چكيده به لاتين
Industrial gas turbines play a fundamental role in power plant, oil, gas, and petrochemical industries. Accurate fault diagnosis and identification are critical factors in a gas turbine's faults that may cause serious consequences such as high repair cost, long and unwanted shot downs. The troubleshooting approach used in this study is based on gas path analysis (GPA). Among the fault diagnosis methods based on GPA, Fuzzy logic can be one of the best options due to its nonlinearity, high reliability, and generalizability. Its algorithms must be developed and hardware-implemented to use this troubleshooting method in real-life applications in the industry. The hardware enables the connection among the turbine's operational inputs, which are analog and measured by the sensor, its troubleshooting algorithm, and the output of the serial displayed values, i.e., type, intensity, and location. In this thesis, the hardware implementation of a gas turbine fault detection algorithm based on fuzzy logic is studied.
To achieve this goal, first, the troubleshooting algorithm is preferably implemented in MATLAB software, then run as manual code in Arduino hardware, and next, the two are compared to verify the hardware code. The defective thermodynamic model of the studied turbine is simulated in Simulink software to approach the real-life system. This turbine's primary and faultless model is also presented in the hardware to compare its performance with the defective model. The output of the hardware is monitored by a separate system using Labview software. The connections between the defective model, hardware, and Labview monitoring software are described in detail in this thesis. The whole work is implemented in the University of Science and Technology gas turbine research institute.The results show 100% conformity of fault location detection and 97% conformity of fault type detection in the gas turbine's location by hardware. Moreover, the hardware has performed well in detecting the faults' severity percentage in each part of the turbine.
In conclusion, this thesis suggests that hardware implementation of fuzzy logic on a small microcontroller is a promising method, with high accuracy and efficiency, suitable for fault diagnosis in real-life industrial applications.
