摘要
针对风电机组基于SCADA数据预测分析故障效率低、准确度差的问题,提出一种融合长短期记忆网络(LSTM)和麻雀搜索算法(SSA)优化支持向量机(SVM)的机器学习模型.首先,利用SCADA历史数据分别训练LSTM和SVM,使其分别具有预测未来数据和判别故障状态的能力;其次,采用SSA优化SVM,保障其寻优能力强、收敛速度快;最后,通过优化的SVM对LSTM预测得到的未来数据进行故障状态识别判定,实现风电机组短期故障预测.通过现场数据测试结果表明:所提出的方法预测风电机组故障准确率为97.90%,能够满足现场对风电机组短期故障预测的实际需求.
In response to the problems of low fault efficiency and poor accuracy in SCADA data prediction and analysis of wind turbines,a machine learning model combining LSTM and SSA to optimize SVM is proposed.Firstly,LSTM and SVM are trained with SCADA historical data to make them have the ability to predict future data and identify fault state respectively;secondly,SSA is used to optimize SVM to ensure its strong optimization ability and fast convergence speed;finally,the optimized SVM is used to identify fault state of LSTM predicted future data to realize short-term fault prediction of wind turbine.The field data test results show that the fault prediction accuracy of the proposed method is 97.90%,which can meet the actual needs of short-term fault prediction of wind turbines.
作者
陈万志
李昊哲
刘恒嘉
王天元
CHEN Wanzhi;LI Haozhe;LIU Hengjia;WANG Tianyuan(College of software,Liaoning Technical University,Huludao 125105,China;School of Electronic and Information Engineering,Liaoning Technical University,Huludao 125105,China;School of Computer Science and Technology,ChangChun University of Science and Technology,Changchun 130022;State Grid Yingkou Electric Power Company,Liaoning Electric Power Supply Company Limited,Yingkou 115000,China)
出处
《辽宁工程技术大学学报（自然科学版）》
CAS
北大核心
2022年第4期379-384,共6页
Journal of Liaoning Technical University (Natural Science)
基金
国家重点研发计划项目(2018YFB1403303)
辽宁省教育厅科学研究面上项目(LJKZ0327)
辽宁工程技术大学双一流学科建设创新团队项目(LNTU20TD-25)
作者简介
陈万志(1977-),男,辽宁黑山人,博士,副教授,主要从事人工智能与智能信息处理、工控软件与数据分析方面的研究.
