Airplane composite material structure surface defect detection method

Abstract

The invention relates to the technical field of composite material detection, in particular to a composite material structure surface defect detection method. The rapid composite material structure surface defect detection method is obtained by adopting a hyperspectral imaging technology, so that the data acquisition and processing speed is high, and the identification precision is high. According to the method, the position and the type of the surface defect of the defective composite can be quickly and accurately detected, and the defect can be classified and judged on the basis, so that the influence of human subjective factors on defect judgment is greatly reduced, meanwhile, the data of the defect position can be directly obtained, secondary metering work is not needed, and the working efficiency is improved. According to the invention, the image dimension and the spectrum dimension of the hyperspectral imaging system are simultaneously utilized to detect the fabric defects, so that the result is more intuitive and accurate.

Description

Technical Field

[0001] The present invention relates to the technical field of composite material detection, and in particular to a method for detecting surface defects of a composite material structure. Background Art

[0002] Since composite materials are made of a mixture of different types of materials in a certain proportion, their internal stability is not as good as that of a single type of material, so they are prone to damage during their processing, which is specifically manifested as cracks, delamination and other phenomena. The current inspection method uses manual inspection, that is, the operator uses tools such as flashlights and reflectors to inspect with the naked eye. However, since aircraft use a large number of composite materials, the composite structures that need to be inspected cover a large area and are large in number, and there is a possibility of omissions in the manual method. In addition, for surface defects such as scratches, the naked eye cannot directly determine the depth and width of the scratches, and it is impossible to determine whether they can continue to be used. It is still necessary to ask the metrology center to measure the surface defects of the composite materials to determine the results. The above results prove that the traditional method has the disadvantages of easy omissions, low efficiency, need for metrology means, and long working time. Therefore, an efficient surface defect detection method for composite material structures is urgently needed. Summary of the invention

[0003] The purpose of the present invention is to solve the problems in the prior art and propose a method for detecting surface defects of composite materials. The method adopts hyperspectral remote sensing imaging technology, a parallel classification algorithm of remote sensing image spectral angles and a neural network structure model to achieve accurate discrimination of defect images, thereby realizing rapid online detection of surface defects of aircraft composite materials.

[0004] To achieve the above object, the present invention proposes a method for detecting surface defects of a composite material structure, comprising the following steps:

[0005] Step 1: Under the same working environment, a hyperspectral remote sensing imaging system is used to collect hyperspectral remote sensing data and images of several composite materials with various surface defects to form a database. The composite materials to be tested include aircraft skins, covers, rudders, doors, etc. The surface defects of the above composite materials include stains, scratches, bumps, bulges, etc.

[0006] The difference in data between the composite material hyperspectral remote sensing defect image and the defect-free image is reflected in the composite material defect quantification value. The data and image acquisition equipment obtain the average brightness of the composite material defect part in the composite material structure defect area and the average brightness of the composite material background in the background area of ​​the composite material defect area. The image contrast of the composite material structure defect area is calculated by the two, and the defect degree of the composite material defect area is calculated by the area and defect depth of the composite material defect area, so as to determine the composite material defect quantification value of the composite material defect area. Specifically, the calculation formula of the composite material defect quantification value is as follows:

[0007]

[0008] Among them, Q fc is the composite material defect quantification value of the composite material structure defect area, K s is the composite defect area correction coefficient (varies according to different composite material types), L q is the average brightness of the defective parts of the composite material, L b is the average background brightness of the composite material, S q is the area value of the defective part of the composite material, K h is the depth correction coefficient of the composite material defect (varies according to the type of composite material), H q is the depth value of the defective part of the composite material.

[0009] The equipment that can meet the needs of hyperspectral remote sensing data and image acquisition of composite materials with various surface defects mainly adopts camera equipment using vision-guided technology.

[0010] Step 2: Since the environment during the actual operation is different from the environment when the database is established, it is necessary to process and convert each hyperspectral remote sensing data and image during actual use, and use remote sensing radiation correction, exposure reduction, image enhancement and other methods to convert it into information that can be recognized and processed by the computer or analysis module based on the completed collection.

[0011] In the process of processing and conversion, remote sensing radiation correction is particularly important. By adjusting and correcting the radiation value in the remote sensing image, the influence of factors such as atmosphere, surface reflectivity and sensor that may be affected by the remote sensing data during the acquisition process is eliminated, so as to avoid the above factors causing inaccurate radiation values ​​in the image or differences that affect the accuracy of the detection results. For this purpose, a correction formula is provided:

[0012]

[0013] Among them, P m is the hyperspectral remote sensing image after remote sensing radiation correction, P l is the hyperspectral remote sensing image before remote sensing radiation correction, A is the theoretical image of the measured composite structure, k y is the correction factor for remote sensing radiation correction.

[0014] Step 3: Use the remote sensing image spectral angle parallel classification algorithm to make each processed and converted hyperspectral remote sensing data and image form fitting image information to display the surface defects of the composite material structure. The principle of the remote sensing image spectral angle parallel classification algorithm is to project the spectrum as a vector into the N-dimensional space, where the dimension N is the number of calculated bands. In the N-dimensional space, the spectral curve is a vector with direction and length. The angle formed between different vectors is the spectral angle. The smaller the spectral angle between different spectral curves, the more similar they are.

[0015]

[0016] Among them, α is the spectral angle, k is the correction coefficient, n is the total number of bands, Xa is the measured spectral curve, and Xb is the theoretical spectral curve of different defect forms.

[0017] Remote sensing image processing software was used to obtain the depth and length of the injury site.

[0018] Step 4: Establish a linear neural network structure model

[0019] A linear neural network model is established, which is similar to a neuron forming a multi-input and single-output information processing unit. According to the characteristics and functions of neurons, neurons can be abstracted into a simple mathematical model algorithm, which includes an input layer, a judgment layer and an output layer. The hyperspectral remote sensing data and images enter the judgment layer through the input layer (the content described in step one), and then reach the output layer through the judgment layer (step three), and the surface defect type and damage degree of the composite structure are obtained.

[0020] Step 5: Use the established linear neural network structure model to detect the defect type of the sample to be tested and formulate a treatment plan

[0021] The composite structure sample to be tested is operated according to steps one to three, and then the hyperspectral remote sensing image obtained in step two and the depth and length of the damaged part obtained in step three are added to the established linear neural network structure model to obtain the type and degree of damage on the surface of the composite structure.

[0022] Beneficial effects of the present invention:

[0023] The present invention uses hyperspectral imaging technology to obtain a rapid composite material structure surface defect detection method, which makes data collection and processing fast and identification accuracy high. The present invention can not only quickly and accurately detect the location and type of composite material surface defects, but also classify and determine them on this basis, thereby greatly reducing the influence of human subjective factors on defect determination. At the same time, the data of the defect location can be directly obtained without secondary measurement work, thereby improving work efficiency. The present invention simultaneously utilizes the image dimension and spectral dimension of the hyperspectral imaging system to detect fabric defects, which can make the results more intuitive and accurate. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 It is a workflow diagram of the surface defect detection method of composite material structure;

[0025] Figure 2 It is an algorithm flow chart of a linear neural network structure model of a composite material structure surface defect detection method according to the present invention;

[0026] Figure 3It is a schematic diagram of the training of the linear neural network structure model of the composite material structure surface defect detection method according to the present invention. DETAILED DESCRIPTION

[0027] Reference Figure 1 The present invention is based on a method for detecting surface defects of a composite material structure, comprising the following steps:

[0028] S1. Use the hyperspectral imaging system to collect hyperspectral data and images of several types of fabrics with defects:

[0029] Firstly, under the same working environment, a hyperspectral remote sensing imaging system was used to collect hyperspectral remote sensing data and images of several types of composite materials with various surface defects to form a database. The composite structures to be tested included aircraft skins, covers, rudders, doors, etc. The surface defects of the above composite structures included stains, scratches, bumps, bulges, etc. In this way, a collection of composite structure defects was formed for subsequent missing parts detection and comparison.

[0030] The difference in data between the composite material hyperspectral remote sensing defect image and the defect-free image is reflected in the composite material defect quantification value. The data and image acquisition equipment obtain the average brightness of the composite material defect part in the composite material structure defect area and the average brightness of the composite material background in the background area of ​​the composite material defect area. The image contrast of the composite material structure defect area is calculated by the two, and the defect degree of the composite material defect area is calculated by the area and defect depth of the composite material defect area, so as to determine the composite material defect quantification value of the composite material defect area. The defect quantification value calculation formula is as follows:

[0031]

[0032] An embodiment is selected: the scratch image and data with a length of 10 mm, a width of 2 mm, and a depth of 1 mm and the dent image and data with a length of 8 mm, a width of 4 mm, and a depth of 6 mm on the surface of the composite material are collected to form a database.

[0033] S2. Process and convert the measured hyperspectral remote sensing data and images:

[0034] Mainly adopt remote sensing radiation correction, exposure reduction, image enhancement and other methods to convert into information that can be recognized and processed by the computer or analysis module based on the acquisition. In the process of processing and conversion, remote sensing radiation correction is particularly important. By adjusting and correcting the radiation value in the remote sensing image, the influence of factors such as atmosphere, surface reflectivity and sensor that may be affected by the remote sensing data in the acquisition process is eliminated, so as to avoid the above factors causing inaccurate radiation values ​​in the image or differences that affect the accuracy of the detection results. For this reason, the correction formula is adopted:

[0035]

[0036] Step 3: Use the remote sensing image spectral angle parallel classification algorithm to make each processed and converted hyperspectral remote sensing data and image form fitting image information to display the surface defects of the composite material structure. The principle of the remote sensing image spectral angle parallel classification algorithm is to project the spectrum as a vector onto the N-dimensional space, where the dimension N is the number of calculated bands. In the N-dimensional space, the spectral curve is a vector with direction and length. The angle formed between different vectors is the spectral angle. The smaller the spectral angle between different spectral curves, the more similar they are:

[0037]

[0038] An embodiment is selected: remote sensing image processing software is used to obtain the depth and length of the damaged part: the measured surface defect information of the composite material is a scratch image with a length of 8 mm, a width of 1.5 mm, and a depth of 0.5 mm.

[0039] Step 4: Establish a linear neural network structure model

[0040] A linear neural network model is established, which is similar to a neuron forming a multi-input and single-output information processing unit. According to the characteristics and functions of neurons, neurons can be abstracted into a simple mathematical model algorithm, which includes an input layer, a judgment layer and an output layer. The hyperspectral remote sensing data and images enter the judgment layer through the input layer (the content described in step one), and then reach the output layer through the judgment layer (step three), and the surface defect type and damage degree of the composite structure are obtained.

[0041] An embodiment is selected: by comparing the measured surface defect information of the composite material with the defect database information, the defect type is obtained as "scratch" and the defect degree is "minor"

[0042] Step 5: Use the established linear neural network structure model to detect the defect type of the sample to be tested and formulate a treatment plan

[0043] The composite structure sample to be tested is operated according to steps one to three, and then the hyperspectral remote sensing image obtained in step two and the depth and length of the damaged part obtained in step three are added to the established linear neural network structure model to obtain the type and degree of damage on the surface of the composite structure.

Claims

1. A method for detecting surface defects of aircraft composite material structures, characterized in that: The steps include: Step 1: Under the same working environment, a hyperspectral remote sensing imaging system is used to collect hyperspectral remote sensing data and images of several composite materials with various surface defects to form a database; The difference in data between the composite material hyperspectral remote sensing defect image and the defect-free image is reflected in the composite material defect quantification value. The calculation formula of the composite material defect quantification value is as follows: Among them, Q fc is the composite material defect quantification value of the composite material structure defect area, K s is the composite material defect area correction factor, L q is the average brightness of the defective parts of the composite material, L b is the average background brightness of the composite material, S q is the area value of the defective part of the composite material, K h is the depth correction coefficient of composite material defect, H q is the depth value of the defective part of the composite material; Step 2: Process and convert each hyperspectral remote sensing data and image, using remote sensing radiation correction, exposure reduction, and image enhancement methods to convert them into information that can be recognized and processed by the computer or analysis module based on the collection completion; Step 3: Using the remote sensing image spectral angle parallel classification algorithm to make each processed and converted hyperspectral remote sensing data and image form fitting image information to display the surface defects of the composite material structure; the principle of the remote sensing image spectral angle parallel classification algorithm is to project the spectrum as a vector onto an N-dimensional space, where the dimension N is the number of calculated bands, and in the N-dimensional space, the spectral curve is a vector with direction and length, and the angle formed between different vectors is the spectral angle, and the smaller the spectral angle between different spectral curves, the more similar they are; Among them, α is the spectral angle, k is the correction coefficient, n is the total number of bands, Xa is the measured spectrum curve, and Xb is the theoretical spectrum curve of different defect forms; Use remote sensing image processing software to obtain the depth and length of the injury site; Step 4: Establish a linear neural network structure model A linear neural network model is established, which includes an input layer, a decision layer and an output layer. The hyperspectral remote sensing data and images enter the decision layer through the input layer, and then reach the output layer through the decision layer to obtain the surface defect type and damage degree of the composite structure. Step 5: Use the established linear neural network structure model to detect the defect type of the sample to be tested and formulate a treatment plan.

2. The method for detecting surface defects of aircraft composite material structures according to claim 1, characterized in that: The composite material structure to be tested includes aircraft skin, flap, control surface, and cabin door, and the surface defects of the composite material structure include stains, scratches, bumps, and bulges.

3. The method for detecting surface defects of aircraft composite material structures according to claim 1 or 2, characterized in that: In the second step, during the conversion process, the radiation value in the remote sensing image is adjusted and corrected to provide a correction formula: Among them, P m is the hyperspectral remote sensing image after remote sensing radiation correction, P l is the hyperspectral remote sensing image before remote sensing radiation correction, A is the theoretical image of the measured composite structure, k y is the correction factor for remote sensing radiation correction.

4. The method for detecting surface defects of aircraft composite material structures according to claim 1 or 2, characterized in that: The equipment for collecting composite material hyperspectral remote sensing data and images adopts a camera device using visual guidance technology.

5. The method for detecting surface defects of aircraft composite material structures according to claim 3, characterized in that: The equipment for collecting composite material hyperspectral remote sensing data and images adopts a camera device using visual guidance technology.

6. The method for detecting surface defects of aircraft composite material structures according to claim 1, 2 or 5, characterized in that: The composite structure sample to be tested is operated according to steps one to three, and then the hyperspectral remote sensing image obtained in step two and the depth and length of the damaged part obtained in step three are added to the established linear neural network structure model to obtain the type and degree of damage on the surface of the composite structure.

7. The method for detecting surface defects of aircraft composite material structures according to claim 3, characterized in that: The composite structure sample to be tested is operated according to steps one to three, and then the hyperspectral remote sensing image obtained in step two and the depth and length of the damaged part obtained in step three are added to the established linear neural network structure model to obtain the type and degree of damage on the surface of the composite structure.

8. The method for detecting surface defects of aircraft composite material structures according to claim 4, characterized in that: The composite structure sample to be tested is operated according to steps one to three, and then the hyperspectral remote sensing image obtained in step two and the depth and length of the damaged part obtained in step three are added to the established linear neural network structure model to obtain the type and degree of damage on the surface of the composite structure.

Images