摘要
为了准确监测到风电机组齿轮箱的运行状态以实现其早期预警,提出了一种基于改进深度置信网络(Deep Belief Net-works,DBN)的运行状态识别方法。首先,利用粒子群算法(Particle Swarm Optimization,PSO)优化DBN网络的结构参数,运用最优的DBN网络结构提取样本数据特征。将特征通过多维尺度分析(Multidimensional Scaling,MDS)算法映射到低维空间,在低维空间内依据欧氏距离构建齿轮箱状态指标,结合状态指标实现样本数据标签化。再采用标签化的样本数据训练极限学习机(Extreme Learning Machine,ELM)模型,识别齿轮箱的运行状态。结果表明,该方法的识别准确率达到95.61%,不仅深度挖掘到样本数据的特征信息,还通过构建状态指标为无标签的样本数据处理提供了参考。
In order to accurately monitor the operating status of wind turbine gearbox for early warning,an operating status identification method based on improved DBN is proposed.Firstly,structural parameters of DBN networks are optimized by using particle swarm algorithm.Sample data features are extracted by the optimal DBN network structure.The features are mapped to the low-dimensional space by using MDS algorithm.Gearbox state index is constructed based on the Euclidean distance in the low-dimensional space.Combined with the status indicators,sample data are labeled.The labeled sample data is then used to train the ELM model to identify the operating state of the gearbox.The results show that the recognition accuracy of the proposed method is 95.61%.The characteristic information of sample data is deeply mined.A reference for the processing of the unlabeled sample data through constructing status indicators is provided.
作者
刘杰
付雪娇
孙兴伟
LIU Jie;FU Xuejiao;SUN Xingwei(School of Mechanical Engineering,Shenyang University of Technology,Shenyang Liaoning 110870,China)
出处
《传感技术学报》
CAS
CSCD
北大核心
2023年第3期434-440,共7页
Chinese Journal of Sensors and Actuators
基金
辽宁省教育厅(LQGD2020016)
辽宁省“兴辽英才计划”资助项目(XLYC1905003)。
关键词
风电机组
状态识别
深度置信网络
多维尺度分析
极限学习机
wind turbine
status identification
deep belief networks
multidimensional scaling
extreme learning machine
