| The research object of this thesis is rolling bearing deterioration trend prediction based on vibration signals.For the characteristics of vibration signals is nonlinear and random,the Variational Mode Decomposition(VMD)is used to extract the energy spectrum entropy of the vibration signal.Glowworm Swarm Optimization(GSO)is used to optimize Extreme Learning Machine(ELM)to find the best weighted value and threshold.The GSO-ELM is proposed to predict the energy spectrum entropy time series.In this paper,the deterioration trend prediction of rolling bearing based on VMD and ELM is studied.The main research contents of the thesis are listed as follows:(1)The frequency components of the vibration signals based on VMD are analyzed.The energy of some IMFs is changed during the bearing deterioration.According to the information entropy theory,the characteristic parameter of VMD spectral entropy is constructed as the characteristic parameter of rolling bearing.(2)The phase space of the VMD energy spectral entropy time series is reconstructed,then the one-dimensional time series is mapped into the high dimensional space.ELM is used to predict the VMD energy time series with the phase point.The influence of different historical data and prediction length on the prediction accuracy is studied.(3)GSO-ELM model combining the GSO with ELM is established to avoid the random setting of model parameters.the state parameters time series of the rolling bearing is predicted based on GSO-ELM.In addition,the remaining life of bearing is predicted based on the trend item of deterioration parameters time series extracted by VMD. |
