关键词: 梯级水库； 防洪调度知识库； 场景推理； 知识图谱； 场景快速匹配； 乌江流域

Abstract: Focusing on the flood dispatching of wujiang cascade Reservoir in Guizhou Province ,a research on intelligent flood dispatching based on knowledge base and rapid scenario reasoning was carried out. First ,based on the calculation of massive operation plans , the operation experience and knowledge were organized and reconstructed. Besides , according to the overall idea of " unit- network-graph " , a preliminary knowledge base for flood dispatching was constructed. On this basis , according to the characteristics of the flood dispatching plans ,the feature information of numerical type ,essential type and fuzzy type used for scene matching was determined. By calculating the weight of each characteristic information of the flood dispatching plan ,and based on fast scene matching and scenario reasoning ,a typical plan with the highest matching degree was selected ,thus the intelligent flood dispatching of basins was implemented. Taking cascade reservoirs on wujiang River as an example , an application analysis was conducted ,revealing that under typical flood scenario ,the peak flow reduction reached 47% . The time taken for solving the scenario was 5.3 seconds using proposed method ,while dynamic programming took 21 . 1 seconds ,demonstrating that the efficiency of the solution was significantly improved. The proposed method can solve the timeliness problem of real - time solution calculation in traditional operation. The proposed solution incorporates scheduling decision -making experience ,making the dispatch process more adaptable to changes in actual flood conditions.

Key words: cascade reservoirs ; knowledge base for flood dispatching ; scenario reasoning ; knowledge graph ; fast scene matc - hing ; Wujiang River Basin