under electric fields further demonstrates bending-mediated control over domain evolution pathways and reversal dynamics. These findings not only elucidate profound bending effects on flexible ferroelectrics’ domain architectures and functional properties but also provide theoretical guidance for designing strain-programmable ferroelectric memories, adaptive sensors, and neuromorphic electronics.