| With the rapid development of data analysis and industrial IoT technology,device fault prediction and health management technology(PHM)based on data driven and physical system models are favored.It enables the equipment to be transformed from conditional maintenance to condition monitoring and health management,with the ability to diagnose faults and predict remaining life.However,most of the current PHM systems for rolling bearings are analyzed based on a large amount of state data,and the health management capabilities are enhanced by strengthening the algorithm,but the organic integration with the bearing’s own service performance and failure mechanism is neglected.Therefore,this paper builds a PHM system for intelligent manufacturing by strengthening the combination of physical systems and data-driven methods.Firstly,the functional system of the rolling bearing PHM system is designed.This system can be used for the full life cycle of the rolling bearing service.It has the system state judgment based on the bearing physical model in the initial stage without historical data,based on the existing data analysis and the bearing physical model in the service process.Combined real-time state management,post-service state-of-service analysis based on state analysis and life-life development model.Through the input bearing basic parameters and working condition parameters,the bearing can be solved for temperature,pseudo-dynamics and fault characteristic frequency.The analysis results can be compared with real-time monitoring signals to determine the state of the bearing in the initial stage of service;the failure based on rolling bearing is established.Fault tree of form and fault symptoms,when the bearing condition monitoring signal is abnormal,the bearing can be initially analyzed according to the fault tree,and the fault cause and improvement measures can be found according to the fault diagnosis knowledge base.Then,the bearing single point failure model is defined according to the lubrication state,and the fatigue failure and wear failure boundary are defined.According to the fatigue development model and the wear development model,the relationship between fatigue crack and wear amount with time is calculated,and the fatigue defect is simulated.The vibration characteristic response of wear defects;the above model is used to build the residual life prediction module of the rolling bearing PHM system.Finally,through the particle filter method,the fatigue development model and the wear development model are combined with the vibration eigenvalues to establish the state space model,and the pre-processing and feature extraction are carried out based on the existing bearing lifetime vibration data of the network,and selected according to the Fisher criterion.The optimal eigenvalues were analyzed.The fatigue degradation analysis and wear degradation analysis of the bearing were carried out.The rolling bearing combination degradation model was established by BP neural network.The rolling bearing PHM platform was built based on the C# language and Matlab language.Fault monitoring and state management of rolling bearings are realized. |
