Modeling And Application Of Wind Turbine Digital Twin System

With the goal of achieving emission peak and carbon neutrality,wind power has achieved rapid development,and large-scale grid-connected wind power will become the main power source of the new power system in the future.Along with the rapid growth of wind turbine installed capacity,conducting intelligent and digital condition monitoring and fault diagnosis research in the field of wind power,establishing practical and available mechanisms and data driven models,sensing the operating status of wind turbines in real-time,detecting potential fault symptoms as soon as possible,reducing fault rates,and reducing operation and maintenance costs have important practical significance for the safe and efficient power generation and operation of wind turbines and power systems.Based on above objectives,this paper has conducted the following researches:(1)Based on the idea of information-physical real-time mapping,a digital twin system for wind turbine is constructed and implemented.According to the actual needs of intelligent operation and maintenance of wind power,the overall functional architecture of the digital twin system is designed;The overall technical architecture of the digital twin system has been systematically constructed.Through effective organization of data and information flows,real-time communication,accurate mapping,and visual presentation between wind turbines and the digital twin platform have been achieved.(2)Starting from the mechanism modeling of key components of wind turbines.a digital twin model for wind turbines is established using multi-model digital thread interaction technology.By comparing the dynamic performance with the actual operation data of the wind turbine,the high fidelity of the digital twin model is verified,and real-time simulation of the key parameters and dynamic loads of wind turbines are realized.At the same time,the application value in condition monitoring and fault diagnosis is expounded.(3)A variational Bayesian LSTM temperature forecasting model including an input attention mechanism encoder and a temporary attention mechanism decoder was constructed for the nacelle temperature monitoring application scenario.The accuracy of the model is verified based on the measured data from wind turbine SCADA system.The practical value and significance of temperature probability forecasting are discussed in conjunction with the nacelle temperature status monitoring and fault warning interface of the digital twin system.(4)Based on the overall wind turbine digital twin model and the model development management system,the application of the digital twin platform in advanced model development scenario was studied.An adaptive feedback sliding mode control strategy for blade pitch control is established.Compared with traditional PID pitch control,the multi-dimensional comparison verifies that the model has better robustness and rapidity under extreme wind conditions,as well as excellent performance in reducing mechanical load of the equipments and smoothing power output.