Research On Noise Reduction And Fault Feature Extraction For Vibration Signal Of Planetary Gearbox

Laboratory DDS test bench can simulate the common faults of planetary gear box sun wheel and planetary bearing,and can be combined into single and two-stage planetary gear boxes with adjustable speed and load.Preset common faults in single or double planetary gearboxes and their combination,a series of experiments are carried out to obtain the vibration acceleration signal of a single or two-stage planetary gearbox.On this basis to explore the planet gearbox vibration signal noise reduction and feature extraction,which assumes that the fault type and location is unknown,choose common diagnostic methods for diagnosis,compared with the preset fault type and location,the effectiveness of the proposed method and feature extraction method is verified in the end.The specific work is as follows:?1?Aiming at the deficiency of the existing threshold function noise reduction method,a two-parameter double threshold denoising method is proposed.Through the analysis of the features such as continuity and deviation,this threshold method not only keep the advantage of the existing threshold method,and considering the influence of the energy distribution after wavelet packet decomposition,according to the actual signal adaptively adjust the function of noise reduction.Through simulation and experimental verification,the improved threshold method is superior to the existing threshold method in improving signal SNR and smoothness.?2?On the basis of analyzing the limitations of feature extraction methods in different fields,according to the characteristics of the planetary gear box operation,the nonlinear characteristics of different types of fault vibration signals,and considering the influence of signal transmission path,to construct a new characteristic parameters-SE-Rms.To measure a single common characteristic parameters of gear box fault distinguish ability,defines a new index,"discrimination"?Degree of discrimination is greater than 1,said characteristic parameter can effectively distinguish between several kinds of common laboratory bench preset failure?.Based on the common fault vibration signal of planetary gear box solar wheel,analysis of discrimination,volatility and noise resistance performance characteristics in the SE-Rms.The results show that,1)The discrimination of SE-Rms for several common faults of single and double planetary gear boxes is greater than 1.4,and SE-Rms is independent of the time origin of sample data;2)Good anti noise performance of SE-Rms:In the same signal-to-noise ratio,the discrimination of SE-Rms is significantly higher than other time-domain or frequency-domain parameters for single planetary gear box solar wheel.4)There is a"minimum sample length T₀"associated with the frequency of the fault.The discrimination of SE-Rms is calculated by selecting different sample length T.The results show that,as long as T is greater than T₀,the discrimination of SE-Rms is greater than 1;Using T₀ or its integer multiples as the sample length?i.e.the whole period truncation?,the discrimination degree will obtain the local maximum value;The larger the sample length,the higher the discrimination.?3?The multi-parameter matrix is constructed with SE-Rms as the input of support vector machine,based on single point data,The results show that,compared with other characteristic matrix diagnosis results,the multi parameter matrix with the SE-Rms is the best,and the SE-Rms has a greater contribution to the diagnosis results.Secondly,DS evidence theory is used to fuse the information of the three points,the results show that,with other measures,the SE-Rms as the main characteristic parameter can further improve the diagnosis rate.Finally,the SE-Rms is applied to the fault diagnosis of the two-stage planetary gearbox,the results of single measurement point data and three measurement points information fusion are better.The applicability of SE-Rms to fault diagnosis of two-stage planetary gear box is proved.