Fatigue crack propagation on-line monitoring method under variable amplitude load based on fiber grating sensor

Through multi-sensor layout and deep learning model, fiber grating sensors are used to extract multiple reflection spectrum characteristic values ​​under variable amplitude load to build a damage feature matrix, realizing online quantitative monitoring of fatigue cracks in aero engine parts, and solving the existing technology. Monitoring problems under variable amplitude load support crack propagation life analysis.

CN119043961APending Publication Date: 2024-11-29BEIHANG UNIV
View PDF 0 Cites 4 Cited by

Patent Information

Application Number
CN202411235894.2
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-09-04
Publication Date
2024-11-29

AI Technical Summary

Technical Problem

Existing fiber grating sensors are difficult to effectively monitor the fatigue crack propagation of aircraft engine components under variable amplitude load conditions, and cannot achieve online quantitative monitoring, making it difficult to ensure the safety and reliability of the structure.

Method used

Using a multi-sensor layout, the FBG distortion spectrum damage feature matrix is ​​constructed by extracting multiple reflection spectrum eigenvalues, and a convolutional neural network model is established using deep learning to realize online quantitative monitoring of fatigue cracks under variable amplitude load.

Benefits of technology

Online quantitative monitoring of fatigue crack length under variable amplitude load conditions is realized, which can support crack propagation life analysis, and solves the problem that the prior art is difficult to apply under constant amplitude load.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119043961A_ABST
    Figure CN119043961A_ABST
Patent Text Reader

Abstract

The invention relates to an on-line monitoring method for fatigue crack propagation under variable amplitude load based on fiber bragg grating sensors. The method comprises the following steps: selecting n fiber bragg grating sensors with grating length of Lmm, and attaching the n fiber bragg grating sensors to the surface of a sample according to design; installing the sample pasted with the FBG on a fatigue testing machine to carry out a variable-amplitude load crack propagation test, reading the crack length at a certain cycle interval, and collecting an FBG spectral image; processing the signal acquired by the FBG sensor; constructing the m characteristic values of the n sensors under a certain crack length into an FBG distortion spectrum damage characteristic matrix under the current crack length; constructing a data set by taking the FBG distortion spectrum damage characteristic matrix as an input sample and the crack length value as an output value; dividing the data set into a training set and a test set, training, and constructing a variable-amplitude load fatigue crack propagation monitoring model; and testing by using the test set. According to the invention, the function of monitoring the crack propagation length on line is realized.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the technical field of aerospace engines, and specifically relates to an online monitoring method for fatigue crack extension under variable amplitude load based on a fiber grating sensor (Fiber Grating Sensor), which can realize online monitoring of fatigue crack damage of aviation equipment structure under variable amplitude load in actual flight. Background Art

[0002] Advanced aircraft engine parts have complex structures and are subjected to complex, alternating variable amplitude fatigue loads for a long time. Fatigue cracks are easily generated and extended at the stress concentration parts of the structure, which leads to a decrease in the structural bearing capacity or even fracture failure, causing safety accidents. Therefore, online monitoring of structural fatigue crack extension under variable amplitude load is an important technology to reduce the risk of fatigue failure of aircraft engines under service conditions and ensure the structural safety and reliability of aircraft engine parts.

[0003] Fiber Bragg Grating (FBG) sensors have the advantages of high temperature resistance, small size, light weight, anti-electromagnetic interference, wavelength division multiplexing, etc. It can etch multiple grating measurement points on an optical fiber to achieve quasi-distributed measurement. By attaching FBG sensors to the surface of the structure, structural state monitoring can be achieved. They are widely used in structural health monitoring in aerospace, petrochemical and other fields. FBG sensors are strain-sensitive elements. When cracks appear in the structure, the non-uniform stress / strain field caused by the crack tip will cause the reflection spectrum of the FBG sensor attached near it to drift and distort. Therefore, characteristic parameters that can reflect the spectral characteristics (such as central wavelength, spectral bandwidth, fractal dimension, spectral area, etc.) are extracted from the distorted spectrum, and the relationship between the FBG spectral characteristic parameters and crack extension under constant amplitude load is studied to achieve crack recognition and single crack tip position monitoring; in addition, with the development of computer technology, neural networks, machine learning and other methods are also used to train the relationship between spectral characteristics and cracks to achieve quantitative monitoring of crack extension under constant amplitude load. However, aircraft engine parts are subjected to variable amplitude loads for a long time during actual service, and since the essence of FBG sensor monitoring is strain, variable amplitude loads will cause irregular changes in structural stress, making the FBG-based crack monitoring method under constant amplitude load difficult to apply. Therefore, it is necessary to develop a quantitative monitoring method for fatigue crack growth under variable amplitude load based on FBG sensors to ensure the structural safety and reliability of aircraft engine parts.

[0004] Existing Chinese patent applications "CN106767475 A", "CN106596255 A", and "CN 106840869 A" provide hole edge crack diagnosis methods based on fiber Bragg grating spectral image analysis based on different sensor subsidy methods, which diagnose the length position of the hole edge crack of the aluminum alloy plate by extracting the sub-peak position of the reflection spectrum as the characteristic value. However, on the one hand, the above methods only use a single reflection spectrum feature, which can only identify the presence or absence of cracks or determine the position of a single crack, and cannot monitor the entire fatigue crack propagation process; on the other hand, they are only applicable to constant amplitude load conditions, and cannot be applied to variable amplitude load conditions. The existing paper "Wang Tiantian, Wang Qinmin, Yang Jinsong, et al. Research on hole edge fatigue crack monitoring based on fiber Bragg grating [J]. Journal of Central South University (Science and Technology), 2022, 53(5): 1614-1625." proposed a crack length monitoring method based on fiber Bragg grating sensor. It deploys multiple FBG sensors on the surface of the structure, extracts multiple eigenvalues ​​from the reflection spectrum, and uses the support vector regression method to establish a diagnostic model between the damage eigenvalue and the crack length to realize the monitoring of the crack length. However, it only considers the constant amplitude load condition.

[0005] In summary, most of the existing fiber Bragg grating sensor fatigue crack monitoring methods are only applicable to constant amplitude load scenarios, and it is difficult to solve the problem of fatigue crack monitoring under variable amplitude load. Therefore, in order to meet the needs of online monitoring of structural fatigue crack damage of key components in the aerospace field, it is necessary to develop a fatigue crack propagation monitoring method based on fiber Bragg grating sensors suitable for variable amplitude load conditions. Summary of the invention

[0006] In order to overcome the shortcomings of the existing FBG-based online monitoring method for aircraft structure crack damage, the present invention provides an online monitoring method for fatigue crack extension under variable amplitude load based on fiber grating sensor, which can meet the online quantitative monitoring of crack damage in aircraft structure and can serve and support crack extension life analysis. Based on test data, the present invention constructs the FBG distorted spectrum damage feature matrix with multi-sensor multi-eigenvalues, establishes a fatigue crack extension monitoring model based on deep learning, and realizes online quantitative monitoring of fatigue crack extension under variable amplitude load.

[0007] In order to achieve the above object, the present invention adopts the following technical scheme: An online monitoring method for fatigue crack growth under variable amplitude load based on fiber grating sensor comprises the following steps: Step 1: Select n FBG sensors with a grating length of L mm and paste them on the surface of the sample. Each fiber grating has a different central wavelength and is etched on the same optical fiber to achieve quasi-distributed measurement; Step 2: Install the sample with FBG sensor on the fatigue testing machine for crack extension test. Collect the FBG initial signal before crack extension. After applying variable amplitude load, fatigue crack starts to extend. Use optical microscope to observe crack extension. Read the crack length and collect FBG spectrum image at certain cycle intervals. Step 3: Process the signals collected by the FBG sensor, extract m eigenvalues ​​including the central wavelength and broadening of the FBG reflection spectrum under each crack length, and analyze the relationship between each eigenvalue and the crack length; Step 4: construct the m eigenvalues ​​of n sensors under a certain crack length into the FBG distortion spectrum damage characteristic matrix under the current crack length; Step 5: Use the FBG distortion spectrum damage feature matrix as the input sample and the crack length value as the output value to construct a data set, and design a neural network model for predicting the crack length value according to the form of the data set; divide the data set into a training set and a test set, use the training set to train the neural network model, and construct a variable amplitude load fatigue crack growth monitoring model; Step 6: Use the test set to test the variable amplitude load fatigue crack growth monitoring model, input the FBG distortion spectrum damage feature matrix, and the variable amplitude load fatigue crack growth monitoring model gives a predicted crack length value, thereby monitoring the crack growth length in real time.

[0008] The beneficial effects of the present invention compared with the prior art are: The present invention creates a typical variable amplitude load scenario, conducts a variable amplitude load fatigue crack growth test, adopts a multi-sensor layout, extracts multiple eigenvalues ​​in the simulated reflection spectrum, establishes a fatigue crack monitoring model based on a convolutional neural network, and uses the FBG distortion spectrum damage feature matrix to monitor the crack length. This method solves the problem of FBG sensor monitoring cracks under variable amplitude loads. At the same time, compared with methods that can only roughly locate cracks, this method can realize online quantitative monitoring of crack length, which can serve and support crack growth life analysis. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] Figure 1 It is a flow chart of an online monitoring method of fatigue crack propagation under variable amplitude load based on fiber grating sensor of the present invention; Figure 2 This is the CT sample image of the present invention; Figure 3 It is a schematic diagram of the arrangement of the fiber Bragg grating sensor of the present invention; Figure 4 The load spectrum of the variable amplitude load fatigue crack growth test of the present invention; Figure 5 It is a schematic diagram of the reflection spectrum of the FBG test of the present invention; Figure 6 It is a schematic diagram of the change of the characteristic value of the reflection spectrum with the crack length of the present invention; Figure 7 Schematic diagram of FBG distortion spectrum damage characteristic matrix; Figure 8 It is a schematic diagram of the monitoring results using the present invention. DETAILED DESCRIPTION

[0010] The present invention will be further described below in conjunction with the accompanying drawings.

[0011] like Figure 1 As shown, a method for online monitoring of fatigue crack propagation under variable amplitude load based on a fiber grating sensor of the present invention comprises the following steps: Step 1: Select the design specimen, including: Design as Figure 2 The CT specimen of 7075-T76 aluminum alloy shown in the figure has an overall size of 125mm×120mm×10mm, a loading hole diameter of 15mm, and the distances from the center of the loading hole to the edge of the specimen are 100mm, 32.5mm, and 25mm, respectively. The upper and lower loading holes are symmetrical, and the notch tip is 12mm away from the center of the loading hole. A 2mm crack is prefabricated. Five FBG sensors with a grid length of 5mm are selected. The FBG sensor placement positions are as follows: Figure 3 As shown. FBG1, FBG2, and FBG3 are longitudinally mounted, with their axial direction perpendicular to the crack propagation direction, and their distances from the initial crack tip are 10mm, 20mm, and 30mm, respectively, and their distances from the prefabricated crack in the Y direction are 2mm; FBG1-2 sensors are transversely mounted, with their distances from the initial crack tip are 10mm and 20mm, respectively, and their distances from the prefabricated crack in the Y direction are 5mm. FBG1, FBG2, FBG3, FBG4, and FBG5 are all etched on the same optical fiber, with central wavelengths of 1510nm, 1520nm, 1530nm, 1540nm, and 1550nm, respectively, which can achieve quasi-distributed measurement.

[0012] Step 2: Lay out the FBG sensor and conduct fatigue crack growth test, including: The CT specimen with FBG sensor was installed on the fatigue testing machine for crack extension test. The initial FBG signal was collected before the crack extension. Figure 4 The variable amplitude load of the periodic single overload is shown. After every 10,000 cycles of constant amplitude load, an overload with an overload ratio of 1.5 is applied. After loading, the fatigue crack begins to grow. The crack growth is observed using an optical microscope. The crack length is recorded every 500 cycles and the FBG spectrum image is collected. Figure 5 Shown is a partial spectrum image at a certain crack length.

[0013] Step 3: FBG reflection spectrum acquisition, extraction of reflection spectrum features, including: The change of strain field causes the change of grating period and effective refractive index of FBG sensor, thus affecting the change of central wavelength of reflection spectrum. Meanwhile, non-uniform strain also distorts the reflection spectrum, resulting in the change of characteristic value of reflection spectrum. Five characteristic values ​​are extracted respectively, including central wavelength, quarter width, spectral area, overlap area between deformation spectrum and original spectrum, and fractal dimension of simulated reflection spectrum at each crack length, as shown in Fig. Figure 6 As shown. Among them, the width corresponding to the maximum reflectivity of the reflection spectrum is the quarter broadening of the reflection spectrum; the area enclosed by the reflection spectrum and the coordinate axis is the reflection spectrum area; the overlapping area of ​​the reflection spectrum measured in the healthy state and the reflection spectrum in the damaged state is defined as the overlapping area; the fractal dimension is an indicator to measure the degree of irregularity of the shape of the figure. Analyze the relationship between each eigenvalue and the crack length.

[0014] Step 4: Construct the FBG distortion spectrum damage feature matrix, including: The five characteristic values ​​of the five sensors under a certain crack length are calculated as follows: Figure 7 The figure shows the damage feature matrix of the FBG distortion spectrum under the current crack length, where i represents the FBG sensor number, j represents the type number of the damage feature, I is the total number of FBG sensors, and J is the number of damage features extracted for each FBG sensor. Each crack length corresponds to 25 damage features in the form of 5×5.

[0015] Step 5: Establish a fatigue crack monitoring method based on convolutional neural network, including: The FBG distortion spectrum damage feature matrix is ​​used as the input sample and the crack length value is used as the output value to construct a data set. According to the form of the data set, a convolutional neural network model for predicting the crack length value is designed. The convolutional neural network model includes convolutional layer, flattening layer, fully connected layer and input and output layer.

[0016] The convolution layer performs convolution operation, and the transfer function between convolution layers performs nonlinear transformation on the input data. The transfer function selects a semilinear function f (x) with a leakage unit: ; Where, leakage unit ρ = 0.001; x is the input value.

[0017] The flattening layer spreads the output of the last stage into a one-dimensional vector, and the fully connected layer fuses and outputs the features extracted by the convolutional layer and the pooling layer.

[0018] In order to better transmit information or data in the neural network, the variance of each layer output is kept consistent as much as possible, and the network parameters are initialized using the Xavier method shown in the following formula: ; In the formula, U represents uniform distribution, and the weight w of the mth layer of the convolutional neural network is m Initialize with the uniform distribution U, e m is the number of neurons in the mth layer of the neural network, e m-1 is the number of neurons in the m-1th layer of the neural network, and the bias values ​​of each layer are initialized to 0.

[0019] 80% of the data set is divided into a training set and the remaining 20% ​​is divided into a test set. The training data is input into the convolutional neural network. The convolutional neural network is trained through forward propagation, sensitivity back propagation and parameter update. The learning rate is set to 0.001, the batch size is set to 1, and the mean square error (MSE) is selected as the loss function to construct a variable amplitude load fatigue crack growth monitoring model.

[0020] Step 6: Real-time quantitative monitoring of crack length, including: The variable amplitude load fatigue crack growth monitoring model was tested using the test set. The FBG distortion spectrum damage feature matrix was input and the monitoring model gave a predicted crack length value, thereby realizing online quantitative monitoring of the crack growth length. The monitoring results are shown in Figure 2. Figure 8 shown.

[0021] The above embodiments are provided only for the purpose of describing the present invention, and are not intended to limit the scope of the present invention. The scope of the present invention is defined by the appended claims. Various equivalent substitutions and modifications made without departing from the spirit and principles of the present invention should all be included within the scope of the present invention.

Claims

1. An online monitoring method for fatigue crack growth under variable amplitude load based on fiber grating sensor, It is characterized in that The steps include: Step 1: Select n FBG sensors with a gate length of L mm and paste them on the surface of the sample. Each FBG has a different central wavelength and is etched on the same optical fiber to achieve quasi-distributed measurement; Step 2: Install the sample with FBG sensor on the fatigue testing machine for crack extension test. Collect the FBG initial signal before crack extension. After applying variable amplitude load, fatigue crack starts to extend. Use optical microscope to observe crack extension. Read the crack length and collect FBG spectrum image at certain cycle intervals. Step 3: Process the collected FBG sensor signals, extract m eigenvalues ​​under each crack length, and analyze the relationship between each eigenvalue and the crack length; Step 4: construct the m eigenvalues ​​of n sensors under a certain crack length into the FBG distortion spectrum damage characteristic matrix under the current crack length; Step 5: Use the FBG distortion spectrum damage feature matrix as the input sample and the crack length value as the output value to construct a data set, and design a neural network model for predicting the crack length value according to the form of the data set; divide the data set into a training set and a test set, use the training set to train the neural network model, and construct a variable amplitude load fatigue crack growth monitoring model; Step 6: Use the test set to test the variable amplitude load fatigue crack growth monitoring model, input the FBG distortion spectrum damage feature matrix, and the variable amplitude load fatigue crack growth monitoring model gives a predicted crack length value, thereby monitoring the crack growth length in real time.

2. According to claim 1, a method for online monitoring of fatigue crack growth under variable amplitude load based on fiber grating sensor, Features: The characteristic values ​​in step three include: the central wavelength of the simulated reflection spectrum at each crack length, the quarter width, the spectrum area, the overlap area between the deformed spectrum and the original spectrum, the fractal dimension, etc.

3. The method for online monitoring of fatigue crack growth under variable amplitude load based on fiber grating sensor according to claim 2, Features: The width corresponding to the quarter maximum reflectivity of the reflection spectrum is the quarter broadening of the reflection spectrum; the area enclosed by the reflection spectrum and the coordinate axis is the reflection spectrum area; the overlapping area between the reflection spectrum measured in a healthy state and the reflection spectrum in a damaged state is defined as the overlapping area; the fractal dimension is an indicator that measures the degree of irregularity of the shape of a figure, and is used to analyze the relationship between each eigenvalue and the crack length.

4. The method for online monitoring of fatigue crack growth under variable amplitude load based on fiber grating sensor according to claim 1, Features: The neural network model in step five is a convolutional neural network model, which includes a convolutional layer, a flattening layer, a fully connected layer, and an input and output layer.

5. The method for online monitoring of fatigue crack growth under variable amplitude load based on fiber grating sensor according to claim 4, Features: The convolution layer performs convolution operation, and the transfer function between convolution layers performs nonlinear transformation on the input data. The transfer function selects a semilinear function f (x) with a leakage unit: ; Where, leakage unit ρ=0.001; x is the input value; The flattening layer spreads the output of the last stage into a one-dimensional vector, and the fully connected layer fuses and outputs the features extracted by the convolutional layer and the pooling layer; The network parameters are initialized using the Xavier method shown below: ; In the formula, U represents uniform distribution, and the weight w of the mth layer of the convolutional neural network is m Initialize with the uniform distribution U, e m is the number of neurons in the mth layer of the neural network, e m-1 is the number of neurons in the m-1th layer of the neural network, and the bias values ​​of each layer are initialized to 0.

Citation Information

Cited By

  • Hole edge crack monitoring method, device, equipment and medium

    CN120911165A

  • Method for quantitatively characterizing fatigue crack full-life characteristics

    CN121347292A

  • Multi-parameter bridge structure state monitoring method and system based on self-sensing material

    CN122361530A

  • A Multi-parameter Bridge Structural Status Monitoring Method and System Based on Self-Sensing Materials

    CN122361530B