中国电力2025，Vol.58Issue(8)：94-102,9.DOI:10.11930/j.issn.1004-9649.202412081
基于注意力机制和RCN-BiLSTM融合的风电机组故障识别
Fault Identification for Wind Turbine Based on Attention Mechanism and RCN-BiLSTM Fusion
摘要
Abstract
To improve the identification accuracy of wind turbine faults,an RCN-BiLSTM-Attention wind turbine fault identification method based on the attention mechanism(AM)and the fusion of residual capsule network(RCN)and bidirectional long short-term memory(BiLSTM)network is proposed.First,the abnormal values in the wind turbine supervisory control and data acquisition(SCADA)system are eliminated by the density-based spatial clustering of applications with noise(DBSCAN)algorithm.Then,RCN is employed to extract spatial relational features from the fault data,the BiLSTM is performed to dynamically capture the hierarchical temporal dependencies of the spatial features extracted by RCN to obtain the temporal information of multiple faults,and fusion AM assigns different weights to the outputs of BiLSTM to enhance the accuracy of wind turbine fault identification.Finally,the SCADA data of several wind turbines are used to verify that the proposed model provides high identification accuracy and generalization ability compared with other models.关键词
风电机组/故障识别/双向长短期记忆网络Key words
wind turbine/fault identification/bidirectional long short-term memory引用本文复制引用
陈小乾,尹亮,展宗辉,王放,李旭涛..基于注意力机制和RCN-BiLSTM融合的风电机组故障识别[J].中国电力,2025,58(8):94-102,9.基金项目
宁夏自然科学基金资助项目(2024AAC03755). This work is supported by Ningxia Natural Science Foundation of China(No.2024AAC03755). （2024AAC03755）
