	US 12,411,907 B1
	Method and apparatus for arranging sparse sensor embedded with physical information, and device
Ying Wang, Shenzhen (CN); and Biqi Chen, Shenzhen (CN)
Assigned to Harbin Institute of Technology, Shenzhen, Shenzhen (CN)
Filed by Harbin Institute of Technology, Shenzhen, Shenzhen (CN)
Filed on Jan. 18, 2025, as Appl. No. 19/031,488.
Claims priority of application No. 202410756556.7 (CN), filed on Jun. 13, 2024.
Int. Cl. G06F 17/11 (2006.01); G06F 17/12 (2006.01); G06F 17/16 (2006.01)

CPC G06F 17/11 (2013.01) [G06F 17/12 (2013.01); G06F 17/16 (2013.01)]

6 Claims

1. A method for arranging a sparse sensor embedded with physical information, specifically comprising:

obtaining prior information of a target structure, mapping the prior information onto high-dimensional feature space by using a plurality of kernel functions, to obtain a plurality of high-dimensional feature matrices; performing singular value decomposition (SVD) on a corresponding high-dimensional feature matrix of each kernel function by using a kernel matrix of the kernel function, to obtain a plurality of singular value matrices, concatenating the singular value matrices into a concatenated singular value matrix, and performing the SVD on the concatenated singular value matrix to obtain an adaptive basis matrix of the prior information, wherein the singular value matrices each comprise singular vectors corresponding to top-ranked maximum singular values of a preset quantity;

constructing a problem of an optimal sparse measurement matrix by using a compressed sensing method based on the adaptive basis matrix, and solving the problem of the optimal sparse measurement matrix to obtain the optimal sparse measurement matrix; and

determining, based on the optimal sparse measurement matrix, a sensor arrangement strategy corresponding to the target structure, and arranging a sensor for the target structure, based on the sensor arrangement strategy, to capture a critical dynamic response of the target structure;

wherein when the optimal sparse measurement matrix is determined, a row vector in the adaptive basis matrix is analyzed to select a position with a maximum projection weight in the adaptive basis matrix to arrange the sensor;

wherein the problem of the optimal sparse measurement matrix is as follows:

y=(CΨ_r)·s=⊖·s

wherein y represents a measurement state vector collected by the arranged sensor, C represents a sparse measurement matrix, Ψ_rrepresents the adaptive basis matrix, ⊖ represents a matrix product of the sparse measurement matrix and the adaptive basis matrix, and s represents a sparse vector; and

the solving the problem of the optimal sparse measurement matrix to obtain the optimal sparse measurement matrix specifically comprises:

obtaining a quantity of sensors to be arranged on the target structure; and

when the quantity of sensors is equal to a quantity of rows in the adaptive basis matrix, solving the problem of the optimal sparse measurement matrix by means of decomposition with partial pivoting to obtain the optimal sparse measurement matrix; or

when the quantity of sensors is less than a quantity of rows in the adaptive basis matrix, solving the problem of the optimal sparse measurement matrix by means of SVD to obtain the optimal sparse measurement matrix.