A Fault Diagnosis Method For Rolling Bearings Based On Suffix Tree

In the background of big data,the Internet and the Internet of things are rising rapidly.Therefore,the data scale is finally showing up at the pump,and the data become more and more richer.Big data of machinery also grow up.Mechanical big data not only has the commonness of big data,but also has the characteristics of Mechanical field: large capacity,diversity,timeliness.Rolling bearing is easy to break down and its ability to bear the impact is poor,which has the most extensive use in all kinds of mechanical devices.Once the rolling bearing is damaged,it will inevitably lead to the paralysis of the unit and even great economic losses.Therefore,it is of great practical significance to carry out real-time diagnosis of rolling bearing fault in the early stage.For analyzing the non-stationary quasi periodic vibration signal of rolling bearing under constant speed operation,a fault diagnosis method of rolling bearing based on suffix tree is proposed in this paper,which can determine the fault location in the time domain.Related research work includes:Firstly,a digital coding method is proposed to convert the time domain discrete digital signal with certain amplitude into binary 0/1 digital data.According to the characteristics of energy concentration of time-domain sequence,binary number is divided into blocks by settled coding.We construct the suffix tree of the blocked time series,and find out all the repeated structures suffix by suffix,which improves the anti-noise characteristic of the diagnosis method.Secondly,we study the fault diagnosis rules of rolling bearing based on suffix tree.In the first step,according to the characteristics of rolling bearing carrier wave with low frequency and impact with high frequency.We propose a preliminary diagnosis rule that the high frequency component match the impact and the low frequency component match the repetition.We match the time points with short repetition length with the fault frequency to confirm the fault location.In the second step,analyzing the coding probability of the simulation signal to discuss the jam of noise pollution to the signal at each time,which determine the specific diagnosis rule reference and describe the fault characteristics more accurately.Finally,the typical fault simulation signal model of rolling bearing is established,and the performance of fault diagnosis method based on suffix tree is tested under different SNR.At the same time,the rolling bearing fault simulation test-bed is built and the fault information is collected at different speeds.Through comparison,the real-time performance and good anti noise performance of the fault diagnosis method of rolling bearing based on suffix tree are verified.