Research On Assessment Technology Of Rolling Bearing Performance Degradation

Rolling bearings are the common parts in rotating machineries, and faults of rolling bearings are important reasons for the failure rotating machineries. All along, the research on fault diagnosis of rolling bearings has gained much attention. Bearings often go through a series of different stages of performance degradation when they run from early period of bearings’ damages to the time of complete failure of bearings. Therefore, if we can assess the degree of bearings’ performance degradation effectively during the running time of bearings, better equipment maintenance plans can be drawn up an can be carried out so as to reduce production downtime, improve utilization ratio and reliability of equipments, and decrease life-cycle costs of machines effectively.Based on the vibration signal this paper aims at researches on assessment technology of rolling bearing performance degradation in three phases---filtering signal, extracting feature and constructing evaluation model. The research contents are as follows:(1) Filtering vibration signal. The feature of vibration signal of rolling bearing performance degradation is analyzed. Because damages of bearings can cause periodic impacts in vibration signals, the vibration signals are filtered by the Morlet wavelet filter and periodic impulse signals are obtained for the extracting of features in the next section. The technique for construction of the filter based on particle swarm optimization(PSO) with a fitness function called as the minimum Shannon entropy is presented. Based on the simulation signal, comparative study of the filtering effect of the method is done.(2) Extracting features. The researches on the extraction of features from the perspective of monotonic trend are the key of this thesis. The performance of the envelope spectrum entropy and the wavelet energy ratio which both are features for reflecting the degree of bearing performance degradation are researched. Then the spectral span is proposed as a feature, and its performance is also researched. These researches are based on experimental vibration data responding to several operating conditions. Besides, the impact of the band-pass filtering to the performances of the envelope spectrum entropy and the spectral span is researched. Results show that these features can monotonically reflect the degree of rolling bearing performance degradation when the damage location of bearings and their speed are given.(3) Constructing evaluation model. The principle for constructing evaluation model of the degree of bearing performance degradation and the structure of model are proposed. A model based on the dimensionless features with monotonic trends above is constructed of structural models with monotonic trends in process equipment performance degradation, and the output of the model is an index which can increase monotonically with the increase of the degree of rolling bearings’ performance degradation.The results of this study show that three features above have their pros and cons, but combining the evaluation model and these features can assess the degree of rolling bearings’ performance degradation effectively.