Research On Fault Feature Detection And Short-term Reliability Assessment Of Variable Pitch System

With the rapid growth of total installed capacity and operation time of wind turbines,the fatigue failures and the operations and maintenance costs are dramatically increased.As the core component of wind turbines,pitch system has complex structure and high failure rate,resulting in large costs from maintenance and part replacement.Fault feature detection and reliability assessments of pitch system are the key process of condition monitoring.Through the reliability assessment,each operation stage of wind turbine can be timely feedback to make reasonable maintenance strategies.Therefore,using real-time SCADA system data,this thesis digs out fault features of the pitch system,and makes full use of the fuzzy matter-element method to assess the short-term reliability results from the fault feature detection model.First of all,the basic structure,control principles and related monitoring parameters of pitch system are discussed and analyzed.After describing the fault mechanism of the pitch system,the feature parameters which can reflect the state of the pitch system are mined.After introducing and preprocessing the raw SCADA system data,box-plot is used to analyze the sample data from four aspects and get the baseline reference,which laying the foundation for the subsequent establishment of the fault feature detection model.Then,according to the distribution characteristics of wind speed and rotor speed,this thesis uses 3? rule and quartile analysis method to get the boundary data sets.The ANFIS method is used to fit the boundary data sets in order to establish the nonlinear fault zone.Simulations such as between wind speed and pitch angle model,between wind speed and generator speed model,between rotor speed and the motor drive current model,between rotor speed and pitch speed model are carried out.The average degradation degrees of feature parameters are obtained,which are used as the input of the next step reliability evaluation model.Finally,in accordance with the fuzzy matter-element theory,the state matter-element of pitch system is determined.Paired comparison method is used to calculate the feature parameters of weight.Short-term reliability assessment model of pitch system is obtained by the above process.Combined with the actual operation of wind turbines,average degradation degrees of feature parameters are got into the reliability assessment model,then assessment results are obtained,which verifies the validity and feasibility of this model.At the same time,compared with other assessment methods,the fuzzy matter-element method is more accurate to assess the abnormal state of the sample data.