A platform for causal knowledge representation and inference in industrial fault diagnosis based on cubic DUCG

TL;DR

This paper introduces a fault diagnosis system for industrial systems using cubic DUCG theory, enabling real-time, sequence-based fault detection without relying on sample data, thus improving safety and maintenance efficiency.

Contribution

It develops a novel industrial fault diagnosis model based on cubic DUCG, integrating expert knowledge for real-time, sequence-dependent fault detection without sample data.

Findings

Effective real-time fault diagnosis in industrial systems

No reliance on sample data for fault detection

Improved safety and maintenance decision support

Abstract

The working conditions of large-scale industrial systems are very complex. Once a failure occurs, it will affect industrial production, cause property damage, and even endanger the workers' lives. Therefore, it is important to control the operation of the system to accurately grasp the operation status of the system and find out the failure in time. The occurrence of system failure is a gradual process, and the occurrence of the current system failure may depend on the previous state of the system, which is sequential. The fault diagnosis technology based on time series can monitor the operating status of the system in real-time, detect the abnormal operation of the system within the allowable time interval, diagnose the root cause of the fault and predict the status trend. In order to guide the technical personnel to troubleshoot and solve related faults, in this paper, an industrial fault diagnosis system is implemented based on the cubic DUCG theory. The diagnostic model of the system is constructed based on expert knowledge and experience. At the same time, it can perform real-time fault diagnosis based on time sequence, which solves the problem of fault diagnosis of industrial systems without sample data.