Fault Detection and Localization of Centrifugal Gas Compressor System Using Fuzzy Logic and Hybrid Kernel-SVM Methods

In many industrial sectors, especially the petroleum sector, surveillance activity in rotating machinery is a very complex task and requires a great deal of information and essential data concerning the operation of these processes. The goal of this chapter is to realize a fault detection and localization strategy applied to a centrifugal gas compressor system, where this system is installed in a gas compressor station in Hassi R’mel in the south of Algeria. The proposed strategy is based on the method of hybrid Kernel-SVM method, where this method is used in many industrial sectors for detecting and diagnosing faults system; in our knowledge, this study is applied for the first time in centrifugal gas compressor system using experimental data. According to this proposed method, we can detect and monitor the state of thermodynamics of the studied system and obtain the important information for automatically take different situations for the intervention such as fault-tolerant control.