Intelligent fault diagnosis is realized at the State Grid substation maintenance center to eliminate the need for manual interference.

1. Accumulation and inheritance of expert experiences

Expert experiences in technical standards, fault investigation and analysis reports, and other unstructured documents are annotated and extracted based on the NLP technology for the structurization and systemization of achievements, and extraction and online sharing of knowledge. 

2. Correlation and integration of line fault information

The heterogeneous data from multiple sources including equipment ledgers, fault history, online monitoring systems, and fault fixing records is connected and integrated to create the knowledge graphs of line faults and fixing procedures, which present full pictures of fault cause analysis, locating, troubleshooting and solutions. 

3. Intelligent Q&A on line faults

Interactive intelligent Q&A in the context of various scenarios is supported based on semantic comprehension and intention recognition technologies to provide the applications such as FAQs, line information retrieval, accurate fault search, and fault diagnosis. User input in the verbal Q&A process is analyzed for accurate determination of user needs and quick retrieval of transmission line journals and fault information, with the search time shortened from 10 minutes to 1 minute. 

4. Quick generation of fault forms

When a line fault occurs, the related information is automatically collected and organized using the predefined templates for easy analysis and determination of the cause of fault, with a fault report generated automatically. The workload of information collection for each fault is reduced by 85%, and the intelligent fault diagnosis is realized without the need for manual intervention.