Linear model-based fault detection and isolation for a screw compressor

✍ Scribed by C.James Li; Taehee Kim

• Publisher: Elsevier Science
• Year: 1994
• Tongue: English
• Weight: 403 KB
• Volume: 8
• Category: Article
• ISSN: 0888-3270
• DOI: 10.1006/mssp.1994.1020

✦ Synopsis

This paper presents an approach that includes the implementation of suitable sensors and linear model-based fault detection and isolation (FDI) method to detect incipient faults in a screw compressor. Mathematical models that describe the causal relationships linking measurable system variables are derived from physical laws and identified through system identification techniques. These mathematical models are in the form of linear ordinary differential and difference equations. The parameters of the models are estimated by least square estimators for baseline and faulty conditions. Using baseline models, a fault detection scheme that does not require prior experience of faults is developed to detect abnormalities. The frequency response of models are, in turn, used to isolate the faults. It can be shown that the use of model parameters, which frequently are more directly influenced by faults, will enable early detection and accurate recognition of the compressor's internal faults, such as gaterotor wear, rolling friction and radiator faults.