A realtime fuzzy Petri net diagnoser for detecting progressive faults in PLC based discrete manufacturing system
Academic Article
Overview
Research
Identity
Additional Document Info
Other
View All
Overview
abstract
In this paper, we explored a realtime fuzzy Petri net approach to diagnose progressive faults in discrete manufacturing systems. Progressive faults are usually caused by deterioration or aging and show stochastic properties. Some researchers have reported how to detect abrupt faults in discrete manufacturing systems using Petri net. However, little research has been conducted on Petri net diagnoser to progressive faults in discrete manufacturing event systems. To tackle this problem, we explored an approach including a realtime Petri net model and a fuzzy Petri net diagnoser to replicate the plant and detect faults in discrete manufacturing systems. The realtime Petri net model monitors events generated from the discrete manufacturing system, also compares the outputs and pre-settings. Once a difference is detected, it will start the fuzzy Petri net diagnoser to locate faults. For the purpose of validation, this approach was implemented with Visual Basic for diagnosing a dual robot arm. Evaluation experiments validated the diagnoser's performance on accuracy and diagnosability. It illustrated that the proposed approach can have a high accuracy rate of 93% and maximum diagnosis delay of eight steps; it proves that the approach has the capability of integrating knowledge and handling uncertainties. It also remedies the nonsynchronization between the diagnoser and the plant. The approach to construct the model and diagnoser is systematic; it has an excellent projection on intermittent fault diagnosis and hybrid systems. 2011 Springer-Verlag London Limited.
