摘要
随着我国经济的快速发展、工业现代化进程加速,能源需求正在持续增长,为了响应国家“碳达峰”“碳中和”目标,风能作为清洁、无污染、可再生并且开发技术成熟的清洁能源,需求更是与日俱增。风机装机量在增加的同时,也伴随着故障检修不及时、不全面以及检修人员匮乏等问题。因此本文提出了基于大数据及人工智能技术故障预警方法,使用深度学习堆叠自编码(SAE)算法对风电机组齿轮箱散热及齿轮箱轴承类故障进行预警,模型准确率达73.6%;利用长短型记忆网络(LSTM)模型完成了发电机轴承故障诊断预警;利用机组降容的数据特点与图像识别进行深度融合对风电机组降容诊断及根因分析,模型准确率达到85%。
With rapid economic development of our country and accelerated industrial modernization,the demand for energy is growing continuously.In order to respond to the national goal of"reaching the peak of carbon"and"carbon neutrality",wind energy as clean,pollution-free,renewable and developed technology mature clean energy,the demand is increasing.At the same time of the increase of fan installation,there are also problems such as not timely and comprehensive fault maintenance and lack of maintenance personnel.Therefore,a fault warning method based on big data and artificial intelligence technology is proposed in this paper.Deep learning stack autocoding(SAE)algorithm is used to warn wind turbine gearbox heat dissipation and gearbox bearing faults,and the model accuracy reaches 73.6%.The Long and short memory network(LSTM)model is used to complete the generator bearing fault diagnosis and early warning.The data characteristics of wind turbine capacity reduction and image recognition were used for deep fusion of wind turbine capacity reduction diagnosis and root cause analysis,and the model accuracy reached 85%.
作者
吴吉军
冯江哲
王灿
WU Jijun;FENG Jiangzhe;WANG Can(Longyuan(Beijing)Wind Power Engineering Technology Company,Beijing 100034,China)
出处
《风力发电》
2022年第5期27-29,26,共4页
Wind Power
关键词
大数据与人工智能
堆叠自编码算法
LSTM模型
图像识别
Big data and artificial intelligence
Stacked autocoding algorithm
LSTM model
Image recognition
