Shale reservoirs are characterized by pronounced heterogeneity, rendering conventional homogeneous fracture spacing and experience-based design approaches inadequate for achieving efficient hydraulic fracturing. Meanwhile, numerical simulation methods suffer from high computational cost and limited global optimization capability. To address these challenges, this study proposes an intelligent optimization method for fracture parameter design. First, a reservoir quality index (RQI) is constructed to perform accurate zoning along the horizontal section, enabling targeted fracture parameter design for intervals with distinct reservoir properties. Subsequently, the Tent chaotic mapping, golden sine strategy, and sparrow search algorithm (SSA) are integrated with a backpropagation neural network to develop a TENT-SSA-BPNN model, thereby enhancing the global search capability of fracture parameters. Finally, the net present value (NPV) is adopted as the optimization objective, and the optimal fracture parameter combination is obtained under physical constraints including fracture half-length, fracture conductivity, and stage number. Results demonstrate that the proposed method outperforms conventional models in both accuracy and stability. The optimized design achieves approximately a 30% increase in production, accompanied by a 3 to 5 fold improvement in NPV. The proposed approach provides an efficient and feasible technical pathway for differentiated and precision stimulation design of shale gas horizontal wells.