Anomaly Detection And Degradation Trend Prediction Of Cement Mill Circulation Fan Spindle Bearing

The circulation fan is an important equipment of cement grinding system,and the spindle bearing is a vital component of the circulation fan.The working performance of the spindle bearing will directly affect the stability and safety of the circulating fan.In addition,the spindle bearing is always in a state of high temperature and high speed,so the bearing is prone to progressive damage.Data show that the circulation fan damage caused by bearing fault accounts for about 30% of all fan damage.Therefore,it is very necessary to effectively master the health state of spindle bearing by intelligent maintenance methods such as anomaly detection and degradation trend prediction.And the circulation fan often causes dynamic unbalance due to impeller wear or ash accumulation,resulting in relatively strong vibration.The vibration will cause interference to the vibration signal of the spindle bearing and bring greater difficulty for subsequent analysis.In this paper,based on the vibration signals of the spindle bearing,the anomaly detection methods of the spindle bearing based on variational mode decomposition(VMD)and improved vibrational resonance(VR)are developed;the degradation trend prediction methods of the spindle bearing based on particle filter and Bayesian neural network are developed.A new potential well function is proposed and based on this,an adaptive underdamped hybrid bistable VR system is constructed.In this paper,the basic structure of the circulation fan and the failure mode of the spindle bearing are analyzed.The formation mechanism of weak fault signal and the principle of bearing anomaly detection based on fault characteristic frequency are expounded,which lays a theoretical foundation for the subsequent research.Then,the vibration signals acquisition of the spindle bearing is introduced in detail,which provides data support for the verification of the methods proposed in this paper.Aiming at the anomaly detection of the circulation fan spindle bearing,the anomaly detection models based on VMD and improved VR are constructed respectively in this paper.In the model based on VMD,the first VMD is used to remove most of the highfrequency interference in the vibration signals,then the weak fault signal can be detected by the second VMD.In the model based on VMD,the particle swarm optimization(PSO)algorithm is used to optimize the key parameters of VMD to achieve the optimal decomposition effect.In the model based on the improved VR,a new potential well function is proposed,and based on this,an underdamped hybrid bistable system is constructed by introducing damping coefficient into the VR system.The performance of the system is obviously better than that of the classical bistable system.Then,by cascading the proposed system and optimizing its parameters through PSO,the detection ability of the system for weak fault signals is further improved.By using the proposed method,the anomaly detection of the spindle bearing is realized,and the effectiveness of the proposed method is proved.Aiming at the degradation trend prediction of the circulation fan spindle bearing,this paper firstly constructed the health index of the spindle bearing by analyzing and screening the time domain indexes and frequency domain indexes of the vibration signals.Then the degradation trend prediction models based on particle filter and Bayesian neural network are constructed respectively.In the model based on particle filter,the health index of the spindle bearing is predicted by particle filter algorithm,and the degradation trend prediction of the spindle bearing is realized.In the model based on Bayesian neural network,the appropriate prior distribution is selected according to the data characteristics,and the network structure is determined,then the target bearing is predicted according to the degradation data of the same type bearings and corresponding remaining life.This method is verified based on public bearing datasets.The research in this paper can effectively capture the early abnormal signals of the bearing and predict its degradation trend,which is helpful for enterprise engineers to better grasp the health information of the circulation fan spindle bearing.The methods proposed in this paper can provide a theoretical basis for engineers’ decisions and help enterprises to develop a more reasonable maintenance scheme.