Research On Fault Prognostics Key Method Of Rolling Bearing

As the key component of the mechanical equipment,the running state of the rolling bearing will affect the performance of the whole equipment.According to statistics,about 30% of the malfunction of rotating machinery is caused by the damage of rolling bearing,which can lead to expensive downtime loss and even casualties.Therefore,feature vectors extraction and dimensions reduction,performance degradation assessment and residual life prediction are studied and improved as the key method of rolling bearing fault prognostics,so that reasonable maintenance strategy is made in time and the device is close to or reached zero stop state,which is of great significance.A method for dimensions reduction of feature vectors of rolling bearing based on improved LPP algorithm is proposed,aiming at the problem that the dimensions of the extracted feature vectors is too high and LPP algorithm destroys the manifold problem of high-dimensional data due to ignoring the global structure of the data in the dimensions reduction process.The feature vectors of the time and time frequency domain are extracted from bearing vibration signal and then are reduced the dimensions based on improved LPP algorithm,finally,the CHSMM with input of the feature vectors of the reduced dimensions is used as the classifier to diagnose the rolling bearing.The experimental results show that the improved LPP algorithm overcomes the defects of the LPP algorithm which does not consider the global structure of high dimensional data and significantly improve clustering effect of data;the overall accuracy rate of bearing fault type identification based on improved LPP and CHSMM is 89%,compared with the method based on LPP and CHSMM,it is improved by 21.5%,which further illustrates the effectiveness of the improved LPP algorithm in the application of bearing feature vectors dimensions reduction.A rolling bearing performance degradation assessment method based on AFOA-WSVDD is proposed,aiming at the problem that the SVDD algorithm is not sensitive to rolling bearing early faults and difficult to select a suitable kernel parameter.The feature vectors of the time and time frequency domain are extracted from bearing fault-free stage and then are reduced the dimensions based on improved LPP algorithm.Then,the AFOA-WSVDD model is established in which the wavelet kernel function is introduced to overcome the problem that the existing kernel function is not sensitive to the early fault of the rolling bearing,and kernel parameter is optimized based on the AFOA in which the ratio of the number of support vectors and the total sample is used as fitness function.Finally,feature vectors are input into the WSVDD model,and the bearing performance degradation index is obtained.The experimental results show that the proposed method can accurately predict the early fault of bearing,and it is 17 hours earlier than the SVDD algorithm based on Gauss kernel function.A rolling bearing remaining useful life prognostics method based on improved CHSMM is proposed,aiming at the problem that the CHSMM algorithm prognostics accuracy is low for residual life of rolling bearing.The feature vectors of the time and time frequency domain are extracted from the vibration signal of bearing and then are reduced the dimensions based on improved LPP algorithm.Then,the degradation state recognition model and the residual life prediction model are established based on improved CHSMM into which the gauss mixture probability density function is introduced aiming at solving the low accuracy of residual life prediction caused by the dwell time probability density function which does not conform to reality.Finally,the whole life cycle data of the bearing is input into the model,and the degenerate state and residual life of the bearing are obtained.The experimental results show that the proposed method can accurately predict the remaining useful life of bearing.Compared with original CHSMM algorithm,the accuracy of the degradation state recognition is increased by 12%,and the accuracy of residual life prediction is increased by 23%.