| The safe and stable operation of mechanical equipment plays a key role in the production and manufacturing industry.As the core component of mechanical equipment,the safety of the transmission system is also crucial.Therefore,fault diagnosis technology for the transmission system has played an important role in ensuring the safe and stable operation of mechanical equipment.Fault diagnosis technology based on vibration signal analysis has been widely used in many fields,with the advantages of real-time,online,convenient,accurate,and non-damaging.This article mainly takes gears and bearings in gear boxes as research objects,and processes and analyzes the vibration signals collected for the problem of fault feature extraction,analysis,and recognition.However,due to the harsh working environment and complex structure of gear boxes,some common analysis methods are difficult to extract fault features from signals containing a large amount of noise.Therefore,this article studies and improves empirical mode decomposition,wavelet analysis Singular value decomposition and other signal processing methods,and successively denoise the signal to obtain effective fault features to determine the fault.Firstly,the structural composition and fault types of the gearbox were analyzed,and a detailed theoretical analysis was conducted on the vibration mechanism and signal modulation phenomenon of the gears and bearings.The spectral characteristics of the vibration signals generated when different faults occur in the gears were studied.Secondly,an improved method of parallel extremum mean extension was studied on the basis of the parallel extension method to address the endpoint effect issue of empirical mode decomposition method,which reduce the distortion of each component at the endpoint,reduce the number of IMF components,and reduce processing time.The efficiency and completeness of this method were verified through simulation signal comparison;The adaptive noise complete set empirical mode decomposition method was adopted to address the phenomenon of modal aliasing,effectively suppressing endpoint effects and modal aliasing.Subsequently,the singular value decomposition method was introduced,utilizing its advantages of stability and better representation of time-frequency matrix features.Combined with the improved empirical mode decomposition method,Matlab software was used to process and compare the simulation signal,achieving the extraction of fault features in the signal.Due to the complex dynamic characteristics of gearbox faults,the vibration signal is a nonlinear and non-stationary signal.Wavelet analysis theory with local analysis ability was introduced to preprocess the signal,and multiple attempts were made to decompose the wavelet packet and determine more accurate thresholds and reconstruct the selection of wavelet packets.The selection criteria were determined,and the superiority of wavelet packet decomposition for denoising was demonstrated through simulation experiments,It can obtain the fault characteristics of vibration signals in various frequency bands,more accurately and purposefully depict the required fault features,and improve work efficiency.Finally,a hardware system for collecting gearbox fault diagnosis data and a software control system based on Labview were constructed.Further research was conducted on the selection and installation position of sensors,and simulated experiments were conducted to collect signal data of gears under normal,broken,worn,pitting,and composite fault conditions at different speeds.The signal was processed using the method proposed in this article,and the meshing frequency and sideband characteristics of the collected signal spectrum were analyzed to determine the location and type of fault occurrence,And compared with the results of direct time-frequency analysis of signals,it indicates the practicality and efficiency of the design scheme in this paper.The fault recognition system can make timely and accurate judgments on signals,making fault diagnosis more convenient and efficient. |
