Research On Remaining Useful Life Prediction Of Wind Turbine Based On Deep Learning

Thermal power generation is accompanied by the crisis of energy consumption and environmental pollution,so the use of renewable clean energy power generation has become the mainstream of the times.Driven by the tide of the times,clean and efficient wind power generation ushers in rapid development.Gearbox plays an extremely important role in the wind turbine,but it is also one of the most failure components.Its working state directly affects the life of the wind turbine.The gearbox is installed on the top of the tower of the wind turbine.When there is a fault,the maintenance cost is very high due to the difficulty of maintenance.Therefore,it is of great significance to carry out the research on the prediction of the remaining useful life of the gearbox of the wind turbine to improve the service life of the wind turbine and reduce the maintenance loss.According to the vibration data collected by the sensor from the gearbox test platform,which can represent the degradation state of the gearbox,this paper proposes a prediction method based on deep learning to predict the remaining useful life of the gearbox according to the fault threshold.The main research work is as follows:(1)This paper presents a method to predict the remaining useful life by using the integrated model of convolutional neural network and short-term memory network.In the integrated model,the convolution neural network is used to automatically extract the acceleration signal features of gear vibration from the data point of view,so as to improve the impact of artificial features on the results in the traditional life prediction methods;the long short-term memory network is improved by using parameters selection and optimization,so that it has better ability to process long interval or long delay time series,so as to predict the remaining useful life of gearbox.The validity of the method is verified by experiments.(2)A method for multi-sensor data fusion to predict remaining useful life is proposed.Due to the limited measurement range of a single sensor,multiple sensors can collect more comprehensive and effective information about the operating status of the gear.Therefore,this method combines a convolutional neural network with data fusion theory to combine the gear vibration acceleration signals collected by multiple sensors data fusion with noise signals improves the accuracy of the model in predicting the remaining useful life,and verifies the effectiveness and feasibility of the method.