An adaptive GALM neural model and its application for fault diagnoses
NI Yuan-ping · 2008
The studies analyzed the idea of Levenberg-Marquardt (LM) algorithm, and also improved genetic algorithm (GA), finally developing an adaptive model of neural algorithm through the GA-LM. This model was applied to the fault diagnoses for power transformers. The results show that the adaptive GALM neural model is able to overcome its local minimum and increase the converging speed in comparison with BP. Meanwhile, the model can diagnose the faults of transformers efficiently and also increase the ratio of fault recognition greatly. The model is supposed to have a reference value in fault diagnoses for similar electrical equipment.