Confidential — Stefan Michaelcheck Only

An improved Knowledge Graph Completion Method for the Power Transformer Based on Graph Neural Network and TransR

2026model innovationincrementalmethod

Hongying He, Nan Liu, Jiajun Zhu, Guangwei Luo, Haiwen Chen, Diansheng Luo, Wenju Liang

IEEE Transactions on Industry Applications

https://doi.org/10.1109/tia.2026.3655656OpenAlex: W7124948427
1
URLs Found
0
Internal Citations
7
Authors
usable
Abstract Quality
GPT-5.5 Abstract Analysis

Problems Identified (5)

Incomplete electrical equipment knowledge graph: Knowledge graphs for electrical equipment are incomplete and need further mining to improve graph integrity.

Complex relation modeling limitation: Existing TransE-style knowledge graph embedding is limited because it is only suitable for one-to-one relations and struggles with complex entity relations.

Out-of-graph entity completion: Knowledge graph completion must support representation and completion of out-of-graph entities in power transformer maintenance triples.

Incomplete electrical equipment knowledge graph: Knowledge graphs for electrical equipment are incomplete and need further mining to improve graph integrity.

Complex relation modeling limitation: Existing TransE-style knowledge graph embedding is limited because it is only suitable for one-to-one relations and struggles with complex entity relations.

Proposed Solutions (5)

GNN-TransR power transformer KGC: The paper proposes a dynamic power transformer maintenance knowledge graph completion method combining TransR and Graph Neural Networks.

GNN triple graph embedding: A GNN is designed to embed triples and represent out-of-graph entities through relational and neighborhood features.

TransR spatial relation transformation: TransR with spatial transformations is used to model complex entity relations by aligning entity and relation space vectors.

GNN-TransR power transformer KGC: The paper proposes a dynamic power transformer maintenance knowledge graph completion method combining TransR and Graph Neural Networks.

GNN triple graph embedding: A GNN is designed to embed triples and represent out-of-graph entities through relational and neighborhood features.

Results (3)

Superior power-domain KGC performance:

Improved out-of-graph identification accuracy:

Enhanced graph extrapolation and KGC accuracy:

Research Domain

Power equipment knowledge graph completion / power transformer maintenance

← Back to all papers