Unsupervised Complex Condition Recognition Based on Stochastic Neighborhood Embedding
Modern industrial production equipment usually has a complex structure and runs alternately in different working conditions.Accurate working conditions identification based on monitoring data is the basis of health monitoring of the system,but the monitoring data of the system usually has a high dimension and a large data volume.To identify the complex equipment operating conditions,an unsupervised operating condition identification method based on stochastic neighborhood embedding is proposed.The stochastic neighborhood embedding algorithm can simultaneously preserve the local and global structural characteristics of the data,and also calculate the probability similarity of data points in high-dimensional and low-dimensional space to achieve the dimensionality reduction and unsupervised clustering of equipment high-dimensional monitoring data and to accurately identify the system operating conditions without establishing a system model.The results show that the proposed method can effectively identify complex operating conditions from high-dimensional monitoring data,which is an effective unsupervised clustering learning method.
NETL
NSTL
万方数据
