Based on the Gas Path Analysis (GPA) method, nonlinear estimation and fuzzy classification theories, a comprehensive fault diagnosis system has been developed for an industrial Gas Turbine (GT). The hybrid method consists of two parts, in the first part noisy sensor output changes are translated to changes in the health parameters using an Extended Kalman Filter (EKF). In the second part the outputs of the EKF are used as the inputs of a fuzzy system. This system can isolate and evaluate the physical faults based on the predetermined rules obtained mostly from experimental data and aerothermodynamical simulations. The ratios of changes in different health parameters due to different faults and also the areas in the compressor most affected by these faults are the key factors for developing the rules. The Fuzzy Inference System (FIS) gives the fault locations in the compressor or turbine. Also, operator-friendly suggestions for the time of the compressor washing or components repair are provided. This leads to a hybrid fault detection and isolation solution for the GT, and with pre-filtering the data before use as input of fuzzy inference system, the accuracy of the fault diagnosis system is improved. Nonlinear simulation, estimation and classification results are provided to show the effectiveness of the proposed methodology.
Transient Fault Area Location and Fault Classification for Distribution Systems Based on Wavelet Transform and Adaptive Neuro-Fuzzy Inference System (ANFIS)
