The invention aims to solve the problems that subjectivity and deviation exist in judgment and division of risks due to the fact that existing power communication potential safety
hazard risk level division depends on artificial experience, and in particular, in a large-scale complex power communication network, manual potential safety
hazard risk level division is long in time consumption and low in efficiency. In the prior art, all potential safety
hazard risk factors are difficult to comprehensively identify and consider, and the
knowledge graph-based
electric power communication safety hazard risk grade classification method is invented. Firstly, a BERT-Lattice-LSTM-based
electric power hidden danger entity recognition model is proposed, and boundary division of professional vocabulary entities in the field of
electric power communication potential safety hazards is realized through a mode of fusing word vectors and word vectors. Secondly, designing a simplified neural
network model based on
semantic association according to the characteristics of more professional vocabularies, longer sentences and
high complexity of the electric power communication potential safety hazard text, simplifying the electric power communication potential safety hazard long text, and proposing a rule-based electric power communication potential safety hazard entity
relationship extraction method; the method comprises the following steps: generating a corresponding power communication potential safety hazard
RDF (
resource description framework) triad
data model according to the power communication potential safety hazard
RDF triad
data model; and finally, designing
a domain ontology of the electric power communication potential safety hazard, establishing an electric power communication potential safety hazard
knowledge graph, and providing an improved TextCNN-based electric power communication potential safety hazard
risk level classification model which comprises two parts, namely an electric power communication potential safety hazard
knowledge graph vectorization
presentation layer based on TransH and an electric power communication potential safety hazard risk level classification layer based on the TextCNN. And automatic division of electric power communication potential safety hazard risk levels is realized.