CN-122016709-A - Defect region automatic identification method applied to terahertz transmission imaging

CN122016709ACN 122016709 ACN122016709 ACN 122016709ACN-122016709-A

Abstract

The invention discloses an automatic defect area identification method applied to terahertz transmission imaging, which relates to the technical field of wind power generation blade defect detection and identification, the method ensures the detection coverage integrity through earlier-stage cooperative calibration and path planning, and acquires high-quality original data through background acquisition and terahertz transmission imaging acquisition, the data quality is optimized through preprocessing such as background removal and noise reduction, a suspected defect area is marked based on a pixel response value, accurate defect identification is completed by combining a trained model, and meanwhile detection precision and practicability are further improved through optimization means such as regional hierarchical scanning, temperature and humidity and gesture compensation and defect trend analysis. The method effectively solves the problems of weak anti-interference capability, insufficient precision, poor suitability and lack of trend prejudgement in the traditional detection method, realizes full-flow automation from data acquisition to result output, and provides technical support for efficient maintenance of the wind power generation blade.

Inventors

LIU JIANG
LIU HAIQING
LI GEN
LI ZHENGWEI
MU FENG
LIU ZHENG
WANG HONGBEI

Assignees

安徽中科太赫兹科技有限公司

Dates

Publication Date: 20260512
Application Date: 20260203

Claims (10)

1. The defect area automatic identification method applied to terahertz transmission imaging is characterized by comprising the following steps of: Step S1, pre-preparation, namely cooperatively calibrating the unmanned aerial vehicle, collecting information of the wind power generation blades to be detected, and planning a flight path of the unmanned aerial vehicle; S2, background collection, namely under the condition that a terahertz light source is not started, the unmanned aerial vehicle flies according to a planned path, and background image data of the fan blades are collected; S3, terahertz transmission imaging acquisition, namely turning on a terahertz light source, enabling an unmanned aerial vehicle to cooperatively fly according to a preset path, enabling the unmanned aerial vehicle carrying the terahertz light source to emit terahertz light to a fan blade, enabling the terahertz light to transmit through the fan blade, receiving a transmission light signal by a terahertz camera on another unmanned aerial vehicle, converting the transmission light signal into imaging data, performing differential scanning and data splicing in the imaging acquisition process, performing image generation, performing temperature influence control and attitude influence control in the image generation stage, obtaining a generated image after completion, and entering the next step; S4, preprocessing data, namely removing background, reducing noise and extracting an original response value; S5, marking a suspected defect area, namely setting a threshold range of a normal area response value according to the original response value of the extracted pixel points, and marking the area where the pixel points beyond the threshold range are located as the suspected defect area; S6, defect detection and identification, namely inputting marked suspected defect area data into a pre-trained defect identification model, accurately identifying suspected defect areas, detecting defect trend of the detection blade, and synchronously reflecting the detection result; And S7, outputting a detection result.
2. The method for automatically identifying a defective area applied to terahertz transmission imaging according to claim 1, wherein the process of differentially scanning and performing data stitching is as follows: carrying out regional classification on the detection blade, and collecting structural features, thickness features and importance features of the detection blade, carrying out quantitative parameter analysis on the structural features, the thickness features and the importance features, and comparing corresponding quantitative parameters with actual operation scenes, namely analyzing the bearing, auxiliary bearing and non-bearing modes of the structure; The method comprises the steps of dividing a blade area into a key area, a secondary area and a conventional area according to comparison, dynamically adjusting scanning parameters of a terahertz detection system according to the type of the blade area, wherein the scanning parameters comprise scanning steps, scanning speed and signal acquisition frequency, and automatically switching the parameters in the scanning process through real-time area identification so as to avoid manual intervention.
3. The method for automatically identifying the defect area applied to terahertz transmission imaging according to claim 2, which is characterized in that seamless splicing is performed according to the sectional data of each type of blade area: Extracting inherent structural features of the blades in overlapping areas of adjacent scanning sections to serve as anchor points, correcting errors of space coordinates of segmented data by taking terahertz signal amplitude/phase features of the anchor points as references, carrying out normalization compensation on terahertz amplitude values of different scanning sections based on signal average values of the overlapping areas, and splicing the calibrated segmented data according to coordinates to generate a complete defect map containing all the areas.
4. The method for automatically identifying a defective area applied to terahertz transmission imaging according to claim 1, wherein the temperature influence control process is as follows: When the unmanned aerial vehicle executes terahertz imaging, the temperature sensor is used for collecting real-time environmental temperature and recording temperature values at all moments, and terahertz time-domain signal main pulse amplitudes A Real time (T) and A Comparison (T) of the detection blades and the detection blades of the same type in the same moment and the same space relative position are extracted for collecting; A Real time (t) refers to the peak voltage/current amplitude corresponding to the pulse signal to be measured at time t: A Comparison (t) is expressed as that the reference main pulse amplitude refers to the preset peak voltage/current amplitude corresponding to the reference pulse signal with standard characteristics at the time t.
5. The method for automatically identifying a defective area applied to terahertz transmission imaging according to claim 4, wherein a relative variation amount is calculated for quantifying a degree of deviation of the amplitude of the main pulse to be measured from the amplitude of the reference main pulse, The relative variation is obtained, then the threshold comparison is carried out, if the relative variation is within the set threshold range, the pulse signal to be detected is judged to be effective, if the relative variation is not within the set range, the signal is judged to be interfered or the equipment is abnormal, and the correction mechanism is triggered.
6. The method for automatically identifying a defective area applied to terahertz transmission imaging according to claim 5, wherein an attitude influence control method is as follows; the method comprises the steps of collecting the vertical distance from a terahertz output probe of a real-time unmanned aerial vehicle to the surface of a blade, collecting the attitude angle of the output probe, combining a preset three-dimensional model of a curved surface of the blade, calculating the actual incident angle of terahertz waves, aligning the distance and the incident angle with a terahertz signal collecting timestamp, and ensuring that pose parameters are matched with the time sequence of signals; The method comprises the steps of carrying out log statistics according to a historical detection process, establishing a calibration mapping relation of distance-focusing lens displacement in advance, integrating a miniature electric displacement platform on an output probe, calling a mapping library according to a real-time distance, driving a focusing lens to adjust the displacement in real time, monitoring an amplitude mean value of a terahertz signal in real time, dynamically correcting a focusing parameter to form closed-loop adjustment if the amplitude mean value is lower than an amplitude mean value threshold value, and otherwise, continuously monitoring.
7. The method for automatically identifying a defective area applied to terahertz transmission imaging according to claim 6, wherein the terahertz signal is subjected to two-dimensional compensation for distance attenuation and angular phase distortion: according to the propagation attenuation model of the terahertz wave, calculating an amplitude compensation coefficient corresponding to the distance; The method comprises the steps of detecting main pulse amplitude in real time through an amplitude compensation coefficient, carrying out compensation adjustment on the main pulse amplitude in real time, establishing a fitting model of incident angle-phase distortion compensation quantity, carrying out real-time correction on the phase of a terahertz signal through a digital phase modulator, eliminating phase distortion caused by the incident angle, calculating stability indexes of the compensated signal, and repeating a compensation process if the stability indexes do not meet a set normal range.
8. The automatic defect area identification method applied to terahertz transmission imaging according to claim 1 is characterized by obtaining pixel points exceeding a threshold range in a suspected defect area, marking the pixel points not exceeding the threshold range as abnormal points, and marking the pixel points not exceeding the threshold range as normal points; determining quantization parameters of corresponding structural features of the suspected defective region and the non-defective region, and determining structural feature parameter deviation of the corresponding region according to parameter comparison; If the deviation of the structural characteristic parameter does not exceed the deviation threshold of the structural characteristic parameter, setting the corresponding suspected defect area as a structural variable trend; and determining a bearing mode of the corresponding area position, and carrying out maintenance priority division according to the type of the bearing mode, namely, the priority order is from high to low, namely, a key area, a secondary area and a conventional area.
9. The method for automatically identifying a defective region applied to terahertz transmission imaging of claim 8, wherein a quantization parameter of thickness characteristics corresponding to a suspected defective region and a non-defective region is determined, and a quantization parameter deviation of the thickness characteristics is obtained; When the quantized parameter deviation of the thickness characteristic exceeds the thickness characteristic parameter deviation threshold, namely adjacent detection interval time, if the detection interval time exceeds the interval time threshold, deducing that a suspected defect area has slow high-span defects, setting a slow high-span defect type as a detection result; When the quantitative parameter deviation of the thickness characteristic does not exceed the thickness characteristic parameter deviation threshold, namely adjacent detection interval time, if the detection interval time exceeds the interval time threshold, deducing that slow low-span defects appear in the suspected defect area, setting a slow low-span defect type as a detection result, and if the detection interval time does not exceed the interval time threshold, deducing that quick low-span defects appear in the suspected defect area, setting a quick low-span defect type as a detection result; and (3) carrying out priority division according to the types corresponding to the detection results, wherein the order of the priorities from high to low is rapid high-span defects, rapid low-span defects, slow high-span defects and slow low-span defects.
10. The method for automatically identifying a defective area applied to terahertz transmission imaging according to claim 9, wherein the importance characteristics corresponding to the suspected defective area and the non-defective area are determined, quantization parameters corresponding to the importance characteristics are compared, data deviations of quantization parameters of corresponding types are recorded, and quantization parameter types and corresponding data deviations of the corresponding deviations of the importance characteristics are bound with the corresponding suspected defective area as a detection result.

Description

Defect region automatic identification method applied to terahertz transmission imaging Technical Field The invention relates to the technical field of defect detection and identification of wind power generation blades, in particular to an automatic defect area identification method applied to terahertz transmission imaging. Background Wind power generation is used as an important component of clean energy, the duty ratio of the wind power generation is gradually increased in an energy structure, a wind power generation blade is a core component of the wind power generation, the running state of the wind power generation blade directly influences the power generation efficiency and the equipment safety, and the blade is easily subjected to the influence of environmental factors (such as wind and rain, sand and dust, temperature change) and running load in the long-term use process, so that the defects of surface cracks, internal layering, material aging and the like are easily caused, and if the defects cannot be detected and processed in time, serious safety accidents such as blade breakage and the like can be caused. The existing wind power generation blade defect detection technology has the following problems: firstly, the structural complexity, the bearing requirement and the damage probability of different areas of the blade are large, the fixed scanning parameters either cause precision redundancy in a conventional area and reduce efficiency, or cause omission in a key area due to insufficient precision, and the traditional data splicing is easy to generate splicing marks due to coordinate deviation and inconsistent signals so as to cover or misjudge defects. Secondly, environmental temperature fluctuation in outdoor detection is large, terahertz signals are sensitive to temperature and easy to have amplitude drift, so that defect characteristic distortion is caused, equipment runs for a long time, aging loss exists, signal deviation is further aggravated, the above factors are not considered in traditional detection, and detection results are greatly influenced by the environment and equipment states, and stability is poor. Thirdly, in the flight of the unmanned aerial vehicle, the distance between the probe and the blade is changed and the gesture is deviated easily due to the interference of air flow, terahertz wave focusing is inaccurate, the incident angle is changed, signal amplitude attenuation and phase distortion are caused, the traditional imaging is not compensated for the gesture change, the detection precision of different parts of the curved blade is inconsistent, and the defect identification error is large. Fourth, the traditional detection can only judge the current defect state, the defect development trend can not be prejudged, most maintenance work is remedied afterwards, the risks are difficult to prevent and control in advance, the maintenance priority division basis is lacking, and maintenance staff can not reasonably allocate resources, so that the defect treatment of a critical area is not timely, and the non-critical area is excessively maintained. In view of the above technical drawbacks, a solution is now proposed. Disclosure of Invention The invention aims to solve the problems and provide an automatic defect area identification method applied to terahertz transmission imaging. The aim of the invention can be achieved by the following technical scheme: the defect area automatic identification method applied to terahertz transmission imaging comprises the following steps: Step S1, pre-preparation, namely cooperatively calibrating unmanned aerial vehicles, collecting basic information of wind power generation blades to be detected, and planning a flight path of the unmanned aerial vehicles; S2, background collection, namely under the condition that a terahertz light source is not started, the unmanned aerial vehicle flies according to a planned path, and background image data of the fan blades are collected; S3, terahertz transmission imaging acquisition, namely turning on a terahertz light source, enabling an unmanned aerial vehicle to cooperatively fly according to a preset path, enabling the unmanned aerial vehicle carrying the terahertz light source to emit terahertz light to a fan blade, enabling the terahertz light to transmit through the fan blade, receiving a transmission light signal by a terahertz camera on another unmanned aerial vehicle, converting the transmission light signal into imaging data, performing differential scanning and data splicing in the imaging acquisition process, performing image generation, performing temperature influence control and attitude influence control in the image generation stage, obtaining a generated image after completion, and entering the next step; S4, preprocessing data, namely removing background, reducing noise and extracting an original response value; S5, marking a suspected defect area, namely setting a threshold range