Research On Fan Blade Fault Diagnosis System Based On SCADA Data

As global warming becomes more serious,the country has increased its use of clean energy.As a clean and renewable energy source,wind energy is playing an increasingly important role in the field of power generation.In order to obtain high-quality wind energy resources,wind turbines are often arranged in relatively desolate places such as coastal areas,high mountain areas,and wasteland,and their distribution is relatively scattered.This makes it inconvenient for wind power companies to perform maintenance on the fans.Due to its special working environment,special manufacturing process,and important work responsibilities,fan blades occupy a lot of use and maintenance costs for wind power generation companies.Because the damage to the blades cannot be found in time,it often causes wind turbine generator equipment to fall down,or the blades fall during work and accidentally hurt the maintenance personnel.Therefore,it is necessary to develop a system of accurate fault diagnosis system for fan blades.This paper designs a fan blade fault diagnosis system and implements the main functional modules of the fan blade fault diagnosis system.The system can collect the data generated in the SCADA system of the wind turbine farm,clean SCADA data,and store the cleaned data in the distributed storage system HDFS.By analyzing the collected data,it is possible to detect whether the blade of the wind turbine has failed.The fault diagnosis module of this system mainly relies on the hybrid model of fan blade fault diagnosis based on Light GBM + XGBoost proposed by the author.The model uses Bayesian optimization to automatically select model parameters,uses Light GBM and XGBoost two existing machine learning techniques to train the model,and uses five-fold cross-validation to optimize the model.The model showed higher accuracy,precision,recall,and F1 value during the experimental verification phase.Using this system can better diagnose whether the fan blade is faulty.