Gas turbine fault diagnosis based on ART2 Neural Network

Abstract

The gas turbine unit is developing larger and automatic continuously. It has become the major units of electric network. System monitor and fault diagnosis technology have been thought much of increasing due to the complexity of system. Conventional fault diagnosis technology is replacing by intelligent fault diagnosis technology gradually. ANN (Artificial Neural Network) has been widely used in pattern recognition, prediction and classification due to its merits. ART2 (Adaptive Resonance Theory 2) network performs unsupervised learning, and overcome the disadvantage of most forward network which is lost in local minimum easily. Combine quick learning with iteration learning algorithm to solve mode shifting of ART2. ART2 neural network learns the 10 typical faults of gas turbine, and then it will be used to pattern recognition, the simulation result show that the proposed gas turbine fault diagnostic model based on ART2 neural networks can diagnose the fault of gas turbine effectively. The method can be generalized to other fault diagnosis also.