<p>Graph Convolutional Networks (GCNs) are effective at processing graph-structured data and are widely used in fields such as intelligent transportation and network analysis. However, as the scale of models and data continues to expand, their computational cost and memory requirements increase significantly, becoming a major bottleneck that limits their application performance. To address this issue, we propose an optimized GCN framework tailored for intelligent transportation applications, integrating a Hadamard product-based feature update mechanism and a virtual edge feature enhancement strategy. The Hadamard product enables element-wise multiplication between node features and edge attributes, effectively reducing the reliance on costly matrix transformations and computations, thereby lowering overall computational overhead. Additionally, the system incorporates a core node identification and subgraph partitioning strategy, which segments the graph to reduce complexity and improve localized computational efficiency. To enhance inter-subgraph message propagation, a virtual edge weighting mechanism is introduced, capturing both topological and semantic influence. In the graph convolution layer, both physical and virtual edge features are jointly considered. Through this comprehensive integration, the proposed model significantly improves GCN computational efficiency and representational capability in route planning tasks, thereby enhancing the practical utility of intelligent transportation systems in real-world scenarios.</p>