Residual Life Prediction Of High Speed Shaft Bearing Of Wind Turbine Drive Chain

The goal of "Carbon Neutral" and "Carbon Peak" will promote the power industry to build more new energy generating units,and the high-speed bearing of wind turbine is often in the bad working environment of variable load and high speed,which leads to a high failure rate.Once the bearing fails,it will often cause system level damage and bring great economic losses and security risks.Therefore,it is of great engineering significance to predict the residual life of high-speed bearing based on the monitoring data of wind turbine.In view of the limitations of the existing residual life prediction methods of rolling bearings in practical application,this paper puts forward a method to predict the residual life of rolling bearings only by using historical vibration data,and verifies and analyzes the research results through the data of pronostia test bench and the whole life data of high-speed bearings of 8 domestic wind turbines in service.The main contents of this paper are as follows:(1)The specific working position of high-speed bearing of wind turbine is defined,the common types and failure forms of high-speed bearing are analyzed statistically,and the common vibration analysis methods of high-speed bearing are explained;The extraction and selection of sensitive features in the process of wind turbine high-speed bearing degradation are completed.The initial feature set which can represent the bearing degradation process is constructed,and a new monotonicity trend evaluation index is proposed for sensitive feature selection.(2)In order to eliminate the adverse effects of different dimensions between different features on the construction of health indicators,a method of integral translation difference for sensitive features is proposed to measure the bearing degradation performance;In view of the fact that the monotony of most existing health indicators is unreasonable,a new idea of health indicator construction is proposed,that is,to calculate the deepening degree of bearing fault at different times,so as to obtain the construction method of recursive health indicators;In order to determine the health index failure threshold,combined with the research results of relevant literature,the C value curve was constructed,and the failure C value was determined by statistical analysis of the C value curve of bearings at the same working position,so as to obtain the health index failure threshold.(3)The principle,advantages and disadvantages of Particle Filter and Gaussian process regression algorithm are analyzed.By using Gaussian process regression prediction results as Particle Filter observations and combining with Particle Filter to modify the prediction results,a PF-GPR fusion residual life prediction model is constructed;The feasibility of the proposed health index construction method and the PF-GPR prediction model is verified by the data of pronostia test bench and the measured whole life data of high-speed bearing of in-service units;The results show that the research results of this paper are not only better than other methods in the test-bed data,but also can achieve good results in the residual life prediction of high-speed bearings of in-service units only using historical vibration data.