Rotating machine high-dimensional data-time-space fusion state monitoring method

• Main Authors: LI ZHECONG, HE YUANBIAO, QIAO ZIJIAN, YANG CHANGPIAO, LI HAO, ZHANG CHENGLONG, XIE BIAOBIAO, BIAN WANLIANG
• Format: Patent
• Language: Chinese; English
• Published: 19.03.2024
• Subjects: CALCULATING; COMPUTING; COUNTING; ELECTRIC DIGITAL DATA PROCESSING; PHYSICS

The invention provides a rotating machine high-dimensional data-time-space fusion state monitoring method, which comprises the following steps: taking a collected one-dimensional vibration signal as a sample, constructing n rows of Hankel matrixes according to a sequence with an interval of 1, and realizing data dimension raising; based on an empirical orthogonal decomposition method, firstly, a covariance matrix of a Hankel matrix of a vibration signal is obtained, all feature vectors and feature values of the covariance matrix are obtained through solving, and a spatial distribution matrix of a sample is formed by combining all the feature vectors; the information contribution rate of each dimension of the sample to all dimensions is determined by the eigenvalue of the dimension, all the eigenvalues are combined to obtain a contribution rate vector, and a Hankel matrix and a spatial distribution matrix are utilized to obtain a time distribution matrix; solving a norm of the contribution rate vector, and tak