System and method for automatically storing boring logs information using artificial intelligence

Patent metadata

Field	Value
Publication number	US-12529809-B2
Application number	US-202217993050-A
Country	US
Kind code	B2
Filing date	Nov 23, 2022
Priority date	Nov 26, 2021
Publication date	Jan 20, 2026
Grant date	Jan 20, 2026

Abstract

The present invention relates to a system and method for automatically storing boring logs information using artificial intelligence. The present invention can make a database of boring logs with high reliability without input errors by training a classification model using various forms of boring logs in advance by using artificial intelligence, identifying a form of a boring log to be actually stored in a database using the classification model when the boring log is input, and then extracting data from the boring log according to the identified form.

First claim

What is claimed is:

1 . A method for automatically storing boring logs information using artificial intelligence that is performed in a system for automatically storing boring logs information, the method comprising:
(a) training a deep learning-based classification model by converting a plurality of boring log files into images; (b) generating form information by applying, to the classification model, a boring log input for information extraction to identify a form of the boring log; (c) selecting information to be extracted from among standard penetration test (SPT) information and stratum information and generating a data frame including basic information and selected information; (d) extracting basic information and extracting at least one of the standard penetration test (SPT) information and the stratum information, from the boring log according to the form information; and (e) generating data to be stored in a database using the extracted information, wherein the step (d) comprises extracting the basic information and the information selected in the step (c), the step (e) comprises generating data to be stored in the database by inserting the information extracted in the step (d) into the data frame generated in the step (c), the step (c) comprises converting the boring log into a text file, and the step (d) comprises extracting information by searching for a character string of an item corresponding to data to be extracted and determining a location of the data to be extracted according to the form information.

2 . The method of claim 1 , wherein the step (d) comprises extracting unified soil classification system (USCS) information by estimating a USCS from a major constituent material and a minor constituent material in a strata detailed description when the USCS information is not included in the stratum information in the boring log.

3 . The method of claim 2 , wherein the step (d) comprises extracting the USCS information by estimating the USCS by substituting the major constituent material and the minor constituent material in the strata detailed description into the following table when the USCS information is not included in the stratum information in the boring log:

Major
Major
Minor
Minor

constituent
classification
constituent
classification
USCS

Gravel
G
Sand
W (or P)
GW (or GP)

Silt
M
GM

Clay
C
GC

Sand
S
Gravel
W (or P)
SW (or SP)

Silt
M
SM

Clay
C
SC

Silt
M
Gravel
L
ML

Sand
L
ML

Clay
H
MH

Clay
C
Gravel
L
CL

Sand
L
CL

Silt
H
CH.

4 . The method of claim 1 , wherein the step (a) comprises converting a plurality of boring log portable document format (PDF) files to JPG image files, performing the training in a k-fold validation method by inputting the converted JPG image files into a convolution neural network (CNN) type classification model, and verifying performance by using a confusion matrix.

5 . A computer program that is stored in a non-transitory storage medium and executed in a computer including a processor, the computer program performing the method for automatically storing boring logs information using artificial intelligence according to claim 1 .

6 . A system for automatically storing boring logs information using artificial intelligence, the system comprising:
a processor; and a memory configured to store predetermined instructions, wherein the processor executing the instructions stored in the memory executes: a step (a) of training a deep learning-based classification model by converting a plurality of boring log files into images; a step (b) of generating form information by applying, to the classification model, a boring log input for information extraction to identify a form of the boring log; a step (c) of selecting information to be extracted from among standard penetration test (SPT) information and stratum information and generating a data frame including basic information and selected information; a step (d) of extracting basic information and extracting at least one of the standard penetration test (SPT) information and the stratum information, from the boring log according to the form information; and a step (e) of generating data to be stored in a database using the extracted information, wherein in the step (d), the processor extracts the basic information and the information selected in the step (c), in the step (e), the processor generates data to be stored in the database by inserting the information extracted in the step (d) into the data frame generated in the step (c), in the step (c), the processor converts the boring log into a text file, and in the step (d), the processor extracts information by searching for a character string of an item corresponding to data to be extracted and determining a location of the data to be extracted according to the form information.

7 . The system of claim 6 , wherein in the step (d), the processor extracts unified soil classification system (USCS) information by estimating a USCS from a major constituent material and a minor constituent material in a strata detailed description when the USCS information is not included in the stratum information in the boring log.

8 . The system of claim 7 , wherein in the step (d), the processor extracts the USCS information by estimating the USCS by substituting the major constituent material and the minor constituent material in the strata detailed description into the following table when the USCS information is not inc

Assignees

• Korea Inst Civil Eng & Building Tech

Inventors

• Park Ka-Hyun
• Han Jin-Tae
• Yoon Youngno

Classifications

• G06V10/764

  using classification, e.g. of video objects · CPC title

• G06V10/454

  Integrating the filters into a hierarchical structure, e.g. convolutional neural networks [CNN] · CPC title

• G06N3/08

  Learning methods · CPC title

• G06N3/04

  Architecture, e.g. interconnection topology · CPC title

• G01N33/24

  Earth materials (G01N33/42 takes precedence) · CPC title