Statistical significance and robustness testing.

This paper proposes a multi-path effect analysis model for major economies in the global trade network based on the MH-DP Transformer architecture. The global trade network comprises interconnected nodes, with interactions among major economies (China, the US, Japan, and Europe) being crucial. A trade network model with four regional nodes and their connections is established to simulate trade flow propagation and capture the nonlinear feedback mechanism through an iterative process. The experimental results demonstrate that trade volume changes exhibit significant nonlinear trends, and mutual influences between economies produce strong cascading effects. For instance, China experiences significant fluctuations when interacting with the US and Europe, while Europe shows strong chain reactions with Japan. The model outperforms other comparative models, achieving the best path prediction accuracy (82.5%) and convergence speed, effectively capturing nonlinear effects. Furthermore, the paper uses 3D charts and heat maps to visualize global trade flow changes, offering a quantitative perspective on understanding the complexity of global trade interactions. The experimental results validate the model’s strong ability to analyze multi-path effects and dynamic chain reactions in global trade networks, providing a powerful tool for future economic policy formulation and interactive prediction in regional economic integration.本文基于MH-DP Transformer架构，提出了一款面向全球贸易网络主要经济体的多路径效应分析模型。全球贸易网络由相互联结的节点构成，其中中国、美国、日本与欧洲各经济体间的互动往来至关重要。本文构建了包含四个区域节点及其关联关系的贸易网络模型，通过迭代流程模拟贸易流的传播过程，并捕捉非线性反馈机制。实验结果表明，贸易额变化呈现显著的非线性趋势，经济体间的相互影响会产生强烈的级联效应。例如，中国在与美国及欧洲地区开展贸易互动时会出现显著波动，而欧洲与日本的贸易往来则会引发强烈的链式反应。本模型优于其他对比模型，不仅实现了最高的路径预测准确率（82.5%）与最快的收敛速度，还能有效捕捉非线性效应。此外，本文采用三维图表与热图对全球贸易流变化进行可视化呈现，为理解全球贸易互动的复杂性提供了量化视角。实验结果验证了本模型在分析全球贸易网络多路径效应与动态链式反应方面的优异性能，可为未来区域经济一体化背景下的经济政策制定与交互预测提供强有力的分析工具。