Search

CN-121978222-A - Modal fusion ultrasonic defect signal generation method and system

CN121978222ACN 121978222 ACN121978222 ACN 121978222ACN-121978222-A

Abstract

The invention provides a method and a system for generating ultrasonic defect signals by mode fusion, wherein the method comprises the steps of scanning a detected workpiece by using ultrasonic detection equipment, collecting original A signals, generating B scanning images and C scanning images, determining the defect A signals, selecting a plurality of defect A signals from the determined defect A signals, randomly selecting a plurality of non-defect A signals, converting the defect A signals and the non-defect A signals into n multiplied n images, inputting the defect A signals into signal branches of a twinning type countermeasure network method frame, inputting the generated images into image branches of the twinning type countermeasure network method frame, generating a first group of A signals through the signal branches, generating a second group of A signals through the image branches, and inputting the first group of A signals and the second group of A signals into a discriminator for discrimination after denoising, filtering and merging, so as to keep the defect A signals. The invention can relieve the unbalanced problem of ultrasonic data in training.

Inventors

  • ZHOU LUSHENG
  • LI YILAN
  • ZHANG YI
  • CUI TAO
  • FU ZHONGMING
  • WU MINGMING
  • LIANG GUOJIAN
  • ZHENG YITING
  • LI XINGFEN

Assignees

  • 国核电站运行服务技术有限公司
  • 江西核电有限公司

Dates

Publication Date
20260505
Application Date
20241030

Claims (10)

  1. 1. The method for generating the ultrasonic defect signal by the mode fusion is characterized by comprising the following steps of: S 1 , scanning a detected workpiece by using ultrasonic detection equipment, collecting an original A signal, and generating a B scanning image and a C scanning image; S 2 , combining the B scanning image and the C scanning image, determining a defect A signal, selecting a plurality of defect A signals from the determined defect A signals, and randomly selecting a plurality of non-defect A signals; S 3 , converting the defect A signal and the non-defect A signal into n multiplied by n images; S 4 , inputting the defect A signal into a signal branch of the twin generation type countermeasure network method frame, and inputting the generated image into an image branch of the twin generation type countermeasure network method frame; s 5 , generating a first group A signal via a signal branch; S 6 , after denoising, filtering and combining the first group A signal and the second group A signal, inputting the signals into a discriminator for discrimination, and reserving the defect A signal.
  2. 2. The method for generating an ultrasonic defect signal by mode fusion according to claim 1, wherein in step S 2 , a plurality of defect A signals with signal-to-noise ratio greater than 6dB and tip diffraction characteristics and a plurality of defect A signals with signal-to-noise ratio less than 4dB are selected from the determined defect A signals, and a plurality of non-defect A signals are randomly selected.
  3. 3. The method for generating a modality fused ultrasonic defect signal as claimed in claim 1, wherein the step S 3 includes the steps of: S 31 , determining N value, calculating maximum N value N, and setting length of A signal as L A Wherein, the Representing a downward rounding; S 32 , filling the A signal into the n multiplied by n image.
  4. 4. A method of generating a modality fused ultrasound defect signal as claimed in claim 3, wherein step S 32 includes the steps of: S 321 , calculating the span of the snapshot area: s 322 , doubling the snapshot area span by m= 2*M; S 323 , calculating the number N A ＝L A /M of the areas; s 324 , extracting the maximum amplitude and the minimum amplitude in each region to serve as filling data without losing the characteristics of the ultrasonic A signal; S 325 , image filling is carried out.
  5. 5. The method for generating an ultrasound defect signal by modality fusion according to claim 4, wherein in the step S 325 , when image filling is performed, the first filling is not performed, the end filling is performed, and the remaining position is filled with an intermediate value.
  6. 6. The method for generating a mode-fused ultrasonic defect signal according to claim 1, wherein in the step S 5 , noise is added to the signal branch, a first group a signal is generated by using a sequence generation network, noise is added to the image branch, a batch of images is generated by passing through the image generation network, and then the images are reversely converted into signals, so that a second group a signal is generated.
  7. 7. A modality fused ultrasonic defect signal generation system employing the modality fused ultrasonic defect signal generation method of any one of claims 1 to 6, the ultrasonic defect signal generation system comprising: the signal detection device is used for collecting the original A signal and generating a B scanning image and a C scanning image; the signal screening device is used for combining the B scanning image and the C scanning image, determining a defect A signal, selecting a plurality of defect A signals from the determined defect A signals, and randomly selecting a plurality of non-defect A signals; Signal conversion means for converting the defective a signal and the non-defective a signal into an n×n image; Signal processing means for inputting the defect a signal into a signal branch of the twinning-generation type countermeasure network method frame, inputting the generated image into an image branch of the twinning-generation type countermeasure network method frame; And the signal discrimination device is used for carrying out denoising and filtering on the first group A signal and the second group A signal, inputting the signals into the discriminator for discrimination, and reserving the defect A signal.
  8. 8. The system for generating a mode-fused ultrasonic defect signal according to claim 7, wherein said method for converting signals by said signal converting means comprises the steps of determining N value, calculating the maximum N value N, and setting the length of the A signal to be L A Wherein, the Representing rounding down, and filling the a signal into the nxn image.
  9. 9. An electronic device comprising a processor and a memory storing a program or instructions executable on the processor, the program or instructions being executable by the processor to perform an ultrasound defect signal generating method that implements the modality fusion of any of claims 1-6.
  10. 10. A readable storage medium, characterized in that a program or instructions is stored on the readable storage medium, which when executed by a processor implements the method of generating an ultrasound defect signal for modality fusion according to any of claims 1-6.

Description

Modal fusion ultrasonic defect signal generation method and system Technical Field The invention relates to the technical field of nondestructive testing, in particular to a method and a system for generating an ultrasonic defect signal by mode fusion. Background In modern industrial production, workpieces such as welded parts are prone to defects in the production or long-time use process, accidents are prone to occur, and therefore defect detection plays an important role in ensuring safe production. In practice, it is often necessary to perform a nondestructive inspection of the workpiece to ensure proper work-piece use. Ultrasonic detection is an important technical means for nondestructive detection, mainly relies on an ultrasonic probe to collect ultrasonic signals, then generates images according to an imaging algorithm, and then determines defects by using the technical means or manually. Artificial intelligence has made important progress in various fields such as face recognition, chatGPT, image semantic segmentation. However, the progress of defect recognition is slow, and on one hand, the material, defect type, ultrasonic probe frequency and noise have important influences on imaging, and on the other hand, defects are very rare for the whole workpiece, namely, defect recognition is a typical extremely unbalanced problem. Therefore, a large amount of defect data, especially raw defect data, i.e., defect data at the ultrasonic signal (a signal) level, is required. At present, the generation of ultrasonic defect signals is mainly generated according to inversion, translation and other technologies, so that the defect A signal (A-scan signal) is relieved to a certain extent, and the method can only aim at part of materials and part of defect types. In view of this, the present inventors devised a method and a system for generating an ultrasonic defect signal by mode fusion, so as to overcome the above technical problems. Disclosure of Invention The invention aims to overcome the defect that an automatic identification means for ultrasonic defects of a workpiece is lacked in the prior art, and provides a method and a system for generating ultrasonic defect signals by mode fusion. The invention solves the technical problems by the following technical proposal: The invention provides a mode-fused ultrasonic defect signal generation method which is characterized by comprising the following steps of S 1, scanning a detected workpiece by using ultrasonic detection equipment, collecting an original A signal, generating a B scanning image and a C scanning image, S 2, combining the B scanning image and the C scanning image, determining a defect A signal, selecting a plurality of defect A signals from the determined defect A signals, randomly selecting a plurality of non-defect A signals, S 3, converting the defect A signal and the non-defect A signal into n multiplied n images, S 4, inputting the defect A signal into a signal branch of a twinning generation type countermeasure network method frame, inputting the generated image into an image branch of the twinning generation type countermeasure network method frame, S 5, generating a first group of A signals through the signal branch, generating a second group of A signals through the image branch, S 6, carrying out noise removal, filtering and merging on the first group of A signals and the second group of A signals, and then carrying out identification by an input device, and reserving the defect A signals. According to one or more embodiments of the present invention, in step S 2, a plurality of defect A signals having a signal-to-noise ratio greater than 6dB and having a tip diffraction characteristic, and a plurality of defect A signals having a signal-to-noise ratio less than 4dB are selected from the determined defect A signals, and a plurality of non-defect A signals are randomly selected. According to one or more embodiments of the present invention, the step S 3 includes the steps of S 31, determining the value of N, calculating the maximum value of N, setting the length of the A signal to L A, thenWherein, the Representing a downward rounding, S 32, filling the a signal into the nxn image. According to one or more embodiments of the present invention, the step S 32 includes the steps of S 321, calculating the snapshot region span: s 322, doubling the snapshot area span by m= 2*M; s 323, calculating the number N A＝LA/M;S324 of the areas, extracting the maximum amplitude and the minimum amplitude in each area as filling data without losing the characteristics of the ultrasonic A signal, and S 325, performing image filling. In accordance with one or more embodiments of the present invention, in the step S 325, when image filling is performed, the first is not filled, the end is filled, and the remaining positions are filled to an intermediate value. According to one or more embodiments of the present invention, noise is added to the signal branch in