基于TCN的风电机组变流器故障预测研究  

TCN-based Wind Turbine Converter Fault Prediction

作　　者：肖成 褚越强 刘博天 赵嗣彪 Xiao Cheng;Chu Yueqiang;Liu Botian;Zhao Sibiao(School of Electronic and Control Engineering,North China Institute of Aerospace Engineering,Langfang 065000,China)

机构地区：[1]北华航天工业学院电子与控制工程学院,河北廊坊065000

出　　处：《北华航天工业学院学报》2024年第4期9-11,共3页Journal of North China Institute of Aerospace Engineering

基　　金：河北省教育厅重点项目(ZD2022089);北华航天工业学院博士基金项目(BKY-2023-03)。

摘　　要：以风电机组网侧变流器电压故障为研究对象,通过分析SCADA系统数据的特点,基于先验知识对缺失数据、异常值数据、离散异常数据进行了常规的数据补齐和删除处理,对较难识别的堆积数据采用基于最小二乘法的变点分组法进行了清洗。基于经验识别方法,选取了风电机组中变流器故障的故障特征变量,应用TCN深度学习网络算法,对具有时序特征的风电机组变流器SCADA数据进行分析,在故障特征变量识别的基础上,进行故障预测,预测准确率达到96.56%。Taking wind turbine grid-side converter voltage faults as the research target,by analyzing the features of SCADA system data,the conventional data complementation and deletion processes for missing data,outlier data,and discrete abnormal data were conducted on priori knowledge,and the unrecognizable stacked data were cleaned by the least squares-based variable point grouping method,and fault characteristic variables for converter faults in wind turbines were selected with empirical identification method. The TCN deep learning network algorithm was applied to analyze the wind turbine converter SCADA data with temporal characteristics,and the fault prediction was performed based on the fault feature identification,and the prediction accuracy reached 96.56%.

关 键 词：风电机组 网侧变流器电压故障 故障预测 TCN 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]