Intelligent residual stress detection method and system based on quantum light sensation

Abstract

The embodiment of the invention provides an intelligent residual stress detection method and system based on quantum light sensation, and belongs to the technical field of residual stress detection. The method comprises the following steps: acquiring multispectral image data and temperature and humidity data of a quantum dot coating on the surface of an area to be detected; inputting the multispectral image data, the temperature and humidity data and the material performance parameters of the to-be-detected area into a residual stress detection model to obtain a residual stress distribution diagram of the to-be-detected area; and visually outputting the residual stress distribution diagram. According to the residual stress intelligent detection method and system based on quantum light sensation, the light-force coupling effect of the quantum dot coating is combined with the machine learning algorithm, and based on multispectral dynamic imaging and environmental data compensation, residual stress visual detection under complex working conditions is realized, and the detection accuracy is improved. And a high-resolution, non-destructive and multi-material adaptive detection technology is realized, and a pioneering technical thought is provided for intelligent manufacturing and structural health.

Description

TECHNICAL FIELD

[0001] Embodiments of the present disclosure belong to the technical field of residual stress detection, and particularly relate to a residual stress intelligent detection method and system based on quantum light sensing. BACKGROUND

[0002] Residual stress can be divided into residual tensile stress and residual compressive stress according to stress properties. Generally, residual tensile stress is not conducive to the service of the structure, and high-level residual tensile stress provides driving force for the crack initiation and propagation process, greatly accelerating the failure process of the structure. Residual tensile stress widely exists in structures. For example, in the metal pipeline system of the power system, the welding between pipes in the manufacturing process of elbow pipes and tees introduces high-level residual tensile stress, which causes a safety hazard to the safe service of the entire structure. Precise grasp of the residual stress level in such areas in order to make appropriate disposal in time has important engineering value for ensuring the safe and long-term operation of the structure. At the same time, in the fields of aerospace, nuclear power equipment and precision manufacturing, residual stress is also an important factor affecting the fatigue life and dimensional stability, and the residual stress level often needs to be strictly controlled. One of the prerequisites for controlling residual stress is to accurately characterize the residual stress in order to choose the most appropriate disposal method.

[0003] Traditional residual stress detection can be divided into non-destructive detection, semi-destructive detection and destructive detection. Among them, semi-destructive and destructive detection damage the integrity of the equipment itself and are not suitable for residual stress detection of in-service equipment. Non-destructive detection generally uses X-ray method to detect residual stress, but it is limited by the characteristics of crystalline materials and is not suitable for non-crystalline materials. At the same time, there is a risk of radiation, the cost of equipment is high, the flexibility is insufficient, the detection accuracy is significantly affected by the surface state, and strict surface treatment is required. In summary, there are still many inconveniences. Traditional photoelastic coating detection of residual stress is easily affected by external environment and has insufficient resolution. In summary, although there are many residual stress detection methods at present, it is still difficult to meet the detection needs of non-destructive, high-precision, multi-material compatibility and no safety risk for in-service equipment. SUMMARY

[0004] Embodiments of the present disclosure aim to at least solve one of the technical problems existing in the prior art, and provide a residual stress intelligent detection method and system based on quantum light sensing.

[0005] One aspect of the present disclosure provides a residual stress intelligent detection method based on quantum light sensing, the method comprising:

[0006] Collecting multi-spectral image data and temperature and humidity data of a quantum dot coating on the surface of a region to be measured; wherein the quantum dot coating is made of quantum dots and fluorescent materials.

[0007] inputting the multispectral image data, the temperature and humidity data, and material performance parameters of the to-be-tested region into a residual stress detection model to obtain a residual stress distribution map of the to-be-tested region; wherein the residual stress detection model is pre-established by training a ResNet-50 model based on finite element simulation data and standard specimen test data;

[0008] visualizing and outputting the residual stress distribution map.

[0009] Further, before the multispectral image data and the temperature and humidity data of the quantum dot coating on the surface of the to-be-tested region are collected, the method further comprises:

[0010] forming a quantum dot coating on the surface of the to-be-tested region and embedding a temperature and humidity sensor in the quantum dot coating.

[0011] Further, the forming of the quantum dot coating on the surface of the to-be-tested region comprises:

[0012] uniformly spraying quantum dot coating material onto the surface of the to-be-tested region;

[0013] curing the quantum dot coating material to form the quantum dot coating.

[0014] Further, the temperature and humidity sensor is a fiber grating sensor.

[0015] Further, the collection of the multispectral image data of the to-be-tested region comprises:

[0016] using a pulsed laser to excite a fluorescence signal of the quantum dot coating;

[0017] using a hyperspectral camera to collect the fluorescence signal to obtain multispectral image data.

[0018] Further, the residual stress detection model is pre-established by the following steps:

[0019] generating simulation data of different materials under different loads by a finite element model and obtaining test data by standard specimen tests;

[0020] mixing the simulation data and the test data to obtain a residual stress detection dataset; the dataset comprises multispectral image data, temperature and humidity data, material performance parameters, and a real stress distribution map;

[0021] training a ResNet-50 model by taking the multispectral image data, the temperature and humidity data, and the material performance parameters as inputs and taking the real stress distribution map as output to obtain the residual stress detection model.

[0022] Further, the training of the ResNet-50 model comprises:

[0023] adding a Hooke's law constraint and a boundary condition constraint.

[0024] Further, the training of the ResNet-50 model comprises:

[0025] updating the ResNet-50 model parameters by using one or more of transfer learning, attention mechanism and dynamic environment compensation mechanism.

[0026] Another aspect of the present disclosure provides a quantum light sensing based residual stress intelligent detection system, comprising:

[0027] An acquisition module is configured to acquire multi-spectral image data and temperature and humidity data of a quantum dot coating on a surface of a to-be-detected area, wherein the quantum dot coating is made of quantum dots and fluorescent materials.

[0028] A detection module is configured to input the multi-spectral image data, the temperature and humidity data and material performance parameters of the to-be-detected area into a residual stress detection model to obtain a residual stress distribution map of the to-be-detected area, wherein the residual stress detection model is pre-established based on training of a ResNet-50 model by using finite element simulation data and standard test specimen test data.

[0029] An output module is configured to visually output the residual stress distribution map.

[0030] Further, the acquisition module comprises:

[0031] A pulsed laser is configured to excite a fluorescent signal of the quantum dot coating.

[0032] A hyperspectral camera is configured to acquire the fluorescent signal to obtain multi-spectral image data.

[0033] The quantum light sensing based residual stress intelligent detection method and system according to an embodiment of the present disclosure combines the light-force coupling effect of the quantum dot coating with a machine learning algorithm, and realizes visual detection of residual stress under complex working conditions based on multi-spectral dynamic imaging and environment data compensation, thereby realizing high-resolution, non-destructive and multi-material adaptive detection technology and providing an innovative technical idea for intelligent manufacturing and structural health. BRIEF DESCRIPTION OF DRAWINGS

[0034] Figure 1 FIG. 1 is a flowchart of a quantum light sensing based residual stress intelligent detection method according to an embodiment of the present disclosure;

[0035] Figure 2 FIG. 2 is a structural diagram of an acquisition module according to another embodiment of the present disclosure;

[0036] Figure 3A structural schematic diagram of a quantum light sensing-based residual stress intelligent detection system according to another embodiment of the present disclosure. DETAILED DESCRIPTION

[0037] The technical solutions in the embodiments of the present disclosure will be clearly and completely described in combination with the drawings in the embodiments of the present disclosure. Obviously, the described embodiments are only part of the embodiments of the present disclosure, rather than all the embodiments. Based on the embodiments in the present disclosure, all other embodiments obtained by a person of ordinary skill in the art without creative labor fall within the protection scope of the present disclosure.

[0038] In addition, the described features, structures, or characteristics can be combined in any suitable manner in one or more embodiments. In the following description, numerous specific details are provided to give a thorough understanding of embodiments of the present disclosure. One skilled in the relevant art will recognize, however, that the technical solutions of the present disclosure can be practiced without one or more of the specific details, or with other methods, components, devices, steps, etc. In other instances, well-known methods, devices, implementations, or operations are not shown or described in detail to avoid obscuring aspects of the present disclosure.

[0039] The flowcharts shown in the drawings are only exemplary illustrations, and do not necessarily include all contents and operations / steps, nor are they necessarily executed in the order described. For example, some operations / steps can be further decomposed, and some operations / steps can be combined or partially combined, so the actual execution order can be changed according to actual conditions.

[0040] It should be understood that although the terms first, second, third, etc. can be used in the present disclosure to describe various components, these components should not be limited by these terms. These terms are used to distinguish one component from another component. Therefore, the first component discussed below can be referred to as the second component without departing from the teachings of the present disclosure concept. As used in the present disclosure, the term "and / or" includes all combinations of any one and one or more of the associated listed items.

[0041] Those skilled in the art can understand that the drawings are only schematic diagrams of exemplary embodiments, and the modules or flows in the drawings are not necessarily required to implement the present disclosure, and therefore cannot be used to limit the protection scope of the present disclosure.

[0042] As Figure 1 shown, one embodiment of the present disclosure provides a quantum light sensing-based residual stress intelligent detection method, which comprises:

[0043] Step S1, acquiring multi-spectral image data and temperature and humidity data of a quantum dot coating on a surface of a region to be measured.

[0044] Specifically, the quantum dot coating is made of quantum dots and fluorescent materials, and the crystal lattice structure of the quantum dots changes under stress, causing quantum confinement effect, thereby providing signal data for residual stress detection technology. The coating material can adopt a core-shell structure, and the inner core material can be composed of CdSe, Inp and perovskite. CdSe has high fluorescence efficiency, Inp is used to form a cadmium-free environmental protection measure, and perovskite has high luminous intensity. The outer shell can be coated with ZnS material to reduce surface defects and improve quantum yield. The diameter of the quantum dots can be controlled to be 2nm-5nm. The base material can be epoxy resin, which can form a moderate stiffness difference with the metal substrate to be measured to ensure effective transmission of strain. Through the core layer structure and epoxy resin packaging, environmental noise can be resisted. In addition, when specific functions are required, appropriate amount of silica particles can be added to resist humidity, or UV absorbers can be added to resist light decay, etc.

[0045] The tensile stress and compressive stress can be identified and judged by the blue shift and red shift of the fluorescence wavelength. When the quantum dot lattice is stretched, the band gap energy decreases and the wavelength shortens, resulting in red shift, which is tensile stress. Conversely, it is compressive stress. The wavelength shift sensitivity of quantum dots can reach 0.1nm / MPa (the resistance change sensitivity of traditional strain gauges is about 0.01% / MPa, which is more than 10 times higher than that of traditional strain gauges). Through experiments, a stress-wavelength shift curve is established. By the wavelength shift of the quantum dot coating and the known stress standard test block on the same test block, a set of stress-wavelength shift data can be obtained. Through repeated operation, multiple sets of data are obtained, and the stress-wavelength shift curve is established.

[0046] Before residual stress detection, the surface of the area to be measured is cleaned to ensure that there is no rust, oil stains, etc. Then, electrostatic spraying can be used to realize uniform dispersion of the coating and accurate control of the thickness, and the quantum dot coating is sprayed onto the surface of the area to be measured. The coating thickness can be calculated based on the shear hysteresis model, and the thickness is controlled in the range of 25μm-35μm to ensure that the strain transmission efficiency is at a high level. At the same time, the optical fiber grating sensor is embedded in the coating, which acts as a temperature and humidity sensor. Finally, the coating material is allowed to cure by itself to form a quantum dot coating on the surface of the area to be measured.

[0047] The quantum dot coating of the embodiment has quantum confinement characteristics and high stress sensitivity. The change of the optical properties of the quantum dot material has a response relationship with the local stress. When the measured area has residual stress, the quantum dot lattice is deformed under the action of residual stress, causing a change in the band gap energy and a shift in the fluorescence wavelength. By capturing the wavelength change through a high-resolution hyperspectral camera, multi-spectral image data representing residual stress can be collected.

[0048] When performing residual stress detection, a quantum dot is excited from a ground state to a fluorescent state using a pulsed laser, the wavelength can be set to 405 nm to match the quantum dot absorption peak of CdSe, the power is adjusted according to actual needs, and the pulse width can be set to 10 ns to reduce thermal effects and avoid damage to the coating. Then, a hyperspectral camera is used to collect 256-band fluorescence signals, the spectral range is 400 nm-950 nm, the resolution is 5 nm, and the spatial resolution is 5 μm scanning imaging. A filter is installed on the hyperspectral camera to separate the quantum dot fluorescence from the ambient light (noise). At the same time, temperature, humidity and other environmental data are collected through the signal collector corresponding to the fiber grating sensor as compensation parameters for the residual stress detection model.

[0049] As a specific example, as shown in Figure 2 The above-mentioned pulsed laser 210, hyperspectral camera 220 and signal collector 230 can be integrated into an optical collection module 200 in the form of a cylinder. The collection module 200 has an aluminum alloy shell 240 and an optical window 250 on one side, which can be made of fused quartz glass and an antireflection film, has high light transmittance, low scattering loss and excellent thermal stability, has little effect on the signal, and the window diameter matches or is slightly smaller than the diameter of the collection module 200. The above-mentioned pulsed laser 210, hyperspectral camera 220 and signal collector 230 are built-in on the side of the optical window 250 of the collection module 200. When collecting data, the side of the optical window 250 of the collection module 200 is opposite to the quantum dot coating on the surface of the region to be measured, and is moved gradually in regions until the overall scanning and data collection of the region to be measured are completed. In addition, the collection module 200 also integrates a signal transceiver unit 260 for sending the collected multi-spectral image data and temperature and humidity data to the terminal device through wireless transmission. The signal transceiver unit 260 can be connected with a bamboo joint antenna 270 extending out of the aluminum alloy shell 240 to improve the stability of communication and the portability of the collection module 200.

[0050] Step S2, inputting the multi-spectral image data, the temperature and humidity data and the material performance parameters of the region to be measured into the residual stress detection model to obtain a residual stress distribution map of the region to be measured.

[0051] Specifically, a corresponding machine learning model, i.e. a residual stress detection model, is configured in the terminal device. The multi-spectral image data obtained by the hyperspectral camera is used as the physical data basis for detection, the temperature and humidity data fed back from the quantum dot coating are obtained by the fiber grating signal collector as the environmental compensation term, and the material performance parameters (such as elastic modulus, Poisson's ratio, etc.) are imported in advance as the material performance supplement term. The residual stress detection model automatically identifies stress-sensitive characteristic variables such as fluorescence main peak shift from the above data, establishes a mapping relationship between optical signals, surface strain and internal stress through deep learning, and finally obtains the residual stress of the region to be measured.

[0052] The residual stress detection model is pre-established based on training of a ResNet-50 model on finite element simulation data and standard specimen test data. The specific steps are as follows:

[0053] 1. Generate simulation data of different materials under different loads by a finite element model, and obtain test data by a standard specimen test. First, simulate the stress distribution of different materials under different loads by a finite element simulation software (such as ABAQUS) to generate more than 50,000 simulation data pairs, including stress distribution maps, strain distribution maps, and material performance parameters (elastic modulus, Poisson's ratio, etc.); second, perform a residual stress detection test on a standard specimen (such as an ASTM E837 standard specimen) by a quantum dot coating method to obtain a real stress distribution map, establish a correspondence between the real stress and the spectral signal, and obtain test data pairs, including multispectral image data, temperature and humidity data, material performance parameters, and real stress distribution maps. In addition, the same specimen can be tested for residual stress by drilling and quantum dot coating methods respectively to compare and verify the detection accuracy of the two methods.

[0054] 2. Mix the simulation data and the test data to obtain a residual stress detection dataset. The simulation data and the experimental data are mixed in a ratio of 9:1 to obtain a residual stress detection dataset including multispectral image data, temperature and humidity data, material performance parameters, and real stress distribution maps.

[0055] 3. Train the ResNet-50 model by taking the multispectral image data, temperature and humidity data, and material performance parameters as inputs and the real stress distribution map as output to obtain a residual stress detection model. The above residual stress detection dataset is input into the ResNet-50 model for training to obtain a preliminary residual stress detection model. The model architecture is as follows:

[0056] Input layer: spectrum x environment x material, input shape reference H x W x 260 (H and W are image sizes, 260 = 256 spectrum x 2 environment (temperature, humidity) x 2 material (elastic modulus, Poisson's ratio));

[0057] Main network: improved ResNet-50, input channel expanded from 3 (RGB) to 260, first layer convolution kernel adjusted to 7 x 7 x 260;

[0058] Attention module: CBAM (channel + spatial attention), inserted after each residual block of ResNet, dynamically focusing on key features;

[0059] Output layer: output stress distribution matrix (H x W x 1 (stress));

[0060] Relevant physical constraints: Hooke's law constraint (extracting local strain from hyperspectral image data and calculating the displacement of adjacent pixels to force the model prediction results to conform to the linear elastic constitutive relation), boundary condition constraint (adding constraint terms (predicted stress ≤ 10 MPa) to boundary regions without stress).

[0061] After obtaining the preliminary residual stress detection model through the above steps, actual environmental noise data can be collected through preliminary field tests for further optimization and training. Later, a real-time training and optimization approach based on collected data can be adopted to continuously improve the accuracy of the residual stress detection model. Additionally, transfer learning can be performed later: the basic model is first trained on common materials (such as steel and ceramics), the backbone network is fixed, and only the output layer and attention module are trained, allowing for fine-tuning of the model parameters.

[0062] In the later adjustment stage, the 256 bands captured by the hyperspectral camera can be reduced to the required characteristic bands (such as 500nm band and 480nm band) as needed, removing redundant band data and reducing the hardware burden.

[0063] Noise can be actively injected during model training. For example, stress distribution of complex-shaped parts can be simulated through geometric transformations (rotation, scaling) during ABAQUS simulations, and extreme working condition data (such as coating peeling, strong vibration interference) can be generated using GANs. By mixing normal data with extreme data (e.g., 80% normal data + 20% extreme data), the generated simulation data can be made closer to the real environment.

[0064] Step S3: Visualize and output the residual stress distribution map.

[0065] Specifically, the residual stress distribution map output by the residual stress detection model is presented on a visualization device, such as a monitor, so that relevant personnel can view the results of the intelligent residual stress detection.

[0066] This disclosure discloses a residual stress intelligent detection method based on quantum optical sensing. By combining the optical-mechanical coupling effect of quantum dot coatings with machine learning algorithms, and based on multispectral dynamic imaging and environmental data compensation, it realizes the visual detection of residual stress under complex working conditions. It achieves high-resolution, non-destructive, and multi-material adaptable detection technology, providing pioneering technical ideas for intelligent manufacturing and structural health.

[0067] like Figure 3 As shown, another embodiment of this disclosure provides a residual stress intelligent detection system based on quantum optical sensing, the system comprising:

[0068] The acquisition module 200 is used to acquire multispectral image data and temperature and humidity data of the quantum dot coating on the surface of the area to be tested; wherein, the quantum dot coating is made of quantum dots and fluorescent materials.

[0069] The detection module 300 is configured to input the multispectral image data, the temperature and humidity data, and the material performance parameter of the to-be-detected region into a residual stress detection model to obtain a residual stress distribution map of the to-be-detected region, wherein the residual stress detection model is pre-established by training a ResNet-50 model based on finite element simulation data and standard test specimen test data.

[0070] The output module 400 is configured to visually output the residual stress distribution map.

[0071] As shown in the example of FIG. 2, the acquisition module 200 includes: Figure 2

[0072] The pulsed laser 210 is configured to excite a fluorescence signal of the quantum dot coating.

[0073] The hyperspectral camera 220 is configured to acquire the fluorescence signal to obtain multispectral image data.

[0074] Specifically, the residual stress intelligent detection system based on quantum light sensing according to the embodiments of the present disclosure is used to implement the residual stress intelligent detection method based on quantum light sensing in the above embodiments, and the specific implementation process has been described in detail in the above embodiments, which will not be repeated here.

[0075] The residual stress intelligent detection system based on quantum light sensing according to the embodiments of the present disclosure combines the light-force coupling effect of the quantum dot coating with the machine learning algorithm, realizes the residual stress visual detection under complex working conditions based on multispectral dynamic imaging and environmental data compensation, and realizes the detection technology of high resolution, non-destructive, and multi-material adaptation, which provides an innovative technical idea for intelligent manufacturing and structural health.

[0076] It can be understood that the above embodiments are only exemplary embodiments adopted to illustrate the principles of the present disclosure, but the present disclosure is not limited thereto. Various modifications and improvements can be made by those of ordinary skill in the art without departing from the spirit and essence of the present disclosure, and these modifications and improvements are also considered to be within the protection scope of the present disclosure.​

Claims

1. A method for intelligent detection of residual stress based on quantum optical sensing, characterized in that, The method includes: Multispectral image data and temperature and humidity data of the quantum dot coating on the surface of the area to be tested are collected; wherein, the quantum dot coating is made of quantum dots and fluorescent materials composite; The multispectral image data, the temperature and humidity data, and the material property parameters of the area to be tested are input into the residual stress detection model to obtain the residual stress distribution map of the area to be tested; wherein, the residual stress detection model is pre-established based on the ResNet-50 model trained with finite element simulation data and standard specimen test data; The residual stress distribution map is visualized and output.

2. The method according to claim 1, characterized in that, Before acquiring multispectral image data and temperature and humidity data of the quantum dot coating on the surface of the area to be tested, the method further includes: A quantum dot coating is formed on the surface of the area to be tested, and a temperature and humidity sensor is embedded in the quantum dot coating.

3. The method according to claim 2, characterized in that, The process of forming a quantum dot coating on the surface of the area to be tested includes: The quantum dot coating material is uniformly sprayed onto the surface of the area to be tested; The quantum dot coating material is cured to form a quantum dot coating.

4. The method according to claim 2, characterized in that, The temperature and humidity sensor is a fiber optic grating sensor.

5. The method according to claim 1, characterized in that, The acquisition of multispectral image data of the area to be measured includes: The fluorescence signal of the quantum dot coating is excited using a pulsed laser; The fluorescence signal was acquired using a hyperspectral camera to obtain multispectral image data.

6. The method according to any one of claims 1 to 5, characterized in that, The residual stress detection model is pre-established through the following steps: Simulation data of different materials under different loads are generated by finite element modeling, and test data are obtained by standard specimen testing. By combining the simulation data and the experimental data, a residual stress detection dataset is obtained; the dataset includes multispectral image data, temperature and humidity data, material property parameters, and a true stress distribution map; Using multispectral image data, temperature and humidity data, and material performance parameters as inputs, and the real stress distribution map as output, the ResNet-50 model is trained to obtain a residual stress detection model.

7. The method according to claim 6, characterized in that, The training of the ResNet-50 model includes: Add Hooke's Law constraints and boundary condition constraints.

8. The method according to claim 6, characterized in that, The training of the ResNet-50 model includes: Update the ResNet-50 model parameters using one or more of the following: transfer learning, attention mechanism, and dynamic environment compensation mechanism.

9. A residual stress intelligent detection system based on quantum optical sensing, characterized in that, The system includes: The acquisition module is used to acquire multispectral image data and temperature and humidity data of the quantum dot coating on the surface of the area to be tested; wherein, the quantum dot coating is made of quantum dots and fluorescent materials. The detection module is used to input the multispectral image data, the temperature and humidity data, and the material property parameters of the area to be tested into the residual stress detection model to obtain the residual stress distribution map of the area to be tested; wherein, the residual stress detection model is pre-established based on the ResNet-50 model trained with finite element simulation data and standard specimen test data; The output module is used to visualize the residual stress distribution map.

10. The system according to claim 9, characterized in that, The acquisition module includes: A pulsed laser is used to excite the fluorescence signal of the quantum dot coating; A hyperspectral camera is used to acquire the fluorescence signal and obtain multispectral image data.

Images