Research On Fault Diagnosis Methods For Rolling Bearing Of Vertical Roller Mill Based On Information Fusion

Vertical roller mill is a new type of milling equipment with high efficiency and it is widely used in the cement production industry for grinding raw material and clinker.Poor working conditions and long-term full load operation can lead to its failure,so it is of great practical significance to timely and accurately diagnose the equipment.However,the fault mechanism of the vertical roller mill is complex and diverse,and the information measured by a single sensor is often fuzzy and uncertain.Therefore,it is necessary to use the technology of information fusion to make full use of the fault information of multiple sensors in order to achieve more comprehensive fault diagnosis of the equipment and improve the reliability and stability of fault diagnosis.In this paper,the rolling bearing of the vertical roller mill is studied.According to its fault characteristics,this paper proposes a diagnostic model of information fusion based on the feature level and the decision-making level.The feature level conducts the local fault diagnosis by several parallel PSO-BP neural networks based on the features extracted from the equipment;the decision-making level uses the D-S evidential theory based on LPMCS to fuse all the local fault diagnosis results and makes the final decision.Finally,the diagnosis goal is realized.The main works are as follows:(1)A feature level fault diagnosis method based on PSO-BP neural network is studied.BP neural network is easy to fall into local minima value in the process of learning and training,so this paper takes advantage of the PSO's ability of global optimization to optimize BP neural network's weights and thresholds to establish the PSO-BP neural network to make sure the network gets the global optimum solution.Then the proposed method is applied to the fault diagnosis of the acceleration signal of the rolling bearing of the vertical roller mill,and the effectiveness of the fault diagnosis method based on PSO-BP neural network is verified by analyzing the diagnosis.(2)A decision-making level fault diagnosis method based on LPMCS and D-S evidential theory is studied.As the traditional conflict factor in the D-S evidential theory can't actually reflect the level of the conflict between evidential sources,this paper proposes a new method for measuring it.Then the parameter calculated by the method proposed above is used as an adaptive threshold for the D-S evidential theory based on the LPMCS.The simulation results show that the parameter can effectively divide the evidential sources into LPMCS and then a final decision can be made.(3)Application of information fusion technology in fault diagnosis of the rolling bearing of the vertical roller mill is studied.Firstly,extract features of signal from several accelerations of the rolling bearing,then input the features into the PSO-BP neural networks,obtaining the local diagnosis results,finally use the D-S evidential theory based on LPMCS to fuse all the local results to make the final decision.The results show that fault diagnosis method based on the two-stage fusion can effectively improve the reliability of fault diagnosis and reflect the working status of the equipment more comprehensively.Based on the research above,we design and complete the fault diagnosis system for the rolling bearing of the vertical roller miller.