Application Of Blind Source Separation In The Fault Diagnosis Of Gear Box Analysis

As a common component in mechanical equipment,the gear box mainly realizes the function of transmission when the equipment is working,and the performance of the gear box plays a crucial role in the safe operation of the equipment.Due to the influence of the working environment and working intensity of the equipment,the gear box is also one of the parts that are prone to failure,and it is prone to multiple failures,which is mainly reflected in the wear and cracks of gears and bearings.When analyzing the actually collected vibration signal,mixed fault information is often included.Therefore,it is necessary to find a suitable signal processing technology and separate the useful fault information from the complex vibration signal.Blind source separation technology rose in the 1990 s and has been widely used in signal recognition and processing in recent years.Blind source separation has universal applicability to the vibration signal analysis of gearboxes.The paper aims at the mixed failure of gears and rolling bearings common in gearboxes.Under the condition of single-channel signal,the single-channel blind separation method based on EEMD pre-processing and CICA blind source separation technology was studied to achieve separation and analysis of mixed faults.The effectiveness of the proposed method is verified by simulation experiments,and it is successfully applied to the separation of single-channel mixed fault signals and the extraction of fault features in the gear box.The paper mainly includes the followings:First,the vibration mechanism and fault characteristics of gears and rolling bearings in gearboxes are analyzed,and simulation signals are established.According to the different characteristics of gear and bearing fault characteristics,the corresponding failure frequency analysis and demodulation algorithm is introduced.And using simulation signals to verify the blind source separation effect of Fast ICA.Secondly,the single-channel signal preprocessing is to achieve channel expansion and noise reduction.The EEMD algorithm is used to process single-channel signals.The method of selecting the IMF component after EEMD is mainly studied,and the selection criterion based on the combination of white noise statistical characteristics and kurtosis value is proposed to extract the components that containing the fault information more accurately,which lays the foundation for the next step in the blind source separation of signals.Finally,the advantages of EEMD and blind source separation algorithm in single-channel mixed fault diagnosis are analyzed,and an EEMD-CICA combining algorithm is proposed.The simulation signal is analyzed to verify the feasibility and effectiveness of the proposed method for single channel mixed fault signal feature extraction.The multi-failure measured data of DDS experimental equipment was analyzed,and the superiority of EEMD-CICA method in signal extraction of single-channel hybrid fault was verified.