Research On Wind Turbine On-line Monitoring And Analysis System Based On Reconfigurable Component Technology

A wind turbine data collection and online surveillance system is essential to ensure the long-term safe and reliable operation of wind turbines.Wind turbine data collection and online surveillance systems are mostly developed and supplied by professional manufacturers other than the complete machine enterprises,faced with many unit models,diverse sensor data,data is only used for reports and lack of analysis and prediction function.In view of the above problems and considering the demand of unattended wind farms in the future,the framework of online surveillance and evaluation system of wind turbines for regional centralized control is studied,and two key technologies are mainly studied,namely reconfigurable component technology for heterogeneous wind turbines and wind turbine prediction technology based on system acquisition data.Firstly,utilizing the concept of regional centralized control for unstaffed wind farms,the architecture of an online monitoring and analysis system for wind turbines is formulated,followed by the comprehensive planning and design of the system using the C/S development mode,and concluding with the establishment of system function and database planning.Secondly,in view of the heterogeneity of structure and sensor data of different types of wind turbines,the reconfigurability of the online monitoring and analysis system of wind turbines adapted to different data acquisition object systems was analyzed.Designed with a component-based approach,the reconfigurable data structure components and interfaces are developed to accommodate the dynamic configuration requirements of various regions,wind farms,and wind turbine monitoring systems.Then,aiming at the problem of "only monitoring and little analysis" in traditional wind turbine data acquisition and online monitoring system,this thesis tries to use neural network to study the analysis and prediction function based on the data collected by the system.A deep learning model based on time series is established for the data collection of the online monitoring and analysis system of wind turbines.A wind turbine state assessment index model is suggested utilizing information integration approach,and the wind turbine power is forecasted utilizing CNN-LSTM.Based on the above theoretical and technical research results,this thesis participated in the development of Hua Neng New Energy group’s wind turbine regional centralized control system,and designed and constructed a simulated testing environment,to confirm the practicality and effectiveness of the above research results.