This study constructed a multi-dimensional coupled risk network for the Yangtze River Basin’s “meteorology-hydrology-socioeconomic-ecology-water conservancy projects” using complex network theory and causal diagnosis technology. Based on network topological indices(e.g., node degree, small world coefficient), the influence mechanisms of typical water conservancy projects including the South to North Water Diversion Project, Three Gorges Project, Water Diversion from the Yangtze River to Taihu Lake, and Niulanjiang River Water Supplement Project on system characteristics were systematically revealed. Results show that: The risk network of the Yangtze River Basin’s complex water resources system contains 368 nodes and 1208 edges, exhibiting significant small world network characteristics. It displays differentiated spatial distribution patterns of “meteorological/ecological factors driving the upstream and socio economic factors driving the middle and lower reaches.” The network features 5 long chains, presenting bimodal topological structural characteristics: “local conduction via short chains in the upstream and cross regional coercion via long chains in the middle and lower reaches.”Quantitative analysis of typical scenarios reveals that: The South to North Water Diversion Project, Water Diversion from the Yangtze River to Taihu Lake, and Niulanjiang River Water Supplement Project increased the small world coefficient by 5.45%, 4.43%, and 0.99%, respectively, playing positive roles in optimizing network structure and maintaining system stability. The Three Gorges Project led to a 1.23% decrease in the small world coefficient due to hydrological path reconstruction and was identified as the core control project of the basin, which helps enhance local robustness and improve the system’s adaptability to failures of key nodes. The Water Diversion from the Yangtze River to Taihu Lake alleviates the ecological cascade effect in the Taihu Lake region by connecting new causal chains, while the South to North Water Diversion Project enhances the robustness of the cross regional network. Based on these findings, multi scale blocking strategies integrating node level early warning, path interception, and network wide coordination are proposed.