Research On Remaining Useful Life Prediction Of Wind Turbine Based On Deep Belief Network

In order to alleviate the environmental problems caused by the energy crisis and unreasonable exploitation of resources,green and pollution-free energy such as wind energy has been widely used.Through wind power generation,on the one hand,it can reduce the use of coal,on the other hand,it can protect the environment.With the increase of the service time of the unit,accidents of wind turbines occur frequently in recent years,resulting in system fault,economic loss,and even life-threatening.Wind turbines are usually installed in remote places,and there are many internal components in the wind turbines,which are much more difficult to maintain.As a key component of the wind turbine,the gearbox is connected to the whole transmission,so it is difficult to maintain once the fault occurs.Therefore,we can formulate the corresponding maintenance strategy in advance by predicting the remaining useful life of the gearbox,so as to save the maintenance cost and reduce the failure time.In this thesis,based on the background of the change of gear vibration characteristic value in the gearbox with the running time,a model is established to predict the remaining useful life of the gear.The main contents are as follows:(1)In order to solve the problem that a single model often does not have good generalization performance,a method of deep belief network ensemble is proposed.The method of negative correlation learning combined with genetic algorithm is used to train multiple neural networks at the same time to make the model learn the rules of the data more accurately.Finally,the average value of the output of multiple individual networks is taken as the output of the model.Taking the vibration acceleration data of the gear as an example,compared with the single deep belief network,the accuracy of the model is improved,and the effectiveness of the proposed method is verified.(2)Niche technology is introduced to improve population diversity,and a deep belief network ensemble model combined with niche technology is established.The niche technology of sharing mechanism is used to reduce theprobability of selection of individual networks with similar fitness,so as to maintain population diversity.Compared the ensemble model with the model of no niche technology,the population diversity is improved after adding niche technology,and it is verified that the proposed method can be applied to the ensemble model to improve population diversity.