Cableway electric control system fault intelligent diagnosis method based on electrical drawing self-learning
By automatically generating device trees and Bayesian network fault reasoning models, the problem of time-consuming and error-prone fault diagnosis of cableway circuits has been solved, and efficient and accurate fault location has been achieved.
Patent Information
- Application Number
- CN202510980062.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-07-16
- Publication Date
- 2025-10-28
AI Technical Summary
Cableway circuit fault diagnosis is time-consuming and error-prone, lacks system-level fault logic analysis capabilities, and maintenance relies on manual experience, resulting in low efficiency.
By uploading electrical drawings of the power control system to the drawing intelligent parsing engine, a device tree is automatically generated, preprocessed and hierarchically parsed, and a Bayesian network fault reasoning model is used for topology analysis to generate detection paths and perform dynamic measurement verification, ultimately locating the faulty components.
It improved fault location efficiency by 300%, reduced the false alarm rate by 80%, and increased measurement speed by 8 times, achieving fast and accurate fault diagnosis.
Smart Images
Figure CN120850896A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of cableway fault diagnosis technology, and specifically relates to an intelligent fault diagnosis method for cableway electrical control systems based on self-learning of electrical drawings. Background Technology
[0002] Cableways are used as transportation in industrial and mining areas, cities, or scenic tourist areas to transport people or goods. Cableways frequently experience malfunctions during operation. Mechanical faults are relatively easy to diagnose, as the location can be easily identified and repairs can be dispatched immediately. However, electrical faults are much more difficult to pinpoint quickly. Currently, cableway electrical fault diagnosis relies heavily on the experience of maintenance personnel to deduce the fault path sequentially, wasting significant time and manpower. Because the control circuit is a large and complex integrated system with interconnected systems, it involves numerous electrical diagrams. Searching through each diagram to find the relevant fault diagram is not only time-consuming but also prone to errors. Maintenance personnel cannot rely on previous experience from other personnel in handling the same fault, lacking the ability to perform system-level fault logic analysis. Summary of the Invention
[0003] To overcome the shortcomings of existing technologies, this invention provides an intelligent fault diagnosis method for cableway electrical control systems based on self-learning of electrical drawings.
[0004] This invention is achieved through the following technical solution: A method for intelligent fault diagnosis of cableway electrical control system based on self-learning from electrical drawings, characterized by the following steps: (1) Upload all electrical drawings of the electrical control system to the drawing intelligent parsing engine to automatically generate the device tree; (2) Preprocess the uploaded electrical drawings: format conversion, layered parsing and dynamic symbol library establishment; (3) Fault alarm signal input; (4) Use topology analysis to determine the area affecting the fault and retrieve the corresponding electrical drawings; (5) Generate the detection path according to the drawing; (6) Perform dynamic measurement verification along the detection path; (7) Finally, the component corresponding to the fault was located.
[0005] Preferably, in step (1), the drawing format is PDF.
[0006] Preferably, in step (1), the electrical drawings are uploaded via USB interface or wireless network.
[0007] Preferably, in step (2), the format is converted to extract the vector data of the drawing and the scanned document is subjected to OCR text recognition.
[0008] Preferably, in step (2), the hierarchical parsing includes: Vector layer: Extracts line and geometric data; Text layer: Locates all text tags and comments; Image layer: Noise reduction is performed using a nonlocal means denoising algorithm.
[0009] Preferably, in step (2), the IEC60617 standard symbol set is stored and the new symbol generation training dataset is uploaded.
[0010] Preferably, in step (2), OpenCV's component symbol recognition algorithm is used when extracting the drawing vector data.
[0011] Preferably, in step (4), the analysis is first performed using a Bayesian network fault reasoning model.
[0012] Preferably, in step (6), a dynamic threshold correction algorithm or wave feature extraction is used for analysis.
[0013] Preferably, historical fault cases are generated by steps (3), (4), (5), (6) and (7) and stored for direct reference during learning.
[0014] The beneficial effects of this invention are: by uploading all circuit diagrams of the entire electronic control system, when a fault is encountered, the circuit diagram corresponding to the fault can be retrieved immediately, and the detection path and location can be quickly determined by the corresponding diagram. By analyzing the measurement results, the fault detection range can be narrowed, the fault location efficiency is improved by more than 300%, the false judgment rate is more than 80% lower than that of traditional methods, and the measurement speed is more than 8 times faster than that of traditional methods. Attached Figure Description
[0015] The invention will now be further described with reference to the accompanying drawings.
[0016] Appendix Figure 1 The electrical drawings and circuit diagrams are for the grounding rod faults in the cableway of this invention. Appendix Figure 2 For the appendix Figure 1 The image of the area to be detected obtained from the analysis; Appendix Figure 3 Figure 2 shows a further component inspection diagram obtained after the inspection. Detailed Implementation
[0017] The attached figure shows a specific embodiment of the present invention. This embodiment includes the following steps: (1) Upload all PDF electrical drawings of the electrical control system to the drawing intelligent parsing engine via USB interface or wireless network to automatically generate the device tree; (2) Preprocess the uploaded electrical drawings: convert the format to extract the drawing vector data using the component symbol recognition algorithm of OpenCV, perform OCR text recognition on the scanned parts, and establish a hierarchical parsing and dynamic symbol library: store the IEC60617 standard symbol set and upload the new symbol generation training dataset; wherein, the hierarchical parsing includes: vector layer: extract line and geometric data; text layer: locate all text labels and annotations; image layer: perform noise reduction processing using the non-local mean denoising algorithm; (3) Input the fault alarm signal; (4) First, analyze it through the Bayesian network fault reasoning model, then determine the area affecting the fault through topology analysis, and retrieve the corresponding electrical drawing; (5) Generate the detection path by referring to the drawing; (6) Perform dynamic measurement verification along the detection path, and perform analysis using the dynamic threshold correction algorithm or wave feature extraction; (7) Finally locate the component corresponding to the fault. Each time a fault occurs, historical fault cases are generated by steps (3), (4), (5), (6) and (7) and stored for direct reference during learning.
[0018] Using the method of this invention, when vector lines and scanned images coexist in the same drawing, component symbols may be split into multiple path objects. This can be solved by a hybrid parsing engine, prioritizing the parsing of editable vector elements and using the Pix2PixHD model for GAN image repair of missing parts. When component annotation text is offset or annotations span pages, a spatial association algorithm is used to establish an R-tree spatial index to accelerate proximity search and implement bidirectional matching. When there are dashed line connections and page-spanning connectors, Hough transform is used to detect dashed lines, and a connector completion algorithm is used for intelligent connection tracing. The drawing is modeled hierarchically, establishing a four-layer topology model: a physical connection layer at the conductor level, a functional module layer divided by subsystems, a signal flow layer related to control logic, and a power tree layer related to the power supply network. The bus structure is simplified and represented.
[0019] Specific examples are as follows: Figure 1 The diagram shown is the one retrieved for a grounding rod fault alarm. Analysis indicates that the fault needs to be detected. Figure 2 Check whether the voltage at positions marked 1, 2, 3, 4, 5, and 6 is normal. The reference value is DC 24V. This was confirmed through testing. Figure 3 The positions marked 1, 2, and 3 indicate the circuit where the voltage measurement is normal; this allows for determination. Figure 2 The positions marked 4, 5, and 6 indicate the fault locations. This indicates that the fault is that the 508S2 limit switch is not connected or is damaged. This allows for the inspection of the 508S2, greatly narrowing down the scope of the fault.
Claims
1. A method for intelligent fault diagnosis of cableway electrical control system based on self-learning of electrical drawings, characterized by: Includes the following steps: (1) Upload all electrical drawings of the electrical control system to the drawing intelligent parsing engine to automatically generate the device tree; (2) Preprocess the uploaded electrical drawings: format conversion, layered parsing and dynamic symbol library establishment; (3) Fault alarm signal input; (4) Use topology analysis to determine the area affecting the fault and retrieve the corresponding electrical drawings; (5) Generate the detection path according to the drawings; (6) Perform dynamic measurement verification along the detection path; (7) Finally, the component corresponding to the fault was located.
2. The intelligent fault diagnosis method for cableway electrical control system based on self-learning of electrical drawings as described in claim 1, characterized in that: In step (1), the drawing format is PDF.
3. The intelligent fault diagnosis method for cableway electrical control system based on self-learning of electrical drawings as described in claim 1, characterized in that: In step (1), the electrical drawings are uploaded via USB interface or wireless network.
4. The intelligent fault diagnosis method for cableway electrical control system based on self-learning of electrical drawings as described in claim 1, characterized in that: In step (2), the format is converted to extract the vector data of the drawing, and OCR text recognition is performed on the scanned document.
5. The intelligent fault diagnosis method for cableway electrical control system based on self-learning of electrical drawings as described in claim 1, characterized in that: In step (2), the hierarchical parsing includes: Vector layer: Extracts line and geometric data; Text layer: Locates all text tags and comments; Image layer: Noise reduction is performed using a nonlocal means denoising algorithm.
6. The intelligent fault diagnosis method for cableway electrical control system based on self-learning of electrical drawings as described in claim 1, characterized in that: In step (2), the IEC60617 standard symbol set is stored and the new symbol generation training dataset is uploaded.
7. The intelligent fault diagnosis method for cableway electrical control system based on self-learning of electrical drawings as described in claim 4, characterized in that: In step (2), OpenCV's component symbol recognition algorithm is used when extracting the drawing vector data.
8. The intelligent fault diagnosis method for cableway electrical control system based on self-learning of electrical drawings as described in claim 1, characterized in that: In step (4), the analysis is first performed using a Bayesian network fault reasoning model.
9. The intelligent fault diagnosis method for cableway electrical control system based on self-learning of electrical drawings as described in claim 1, characterized in that: In step (6), a dynamic threshold correction algorithm or wave feature extraction is used for analysis.
10. The intelligent fault diagnosis method for cableway electrical control system based on self-learning of electrical drawings according to claim 1, characterized in that: Historical fault cases are generated by steps (3), (4), (5), (6) and (7) and stored for direct reference during learning.