Nonlinear System Fault Diagnosis Based on Changes of the Output Spectrum Energy

A large class of mechanical objects has nonlinear changes of transfer characteristics when a fault occurs, and then the output spectrum can generate new frequency components than normal status, resulting in energy transfer between different frequencies, such as damping spring systems. This phenomenon can be used for fault diagnosis. This paper first construct a combination of sinusoidal signals in the normal state of a damping spring system, and then to motivate this system, calculating the changes in its spectrum energy. By this way, we can judge whether the fault occurs. Experiment shows that this method is not only effective, but the time required for diagnosis is only 1/8 of that of the traditional particle filter method.