US 12,354,007 B2
Artificial intelligence (AI)-based method for non-contact measurement of sheet resistance of a conductive film material
Yaqing Liu, Taiyuan (CN); Guanyu Han, Taiyuan (CN); and Guizhe Zhao, Taiyuan (CN)
Assigned to NORTH UNIVERSITY OF CHINA, Taiyuan (CN)
Filed by NORTH UNIVERSITY OF CHINA, Taiyuan (CN)
Filed on Jan. 17, 2025, as Appl. No. 19/027,621.
Application 19/027,621 is a continuation of application No. PCT/CN2024/092510, filed on May 11, 2024.
Claims priority of application No. 202410555163.X (CN), filed on May 7, 2024.
Prior Publication US 2025/0156714 A1, May 15, 2025
Int. Cl. G06F 30/18 (2020.01); G06N 3/08 (2023.01)
CPC G06N 3/08 (2013.01) 10 Claims
OG exemplary drawing
 
1. An artificial intelligence (AI)-based method for non-contact measurement of sheet resistance of a conductive film material, comprising:
(S1) based on a material type of a to-be-tested conductive film material, determining a sheet resistance range of the to-be-tested conductive film material as Rsql-Rsqn;
(S2) based on the sheet resistance range Rsq1-Rsqn, a material texture of a first to-be-used test substrate, and a thickness range h1-hn of the first to-be-used test substrate, constructing a structural dataset and training an AI model through steps of:
based on a Salisbury screen I with a conventional electromagnetic wave-absorbing structure Salisbury screen, constituted by the to-be-tested conductive film material, the material texture of the first to-be-used test substrate, and the thickness range h1-hn of the first to-be-used test substrate, setting the structural dataset consisting of the sheet resistance range Rsq1-Rsqn, the thickness range h1-hn, and the material texture of the first to-be-used test substrate as an input set;
obtaining electromagnetic wave reflection losses RLsim respectively corresponding to individual groups of the structural dataset in the input set by using a microwave radio frequency simulation software as an output set; and
training an artificial neural network through a training algorithm; and obtaining adaptively a mapping relationship between the structural dataset of the Salisbury screen I in the input set and a corresponding RLsim in the output set; and obtaining an AI model that is capable of accurately predicting and outputting a corresponding electromagnetic wave reflection loss RLpre data when a first structural data within the structural dataset of the Salisbury screen I is input;
(S3) experimentally obtaining an electromagnetic wave reflection loss RLtest data of a Salisbury screen II of an actual test sample constituted by the to-be-tested conductive film material and a second to-be-used test substrate whose thickness h is within the thickness range h1-hn and material texture is the same as that of the first to-be-used test substrate;
(S4) embedding the AI model obtained in step (S2) into a step-by-step search algorithm program to quickly predict an electromagnetic wave reflection loss RLpre data of a Salisbury screen III to obtain a step-by-step AI-based sheet resistance search program, wherein the Salisbury screen III is temporarily generated during an operation of the step-by-step search algorithm program, and a second structural data consists of a conductive film material with a sheet resistance falling within the sheet resistance range Rsq1-Rsqn and the second to-be-used test substrate;
inputting the sheet resistance range Rsq1-Rsqn and the thickness h of the second to-be-used test substrate of the Salisbury screen II in step (S3) into the step-by-step AI-based sheet resistance search program;
determining a total iteration number j according to a required operation precision i, expressed as

OG Complex Work Unit Math
wherein n is a number of sheet resistance ranges Rsq1k-Rsqnk divided in a single iteration; and
simultaneously importing the RLtest data obtained in step (S3) into the step-by-step AI-based sheet resistance search program, and obtaining the sheet resistance of the to-be-tested conductive film material after j iterations.