Training data generator and method for generation of training datasets

Patent metadata

Field	Value
Publication number	US-12462595-B2
Application number	US-202118035802-A
Country	US
Kind code	B2
Filing date	Oct 19, 2021
Priority date	Nov 16, 2020
Publication date	Nov 4, 2025
Grant date	Nov 4, 2025

Abstract

A training data generator includes an interface to read symbols extracted from digital system plans for technical systems, wherein each of the system plans represents the design and/or the functionality of a technical system by of symbols, and the system plans are identical. The training data generator also includes a storage module which is designed to store the extracted symbols, a selection module which is designed to randomly select a symbol sub-quantity of the stored symbols using a random generator, a generator which is designed to generate at least one synthetic system plan on the basis of the selected symbol sub-quantity, and an output module to output the at least one synthetic system plan as training data in order to train a trainable image detection module, wherein the trainable image detection module is designed to generate a digital system plan using an analog system plan of a technical system.

First claim

The invention claimed is:

1 . A training data generator for generating training data to be used for training a trainable image detection module to generate a digital system plan of a digital system with an aid of an analog system plan of the technical system, said training data generator comprising:
an interface, designed to read in (i) symbols extracted from digital system plans for technical systems and (ii) a positioning rule provided for the positioning of a symbol in a digital system plan, wherein the digital system plans each depict a structure and/or a functionality of the technical system by means of symbols, wherein the digital system plans are of an identical type, wherein the extracted symbols depict the technical system or a technical function, and wherein the at least one positioning rule for the positioning of a symbol in the digital system plan is directed to a relative positioning of the symbol in relation to an annotation, a predetermined coupling of the symbol to a further symbol, and/or a symbol-specific dependence of the symbol on at least one further symbol, a memory module, which is designed in such a way as to store the extracted symbols, a selection module, which is designed in such a way as to select at random by means of a random generator a symbol sub-quantity of the stored symbols, a generator, which is designed in such a way as to generate at least one synthetic system plan as a function of the selected symbol sub-quantity and as a function of the positioning rule, and an output module, which is designed in such a way as to output the at least one synthetic system plan as training data for training the trainable image detection module to generate the digital system plan of digital system with the aid of the analog system plan of the technical system.

2 . The training data generator as claimed in claim 1 , wherein the interface is designed to read in at least one item of additional information assigned to a respective symbol and to transfer the at least one item to the memory module for storing the at least one item, wherein the at least one item comprises:
an item of information assigned to a symbol about a possible connection to another symbol, and/or an item of information about an annotation assigned to a symbol.

3 . The training data generator as claimed in claim 2 , wherein the interface is designed to read in for an annotation a relative position in relation to the symbol and/or dimension of the annotation.

4 . The training data generator as claimed in claim 1 , wherein the generator is designed to generate the at least one synthetic system plan as a function of an item of additional information.

5 . The training data generator as claimed in claim 1 , wherein the generator is designed to generate the synthetic system plan while taking into account at least one predetermined boundary condition of the synthetic system plan.

6 . The training data generator as claimed in claim 1 , wherein the generator is designed to create and/or to modify the synthetic system plan and/or at least one symbol of the synthetic system plan by means of artifacts provided.

7 . The training data generator as claimed in claim 6 , wherein the output module is designed to output the created and/or modified synthetic system plan as an additional synthetic system plan.

8 . The training data generator as claimed in claim 1 , comprising a graphics module, which is designed in such a way as to create a graphical representation of the synthetic system plan and the output module is designed to output the graphical representation of the synthetic system plan.

9 . The training data generator as claimed in claim 1 , wherein the synthetic system plans comprise circuit diagrams, function plans, and/or piping and instrument flow schemes.

10 . A computer-implemented method for generating training data to be used for training a trainable image detection module to generate a digital system plan of a digital system with an aid of an analog system plan of the technical system, said method comprising:
reading in of (i) symbols extracted from digital system plans for technical systems and (ii) a positioning rule provided for the positioning of a symbol in a digital system plan, wherein the digital system plans each depict a structure and/or a functionality of the technical system by means of symbols, wherein the digital system plans are of an identical type, wherein the extracted symbols depict the technical system or a technical function, and wherein the at least one positioning rule for the positioning of a symbol in the digital system plan is directed to a relative positioning of the symbol in relation to an annotation, a predetermined coupling of the symbol to a further symbol, and/or a symbol-specific dependence of the symbol on at least one further symbol, storing the extracted symbols, randomly selecting, by means of a random generator, a symbol sub-quantity of the stored symbols, generating at least one synthetic system plan as a function of the selected symbol sub-quantity and as a function of the positioning rule, and outputting the at least one synthetic system plan as training data for training the trainable image detection module to generate the digital system plan of digital system with the aid of the analog system plan of the technical system.

11 . A computer program product, comprising a computer readable hardware storage device having computer readable program code stored therein, said program code executable by a processor of a computer system to implement a method that is able to be loaded directly into a programmable computer, comprising program code sections that are suitable for carrying out the method as claimed in claim 10 .

12 . The training data generator as claimed in claim 1 , wherein said to generate the digital system plan of digital system with the aid of the analog system plan of the technical system comprises to digitize the synthetic system plan with an aid of a scan of a paper plan.

Assignees

• Siemens Ag

Inventors

• Allmaras Moritz
• Schenk Tim
• Wehrstedt Jan Christoph
• Held Harald
• Limmer Steffen
• Herzog Ulli
• Hess Sascha
• Wendelberger Klaus Walter
• Heinzerling Till
• Knittel Julian

Classifications

• G06T11/26

  Drawing of charts or graphs · CPC title

• G06N3/09

  Supervised learning · CPC title

• G06V30/19147

  Obtaining sets of training patterns; Bootstrap methods, e.g. bagging or boosting · CPC title

• G06N3/0464

  Convolutional networks [CNN, ConvNet] · CPC title

• G06F18/214

  Generating training patterns; Bootstrap methods, e.g. bagging or boosting · CPC title