Study on Self-Learning for Vibration Fault Diagnosis System of Rotating Machinery

For an established fault diagnosis system which is based on it own expert system, it is usually incapable to diagnosis the new operating conditions, of which the knowledge has not been explored by the system. It is the purpose of the paper to develop the approach of identifying new fault and self-learning for diagnosis based on non-linear fractal theorem. It has been generally accepted that the vibration series has obvious fractal feature, which can reflect the essential characteristics of new fault. When the novel fault is taken place in the system, a related sub-net is increased to the system and trained with this sample. We have verified experimentally that the fractal dimensions of the same class faults are distributed approximately around a definite value that can represents the dimension of the standard sample for the novel fault. Based on non-linear theorem, the approach of identifying new fault and self-learning for diagnosing is put forward.