Power dispatching service feature extraction method based on named entity recognition technology

The invention relates to the technical field of power dispatching, in particular to a named entity recognition technology-based power dispatching service feature extraction method, which comprises the following steps of S1, collecting unstructured text data of an operation log, a maintenance record and an accident report of a power system; s2, carrying out data cleaning, labeling and format standardization on the integrated unstructured text data and real-time data; s3, analyzing the preprocessed data by applying a pre-trained named entity recognition model; s4, time sequence analysis is carried out to capture dynamic characteristics of the time-varying object; s5, analyzing a context relationship between the identified entity and the event through a graph theory method; and S6, constructing a feature vector for state evaluation and fault prediction of the power system. According to the method, more accurate and comprehensive state evaluation and fault prediction of the power system are realized by comprehens