The Study Of Multi-parameters Integration Of Feature-extraction Method Of Rolling Bearing Fault Diagnosis

ABSTRACT:Condition-monitoring and Fault-diagnosing of equipment is a kind of technology whose destination is to figure out and master the condition during serving process and then to monitor abnormal state and judge the class of fault. The team author belongs to has worked with the Zhungeer open pit coal mine, Shenhua Group on project of diagnosing of coaled equipment. We found that mining shovel, the typical equipment in coal pit, has some feathers:tough environment, complexity of working condition and multiplicity of working model. Among mining shovel, roll bearing is one of the most important part. Because of the pressure on it is most significant and the working condition of it is changing rotate speed, roll bearing would break easily. Above all, the research of fault diagnosing method of rolling bear under changing rotate speed is necessary and worthy.The main research content and work is narrated as follow:At first, the failure mode and fault diagnosing method of rolling bearing is analyzed. Kurtosis indicator is chosen as efficient parameter for fault diagnosing through comparing the sensitivity of characteristic parameters towards fault information.At second, vibration signals of rolling bearing under changing rotate speeds are analyzed. Considering Kurtosis indicator can not reflect the complexity of signal waveform, approximate entropy and spectral entropy are added as supplements. Then, a method of multi-parameters integration of feature-extraction is put forward to cater the demand mentioned above:confirming fault class and level by figuring the distance between unmeasured waveform and formal one. This method can not only accurately distinguish the class of faults but also get rid of the impact of speed changing, and is feasible and effective proved by an amount of tests.At third, comparative analysis on the diagnosis effect between distance of three-dimensional and six-dimensional characteristics parameters are made respectively in this dissertation. In addition the theoretical and experimental analyses of the impact on the diagnosis by the increase of dimension are performed too. Above, the weighted optimization method is proposed to improve the diagnostic reliability.At last, simulated waveforms of changing rotate speed are made in lab. Using the method of Distance of the Characteristic Parameters which is put forward in this dissertation, we gain good outcome that this method can deal with the fault diagnose under condition of changing rotate speed.