Fault Diagnosis and Simulation of Regenerative System Based on Fuzzy RBF
Kuang Yin · 2010
Regenerative system is the most important and complicated system in modern fossil power unit. According to common faults of regenerative system and actual experience, fault set and symptom parameter set were set up. Faults were concluded as thirteen types and fault symptom parameters were divided into five grades. Through fuzzy processing and normalization, faults and symptom parameters are suitable for artificial neural networks (ANN). Using radial basis function (RBF) provided by MATLAB, a diagnosis model was set up for regenerative system. After learning and training, the model has the capacity of diagnosis and identification. The results of simulation had proved that the model could identify faults accurately. At the same time, the fuzzy processing of symptom parameters and thresholds had improved the convergence of RBF. Now this fault diagnosis system has been used in a steam power station for about one year and the actual performance can meet the needs.