Vibration Characteristics Analysis And Experimental Study Of Parallel-stage Tooth Root Crack Fault Of Wind Turbine Gearbox

Wind turbine gearbox is the key core transmission component of wind turbine,which is often installed in the narrow cabin.It needs long-term operation in harsh environments such as gust and turbulent wind.It will be affected by complex multi-source internal and external excitations such as input torque fluctuation,time-varying meshing stiffness and time-varying transmission error,resulting in obvious alternating loads on gears,shafts and other components of wind turbine gearbox transmission system.Among them,the parallel gear of the gearbox has high speed and frequent load,and the bending stress cycle and stress concentration effect are prone to crack or even break the tooth root.According to statistics,38% of the faults of the wind turbine gearbox occur in the parallel stage transmission system.Therefore,the vibration characteristics analysis of the parallel stage tooth root crack of the wind turbine gearbox is helpful to find the early faults of the wind turbine gearbox in advance and ensure the safe operation of the wind turbine.Taking a certain type of 5 MW wind turbine gearbox parallel stage as the research object,the time-varying meshing stiffness equation of gear with penetrating / non-penetrating tooth root crack is derived,and the influence of tooth root crack propagation direction and propagation path on it is considered respectively.Combined with the time-varying transmission error of gear pair,the flexibility of transmission shaft and the stiffness of bearing support,the dynamic model of wind turbine gearbox parallel stage system is established.The influence of tooth root crack mode on the dynamic characteristics of wind turbine gearbox parallel stage is analyzed.The fault feature recognition method of wind turbine gearbox parallel stage with tooth root crack fault is studied,and the theoretical analysis is verified by experiments.The research results have important theoretical guiding significance for early fault diagnosis of wind turbine gearbox.The main research work of this paper is as follows :(1)Taking the parallel stage of a 5 MW wind turbine gearbox as the research object,and considering the tooth root transition arc,the time-varying meshing stiffness equations of the gear with penetrating tooth root crack / non-penetrating tooth root crack are derived by using the potential energy method.The finite element method is used to establish the dynamic model of wind turbine gearbox parallel stage system considering the flexibility of transmission shaft,bearing support stiffness,time-varying meshing stiffness of gear pair and transmission error.(2)The influence of the propagation depth of the penetrating tooth root crack and the variation of the propagation width of the non-penetrating tooth root crack along the tooth width direction on the time-varying meshing stiffness of the gear pair is analyzed;based on the dynamic model of the wind turbine,the input loads of the parallel stage system of the wind turbine gearbox under different wind conditions are obtained.The influence of the root crack fault mode on the dynamic characteristics of the parallel stage pinion node of the wind turbine gearbox under steady and turbulent wind conditions is studied.(3)Based on the time-frequency domain characteristics of displacement vibration signal,the fault vibration feature set of the tooth root crack of the parallel stage of the wind turbine gearbox is constructed;a pattern recognition method of tooth root crack based on generalized BP neural network model is proposed,and the effect of generalized BP neural network on pattern recognition of parallel gear root crack of wind turbine gearbox under steady and turbulent wind conditions is analyzed.(4)The vibration response experiment of parallel-stage gear system with tooth root crack fault was carried out.The vibration response test system of parallel-axis gear system was built,and the vibration response characteristics under different crack degrees were analyzed.The generalized BP neural network model was used to identify the fault characteristics of the measured signals with different crack degrees,and the accuracy and effectiveness of the model were verified.