Search

EP-4742076-A1 - WIRING DIAGRAM DETECTION USING MACHINE LEARNING

EP4742076A1EP 4742076 A1EP4742076 A1EP 4742076A1EP-4742076-A1

Abstract

Offenbart is a method for generating a machine-readable representation of a circuit diagram document, as well as an associated data processing device and a computer program.

Inventors

  • Binder, Sven
  • Pfestorf, Sebastian

Assignees

  • Murrelektronik GmbH

Dates

Publication Date
20260513
Application Date
20241111

Claims (15)

  1. A method (100) for generating a machine-readable representation of a circuit diagram document (102a), wherein the circuit diagram document (102a) represents a decentralized electrical system, wherein the method comprises at least the following steps: - Receiving the circuit diagram document (102a), wherein the circuit diagram document (102a) includes graphical circuit diagram information (102b); and - Generating the machine-readable representation at least partially based on the graphical circuit diagram information (102b) using machine learning; - wherein the machine-readable representation comprises a graph that can be displayed on a graphical user interface and that shows at least one electrical device and at least one electrical connection shown in the circuit diagram document (102a).
  2. The method according to claim 1, wherein generating the machine-readable representation comprises: recognizing (104c) the at least one electrical device.
  3. The method according to claim 2, wherein the detection (104c) of the at least one electrical device comprises: detection (104a) of at least one graphic object in the circuit diagram document (102a) by means of an object recognition model.
  4. The method according to claim 2 or 3, wherein the recognition (104c) of the at least one electrical device comprises: recognition (104b) of at least one text string in the circuit diagram document by means of a text recognition model.
  5. The method according to claims 3 and 4, wherein the detection (104c) of the at least one electrical device further comprises: determining, at least partially based on the detected at least one text string, that the at least one graphic object is an electrical device.
  6. The method according to any one of the preceding claims 1-5, wherein generating the machine-readable representation comprises: recognizing (106) at least one electrical terminal by means of a terminal recognition model.
  7. The method according to any of the preceding claims 2-5 combined with claim 6, wherein generating the machine-readable representation further comprises: determining (108), at least partially based on the detected at least one terminal and the detected at least one electrical device, at least one connection of the electrical device.
  8. The method according to any one of the preceding claims 1-7, wherein generating the machine-readable representation further comprises: detecting (110) the at least one electrical connection by means of a connection detection model.
  9. The method according to any of the preceding claims 1-7 combined with claim 8, wherein generating the machine-readable representation further comprises: assigning (112) the at least one electrical connection to the at least one electrical device.
  10. The method according to any one of the preceding claims 3-9, wherein the detection (104c) of the at least one electrical device further comprises: Matching at least one graphic object with one or more reference objects, wherein a reference object corresponds to an electrical device whose shape, structure and/or function substantially corresponds to those of the graphic object within the circuit diagram document (102a).
  11. The method according to any one of the preceding claims 1-10, wherein the generated machine-readable representation incorrectly and/or not represents at least one electrical device within the circuit diagram document (102a), and wherein the method further comprises: - Receiving a corrective user input to properly display the incorrectly displayed electrical equipment; and/or - Receiving a reference object to represent the electrical equipment not shown.
  12. The method according to any of the preceding claims 3-11, wherein the object recognition model, the text recognition model, the terminal recognition model and/or the connection recognition model can be trained separately.
  13. A data structure comprising a machine-readable representation of a circuit diagram document (102a) generated by the method according to any of the preceding claims 1-12.
  14. A data processing device comprising means for carrying out the method according to any of the preceding claims 1-12.
  15. A computer program or a computer-readable medium on which a computer program is stored, wherein the computer program comprises instructions which, when the computer program is executed by a data processing device, cause the data processing device to execute the method according to any one of the preceding claims 1-12.

Description

TECHNICAL AREA The present invention relates generally to the field of planning, installation and setup of electrical systems in automation technology and in particular to a method for generating a machine-readable representation of a circuit diagram document using machine learning. BACKGROUND Planning, installing, and setting up electrical automation technology is a complex task. This applies to centralized control cabinet architectures as well as to decentralized architectures where control modules are mounted directly on the equipment. In modern industrial automation, electrical circuit diagrams are increasingly created using digital tools. While these tools enable efficient graphical representation of electrical components and their connections, they often offer only limited possibilities for semantic verification and automated processing of the diagrams. The circuit diagrams are essentially "drawn," with the electrician defining the components and connections according to their own discretion. Automated verification of logical consistency and compliance with standards and safety guidelines is generally not performed. This can lead to various problems. Faulty or inconsistent circuit diagrams, for example, can cause malfunctions, production downtime, or even safety risks. Manually checking the diagrams is time-consuming and prone to errors, especially in complex systems with numerous components and interconnections. It is therefore an objective of the present invention to improve the semantic analysis of electrical circuit diagrams in order to at least partially overcome the above-mentioned limitations. SUMMARY OF THE INVENTION This is achieved by the subject matter defined in the independent claims. Advantageous further developments of embodiments of the present disclosure are defined in the dependent claims, as well as in the description and the figures. According to one aspect of the invention, a method for generating a machine-readable representation of a circuit diagram document is provided. The circuit diagram document can represent an electrical system, in particular a distributed electrical system. The method can be computer-implemented. The electrical system can be designed as at least one of the following: an automation system, a production system, a logistics system, a production line, a machining center, an industrial robot, a manufacturing plant, a unit, an electrical device, or combinations thereof. A "decentralized system" can be designed as a modular system or as a mobile or movable system, where individual components are installed modularly at different locations. This installation is performed, at least partially, manually by a user such as a worker. Unlike a centralized system, where electrical equipment is connected via a point-to-point connection (i.e., the start point, connection, and endpoint are clearly defined), electrical equipment in a decentralized system is typically connected via a module-switch-module-hub-point connection. This means that in decentralized systems, multiple connection modules are typically arranged between the start point and the endpoint of an electrical device to establish the connection via the local distribution system. Due to this increased complexity, the planning, design, and maintenance of decentralized systems are significantly more demanding than those of centralized systems. The process may include receiving the circuit diagram document. The circuit diagram document may contain graphical circuit diagram information. The process can include generating the machine-readable representation. This can be done, at least partially, using machine learning based on the graphical circuit diagram information. The machine-readable representation can include a graph that can be displayed on a graphical user interface. The graph can include at least one electrical device, which is shown in the circuit diagram document, and at least one electrical connection which is shown in the circuit diagram document. The resulting machine-readable representation provides a semantic understanding of a circuit diagram, enabling a software-supported method for the efficient planning, design, and operation of an electrical system. This is particularly relevant for decentralized systems, as their complexity is significantly higher compared to centralized systems, making their planning, design, and operation considerably more demanding. However, the method according to the present disclosure can also be applied to centralized systems, since, in principle, methods for decentralized systems are also applicable to centralized systems, whereas methods for centralized systems are typically not directly applicable to methods for decentralized systems. The reason for this, as described above, is the greater complexity of decentralized systems. One application example is the redesign or conversion of a centralized system into a decentralized system. In this case, the method would first b