Vibration Signal Analysis And Pattern Recognition Of Shearer Rocker Arm Based On Optimized VMD-ResNet

As the main part of the power transmission of the shearer cutting part,the shearer rocker arm often bears the cutting load in the fully mechanized mining operation.At the same time,under the interference of the complex working environment,the vibration signal of the rocker arm transmission system often shows strong nonlinearity and coupling.In order to achieve the development of intelligent mining in fully mechanized coal mining towards "unmanned" or "less manned",it is necessary to conduct in-depth research on the vibration signal and cutting mode recognition of the coal mining machine’s rocker arm.This thesis takes the coal mining machine rocker arm of model MG160-375-WD as the research object,and conducts research based on the advantages of optimized VMD algorithm combined with deep learning algorithm.Firstly,the particle swarm optimization(PSO)algorithm was used to optimize the penalty factors for key parameters in variational mode decomposition(VMD)α Number of decomposition layers Κ,Determine the critical correlation coefficient based on cross correlation analysis,select components from all modal components,and perform signal processing for wavelet packet threshold denoising.Secondly,theoretical calculations were conducted on the rotation frequency of each shaft and gear of the rocker arm cutting section of the coal mining machine,as well as the meshing frequency of the gears;Analyze the vibration signals collected by the coal mining machine using the optimized VMD algorithm,and verify the feasibility of the optimization algorithm based on the optimized VMD through a combination of theory and experiment.Finally,in order to improve the accuracy of coal mining machine cutting pattern recognition,an improved deep residual(VMD ResNet)network recognition method was proposed.The vibration signals of the coal mining machine rocker arm were tested and tested,and a total of 18 working conditions were collected.The vibration signals of 5 typical working conditions,including single no-load,single cutting,and single walking,were analyzed and compared with classical CNN,ResNett,and VMD-BP network models.The experiment shows that the optimized VMD algorithm overcomes the blindness of traditional VMD denoising,preserves the effective components of the signal,and ensures the authenticity of the denoised signal.Using optimized VMD modal components as training and testing samples,the improved ResNet was used to identify the cutting mode of the shearer rocker arm,achieving high accuracy.Four network model recognition methods were used,and the comparison results showed that the average accuracy of cutting pattern recognition based on optimized VMD-ResNet was 97.5%,while the average accuracy of ResNet,CNN,and VMD-BP were 91.2%,87.3%,and 72%,respectively.Theoretical analysis and experimental results demonstrate the advantages and effectiveness of the proposed method for optimizing VMD-ResNet.