Life Prediction Of Power Machinery Based On Particle Filter

Power machinery is a type of mechanical device that is supported by rolling or sliding bearings and relies on blades,gears and other components for power transmission or energy conversion.In power systems,steam turbines,gas turbines,and wind turbines are typical power machines.After years of development,the power machinery in the industry has the characteristics of large-scale,automation and precision,which is beneficial to the smooth progress of national production.The failure of such electric power machinery will inevitably cause greater economic losses and even more serious consequences.Therefore,it is of great theoretical significance and engineering value to carry out state monitoring,fault warning and life prediction of power machinery.The failure modes of the rotating machinery are various.Specifically,the blade cracks of the turbine and gas turbine shafts,the uneven thermal stress,and the combustion oscillation will cause the unit to stop.The fracture and wear of the transmission gears and bearings in the wind turbine need to be repaired and replaced,cause the loss of power generation.Vibration analysis is an effective means to carry out the above-mentioned equipment fault trend monitoring.By collecting vibration signals,extracting fault features reflecting fault trends,and combining life prediction methods,faults can be effectively found and the unit's failure time can be predicted,providing technology for unit maintenance.Particle filtering is a recursive Bayesian analysis method.It performs state prediction through the state transition matrix of the system,and combines a large number of known distributed particles,observation matrices and resampling techniques to update the system state.The particle filter-based state estimation method has been widely used in the field of remaining life prediction because of its low demand historical data,wide application range and relatively small computational complexity.In this paper,the research on residual life prediction method of rotating machinery based on particle filter is carried out.The resampling part of particle filter and the construction of system model are improved to some extent.The improved particle filter is more resistant to particle starvation.The adaptability and stability are also improved.Based on the improved particle filter and the unscented Kalman filter,the residual life prediction method based on the unscented particle filter is proposed.The particle filter group can be optimized by the unscented Kalman filter to further improve the stability and prediction accuracy of the filter.The vibration trend prediction of gas turbine and the bearing life prediction of wind turbine generators verify the effectiveness of the proposed method.