Industrial Gas Turbine Fault Detection and Diagnosis For Health Monitoring


The Process faults can be detected by two mechanisms, alarm and isolation methods. A potential solution to this problem is the use of neural network methodology. The fault detection and diagnosis model for bearing system of gas turbine is developed utilizing the MATLAB software. The second unit of south of Baghdad power station is chosen as a case study for engine monitoring. Data of (temperature, pressure at inlet and outlet of compressor and turbine, vibration parameter in x and y axis and compressor casing vibration) are taken for normal operation. Plant data are taken every two hour. These are compared with the fault reading to locate any sign of faults. The standard vibration parameters limit, which given by the company manufacturing of gas turbine unit are 15mm/s for over vibration and 20 mm/s for dangerous vibration , while for compressor casing vibration velocity limit is 12.5 mm/s , therefore . These limits are -considered in the present work. The fault detection and diagnosis results was found that vibration limitation are similar for both fault diagnosis simulation and standard limitation.