Research On Intelligent Condition Monitoring Method For Wind Turbine Planetary Gearbox

Wind power,as the main component of clean energy,plays a crucial role in optimizing the energy structure reform,controlling carbon emissions and protecting the lucid waters and lush mountains.As a part of wind turbines with high fault frequency and long downtime,planetary gearbox can be reliably and effectively monitored by judging and predicting its current status and development tendency,meanwhile it can guide the maintenance work of planetary gearbox of wind turbines accordingly.Compared with the current post-mortem maintenance mode and regular maintenance mode commonly used in wind farms,identifying and predicting the fault status of planetary gearbox of wind turbines by condition monitoring method,and then carrying out corresponding maintenance can get better performance in ensuring its economic benefits and production safety,which is of great significance for the further development of wind power technology.Taking double-feed wind turbine planetary gearbox as the main research object,the condition monitoring of gearbox is studied by using artificial intelligence approach and vibration signal of gearbox.The fault type and severity of the gearbox are identified by the vibration signal at first,then the fault trend and remaining useful life of the gearbox in the fault state are predicted,and the condition monitoring of the gearbox is realized by integrating wireless sensor technology.The main contents of this paper include:(1)A deep hierarchical method for early fault type and severity diagnosis of wind turbine planetary gearbox is presented.Compressed sensing(CS)method is adopted to compress the raw vibration signal in the pre-processing procedure of the input gearbox vibration signal.The compressed sampling data is employed as the input of deep hierarchical feature extraction structure,which combines the hierarchical diagnostic network with the deep belief network(DBN).After extracting the feature vectors that represent the failure modes,they are input into least squares support vector machine(LSSVM)to discriminate the failure modes.At the same time,in order to ensure the optimal performance of feature extraction and fault identification for wind turbine planetary gearbox,the parameters of each DBN and LSSVM in the hierarchical diagnostic structure are optimized by chaotic quantum particle swarm optimization(CQPSO)algorithm,and the optimized stacked diagnosis structure(OSDS)is finally completed,for obtaining better fault diagnosis performance of wind turbine planetary gearbox.The experimental results show that this method can achieve better fault diagnosis accuracy than traditional fault diagnosis methods.(2)An intelligent double-index fault prediction method for wind turbine planetary gearbox is presented.Before fault prediction,the HI_R is constructed by compressed sampling valid values and the kurtosis values of the original vibration signal,and the HI_B is constructed by using the ratio of the Hilbert envelope value at characteristic frequencies,thus completing the double index construction.By placing least squares support vector regression(LSSVR)on the last hidden layer of a single DBN structure,a DBN-LSSVR fault prediction model is constructed,and CQPSO algorithm is adopted to optimize it.Finally,the CQPSO-DBN-LSSVR fault prediction structure is constructed to improve its fault prediction performance for health indicators and the remaining useful life.The experimental results show that this method can obtain a more accurate prediction performance than traditional methods for the built double index HI and the remaining life of planetary gearbox.(3)A method for condition monitoring of wind turbine planetary gearbox fused with wireless sensors technology is presented.Wireless sensor node,which combines radio frequency identification(RFID)with piezoelectric energy harvester,is designed to collect and transmit planetary gearbox vibration signals.Their normal working cycle is guaranteed by the logic control of energy management circuit and microcontroller.Compared with wiring signal collection,using wireless sensor device can reduce the cost of pre-installation and post-maintenance before ensuring normal operation.The experimental results show that the device can complete the data collection and transmission tasks,moreover an example of wind turbine planetary gearbox failure in the actual project shows that the condition monitoring method can obtain more accurate diagnosis and prediction than traditional methods,which proves that the method can provide effective condition monitoring guarantee for the planetary gearbox of wind turbine in the actual project application,and has obtained certain application value.