Domain Knowledge-Driven Multi-Source Data Fusion Method for Power Grid Dispatching Business

Power grid dispatching business critically require the fusion of multi-source data, but the data involved often exhibit significant specialized characteristics, making general data fusion methods unsuitable for direct application to power grid dispatching. This paper proposes a domain knowledge-driven multi-source data fusion method, which realizes dynamic weighted multidimensional similarity calculation, implements synonym replacement via domain knowledge graphs, and verifies data matching results through attributes value similarity calculations. Additionally, the method is applied to fuse various data files related to power grid dispatching. Experimental results indicate that this method significantly improves the accuracy of data fusion in power grid dispatching. Without considering domain knowledge, it can increase accuracy by 38.3-58.5% compared to single-metric similarity calculation methods. When domain knowledge is taken into account, the accuracy can be further improved by 53.8% under certain conditions, achieving precise matching and accurate fusion.