This research addresses the theoretical limitations of conventional single-scale optimization approaches in understanding filler topology regulation mechanisms, molecular alignment coupling effects, and multi-scale cooperative interactions. Through a multi-level collaborative strategy of "filler topology design-molecular ordering-density coordination", we systematically investigate the synergistic control mechanism between filler architectural design and resin matrix modification on the thermal conductive performance of epoxy-based composite materials. Fiber, sheet, and ellipsoid models are constructed based on finite element homogenization theory, revealing the regulation rules of filler volume fraction, aspect ratio, and spatial orientation on the heat transfer network. In conjunction with the reverse non-equilibrium molecular dynamics (RNEMD) method, we propose a novel strategy to enhance the intrinsic thermal conductivity of the resin through the synergistic effects of molecular ordering and density-induced non-bonded interactions. The study shows that the flake filler with 25% volume fraction exhibits an effective thermal conductivity (ETC) of 4.2 W/(m·K). The ordered cross-linked structure and enhanced density-induced non-bond interactions (where non-bonded energy accounts for 60%) increase the intrinsic thermal conductivity of the resin from 0.3 W/(m·K) to 2.85 W/(m·K). Multi-scale synergistic analysis shows that under the coupled effects of 25% filling and density enhancement, the ETC of the doped ordered cross-linked system exceeds 6.8 W/(m·K). The density synergistic effect (ρ>1.5 g/cm³) reduces the standard deviation of heat flux density by 64%. The cross-scale simulation framework established in the study reveals the quantitative correlation between heat transfer network construction and molecular ordering, providing a new theoretical paradigm for designing high thermal conductivity composites.