PHM Method Based On State Estimation And Fusion Prediction

With the rapid development of science and technology today,more and more complex equipment is widely used in aerospace,energy,shipbuilding,manufacturing,and other fields,and the demand for equipment integration,intelligence,and integration is also increasing..Among them,electromechanical equipment plays an important role in the aerospace field,the complexity of its functions is increasing,and the risk,cycle and cost of equipment development are getting higher and higher,which also puts forward the requirement that the equipment should have high reliability.As a technology for monitoring,diagnosing,predicting and managing the health status of equipment,Prognostics and Health Management(PHM)technology predicts the potential failures in the system and the remaining life of the equipment.Improving equipment safety,thereby reducing the impact of failures and logistics maintenance costs as much as possible,and reducing the occurrence of property losses and major accidents,this technology has received increasing attention and attention in many fields.Therefore,in-depth study of PHM technology related to electromechanical equipment has important theoretical value and practical significance.This article focuses on the PHM of the electromechanical servo control system,and carries out researches such as system equivalent mathematical model modeling,feature parameter identification,and health state prediction and evaluation.The main research contents are as follows:(1)The failure mechanism of the servo control system of the Stewart platform driven by the linear motor is analyzed for the semi-physical simulation test platform,the equivalent mathematical model is established,the fault tree is established,and the characteristic parameters of performance degradation are determined.PHM technical solution of the system(2)Apply the Unscented Kalman Filter(UKF)theory to study the characteristic parameter identification method,and give a multi-UKF cyclic push-pull structure observation for the problem that the multi-parameter simultaneous identification cannot be realized due to the lack of state equations The design method of the device realizes the joint identification of multiple parameters(3)Introduce fuzzy membership rules,adopt the DS evidence fusion theory method based on conflict allocation to realize the health status assessment,prediction and remaining useful life(RUL)prediction of the servo system,and give the cause of the failure in the corresponding state,Facilitate auxiliary maintenance decisions.(4)Give a data fusion prediction method based on nonlinear modified ARIMA model and Long Short-Term Memory(LSTM)neural network,improve the long-term prediction accuracy of the data,and help complete the health of the servo system Prediction of status and failure.Based on the above research results,the corresponding simulation verification is completed.The research results show that the multi-UKF cyclic push-pull observer proposed in this paper and the fusion prediction method based on nonlinear modified ARIMA-LSTM can better solve some existing problems.The evaluation method based on the fusion of fuzzy membership function and DS evidence can also give a good health status of the system at present and in the future.