Study of data mining based machinery fault diagnosis

In accordance with the reality of the installation of an online monitoring system to significant equipment and many large-scale databases or data warehouses that have come into being, a new artificial intelligence research approach known as data mining is introduced into the fault diagnosis field in...

Full description

Saved in:
Bibliographic Details
Main Authors: Dong Jiang, Shi-Tao Huang, Wen-Ping Lei, Jin-Yan Shi
Format: Conference Proceeding
Language:English
Subjects:
Artificial intelligence
Bayesian methods
Data mining
Data warehouses
Fault diagnosis
Large-scale systems
Machinery
Monitoring
Statistical learning
Online Access:Request full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:In accordance with the reality of the installation of an online monitoring system to significant equipment and many large-scale databases or data warehouses that have come into being, a new artificial intelligence research approach known as data mining is introduced into the fault diagnosis field in this paper. Based on the Bayesian statistical learning theory and a large number of sample data, which represent the historic running record of the machine, different probability density functions of frequent classes of machine faults are established to determine the current running state. Moreover, the mining results are valuable for domain experts to discover the running regularity of machines, predict the running trend and provide decision supports for senior managers. Experiments indicate that the method is feasible in the fault diagnosis field and effective in distinguishing some frequent rotary machine faults.
DOI:10.1109/ICMLC.2002.1176814