CN-121453926-B - Ultrasonic and electromagnetic composite detection method for internal defects of generator rotor forging

Abstract

The invention relates to the technical field of defect detection, in particular to an ultrasonic and electromagnetic composite detection method for internal defects of a generator rotor forging, which comprises the steps of firstly carrying out self-adaptive path electromagnetic scanning on the surface of the forging, utilizing a multi-frequency eddy current technology and intelligent signal analysis to identify and parametrize a near-surface suspicious region, then carrying out global scanning and local directional fine scanning through a dynamic trigger mechanism based on suspicious region information to generate a defect point cloud data set containing multidimensional attributes, and finally carrying out space-time alignment and feature level correlation on electromagnetic and ultrasonic data through a three-level fusion framework and carrying out accurate judgment and visual output on defect types.

Inventors

• YU SHUN
• Yu Linze

Assignees

• 重庆新颜达机电设备有限公司

Dates

Publication Date

20260512

Application Date

20260107

Claims (7)

1. 1. The ultrasonic and electromagnetic composite detection method for the internal defects of the rotor forgings of the generator is characterized by comprising the following steps of performing electromagnetic scanning by covering the surfaces of the forgings with self-adaptive paths based on a three-dimensional model of the rotor forgings by using an eddy current effect detection mode, identifying abnormal signals by using a dynamic threshold value and a multi-parameter pattern recognition algorithm, and marking region coordinates corresponding to the abnormal signals; based on the information of the marked area, acquiring full waveform radio frequency data according to a master-slave detection mode of global routine scanning and local directional scanning, and generating a defect point cloud data set containing space coordinates and multidimensional attributes after preprocessing and feature extraction; feature level fusion is carried out on the marking area and the defect point cloud data set, the generated defect comprehensive feature vector fused with each defect ID is analyzed, the defect type is judged, and a three-dimensional fusion detection report is output; Electromagnetic scanning is carried out by covering the surface of the forging piece with a self-adaptive path based on a three-dimensional model of the rotor forging piece by utilizing an eddy current effect detection mode, abnormal signals are identified through a dynamic threshold value and a multi-parameter pattern recognition algorithm, and region coordinates corresponding to the abnormal signals are marked; exciting an alternating magnetic field by adopting a multi-frequency eddy current probe, and synchronously collecting complex impedance signals; Processing the complex impedance signals in real time, identifying abnormal signals through a dynamic threshold value and a multi-parameter mode identification algorithm, and aggregating and parametrizing abnormal signal points in a set space range by utilizing a spatial clustering algorithm to obtain a marked area; The self-adaptive path is characterized in that an outer cylindrical surface of a forging is unfolded to be a two-dimensional plane, and a constant-pitch spiral line is planned to be used as a basic scanning path; the feature level fusion of the marking area and the defect point cloud data set comprises the steps of calculating the plane projection distance between the center of the marking area and the defect point cloud data set, the included angle between the electromagnetic inferred orientation and the ultrasonic principal component analysis orientation and the proportional relation between the near-surface size and the ultrasonic cross-section size, and judging the marking area and the defect point cloud data set as the same defect entity and assigning a unique fusion defect ID when the marking area and the defect point cloud data set meet the preset threshold.
2. 2. The ultrasonic and electromagnetic composite detection method for internal defects of a generator rotor forging according to claim 1, wherein after the marked areas are obtained, the method further comprises the step of simultaneously recording the center coordinates, boundaries, defect confidence and preliminary type trends of each marked area.
3. 3. The ultrasonic and electromagnetic composite detection method for internal defects of a generator rotor forging according to claim 1 is characterized in that the multi-parameter pattern recognition algorithm specifically comprises the steps of analyzing phase angle differences of complex impedance signals under a plurality of excitation frequencies, shape complexity of impedance plane tracks and signal harmonic components, calculating defect confidence coefficient by matching extracted characteristic parameters with pre-stored defect characteristic spectrums, and carrying out preliminary tendency judgment on crack and inclusion types.
4. 4. The ultrasonic and electromagnetic composite detection method for internal defects of a generator rotor forging according to claim 1 is characterized in that based on the information of a marked area, full waveform radio frequency data is collected according to a master-slave detection mode of global regular scanning and local directional scanning, and after preprocessing and feature extraction, a defect point cloud data set containing space coordinates and multidimensional attributes is generated, and the method comprises the steps of controlling an ultrasonic detection unit to execute global regular scanning and local directional scanning based on the information of the marked area; Triggering a local directional scanning mode when the ultrasonic probe enters a dynamic triggering boundary of a marking area, and executing high-density raster scanning or sector scanning; And extracting geometric, texture and frequency domain characteristics to generate a defect point cloud data set containing space coordinates and multidimensional attributes.
5. 5. The ultrasonic and electromagnetic composite detection method for internal defects of a generator rotor forging according to claim 4, wherein the dynamic triggering boundary is generated by expanding a buffer zone outwards by an original boundary of a marking area, and dynamically calculating the width of the buffer zone based on the estimated size of the marking area; The ultrasonic probe enters the buffer zone and marks a pre-positioning state, when the ultrasonic probe further enters a kernel zone of an original boundary, the ultrasonic probe formally triggers local directional scanning, and when the ultrasonic probe exits from the buffer zone, the ultrasonic probe finishes the local directional scanning.
6. 6. The ultrasonic and electromagnetic composite detection method for internal defects of a generator rotor forging according to claim 4, wherein generating a defect point cloud data set comprises fusing sparse defect indication points of main scanning with three-dimensional reconstruction point clouds of local directional scanning, and registering through an iterative nearest point algorithm; And complementing the detection blind area by using a poisson surface reconstruction algorithm, and finally generating structured point cloud data containing space coordinates, local normal vectors, scattering intensity and frequency response attributes.
7. 7. The ultrasonic and electromagnetic composite detection method for the internal defects of the generator rotor forgings according to claim 1 is characterized by comprising the steps of analyzing a defect comprehensive feature vector of each generated fusion defect ID, performing primary screening according to electromagnetic anisotropy, ultrasonic image texture and three-dimensional spreading ratio features in the defect comprehensive feature vector, comparing defects which do not pass through the primary screening or have low confidence level with a preset historical defect case library, searching for a matching case through graph neural network matching, and comprehensively obtaining a final classification result and a confidence level score.

Description

Ultrasonic and electromagnetic composite detection method for internal defects of generator rotor forging Technical Field The invention relates to the technical field of defect detection, in particular to an ultrasonic and electromagnetic composite detection method for internal defects of a generator rotor forging. Background The generator rotor forging is used as a core component of large-scale power generation equipment, and the internal quality of the generator rotor forging directly influences the safety and reliability of the generator. Internal defects such as inclusions, cracks, porosity, etc. may be generated during forging manufacture due to improper metallurgical, forging, or heat treatment processes. Currently, the industry mainly adopts a single nondestructive detection method to detect defects, such as ultrasonic detection is sensitive to volume type defects, but has limited detection capability on microcracks or near-surface defects, and electromagnetic detection (such as eddy current detection) is sensitive to surface and near-surface defects, but is difficult to detect deep internal defects. The existing single detection method cannot meet the requirements of high sensitivity, full-depth coverage and accurate classification of internal defects of a generator rotor forging, and particularly is difficult to effectively distinguish defect types (such as cracks and nonmetallic inclusions) and spatial distribution thereof, so that detection results are incomplete, and quantitative evaluation and maintenance decision of the defects are affected. Disclosure of Invention The invention aims to provide an ultrasonic and electromagnetic composite detection method for internal defects of a generator rotor forging, which realizes composite detection of the internal defects of the generator rotor forging and can improve the detection rate, positioning accuracy and type recognition capability of the defects. In order to achieve the purpose, the invention provides an ultrasonic and electromagnetic composite detection method for internal defects of a generator rotor forging, which comprises the following steps: Based on a three-dimensional model of the rotor forging, using an eddy current effect detection mode to cover the surface of the forging by a self-adaptive path for electromagnetic scanning, identifying an abnormal signal by a dynamic threshold value and a multi-parameter pattern identification algorithm, and marking a region coordinate corresponding to the abnormal signal; based on the information of the marked area, acquiring full waveform radio frequency data according to a master-slave detection mode of global routine scanning and local directional scanning, and generating a defect point cloud data set containing space coordinates and multidimensional attributes after preprocessing and feature extraction; and carrying out feature level fusion on the marking area and the defect point cloud data set, analyzing the generated defect comprehensive feature vector fused with each defect ID, judging the defect type and outputting a three-dimensional fusion detection report. Based on a three-dimensional model of a rotor forging, electromagnetic scanning is performed by covering the surface of the forging with a self-adaptive path in an eddy current effect detection mode, abnormal signals are identified through a dynamic threshold value and a multi-parameter pattern recognition algorithm, and region coordinates corresponding to the abnormal signals are marked, and the method comprises the following steps: based on the three-dimensional model of the rotor forging, controlling the electromagnetic detection unit to cover the surface of the forging by a self-adaptive path; exciting an alternating magnetic field by adopting a multi-frequency eddy current probe, and synchronously collecting complex impedance signals; And processing the complex impedance signals in real time, identifying abnormal signals through a dynamic threshold value and a multi-parameter mode identification algorithm, and aggregating and parametrizing abnormal signal points in a set space range by utilizing a spatial clustering algorithm to obtain a marked area. Wherein after obtaining the marked area, the method further comprises: And simultaneously recording the center coordinates, boundaries, defect confidence and preliminary type tendency of each marked area. The self-adaptive path planning specifically comprises the following steps: Expanding the outer cylindrical surface of the forging into a two-dimensional plane, and planning a constant-pitch spiral line as a basic scanning path; and according to the geometric characteristics identified by the three-dimensional CAD model of the forging, automatically switching into encryption raster scanning in the journal and slot position area, and ensuring that the probe is vertical in posture and fully covered in scanning. The multi-parameter pattern recognition algorithm specifically comprises th