| The diagnosis of gear fault is a typical subject in the field of mechanical fault. At present the transverse vibration signal is commonly used. There are many ways with which we pick up the transverse vibration signal, but these ways are almost that make use of some sensors being installed in the machine surface. Then we work over the signals by various means, for example, spectrum analysis, cepstrum and wavelet and so on. At last we can identify the gear fault. However the transmission course from the fault source to these signals is very complicated, moreover easily disturbed. So the result isn't satisfied. In this paper, the author advance a new means that utilize the torsion signal to identify the gear fault.Gears can be regarded as a mass-spring system. They produce torsion vibration in the circumference direction by some reasons such as the changes of gear's messing rigidness, assemble error and gear's daminification. So we can test the torsion vibration signal, when load of the mechanical system is stable, not only is the transmission course of signal more directly than the transverse vibration signal but also isn't easily disturbed. So the torsion vibration signal is more sensitive than the reciprocal vibration signal, and its signal-to-noise ratio is high. These character of the means show its superiority in the field of gear fault.So, if we can pick up the torsion vibration signal effectually, we would make better results in the diagnosis of gear fault.However it is very difficult and uneconomical to test the torsion vibration directly by the torque-rotate speed sensor. But the installation of coder can't damage the structure of axis, and as to the 6000-30000p/r coder, its differentiate rate is up to 0.06c-0.012c. It is possible to pick up the torsion vibration signal. Finally, we make use of various means of data-processing, for instance, spectrum analysis, wavelet and so on to achieve the diagnosis of gear fault. |
