A method and system for diagnosing the health state of a water-lubricated bearing

By constructing a knowledge graph and large language model specific to water-lubricated bearings, and by monitoring and generating comprehensive diagnostic reports in real time, the problem of insufficient interpretability in fault diagnosis in existing methods is solved, and understandable fault analysis and maintenance guidance are achieved.

CN122021836BActive Publication Date: 2026-07-03QINGDAO UNIV OF TECH
View PDF 2 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
QINGDAO UNIV OF TECH
Filing Date
2026-04-10
Publication Date
2026-07-03

AI Technical Summary

Technical Problem

Existing methods for diagnosing the health of water-lubricated bearings are insufficient to provide highly interpretable causes of failures and maintenance measures. Methods based on physical models have limited accuracy, while data-driven methods lack transparency and interpretability.

Method used

A dedicated knowledge graph for water-lubricated bearings is constructed. Combined with a large language model, parameters such as film thickness, vibration, temperature, and noise are monitored in real time to generate natural language descriptions. A comprehensive diagnostic report containing phenomena, causes, mechanisms, and maintenance measures is generated through knowledge graph retrieval and template matching.

Benefits of technology

It improves the interpretability of fault diagnosis, provides traceable, understandable, and actionable natural language reports, helps maintenance personnel understand abnormal states, causes, and handling measures, and overcomes the limitations of traditional methods.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122021836B_ABST
    Figure CN122021836B_ABST
Patent Text Reader

Abstract

This invention belongs to the field of intelligent operation and maintenance technology for water-lubricated bearings, specifically a method and system for diagnosing the health status of water-lubricated bearings. It involves acquiring bearing operating parameters such as film thickness, vibration, temperature, and noise, extracting current values ​​and trends, and generating structured state parameter tuples. The acquired parameters are then used to generate natural language state descriptions using a large language model. Abnormal symptoms in the tuples are used as query conditions to retrieve candidate fault modes and their associated paths from a knowledge graph, forming search results. Based on the type, quantity, and trend of abnormal symptoms, the current symptom combination features are determined, a matching inference mode template is selected, and the tuples and search results are filled into the template to generate inference instructions, guiding the large language model to obtain core diagnostic conclusions. The knowledge graph is then retrieved using natural language descriptions as query conditions to extract the knowledge subgraph that best matches the fault mode. Finally, the core conclusions, state descriptions, and knowledge subgraphs are input into the large language model to generate a comprehensive diagnostic report.
Need to check novelty before this filing date? Find Prior Art