Research Of Local Strong Signal Diagnosis For Incipient Gear Fault

Gear transmission systems have been widely used as key mechanical elements in various areas due to advantages in power range, speed range, transmission stability, transmission accuracy and transmission efficiency. However, gear transmission systems often work at high-speed and overload conditions, along with severe environment, rendering gears prone to faults. The results of gear faults range from economics losses to accident casualties. Diagnosing gear faults and taking measures beforehand via condition monitoring techniques can ensure its stable operation, which is of great significance. Wide area signal is the main analysis subject of current fault diagnosis research. But due to complex excitation sources and energy transmission paths inside the gearbox, the fault feature in wide area signal is substantially attenuated. The measuring point of local strong signal is much more closed to the excitation source, which greatly improves the SNR ration and provides a new approach for incipient gear fault diagnosis.According to the concept of local strong signal, the acquisition method of local strong signal is fully investigated. Based on the composition of local strong signal, its signal separation method is designed. To redress the problem of sensitivity variation, a gear meshing signal transform method is proposed. Then, an angular domain signal resampling method designed for local strong signal is proposed, aimed at processing non-stationary signal under speed fluctuation. Combining proposed methods and traditional spectral analysis and statistical analysis, crest factor of order cepstrum is proposed as the feature parameter. The effectiveness of proposed method is validated through experimental result. The main contents of the research are summarized as follows:(1) Based on the requirement of local strong signal acquisition, suitable accelerometers are selected and corresponding signal transmission schemes are designed specifically for gears and bearings. Local strong signal of spur gear is acquired through experiment. The characteristics of both local strong signal and wide area signal are analyzed.(2) The effect of filter phase response on signal separation is investigated, which result in the demand for linear phase separation filter. Based on equiripple approximation and Remez algorithm, the highpass and lowpass linear phase FIR filters are designed to separate rotation and vibration information from local strong signal.(3) Base on the analysis of sensitivity periodic variation during acquisition process, the gear meshing signal transform method via dynamic projection is proposed, through which, the characteristic shock signal is amplified and sensitivity variation is attenuated.(4) To eliminated the negative effect led by the signal non-stationarity under speed fluctuation, angular domain signal acquiring method for local strong signal is proposed, which helps the subsequent feature extraction.(5) According the characteristics of the order spectrum of transformed signal, the crest factor of order cepstrum is proposed as feature parameter. Using this parameter to analyze experimental data, spalling faults of different severity are successfully identified, which proves the effectiveness of this method.