Band Entropy Method and Its Application to Fault Diagnosis of Rolling Bearings
|School||Shanghai Jiaotong University|
|Course||Mechanical Design and Theory|
|Keywords||Troubleshooting Condition Monitoring Time-Frequency Analysis Band entropy Bandpass filtering Genetic Algorithms Rolling|
With scientific and technological progress and increasingly rapid development of modern industry, machinery and equipment to keep the large, complex, high-speed, efficient and reload direction; same time, its work and the operating environment is also more complex and demanding. Once the sudden failure of these devices will not only increase their maintenance costs, reduce production efficiency, but also may cause huge economic losses, and even lead to serious personal injury or death, resulting in adverse social impact. So, how effective the equipment condition monitoring and fault diagnosis is needed to solve the current problem. How to effectively extract reflects the characteristics of the equipment running status and accurately determine the fault type, fault diagnosis has been a hot research field, new methods and theoretical research are endless, to enrich and improve the mechanical fault diagnosis technology has played an important role. In this paper, rolling for the study, proposed band entropy method, and its application in fault diagnosis were studied, bearing condition monitoring aims to provide a new indicator for fault diagnosis signal preprocessing provides a new method, the paper mainly include the following aspects: (1) From the theoretical analysis and practical application point of view, the paper describes the research background and significance. Analysis of the mechanical equipment fault diagnosis method, bearing fault diagnosis, time-frequency analysis and information entropy theory and other aspects of research status, established the content of this study. (2) This paper describes the theoretical basis for several time-frequency analysis method and information entropy theory, drawing on spectral kurtosis band entropy method proposed concepts, definitions entropy for a particular frequency band (band) signal complexity, or uncertainty band entropy gives the basic algorithm, the last point of view of the band from the filter Entropy is extended. (3) describes the rolling bearing vibration mechanism and fault characteristics. Entropy index discussed band possibility for a rolling bearing condition monitoring its robustness of the study demonstrate its insensitivity singular points. Based on the above characteristics entropy band, will be applied to the whole life cycle of rolling data analysis, discusses the band entropy indicators in various stages of performance degradation. Introduced to provide data to support the theory of rolling bearing fault testing and accelerated fatigue life test, through the analysis of the test data indicates that the band entropy condition monitoring indicators as an effective complement. (4) For the resonant bandpass filter center frequency demodulation is difficult to determine the problem, a band entropy method to determine the center frequency. STFT-based band entropy, discusses the discrete frequency points, time-frequency analysis window length, window function type parameters such as the impact on the band entropy; pair of bands based on wavelet packet entropy discussed wavelet packet decomposition levels and wavelet packet functions The choice of the entropy of the band. Finally, the two methods used in simulation and actual bearing fault diagnosis. The results show that the entropy can accurately determine the signal band of the resonance frequency band, to enhance the band-pass filter and envelope demodulated diagnostic results. (5) proposed band entropy and genetic algorithm combined method for resonant bandpass filter demodulation optimal design. Minimum entropy of frequency bands for the genetic algorithm optimization goal, by selection, crossover and mutation operations, the range of values ??to find the optimal combination of center frequency and bandwidth, design optimization filter. Through the simulation signal and different SNR analysis of experimental data, that this method can effectively determine the filter center frequency and bandwidth, thereby enhancing the signal to noise ratio, to achieve the bearing fault diagnosis.