Engineering plastic product internal bubble defect detection method based on laser scattering chromatography

Abstract

The invention relates to the technical field of plastic defect detection, in particular to an engineering plastic product internal bubble defect detection method based on laser scattering chromatography, which comprises the following steps of: acquiring a time domain scattering signal through a multi-angle scattering light receiving array, and performing dynamic compensation in combination with material thickness and environment temperature parameters; and the refractive index interference caused by temperature is eliminated. And based on the compensated scattering characteristic parameter set, constructing a three-dimensional scattering coefficient distribution diagram by adopting a chromatography reconstruction algorithm, visually displaying the internal density gradient change, and presenting a local high scattering coefficient value in a bubble defect area. And through adaptive threshold segmentation and bubble morphological feature screening, accurate identification and authenticity discrimination of a defect area are realized. And finally, a hyperboloid fitting model is adopted to quantify the bubble volume radius, geometric center and distribution density parameters, the product structural integrity is judged according to a preset safety threshold, and the detection precision and reliability are remarkably improved.

Description

Technical Field

[0001] This invention relates to the field of plastic defect detection technology, and in particular to a method for detecting internal bubble defects in engineering plastic products based on laser scattering tomography. Background Technology

[0002] In modern industrial manufacturing, engineering plastics are widely used in key structural components of automobiles, aerospace, electronics, and precision machinery due to their excellent mechanical properties, corrosion resistance, and processability. As product requirements for material performance and structural reliability continue to increase, internal bubble defects that may arise during injection molding and extrusion molding of engineering plastic products are increasingly becoming a key factor affecting their quality and service life. These micron- to millimeter-sized closed bubbles not only cause problems such as reduced apparent density and decreased mechanical strength, but may also lead to derivative defects such as stress concentration and fatigue crack initiation, and in severe cases, cause sudden failure of structural components during use.

[0003] Currently, conventional non-destructive testing techniques such as ultrasonic testing and X-ray tomography have been applied to some extent in the detection of internal defects in engineering plastics. However, ultrasonic testing is limited by the non-uniform acoustic impedance of the material and surface coupling conditions, resulting in low sensitivity for detecting microbubbles and difficulty in achieving high spatial resolution imaging. While X-ray testing has good penetration capabilities, it has high radiation safety requirements, expensive equipment, and limited contrast resolution for microbubbles in low atomic number materials (such as organic polymer plastics), making it difficult to meet the needs of online and rapid detection. In recent years, optical tomography and laser scattering technologies have shown potential in the detection of microstructures inside transparent or semi-transparent media due to their advantages of non-contact, high sensitivity, and high spatiotemporal resolution. However, existing optical methods are generally sensitive to environmental disturbances, especially in actual production environments, where temperature changes causing fluctuations in the refractive index of materials can significantly interfere with the stability of the scattering signal, leading to imaging distortion and misjudgment. In addition, traditional scattering imaging algorithms still have shortcomings in reconstruction accuracy, noise resistance, and the degree of automation in defect identification, making it difficult to effectively integrate the spatial localization, morphological characterization, and engineering safety criteria of microbubbles.

[0004] The above content is only used to help understand the technical solution of the present invention and does not represent an admission that the above content is prior art. Summary of the Invention

[0005] The main objective of this invention is to provide a method for detecting internal bubble defects in engineering plastic products based on laser scattering tomography, aiming to solve the technical problems of insufficient sensitivity to minute defects, susceptibility to environmental temperature interference, low three-dimensional imaging resolution, and weak automated discrimination capability in existing engineering plastic internal bubble defect detection technologies.

[0006] To achieve the above objectives, this invention provides a method for detecting internal bubble defects in engineering plastic products based on laser scattering tomography, the method comprising: S1. A modulated laser beam is emitted toward the engineering plastic product to be tested using a laser emitting device. At the same time, a multi-angle scattered light receiving array is used to synchronously collect a multi-angle time-domain scattered light signal sequence. The multi-angle time-domain scattered light signal sequence characterizes the intensity attenuation characteristics and time distribution characteristics of the laser pulse after being scattered inside the engineering plastic product at different incident angles. S2. Obtain the material thickness parameters and real-time ambient temperature parameters of the engineering plastic product, and perform dynamic compensation processing on the multi-angle time-domain scattered light signal sequence based on the material thickness parameters and real-time ambient temperature parameters to eliminate the interference effect of refractive index fluctuations caused by temperature changes on the scattered signal, and generate a set of compensated scattering characteristic parameters. S3. Based on the compensated scattering characteristic parameter set, a three-dimensional scattering coefficient distribution map of the interior of the engineering plastic product is constructed using a tomographic reconstruction algorithm. The three-dimensional scattering coefficient distribution map reflects the spatial variation characteristics of the density gradient inside the material. The internal bubble defect region is a local high value region with a scattering coefficient value that is significantly higher than that of the engineering plastic matrix material. S4. Implement an adaptive threshold segmentation algorithm on the three-dimensional scattering coefficient distribution map to obtain a set of candidate regions for bubble defects. Then, based on the bubble morphology feature screening rules, perform a true / false judgment on the set of candidate regions for bubble defects to determine the real bubble defect regions. S5. For the actual bubble defect area, a hyperboloid fitting model is used to calculate the volume radius parameter, geometric center coordinates and spatial distribution density parameter of each bubble defect. When the volume radius parameter of any bubble defect exceeds the preset structural safety threshold, it is determined that the engineering plastic product has internal bubble defects that affect the structural integrity.

[0007] Optionally, step S2 specifically includes: S21. Real-time acquisition of surface temperature distribution data of engineering plastic products using an infrared thermal imaging device; querying a pre-constructed temperature-refractive index perturbation mapping table based on the surface temperature distribution data to obtain the refractive index compensation coefficient corresponding to the temperature of each region. S22. For the scattered light signal corresponding to each incident angle, the refractive index compensation coefficient is applied to the scattered light intensity normalization process, and the effective optical path difference compensation amount is calculated in combination with the material thickness parameter of the engineering plastic product. S23. The normalized scattered light signal is fused with the effective optical path difference compensation amount to generate a compensated scattering characteristic parameter set containing scattering phase function parameters and attenuation length parameters.

[0008] Optionally, step S3 specifically includes: S31. Input the compensated scattering characteristic parameter set into the iterative tomography reconstruction model, wherein the iterative tomography reconstruction model has built-in prior constraints based on Monte Carlo scattering simulation. S32. By minimizing the mean square error between the actual measured scattering signal and the simulated scattering signal, the spatial distribution matrix of the scattering coefficient inside the engineering plastic product is iteratively inverted and solved to generate a high-resolution three-dimensional scattering coefficient distribution map. S33. Perform anisotropic diffusion filtering on the three-dimensional scattering coefficient distribution map to effectively suppress image noise generated during tomographic reconstruction while preserving the clarity of bubble edges.

[0009] Optionally, step S4 specifically includes: S41. A dynamic segmentation threshold is determined based on the statistical distribution characteristics of the scattering coefficient in the matrix region of the engineering plastic product. The dynamic segmentation threshold is adaptively adjusted according to the standard deviation of the scattering coefficient in the local region. S42. Perform a connected component labeling algorithm on the three-dimensional scattering coefficient distribution map to obtain an initial set of candidate regions for bubble defects; S43. For each candidate region in the initial set of candidate bubble defect regions, calculate its sphericity feature parameter and surface roughness feature parameter, and compare the calculation results with the pre-constructed bubble morphology feature database for verification. When the sphericity feature parameter is within a preset reasonable range and the surface roughness feature parameter is below the threshold, the candidate region is confirmed as a real bubble defect region.

[0010] Optionally, the process of constructing the bubble morphology feature database includes: B1. Collect bubble defect samples of known size and shape from standard engineering plastic specimens, and accurately measure the sphericity characteristic parameters and surface roughness characteristic parameters of each bubble defect. B2. Establish a hierarchical mapping table of bubble morphological feature parameters based on the bubble volume radius parameter, where small-sized bubbles correspond to the first sphericity range and large-sized bubbles correspond to the second sphericity range. B3. Establish a correlation between the bubble morphology feature parameters in the graded mapping table and the critical influence threshold of the mechanical properties of engineering plastic products to form a bubble morphology feature database for judging the authenticity of bubble defects.

[0011] Optionally, the following steps, performed after step S5, may also be included: S6. Generate a defect spatial distribution heat map based on the spatial distribution density of the actual bubble defect area, and evaluate the residual strength coefficient of the engineering plastic product based on the defect spatial distribution heat map. S7. When the residual strength coefficient is lower than the preset safety threshold, a defect analysis report is generated, which includes the spatial distribution characteristic parameters of bubble defects and corresponding suggestions for adjusting injection molding process parameters.

[0012] Optionally, step S1 specifically includes: S11. Obtain the material type parameters of the engineering plastic product to be tested, and query the pre-built material type-laser wavelength mapping table according to the material type parameters to determine the target wavelength range of the modulated laser beam applicable to the engineering plastic product. S12. Control the laser emitting device to emit a modulated laser beam within the target wavelength range, and synchronously capture multi-angle time-domain scattered light signal sequences using time-correlated single-photon counting technology.

[0013] Optionally, the pulse width of the modulated laser beam is set to a parameter value that matches the typical scattering timescale of bubbles inside the engineering plastic product to ensure that the scattered signal has sufficient time resolution.

[0014] Furthermore, to achieve the above objectives, the present invention also provides a laser scattering tomography-based device for detecting internal bubble defects in engineering plastic products. The device includes: a memory, a processor, and a laser scattering tomography-based program for detecting internal bubble defects in engineering plastic products, stored in the memory and executable on the processor. The laser scattering tomography-based program for detecting internal bubble defects in engineering plastic products is configured to implement the steps of the laser scattering tomography-based method for detecting internal bubble defects in engineering plastic products as described above.

[0015] Furthermore, to achieve the above objectives, the present invention also provides a medium storing a laser scattering tomography-based internal bubble defect detection program for engineering plastic products. When the laser scattering tomography-based internal bubble defect detection program for engineering plastic products is executed by a processor, it implements the steps of the laser scattering tomography-based internal bubble defect detection method for engineering plastic products as described above.

[0016] This invention provides a method for detecting internal bubble defects in engineering plastic products based on laser scattering tomography. The method effectively suppresses the interference of temperature fluctuations and material differences on the detection results through multi-angle time-domain scattered light signal acquisition and dynamic compensation processing. Combined with high-precision tomographic reconstruction and an intelligent discrimination algorithm based on morphological features, it achieves three-dimensional visualization and localization of internal bubble defects in engineering plastics, as well as authenticity identification. Furthermore, it uses a hyperboloid fitting model to accurately quantify and evaluate bubble parameters, significantly improving the detection sensitivity, imaging resolution, and reliability of small bubble defects. This method can accurately identify key bubble defects affecting structural safety, providing a high-precision, non-contact, and automated non-destructive testing method for quality control and process optimization of engineering plastic products. Attached Figure Description

[0017] Figure 1 This is a schematic flowchart of an embodiment of the method for detecting internal bubble defects in engineering plastic products based on laser scattering tomography according to the present invention.

[0018] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0019] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0020] Reference Figure 1 , Figure 1 This is a flowchart illustrating an embodiment of the method for detecting internal bubble defects in engineering plastic products based on laser scattering tomography according to the present invention.

[0021] In one embodiment, a method for detecting internal bubble defects in engineering plastic products based on laser scattering tomography is provided, the method comprising: Step S1: A modulated laser beam is emitted to the engineering plastic product to be tested using a laser emitting device. At the same time, a multi-angle scattered light receiving array is used to synchronously collect multi-angle time-domain scattered light signal sequences. The multi-angle time-domain scattered light signal sequences characterize the intensity attenuation characteristics and time distribution characteristics of the laser pulse after being scattered inside the engineering plastic product at different incident angles.

[0022] The laser emitting device can be an optical device for generating and directionally emitting a modulated laser beam. It can provide a controllable and repeatable incident light source to the engineering plastic product under test to excite an internal scattering response. In this embodiment, the laser emitting device can combine a laser and a modulator to output a pulsed or continuously modulated beam with a specific temporal structure. For example, the laser emitting device can be one or more of semiconductor lasers, fiber lasers, and solid-state lasers. The modulated laser beam can be a time- or frequency-modulated laser output with well-defined temporal characteristics, which can be used to enhance the temporal resolution of the scattering signal and facilitate the differentiation of scattering sources at different depths. The engineering plastic product under test can be a polymer structural component requiring internal bubble defect detection, which can be used as the object of laser scattering; its internal microstructure determines the characteristics of the scattering signal.

[0023] A multi-angle scattered light receiving array can be a signal acquisition system composed of multiple spatially distributed photodetector units. It can be used to capture multi-view temporal signals of laser light scattered by heterogeneous structures such as bubbles within a material, providing raw data for tomographic reconstruction. Furthermore, the multi-angle scattered light receiving array can simultaneously record the intensity and time information of scattered light by arranging high-sensitivity photodetectors at different orientations around the sample. In an exemplary embodiment, the multi-angle scattered light receiving array may include, but is not limited to, a ring photodiode array, a spherically distributed avalanche photodiode array, or a linear scanning multi-channel detector group. The multi-angle temporal scattered light signal sequence can be a time-varying sequence of scattered light intensity data recorded at multiple incident / received angles, which can be used to carry information on the position, size, and refractive index differences of heterogeneous structures (such as bubbles) within the material.

[0024] A modulated laser beam is emitted from a laser emitting device onto the engineering plastic product to be inspected. This can be achieved by activating the laser emitting device to output a laser beam with a specific modulation format and irradiate the sample surface. Furthermore, this operation can achieve precise timing synchronization by controlling the laser drive circuit and the modulation signal source, thereby exciting the characteristic scattering response generated by defects such as bubbles within the material. A multi-angle scattered light receiving array is used to synchronously acquire multi-angle temporal scattered light signal sequences. This can be achieved by detectors distributed at different spatial angles synchronously recording the changes in scattered light intensity over time during laser irradiation. In one specific embodiment, this operation can achieve strict time synchronization of all detection channels through hardware triggering; alternatively, software timestamps can be used to align the acquired data from each channel and perform post-processing synchronization, thereby obtaining high-dimensional scattering information containing both spatial and temporal dimensions and improving defect localization capabilities.

[0025] Step S2: Obtain the material thickness parameters and real-time ambient temperature parameters of the engineering plastic product. Based on the material thickness parameters and real-time ambient temperature parameters, perform dynamic compensation processing on the multi-angle time-domain scattered light signal sequence to eliminate the interference effect of refractive index fluctuations caused by temperature changes on the scattered signal, and generate a set of compensated scattering characteristic parameters.

[0026] The material thickness parameter can be the geometric thickness value of the engineering plastic product under test along the laser propagation path. This can be used to correct signal attenuation caused by optical path differences and assist in dynamic compensation processing. The real-time ambient temperature parameter can be the instantaneous temperature value of the environment in which the engineering plastic product is located during the testing process. This can be used to reflect the degree of material refractive index disturbance and serve as a key input variable for dynamic compensation. The material thickness parameter and real-time ambient temperature parameter of the engineering plastic product can be obtained by acquiring the thickness through a distance sensor or CAD model, and by reading the ambient temperature through a temperature sensor. Furthermore, this operation can be performed by using an infrared thermometer to non-contactly measure the sample surface temperature and combining it with a thermal conduction model to estimate the internal temperature. In an exemplary embodiment, the standard thickness value of the current batch of materials can also be matched using a pre-stored process parameter library, thereby providing the necessary physical parameters for dynamic compensation processing.

[0027] Dynamic compensation processing can be a signal preprocessing operation that corrects the original scattering signal based on material thickness and real-time temperature. It can be used to suppress the impact of temperature-induced refractive index fluctuations on the stability of the scattering signal. Dynamic compensation processing of multi-angle time-domain scattered light signal sequences based on material thickness parameters and real-time ambient temperature parameters can involve establishing a temperature-refractive index-scattering signal mapping model to perform inverse correction on the original signal. For example, this operation can use a lookup table method combined with interpolation for real-time signal compensation; in a specific embodiment, a neural network model can also be constructed to learn the relationship between temperature and signal distortion and perform compensation, thereby reducing signal drift caused by environmental disturbances and improving detection stability. Generating a set of compensated scattering characteristic parameters can be achieved by converting the dynamically compensated multi-angle time-domain signal into a standardized set of scattering feature vectors. Furthermore, this operation can form input data in a unified format suitable for tomographic reconstruction.

[0028] The compensated scattering characteristic parameter set can be a standardized data set reflecting the true scattering characteristics inside the material, obtained after dynamic compensation processing. It can be used to eliminate the interference of temperature fluctuations and material thickness differences on the original scattering signal, improving the accuracy of subsequent reconstruction and identification. In an exemplary embodiment, the compensated scattering characteristic parameter set can be used as input data for a tomographic reconstruction algorithm, directly affecting the quality of the three-dimensional scattering coefficient distribution map.

[0029] Step S3: Based on the compensated scattering characteristic parameter set, a three-dimensional scattering coefficient distribution map of the interior of the engineering plastic product is constructed using a tomographic reconstruction algorithm. The three-dimensional scattering coefficient distribution map reflects the spatial variation characteristics of the density gradient inside the material. Among them, the internal bubble defect region is a local high value region with a scattering coefficient value that is significantly higher than that of the engineering plastic matrix material.

[0030] The tomographic reconstruction algorithm can be a mathematical calculation method that inverts multi-angle projection data into an internal three-dimensional distribution, and can be used to map external measurement signals to the spatial distribution of internal scattering coefficients. Further, the tomographic reconstruction algorithm can include, but is not limited to, filtered back-projection algorithms, iterative algebraic reconstruction algorithms, and maximum entropy reconstruction algorithms. The three-dimensional scattering coefficient distribution map can be a spatial three-dimensional image generated by the tomographic reconstruction algorithm, characterizing the scattering ability of each point inside the material. It can be used to visualize changes in internal density gradients, making bubble defects appear as local high-value regions. For example, the three-dimensional scattering coefficient distribution map can include, but is not limited to, voxelized scattering intensity maps, continuous-field scattering coefficient models, and piecewise constant approximate distribution maps. The scattering coefficient value can be a physical quantity characterizing the ability of a material per unit volume to scatter incident light, and can be used as a criterion to distinguish bubbles from the matrix material; bubble regions exhibit high values ​​due to abrupt changes in refractive index. The engineering plastic matrix material can be a homogeneous polymer phase constituting the main body of the product, and can be used as a scattering background, with a scattering coefficient value lower than that of the bubble defect region. Local high-value regions can be areas where the scattering coefficient is significantly higher than that of the surrounding matrix in a three-dimensional scattering coefficient distribution map, and can be used to indicate the location of potential bubble defects.

[0031] Based on the compensated scattering characteristic parameter set, a tomographic reconstruction algorithm is applied to construct a three-dimensional scattering coefficient distribution map inside engineering plastic products. This can be achieved by using multi-angle scattering data as projection input and calculating the three-dimensional distribution of internal scattering coefficients through an inversion algorithm. In one specific embodiment, this operation can use a filtered back-projection algorithm to quickly generate initial reconstruction results; furthermore, iterative optimization algorithms combined with regularization constraints can be used to improve reconstruction accuracy, thereby enabling three-dimensional visualization imaging of internal bubble defects.

[0032] Step S4: Implement an adaptive threshold segmentation algorithm on the three-dimensional scattering coefficient distribution map to obtain a set of candidate regions for bubble defects. Then, based on the bubble morphology feature screening rules, perform a true / false judgment on the set of candidate regions for bubble defects to determine the real bubble defect regions.

[0033] The adaptive threshold segmentation algorithm can be an image processing method that automatically determines the segmentation threshold based on the local statistical characteristics of the image. It can be used to extract candidate defect regions from the three-dimensional scattering coefficient distribution map, adapting to different noise levels in different regions. The bubble defect candidate region set can be a set of potential defect regions extracted from the three-dimensional scattering coefficient distribution map after adaptive threshold segmentation. It can be used to initially screen voxel regions that may contain bubbles, narrowing the calculation range for subsequent true / false discrimination. In an exemplary embodiment, the bubble defect candidate region set can be used as the input object for bubble morphological feature screening rules, and its completeness affects the detection rate of the final real bubble defect region. The bubble morphological feature screening rules can be a set of discrimination criteria based on the geometric and topological characteristics of bubbles. They can be used to exclude false defects caused by noise or non-bubble heterogeneous bodies, improving recognition accuracy. Furthermore, the bubble morphological feature screening rules can include, but are not limited to, sphericity constraint rules, connectivity verification rules, edge gradient consistency rules, etc. The real bubble defect region can be the three-dimensional spatial region corresponding to the bubble defect that has been confirmed as physically existing after morphological feature screening. It can be used as the processing object of the hyperboloid fitting model to accurately quantify the bubble geometric parameters. For example, real bubble defect areas may include, but are not limited to, isolated spherical bubble areas, connected clustered bubble areas, and interface-attached bubble areas.

[0034] An adaptive threshold segmentation algorithm is applied to the three-dimensional scattering coefficient distribution map to obtain a set of candidate regions for bubble defects. This can be achieved by automatically setting a segmentation threshold based on the statistical characteristics of the local scattering coefficients and extracting connected regions that exceed the threshold. In one specific embodiment, the Otsu method can be used to calculate the optimal threshold within a local window; furthermore, a Gaussian mixture model can be used to fit the background and foreground distributions and determine the segmentation points, thereby initially separating potential defect regions and reducing subsequent computational load. The set of candidate regions for bubble defects is then used to determine whether they are genuine or not based on bubble morphological feature screening rules. This can be achieved by calculating morphological indicators (such as sphericity and surface smoothness) for each candidate region and comparing them with preset rules. For example, a support vector machine can be used to classify and distinguish multi-dimensional morphological features; in one exemplary embodiment, a rule engine can be used to sequentially verify sphericity, volume rationality, and edge gradient consistency, thereby eliminating false defects and improving recognition accuracy.

[0035] Step S5: For the actual bubble defect area, use a hyperboloid fitting model to calculate the volume radius parameter, geometric center coordinates and spatial distribution density parameter of each bubble defect. When the volume radius parameter of any bubble defect exceeds the preset structural safety threshold, it is determined that the engineering plastic product has internal bubble defects that affect the structural integrity.

[0036] The hyperboloid fitting model can be a mathematical surface fitting method used for geometric modeling of real bubble defect regions, enabling high-precision parameterized descriptions of bubble volume radius, geometric center, and spatial orientation. Furthermore, the hyperboloid fitting model can fit discrete data points to a standard hyperboloid equation using least squares or maximum likelihood estimation. For example, the hyperboloid fitting model can include, but is not limited to, single-leaf hyperboloid models, two-leaf hyperboloid models, and hyperboloid of revolution models. The volume radius parameter can be an equivalent bubble radius index obtained by back-calculating from the hyperboloid fitting model, which can be used to assess the potential impact of bubbles on structural strength. The geometric center coordinates can be the coordinates of the bubble defect's center point in three-dimensional space, supporting defect location and distribution analysis. The spatial distribution density parameter can be the number or volume percentage of real bubble defects per unit volume, reflecting the overall defect level of the material and assisting in process quality assessment. The preset structural safety threshold can be an upper limit value for the bubble volume radius set according to engineering application requirements, serving as a basis for determining whether the product is qualified.

[0037] For real bubble defect areas, a hyperboloid fitting model is used to calculate the volume radius parameters, geometric center coordinates, and spatial distribution density parameters of each bubble defect. This can be achieved by fitting the voxel point set of the real bubble region to a hyperboloid equation and then analyzing the geometric parameters. In one specific embodiment, this operation can use the Levenberg-Marquardt algorithm for nonlinear least squares fitting; furthermore, principal component analysis can be performed first to determine the principal axis direction, and then a rotated hyperboloid fitting can be performed, thereby achieving accurate quantitative characterization of bubble defects. When the volume radius parameter of any bubble defect exceeds a preset structural safety threshold, the engineering plastic product is determined to have internal bubble defects that affect structural integrity. This can be achieved by comparing the calculated volume radius parameter with the preset threshold; if it exceeds the limit, it is marked as unqualified, thus completing a closed loop from detection data to engineering safety determination.

[0038] Taking the online quality inspection of injection-molded automotive engine hood liners as an example, the laser scattering tomography-based method for detecting internal bubble defects in engineering plastic products in this embodiment can be implemented by deploying the detection system at the end of the injection molding production line. The laser emitting device emits a modulated laser beam onto the newly demolded polyamide 66 engine hood liner, and a multi-angle scattered light receiving array synchronously collects the scattered signals. The system acquires the product thickness (from mold parameters) and workshop temperature (from environmental sensors) in real time and performs dynamic compensation on the signals. A three-dimensional scattering map is obtained through tomographic reconstruction, and 12 candidate regions are adaptively segmented. Morphological screening confirms that 3 of them are real bubbles. Hyperbolic surface fitting shows that the maximum bubble volume radius is 0.42 mm, which exceeds the preset safety threshold of 0.4 mm. The system automatically determines that the part is unqualified and triggers the rejection mechanism.

[0039] In one embodiment, step S2 specifically includes: S21. Real-time acquisition of surface temperature distribution data of engineering plastic products using infrared thermal imaging device; querying a pre-built temperature-refractive index perturbation mapping table based on surface temperature distribution data; obtaining the refractive index compensation coefficient corresponding to the temperature of each region. The infrared thermal imaging device can be a thermal radiation imaging device used for non-contact acquisition of the surface temperature distribution of an object. It can be used to capture the spatial temperature distribution of the product surface in real time, providing input for regional refractive index perturbation compensation. In this embodiment, the infrared thermal imaging device can detect the infrared radiation intensity emitted by the surface of the engineering plastic product, convert it into a corresponding temperature value, and generate a two-dimensional temperature field image. For example, the infrared thermal imaging device can be one or more of the following, including but not limited to a microbolometer-type thermal imager, a quantum well infrared photodetector array, and a cooled indium antimonide thermal imaging system. The surface temperature distribution data can be two-dimensional spatial data reflecting the temperature values ​​at various points on the surface of the engineering plastic product, which can be used as input for querying the temperature-refractive index perturbation mapping table, supporting regional dynamic compensation.

[0040] The temperature-refractive index perturbation mapping table can be a data mapping table established in advance through calibration experiments, characterizing the refractive index variation of engineering plastics at different temperatures. It can be used to quickly convert measured surface temperatures into corresponding refractive index compensation coefficients, achieving quantitative correction of physical perturbations. In an exemplary embodiment, the temperature-refractive index perturbation mapping table can be fitted into a lookup function by simultaneously measuring the temperature and refractive index at multiple points on a standard sample in a controlled temperature chamber. Furthermore, the temperature-refractive index perturbation mapping table can include, but is not limited to, piecewise linear interpolation tables, polynomial fitting parameter tables, and neural network weight mapping tables. The refractive index compensation coefficient can be a refractive index correction factor obtained from the mapping table based on the local temperature, used to correct the scattering signal in that region. It can be used to compensate for local refractive index shifts caused by temperature fluctuations, improving the accuracy of scattering signal normalization.

[0041] Real-time acquisition of surface temperature distribution data of engineering plastic products using an infrared thermal imaging device can be achieved by activating the device, pointing it at the product, acquiring its surface thermal radiation image, and converting it into a temperature matrix. Furthermore, this operation can be improved by using frame averaging to suppress thermal noise and enhance temperature field stability, or by combining background radiation correction algorithms to eliminate environmental reflection interference, thereby obtaining a spatially resolved surface temperature field that supports regionalized refractive index compensation.

[0042] By querying a pre-constructed temperature-refractive index perturbation mapping table based on surface temperature distribution data, the refractive index compensation coefficient corresponding to the temperature of each region can be obtained. This can be achieved by using the temperature value of each pixel as an index to look up or interpolate the corresponding refractive index compensation coefficient in the mapping table. In a specific embodiment, this operation can estimate the coefficient corresponding to the intermediate temperature between discrete calibration points using bilinear interpolation, or call a pre-trained small regression model to replace the table lookup to improve continuity. This enables the conversion of the physical quantity from temperature to refractive index perturbation, providing a basis for signal correction.

[0043] S22. For the scattered light signal corresponding to each incident angle, the refractive index compensation coefficient is applied to the scattered light intensity normalization process, and the effective optical path difference compensation is calculated in combination with the material thickness parameter of the engineering plastic product. The scattered light signal corresponding to the incident angle can be the original scattered light timing signal collected by a specific angle channel in a multi-angle scattered light receiving array. This signal can be used as a basic processing unit for refractive index compensation and optical path correction, and normalized independently by angle. The scattered light intensity normalization process can be a signal preprocessing procedure that corrects the amplitude of the original scattered light signal according to the local refractive index compensation coefficient. This can be used to eliminate scattering intensity distortion caused by refractive index fluctuations and improve the consistency of multi-angle signals. The effective optical path difference compensation amount can be a compensation parameter calculated by combining material thickness and local refractive index, used to correct differences in the laser propagation path within the material. This can be used to correct optical path differences caused by the combined effects of thickness and refractive index in different regions, improving the alignment accuracy of the time-domain signal.

[0044] For the scattered light signal corresponding to each incident angle, the refractive index compensation coefficient is applied to the scattered light intensity normalization process. This can be achieved by scaling the amplitude of the scattered signal for each angle channel according to the refractive index compensation coefficient of its corresponding irradiated area. Furthermore, this operation can determine the compensation coefficient for the channel by weighted averaging of the temperature in the neighborhood of the laser incident point, or by applying a dynamic compensation factor corresponding to the local temperature point-by-point in the time domain. This can correct the scattering intensity deviation caused by thermally induced refractive index changes and improve the consistency of multi-angle data. The effective optical path difference compensation is calculated by combining the material thickness parameters of the engineering plastic product. This can be achieved by multiplying the local refractive index by the material thickness to calculate the actual optical path of light propagation in the region and comparing it with the reference optical path to obtain the compensation amount. In an exemplary embodiment, this operation can be achieved by using a ray tracing model to calculate the actual optical path under non-perpendicular incidence, or by fusing the thickness parameters with the temperature field to generate a three-dimensional effective optical path distribution map. This can correct the time delay drift caused by the thermal-thickness coupling effect and improve the temporal resolution.

[0045] S23. The normalized scattered light signal is fused with the effective optical path difference compensation amount to generate a compensated scattering characteristic parameter set containing scattering phase function parameters and attenuation length parameters.

[0046] The normalized scattered light signal can be refractive index compensated and amplitude corrected time-series scattered light data, which can be used as the basis signal for fusion with the effective optical path difference compensation amount to construct a high-fidelity parameter set. The scattering phase function parameter can be a function parameter describing the spatial angular distribution characteristics of the scattered light, reflecting the microscopic geometry of the defect. It can be used as a key component of the compensated scattering characteristic parameter set, supporting directional modeling in subsequent tomographic reconstruction. The attenuation length parameter can be a characteristic length quantity characterizing the intensity attenuation rate of laser propagation in a material. It can be used to reflect the combined absorption and scattering effects within the material, and to correct depth-dependent signal attenuation.

[0047] The normalized scattered light signal is fused with the effective optical path difference compensation to generate a compensated set of scattering characteristic parameters, including scattering phase function parameters and attenuation length parameters. This can be achieved by jointly processing the amplitude-normalized signal with the time delay correction to extract scattering directionality and attenuation characteristic parameters. Furthermore, this operation can separate phase and amplitude information using wavelet transform and fuse the compensation separately, or construct a joint optimization objective function to simultaneously estimate the phase function and attenuation length. This allows for the generation of a high-fidelity, interference-resistant set of scattering characteristic parameters, laying the foundation for high-precision tomographic reconstruction.

[0048] For example, in the scenario of online thermal inspection of polyetheretherketone (PEEK) structural components for aerospace applications, the method for detecting internal bubble defects in engineering plastic products based on laser scattering tomography in this embodiment can be as follows: During the cooling stage after injection molding, an infrared thermal imaging device captures the surface temperature distribution of the PEEK component in real time, showing a local temperature difference of 15°C; the system queries the pre-stored PEEK temperature-refractive index mapping table and assigns refractive index compensation coefficients of 1.023 for the high-temperature region and 1.018 for the low-temperature region; the multi-angle scattering signals are normalized according to the corresponding regional coefficients, and the effective optical path difference of each path is calculated in combination with the 3.2mm thickness parameter; the scattering characteristic parameter set generated after fusion contains the accurate forward scattering dominant phase function and a 1.8mm attenuation length; subsequent tomographic reconstruction successfully identifies a microbubble with a diameter of 0.15mm, while the defect in the traditional uncompensated method is submerged in artifacts due to the thermal gradient.

[0049] In one embodiment, step S3 specifically includes: S31. Input the compensated scattering characteristic parameter set into the iterative tomography reconstruction model. The iterative tomography reconstruction model has built-in prior constraints based on Monte Carlo scattering simulation. The iterative tomographic reconstruction model can be a computational framework for tomographic imaging that approximates the true internal structure through multiple iterations of optimization. It can be used to improve reconstruction accuracy and detail fidelity under conditions of limited viewing angle and low signal-to-noise ratio. In this embodiment, the iterative tomographic reconstruction model can take the compensated set of scattering characteristic parameters as input, combine it with physical prior constraints, and gradually correct the scattering coefficient distribution estimate through an error feedback mechanism. Furthermore, the iterative tomographic reconstruction model can include, but is not limited to, one or more of the following: gradient descent-based iterative reconstruction model, conjugate gradient least squares reconstruction model, and fully variational regularized iterative model.

[0050] The prior constraints of Monte Carlo scattering simulation can be statistical laws obtained by simulating the propagation and scattering process of laser light in engineering plastics using the Monte Carlo method. These constraints serve as physical constraints for the reconstruction model and can be used to embed the physical mechanisms of light transmission within the material into the reconstruction process, enhancing sensitivity and geometric consistency for weakly scattering targets (such as microbubbles). In an exemplary embodiment, the prior constraints of Monte Carlo scattering simulation can be generated by simulating a large number of random photon trajectories to create a scattering response database or probability distribution function, given known material optical parameters (such as absorption coefficient and anisotropy factor). Exemplarily, the prior constraints of Monte Carlo scattering simulation can include steady-state Monte Carlo priors, time-domain impulse response Monte Carlo priors, and polarization-resolved Monte Carlo priors. Inputting the compensated scattering characteristic parameter set into the iterative tomographic reconstruction model can be achieved by passing a dataset containing scattering phase function parameters and attenuation length parameters as the initial observation input to the reconstruction model. Furthermore, this operation can be further enhanced by initiating a high-precision reconstruction process that integrates physical priors, ensuring that the reconstruction process possesses material specificity and physical consistency.

[0051] The iterative tomographic reconstruction model incorporates prior constraints based on Monte Carlo scattering simulations. These constraints can be embedded in the model's forward operators or regularization terms, using statistical laws of light scattering obtained from Monte Carlo simulations. In one specific embodiment, this operation can be achieved by constructing a lookup table from the Monte Carlo simulation results and calling it for forward prediction in each iteration; or by training an alternative neural network surrogate model to approximate the Monte Carlo response, thereby accelerating the iteration process and ensuring that the reconstruction process conforms to the physical mechanism of light transmission within the material, thus improving sensitivity to subtle defects.

[0052] S32. By minimizing the mean square error between the actual measured scattering signal and the simulated scattering signal, the spatial distribution matrix of the scattering coefficient inside the engineering plastic product is iteratively inverted and solved to generate a high-resolution three-dimensional scattering coefficient distribution map. The actual measured scattering signal can be measured scattering data acquired by a multi-angle scattering light receiving array and dynamically compensated. It can be used as a target reference for iterative reconstruction, driving the optimization process. The simulated scattering signal can be the theoretical scattering response calculated forward based on the currently estimated scattering coefficient distribution and the Monte Carlo light transmission model. It can be compared with the actual measured scattering signal, forming the error feedback basis for iterative updates. The mean square error can be the average of the squared differences between the actual measured scattering signal and the simulated scattering signal at each point. It can be used as the objective function for iterative optimization, minimizing which drives the reconstruction result to approximate the real internal structure. The spatial distribution matrix of the internal scattering coefficient of the engineering plastic product can be a three-dimensional array represented in voxel form, with each element corresponding to the estimated scattering coefficient value at a certain point in space. It can be used as the solution objective for iterative inversion, ultimately transforming into a high-resolution three-dimensional scattering coefficient distribution map. The high-resolution three-dimensional scattering coefficient distribution map can be a three-dimensional image of the internal scattering coefficient with higher spatial detail fidelity output after iterative optimization and physical constraint reconstruction. It can be used to clearly present the spatial position and boundary morphology of micron-sized bubbles. In one exemplary embodiment, a high-resolution three-dimensional scattering coefficient distribution map is used as the direct input to the anisotropic diffusion filtering process, and its initial quality determines the edge preservation effect after filtering.

[0053] By minimizing the mean square error between the actual measured scattering signal and the simulated scattering signal, the spatial distribution matrix of the scattering coefficients inside the engineering plastic product is iteratively inverted. This can be achieved by calculating the simulated signal corresponding to the currently estimated distribution in each iteration, comparing it with the measured signal, and then updating the scattering coefficient matrix. Furthermore, this operation can employ the Levenberg-Marquardt algorithm to accelerate the minimization of the mean square error, or introduce a total variation regularization term to prevent overfitting and improve the stability of the solution. This allows for adaptive approximation of the true internal structure, generating a high-fidelity scattering coefficient distribution. Generating a high-resolution three-dimensional scattering coefficient distribution map can be achieved by visualizing the converged spatial distribution matrix of the scattering coefficients as a three-dimensional volumetric rendering image, thus providing high-contrast, high-detail images suitable for defect identification.

[0054] S33. Perform anisotropic diffusion filtering on the three-dimensional scattering coefficient distribution map to effectively suppress image noise generated during tomographic reconstruction while preserving the clarity of bubble edges.

[0055] Anisotropic diffusion filtering can be a nonlinear filtering method that adaptively controls the smoothing intensity based on the local gradient direction of the image. It can effectively suppress speckle noise or iterative artifacts introduced by tomographic reconstruction while maintaining the edge sharpness of bubble defects. In a specific embodiment, anisotropic diffusion filtering can control the diffusion coefficient based on the local gradient tensor, suppressing smoothing in the tangential direction and enhancing smoothing in the normal direction. For example, anisotropic diffusion filtering can include Perona-Malik diffusion filtering, coherent enhancement diffusion filtering, and anisotropic diffusion guided by the structure tensor. The bubble edge sharpness can be the steepness of the scattering coefficient gradient at the bubble defect boundary in the three-dimensional scattering coefficient distribution map, which directly affects the accuracy of subsequent threshold segmentation and morphological discrimination, and is a key indicator of image quality. Image noise can be non-structural intensity fluctuations introduced during tomographic reconstruction due to insufficient data, model simplification, or measurement errors. It can be used to describe interference components that may be misjudged as minor defects and needs to be effectively suppressed through filtering.

[0056] Anisotropic diffusion filtering is applied to the 3D scattering coefficient distribution map to effectively suppress image noise generated during tomographic reconstruction while preserving the sharpness of bubble edges. This can be achieved by controlling the diffusion coefficient based on the local gradient tensor, suppressing smoothing in the tangential direction and enhancing smoothing in the normal direction at the edges. Furthermore, this operation can employ an explicit finite difference scheme to implement the 3D anisotropic diffusion equation, or use a fast Fourier transform to accelerate the calculation of the structure tensor and the filtering iteration, thereby maintaining sharp bubble boundaries while denoising and avoiding the detail blurring caused by traditional filtering.

[0057] Taking the inspection of micro-injection molded parts of liquid crystal polymer (LCP) for electronic connectors as an example, the method for detecting internal bubble defects in engineering plastic products based on laser scattering tomography in this embodiment can be as follows: The system inputs the dynamically compensated scattering characteristic parameter set into an iterative reconstruction model with built-in Monte Carlo priors; the model uses the known optical parameters of LCP material to perform photon transmission simulation and construct prior constraints; after 12 iterations, the mean square error decreases to 8% of the initial value, generating a three-dimensional scattering map with a resolution of 50μm; then, a Perona-Malik anisotropic diffusion filter is applied, reducing the standard deviation of image noise by 62%, while the edge gradient preservation rate of the 0.1mm bubble exceeds 90%; the subsequent segmentation algorithm successfully identifies 3 real microbubbles, while the traditional FBP reconstruction misses 2 of them due to artifact interference.

[0058] This embodiment inputs the compensated scattering characteristic parameter set into an iterative tomographic reconstruction model with built-in Monte Carlo scattering simulation prior constraints. By minimizing the mean square error between the actual measured scattering signal and the simulated scattering signal, the spatial distribution matrix of the scattering coefficient inside the engineering plastic product is solved through iterative inversion, and a high-resolution three-dimensional scattering coefficient distribution map is generated. Then, anisotropic diffusion filtering is performed on the map to effectively suppress reconstruction noise while preserving the clarity of bubble edges. This achieves the technical effect of integrating physical mechanism priors and nonlinear optimization, significantly improving reconstruction accuracy and noise resistance, and providing a high-fidelity three-dimensional image basis for defect identification.

[0059] In one embodiment, step S4 specifically includes: S41. The dynamic segmentation threshold is determined based on the statistical distribution characteristics of the scattering coefficient in the matrix region of the engineering plastic product. The dynamic segmentation threshold is adaptively adjusted according to the standard deviation of the scattering coefficient in the local region. The statistical distribution characteristics of the scattering coefficient in the matrix region of the engineering plastic product can be the statistical characteristics such as the mean and variance of the scattering coefficient exhibited by the defect-free engineering plastic body in the three-dimensional scattering map. These can be used as a benchmark for dynamic threshold calculation, reflecting the local background scattering level. Furthermore, the dynamic segmentation threshold can be an adaptively determined segmentation critical value based on local background scattering characteristics, used to distinguish between bubbles and the matrix. This can be used to adapt to material inhomogeneity and noise fluctuations, improve the detection rate of microbubbles, and reduce over-segmentation. In an exemplary embodiment, the dynamic segmentation threshold can be dynamically calculated based on the statistical distribution of the scattering coefficient in the engineering plastic matrix region, combined with the local standard deviation. For example, the dynamic segmentation threshold can include, but is not limited to, one or more of the following: a threshold based on a local mean-weighted standard deviation, a sliding window Otsu adaptive threshold, and a multi-scale variance normalization threshold. The standard deviation of the scattering coefficient in the local region can be a measure of the dispersion of the scattering coefficient in the neighborhood centered at a certain point in the three-dimensional scattering coefficient distribution map. This can be used to characterize the intensity of local noise or material fluctuations and to adjust the sensitivity of the dynamic segmentation threshold.

[0060] Determining a dynamic segmentation threshold based on the statistical distribution characteristics of the scattering coefficient in the matrix region of engineering plastic products can be achieved by analyzing the scattering coefficient distribution in defect-free areas and setting an initial threshold benchmark. Further, this operation can be implemented by estimating the local matrix mean and standard deviation using a sliding window, or by extracting the background matrix region using morphological opening operations followed by statistical analysis, thereby establishing a segmentation benchmark adapted to the material's background characteristics. The dynamic segmentation threshold is adaptively adjusted based on the standard deviation of the scattering coefficient in the local region. This can be achieved by increasing the threshold in high-noise areas and decreasing it in stable areas, dynamically setting the threshold through a linear combination of the local mean and standard deviation. In a specific embodiment, this operation can be implemented by using variable weights to adjust sensitivity with depth decay, or by combining multi-scale standard deviation fusion decision-making, thereby improving sensitivity to weak bubble signals while suppressing false noise detections.

[0061] S42. Perform a connected component labeling algorithm on the three-dimensional scattering coefficient distribution map to obtain an initial set of candidate regions for bubble defects. The connected component labeling algorithm can be a region segmentation method that groups adjacent voxels in a 3D image that meet a threshold condition into regions with the same label. This can be used to extract a structured initial set of candidate bubble defect regions from the segmentation results. In an exemplary embodiment, the connected component labeling algorithm can employ one or more of the following: six-neighbor connected component labeling, twenty-six-neighbor flood filling algorithm, and fast connected component analysis optimized by disjoint-set data structure. The initial set of candidate bubble defect regions can be a set of 3D regions that may contain real bubbles, obtained after dynamic thresholding and connected component labeling, and can be used as input for morphological authenticity determination.

[0062] A connected component labeling algorithm is applied to the 3D scattering coefficient distribution map to obtain an initial set of candidate bubble defect regions. This can be achieved by clustering voxels above a dynamic threshold according to their spatial adjacency and assigning them unique labels. Furthermore, this operation can be implemented using GPU-accelerated parallel connected component labeling or by introducing minimum volume constraints to eliminate individual voxel noise points, thereby generating a structured, indexable list of candidate defect regions.

[0063] S43. For each candidate region in the initial set of candidate bubble defect regions, calculate its sphericity feature parameter and surface roughness feature parameter, and compare the calculation results with the pre-built bubble morphology feature database for verification. When the sphericity feature parameter is within a preset reasonable range and the surface roughness feature parameter is below the threshold, the candidate region is confirmed as a real bubble defect region.

[0064] The sphericity feature parameter can be a dimensionless index that quantifies how close the geometry of a candidate region is to an ideal sphere. It can be used to distinguish near-spherical bubbles formed by surface tension from irregular impurities or artifacts. For example, the sphericity feature parameter may include, but is not limited to, one or more of the following: volume-to-surface-area ratio sphericity, principal component ellipsoid fitting eccentricity, and Fourier descriptor spherical similarity. The surface roughness feature parameter can be a statistical measure characterizing the degree of microscopic undulations on the boundary surface of the candidate region. It can be used when the inner wall of a real bubble is smooth, and its roughness is significantly lower than that of noise or cracks. In a specific embodiment, the surface roughness feature parameter may include, but is not limited to, one or more of the following: root mean square gradient roughness, Laplace energy response roughness, and normal vector variance roughness. The bubble morphology feature database can be a reference knowledge base pre-constructed through statistical analysis of a large number of real bubble samples, containing typical sphericity and surface roughness distribution ranges. It can be used to provide quantitative comparison criteria for judging the authenticity of candidate regions and support automated identification. Furthermore, the bubble morphology feature database may include, but is not limited to, one or more of the following: a bubble feature library specifically for polymer materials, a bubble morphology sub-library categorized by process type, and a multi-scale surface texture feature template library. The preset reasonable range may be an allowable sphericity range set based on the bubble morphology feature database, which can be used as a basis for determining whether a candidate region conforms to bubble geometry characteristics. The surface roughness threshold may be an upper limit value for surface roughness set based on database statistics, which can be used to exclude high-roughness non-bubble interference regions.

[0065] For each candidate region in the initial set of candidate bubble defect regions, its sphericity and surface roughness feature parameters are calculated. This can be achieved by extracting geometric and surface differential attributes for each marked region to quantify its morphological characteristics. Furthermore, this operation can be implemented by reconstructing the surface mesh using MarchingCubes and then calculating curvature statistics, or by directly estimating sphericity using spherical harmonic function expansion in the voxel domain, thus transforming geometric intuition into discriminable numerical features. The calculation results are compared and verified with a pre-built bubble morphology feature database. This can be done by querying the sphericity and roughness distribution boundaries for the corresponding material type in the database to determine whether the current parameters fall within a reliable range. In an exemplary embodiment, this operation can be implemented by using Mahalanobis distance to measure the similarity between the feature vector and the bubble class center, or by calling a lightweight classifier (such as a random forest) instead of a hard threshold comparison, enabling intelligent authenticity discrimination based on physical origin. When the sphericity feature parameter is within a preset reasonable range and the surface roughness feature parameter is below the threshold, the candidate region is confirmed as a real bubble defect region. This can be achieved by executing logic and judgment: if the sphericity falls within the allowable range and the roughness is below the upper limit, it is marked as a real bubble, thus efficiently filtering out non-bubble interference and outputting a high-confidence defect region.

[0066] Taking the quality inspection of polycarbonate (PC) injection molded parts for precision electronic casings as an example, the method for detecting internal bubble defects in engineering plastic products based on laser scattering tomography in this embodiment can be as follows: systematically analyze the scattering coefficient of the PC matrix region, with the local standard deviation varying between 0.03 and 0.08, and adjust the dynamic segmentation threshold accordingly to the mean + 2.1σ; nine initial candidate regions are extracted by connectivity markers; three of these regions have a sphericity of 0.87–0.93 (reasonable range 0.85–0.95) and a surface roughness of less than 0.12 (threshold 0.15), and are confirmed as real bubbles after comparison with a dedicated PC bubble morphology database; the remaining six regions are eliminated because the sphericity is <0.7 or the roughness is >0.2, to avoid misjudging cooling lines or reconstruction artifacts as defects.

[0067] In one embodiment, the process of constructing the bubble morphology feature database includes: B1. Collect bubble defect samples of known size and shape from standard engineering plastic specimens, and accurately measure the sphericity characteristic parameters and surface roughness characteristic parameters of each bubble defect. The standard engineering plastic specimens can be benchmark test samples used to accurately calibrate bubble morphology parameters. They are prepared under controlled conditions and contain known artificial or natural bubble defects to ensure the reliability of the database data source. Bubble defect samples with known size and morphology can be real bubble instances whose geometric parameters have been confirmed by high-precision imaging in the standard specimens. These serve as the original objects for morphological feature extraction, ensuring the accuracy of parameter measurements.

[0068] Collecting bubble defect samples of known size and shape from standard engineering plastic specimens can be achieved by preparing specimens containing controllable bubbles under laboratory conditions and confirming their actual existence and geometric properties through high-resolution imaging. Furthermore, this operation can be accomplished by introducing bubbles of known volume at specific locations using microinjection, or by inducing repeatable bubble defects through accelerated aging or thermal shock and recording the formation conditions, thereby obtaining training and calibration samples with a realistic physical background. Accurately measuring the sphericity and surface roughness characteristic parameters of each bubble defect can be achieved by calculating geometric and surface differential properties based on the 3D reconstruction results. In a specific embodiment, this operation can be achieved by calculating sphericity after fitting the surface using spherical harmonic functions, or by quantifying surface roughness through local curvature variance, thereby generating a high-fidelity morphological feature dataset to support subsequent mapping modeling.

[0069] B2. Establish a hierarchical mapping table of bubble morphological feature parameters based on the bubble volume radius parameter, where small-sized bubbles correspond to the first sphericity range and large-sized bubbles correspond to the second sphericity range. The hierarchical mapping table can be a mapping relationship table that divides morphological features into different scale intervals based on the bubble volume radius parameter. It is used to reflect the size-dependent evolution law of bubble morphology and improve the adaptability of the discrimination model to defects of different scales. The first sphericity range can be a typical sphericity range corresponding to small-sized bubbles (e.g., volume radius < 0.2 mm), reflecting the physical characteristic that small bubbles are closer to an ideal sphere due to the dominance of surface tension. The second sphericity range can be a typical sphericity range corresponding to large-sized bubbles (e.g., volume radius ≥ 0.5 mm), reflecting the process characteristics that large bubbles are easily affected by flow disturbances, cooling contraction, etc., and deviate from a spherical shape. For example, the hierarchical mapping table can include, but is not limited to, a small-sized bubble morphology mapping table, a medium-sized bubble morphology mapping table, and a large-sized bubble morphology mapping table.

[0070] Establishing a hierarchical mapping table for bubble morphological characteristic parameters based on bubble volume radius parameters can be achieved by grouping bubbles according to their volume radius and statistically analyzing their sphericity and roughness distributions to form multi-interval mapping rules. In an exemplary embodiment, this operation can be achieved by automatically discovering the optimal size boundary point using a clustering algorithm, or by introducing a nonlinear regression model to describe the trend of sphericity changing with radius, thereby reflecting the size dependence of bubble morphology and enhancing the physical rationality of the discrimination logic.

[0071] B3. Establish a correlation between the bubble morphology feature parameters in the grading mapping table and the critical influence threshold of the mechanical properties of engineering plastic products to form a bubble morphology feature database for judging the authenticity of bubble defects.

[0072] The critical impact threshold for the mechanical properties of engineering plastic products can be the morphological or dimensional boundary value at which bubble defects cause key mechanical indicators (such as tensile strength and fatigue life) to drop to unacceptable levels. This serves as a bridge connecting morphological characteristics and structural safety, giving the test results engineering decision-making value. The bubble morphology feature database can be a structured knowledge base that links the morphological parameters of bubbles with their impact on the mechanical properties of engineering plastics. It provides a basis for identifying bubble defects that combines physical authenticity and engineering significance, supporting the transformation from image anomalies to structural risks. In this embodiment, the bubble morphology feature database may include, but is not limited to, sub-databases divided by material type, feature libraries classified by process conditions (injection molding / extrusion), and criterion libraries customized by service environment (high temperature / load).

[0073] Establishing a correlation between bubble morphological feature parameters in a grading mapping table and the critical influence threshold of the mechanical properties of engineering plastic products can be achieved by determining the impact of different bubble morphologies on indicators such as strength and toughness through mechanical experiments, and setting the morphological boundary corresponding to the critical point of performance degradation. Furthermore, this operation can be implemented by constructing a surrogate model (such as a Gaussian process) to predict performance degradation under arbitrary bubble parameters, or by using failure probability analysis to determine the threshold within the safety margin. This allows morphological features to possess engineering safety semantics, supporting risk-oriented judgments. A bubble morphological feature database for identifying the authenticity of bubble defects can be formed by integrating the grading mapping table with mechanical threshold association rules and constructing a structured query interface. This provides a discriminative knowledge source for defect authenticity identification that combines authenticity, scale adaptability, and engineering significance.

[0074] For example, in the scenario of safety assessment of nylon 66 injection molded parts for automotive fuel rails, the laser scattering tomography-based method for detecting internal bubble defects in engineering plastic products in this embodiment can be as follows: During the R&D stage, standard nylon 66 samples containing 0.1mm–1.0mm artificial bubbles are prepared, and their sphericity and roughness are measured using micro-CT. A grading mapping table is established: the sphericity range for 0.1–0.3mm bubbles is 0.88–0.95, and for 0.6–1.0mm bubbles it is 0.75–0.88. Tensile testing reveals that when the sphericity of the bubble is <0.8 and the radius is >0.4mm, the strength decreases by more than 12%, and a critical influence threshold is set accordingly. After this database is deployed in the detection system, a 0.45mm defect with a sphericity of 0.78 is identified on the production line. The system determines that it exceeds the safety boundary, triggers a process alarm, and avoids batch failure.

[0075] In one embodiment, the following steps are also included after step S5: S6. Generate a defect spatial distribution heat map based on the spatial distribution density of the actual bubble defect area, and evaluate the residual strength coefficient of the engineering plastic product based on the defect spatial distribution heat map. The spatial distribution density of the actual bubble defect region can be the number of actual bubble defects per unit volume or the proportion of the total volume, reflecting the local defect density. This spatial distribution density can be used as the basic input data for generating a defect spatial distribution heatmap. The defect spatial distribution heatmap can be a two-dimensional or three-dimensional visualization image generated based on the spatial distribution density of the actual bubble defect region, using color or grayscale gradients to represent the defect density. In this embodiment, the defect spatial distribution heatmap can perform kernel density estimation or sliding window statistics on the number or volume proportion of bubbles in the neighborhood of each spatial location, mapping it to continuous intensity values. Furthermore, the defect spatial distribution heatmap can be coordinated with the residual intensity coefficient evaluation process to provide a spatially continuous damage field input. For example, the defect spatial distribution heatmap can include, but is not limited to, one or more of the following: two-dimensional projection heatmap, voxelized three-dimensional thermal field, radially symmetric distribution heatmap, etc.

[0076] The residual strength coefficient can be a dimensionless index characterizing the mechanical load-bearing capacity retained by an engineering plastic product containing bubble defects relative to its defect-free ideal state. In an exemplary embodiment, the residual strength coefficient can be calculated based on a defect spatial distribution heatmap, combined with a material constitutive model and damage mechanics theory to determine the effective load-bearing section or stress concentration factor of the structure. The residual strength coefficient can be used to quantify the comprehensive impact of bubble distribution on structural safety, serving as a criterion for whether to trigger process feedback. In a specific embodiment, the residual strength coefficient may include, but is not limited to, the static tensile residual strength coefficient, fatigue life reduction factor, and impact toughness retention rate.

[0077] Generating a defect spatial distribution heatmap based on the spatial distribution density of real bubble defect regions can be achieved by estimating the spatial density of the geometric center coordinates of the real bubble defect regions, generating a continuous intensity field, and visualizing it. Furthermore, this operation can be achieved by smoothing the bubble center point set using a Gaussian kernel density estimation method to generate the heatmap, or by statistically analyzing the local bubble volume proportion using a three-dimensional voxel mesh and mapping it to color intensity. This transforms discrete defect information into a continuous spatial distribution representation, revealing defect aggregation trends. Evaluating the residual strength coefficient of engineering plastic products based on the defect spatial distribution heatmap can be achieved by inputting the heatmap as a damage field into a mechanical property degradation model to calculate the strength retention ratio of the overall structure or key areas. Further, this operation can be achieved through simulation by applying equivalent material weakening properties to high-density areas using the finite element sub-model method, or by estimating the residual strength coefficient based on an empirical formula by linearly combining the maximum thermal value and the average thermal gradient. This enables a quantitative mapping from defect distribution to the impact on structural performance.

[0078] S7. When the residual strength coefficient is lower than the preset safety threshold, generate a defect analysis report containing the spatial distribution characteristic parameters of bubble defects and corresponding suggestions for adjusting injection molding process parameters.

[0079] The preset safety threshold can be the minimum allowable value of the residual strength coefficient set according to the product's service requirements. The preset safety threshold can be used to determine whether the product meets structural safety requirements. The spatial distribution characteristic parameters of bubble defects can be a set of quantitative indicators describing the spatial distribution pattern of bubbles. These parameters can be used to support defect cause analysis and process correlation modeling. In a specific embodiment, the spatial distribution characteristic parameters of bubble defects may include, but are not limited to, defect cluster radius, depth-direction density gradient, and main distribution axial skew angle. The injection molding process parameter adjustment suggestions can be recommended modification schemes for injection molding process variables based on the current defect distribution pattern. These suggestions can be used to guide the production end to optimize the process to suppress the recurrence of similar defects. For example, these suggestions may include, but are not limited to, suggestions for correcting the pressure-time curve during the holding pressure stage, melt temperature zoning control strategies, and mold cooling water channel flow redistribution schemes. The defect analysis report can be a structured output document containing the spatial distribution characteristic parameters of bubble defects and corresponding injection molding process parameter adjustment suggestions. The defect analysis report can be used to provide traceable and executable data for manufacturing process optimization, achieving a closed loop of detection-diagnosis-improvement. In one exemplary embodiment, the defect analysis report may include, but is not limited to, PDF format diagnostic reports, XML structured data packets, MES system integration interface messages, etc.

[0080] When the residual strength coefficient falls below a preset safety threshold, a defect analysis report is generated, containing spatial distribution characteristic parameters of bubble defects and corresponding injection molding process parameter adjustment suggestions. This can be achieved by extracting the spatial distribution characteristic parameters of bubbles after triggering a condition, matching them with a process knowledge base to generate corresponding adjustment suggestions, and then packaging them into a standard report format. Furthermore, this operation can be implemented by calling a rule engine to recommend local cooling and strengthening measures for the mold based on the location of defect clusters, or by using similar cases to retrieve parameter adjustment suggestions from a historical process-defect database. This establishes a feedback channel between the detection results and the manufacturing process, supporting closed-loop quality control.

[0081] For example, in the scenario of final inspection and process feedback of polyetheretherketone (PEEK) brackets for aerospace applications, the laser scattering tomography-based method for detecting internal bubble defects in engineering plastic products in this embodiment can be as follows: after the system completes step S5, it calculates the spatial distribution density of the seven identified real bubble defects (mainly concentrated in the rib intersection area) and generates a three-dimensional defect heat map. The heat map shows that the density value in this area reaches 0.18 mm⁻³, and based on this, the residual strength coefficient is evaluated as 0.82, which is lower than the safety threshold of 0.85 required for aerospace components. The system automatically generates a defect analysis report, pointing out that "bubbles are gradient-aggregated in the depth direction, and the maximum density is located in the wall thickness transition area," and recommends "extending the holding pressure time by 15% and increasing the cooling rate of the mold in this area by 20%." The report is simultaneously pushed to the injection molding machine control system and the process engineer's terminal.

[0082] This embodiment transforms discrete defect information into a continuous spatial intensity field, quantitatively characterizes the impact of defect distribution on structural load-bearing capacity, and forms a complete closed loop from defect visualization and performance evaluation to process optimization suggestions. This can achieve the technical effect of enhancing the decision support capability of the detection system in intelligent manufacturing.

[0083] In one embodiment, step S1 specifically includes: S11. Obtain the material type parameters of the engineering plastic product to be tested, and query the pre-built material type-laser wavelength mapping table according to the material type parameters to determine the target wavelength range of the modulated laser beam applicable to the engineering plastic product. The material type parameter of the engineering plastic product to be tested can be classification information identifying the chemical composition or brand name of the engineering plastic, which can be used to provide input for adaptive wavelength selection. In this embodiment, the material type parameter of the engineering plastic product to be tested can be obtained by reading the material grade from the production information system or by identifying the type of engineering plastic used in the product through barcode / RFID tags.

[0084] A material type-laser wavelength mapping table can be a pre-established lookup table that associates engineering plastic material types with their optimal detection laser wavelength range. This table can be used to automatically match the optimal excitation wavelength for a specific material, improving the intensity of the scattered signal and the contrast of defects. For example, the material type-laser wavelength mapping table can be used through experimental measurements or optical simulations to screen and structurally store the wavelength range with the optimal signal-to-noise ratio for different polymer materials at multiple wavelengths, considering their penetration depth, scattering contrast, and absorption coefficient. Furthermore, the material type-laser wavelength mapping table can include, but is not limited to, one or more of the following: a wavelength optimization table based on polycarbonate, a wavelength optimization table based on polyamide, and a wavelength optimization table based on polyoxymethylene.

[0085] The target wavelength range of the modulated laser beam can be a laser emission wavelength range suitable for the engineering plastic product under test, determined according to the material type. This range can be used to ensure sufficient penetration depth of the laser in the material and high scattering contrast at the bubble-matrix interface. In an exemplary embodiment, the target wavelength range of the modulated laser beam, as a control parameter of the laser emission device, directly affects the signal-to-noise ratio and saliency of the multi-angle time-domain scattered light signal sequence. Obtaining the material type parameter of the engineering plastic product under test can be achieved by reading the material grade from a production information system or by identifying the type of engineering plastic used in the product through barcodes / RFID tags. Furthermore, obtaining the material type parameter of the engineering plastic product under test can be achieved in the above manner, thereby providing input for adaptive wavelength selection.

[0086] The target wavelength range of the modulated laser beam suitable for the engineering plastic product is determined by querying a pre-built material type-laser wavelength mapping table based on the material type parameter. This can be achieved by using the material type as a key value and retrieving the corresponding optimal wavelength range from the pre-stored mapping table. In one specific embodiment, this operation can be implemented by using a database index to quickly match the material type and wavelength range, or by using an embedded rule engine to perform conditional judgments to output the target wavelength. This allows for material-adaptive optimization of the light source wavelength, improving the quality of the original scattered signal.

[0087] S12. Control the laser emitting device to emit a modulated laser beam within the target wavelength range, and synchronously capture multi-angle time-domain scattered light signal sequences through time-correlated single-photon counting technology.

[0088] Controlling the emission of a modulated laser beam from a laser emitting device within the target wavelength range can be achieved by adjusting the wavelength tuning elements of the laser (such as gratings, filters, or tunable laser sources) to ensure that its output falls within the target wavelength range. For example, this operation can be achieved by directly setting the center wavelength and bandwidth using a tunable semiconductor laser, or by selecting the output band using a fixed-wavelength laser in conjunction with an acousto-optic tunable filter (AOTF), thereby ensuring that the incident light matches the optical properties of the material and enhancing the bubble scattering response. Time-correlated single-photon counting technology can be a highly sensitive time-domain optical signal acquisition method based on single-photon detection and precise timestamp recording. It can be used to achieve picosecond-level time-resolved scattering signal capture, effectively separating multiple scattering paths and enhancing the time-space resolution capability for tiny bubbles. In a specific embodiment, time-correlated single-photon counting technology can utilize a single-photon avalanche diode (SPAD) detector in conjunction with a time-to-digital converter (TDC) to record the arrival time of each scattered photon relative to the laser pulse emission time, accumulating to form a photon time-of-flight distribution histogram. Furthermore, time-correlated single-photon counting techniques may include, but are not limited to, multi-channel TCSPC systems, gated single-photon counting techniques, and time-gated fluorescence lifetime imaging techniques.

[0089] Synchronously capturing multi-angle time-domain scattered light signal sequences using time-correlated single-photon counting technology can be achieved by having single-photon detectors at each angle simultaneously record photon arrival times under laser pulse triggering, accumulating these times to generate a multi-channel time histogram. In an exemplary embodiment, this operation can be achieved by using a multi-channel TDC module to enable independent time recording by each receiving unit, or by using a global clock to synchronize multiple SPAD arrays and share a time reference. This allows for obtaining scattered signals with high temporal resolution and high sensitivity, supporting precise resolution of deep microbubbles.

[0090] Taking the inspection of LCP (liquid crystal polymer) injection molded parts for electronic connectors as an example, the method for detecting internal bubble defects in engineering plastic products based on laser scattering tomography in this embodiment can be as follows: The system reads the product material type parameter as "LCP-123", and after querying the material type-laser wavelength mapping table, determines the target wavelength range as 780–810nm; the laser emitting device switches to an 800nm ​​tunable laser and emits picosecond-level modulated pulses; the SPAD detector in the multi-angle scattered light receiving array synchronously collects the time-of-flight distribution of photons at each angle through time-correlated single-photon counting technology, forming a time-domain scattering sequence with a high signal-to-noise ratio; after subsequent dynamic compensation and tomographic reconstruction, this sequence successfully identifies a closed bubble with a diameter of only 35 micrometers, while the traditional fixed wavelength (such as 650nm) scheme cannot detect this defect because LCP has strong absorption in this band.

[0091] This embodiment obtains the material type parameters of the engineering plastic product to be tested, queries the material type-laser wavelength mapping table to determine the target wavelength range of the modulated laser beam, and controls the laser emitting device to emit the modulated laser beam within this range, so that the incident light matches the optical properties of the material, maximizing the penetration depth and bubble scattering contrast. At the same time, time-correlated single-photon counting technology is used to simultaneously capture multi-angle time-domain scattered light signal sequences. Utilizing its single-photon sensitivity and picosecond-level time resolution, the scattering path and time delay characteristics of deep microbubbles can be accurately analyzed. This can optimize the wavelength adaptability of the excitation light source and the time-domain resolution capability of the detection system, improve the quality of the original signal from the source, and enhance the universality of the detection method under different material systems and the technical effect of resolving micron-sized bubbles.

[0092] In one embodiment, the pulse width of the modulated laser beam is set to a parameter value that matches the typical scattering timescale of bubbles inside the engineering plastic article to ensure that the scattered signal has sufficient time resolution.

[0093] The pulse width of the modulated laser beam can be the duration of a single pulse in the time domain, which can be used to determine the temporal resolution of the incident light and affect the separability of scattering sources at different depths in the scattered signal. In this embodiment, the pulse width of the modulated laser beam can be matched with the typical scattering timescale of bubbles inside the engineering plastic product to optimize the temporal resolution of the scattered signal. The typical scattering timescale of bubbles inside the engineering plastic product can be the magnitude of the time delay characteristic of the laser relative to the direct light after scattering at the bubble interface inside the engineering plastic. It can be used as the physical basis for setting the laser pulse width to guide the design of time resolution parameters. For example, the typical scattering timescale of bubbles inside the engineering plastic product can be determined by the bubble size, location depth, and material refractive index, and is usually in the picosecond to nanosecond range. Further, the typical scattering timescale of bubbles inside the engineering plastic product can include, but is not limited to, one or more of the following: picosecond-level scattering time for micron-sized bubbles, hundreds of picosecond-level scattering time for submillimeter-sized bubbles, and short-delay scattering time for near-surface bubbles. The temporal resolution capability of the scattered signal can be the system's ability to distinguish the components of scattered light arriving at different times. It can directly affect the resolution accuracy of bubble depth and size, and is a prerequisite for achieving high-precision tomographic reconstruction.

[0094] Setting the pulse width of the modulated laser beam to a parameter value that matches the typical scattering timescale of bubbles inside the engineering plastic product can be achieved by adjusting the laser drive circuit or pulse shaping module according to the estimated or calibrated bubble scattering timescale, outputting a laser pulse of the corresponding width. Furthermore, setting the pulse width of the modulated laser beam to a parameter value that matches the typical scattering timescale of bubbles inside the engineering plastic product can be achieved by controlling the rising / falling edge of the laser diode drive current with a programmable pulse generator to adjust the output pulse width, or by using an electro-optic modulator to externally chop the continuous laser to generate an optical pulse of the required width. This avoids the overlap of signals from different scattering paths in the time domain, improving the separability of bubble scattering components. Ensuring sufficient time resolution of the scattering signal can be achieved by matching the pulse width with the scattering timescale, enabling the detection system to clearly distinguish the characteristic time shift caused by the bubble. In one specific embodiment, ensuring sufficient time resolution of the scattering signal enhances the detectability and three-dimensional positioning accuracy of tiny enclosed bubbles.

[0095] For example, in the scenario of detecting internal defects in PEEK structural components for aerospace applications, the laser scattering tomography-based method for detecting internal bubble defects in engineering plastic products in this embodiment can be used for closed bubbles with typical diameters of 50–200 micrometers, whose scattering time delay in PEEK material is approximately 300–800 picoseconds. The system sets the pulse width of the modulated laser beam to 600 picoseconds, which is slightly wider than the minimum resolvable time interval but narrower than the multiple scattering aliasing threshold. Combined with time-correlated single-photon counting technology, the scattered echoes of deep bubbles are successfully separated, avoiding temporal overlap with surface reflections or matrix Rayleigh scattering. This results in clear bubble boundaries and accurate positions in the tomographic reconstruction image, whereas the traditional method using a 2-nanosecond wide pulse causes multiple bubbles to be misjudged as a single large defect due to time aliasing.

[0096] This embodiment adjusts the laser pulse width according to the bubble scattering timescale to avoid overlap of signals from different scattering paths in the time domain. By matching the pulse width with the scattering timescale, the detection system can clearly distinguish the characteristic time shift caused by the bubble. This achieves the goal of avoiding overlap of signals from different scattering paths in the time domain and ensuring that the scattering signal has sufficient time resolution. This matching strategy enables clear separation of the characteristic time delay caused by the bubble. Within a high-sensitivity detection framework such as time-correlated single-photon counting, it effectively improves the detectability and three-dimensional positioning accuracy of micron-sized enclosed bubbles. It provides high-fidelity time-domain information for authenticity determination and security assessment, and solves the technical problem of missed detection or misjudgment of micro-defects caused by insufficient time resolution in traditional optical methods.

[0097] Furthermore, to achieve the above objectives, the present invention also provides a laser scattering tomography-based device for detecting internal bubble defects in engineering plastic products. The device includes: a memory, a processor, and a laser scattering tomography-based program for detecting internal bubble defects in engineering plastic products, stored in the memory and executable on the processor. The laser scattering tomography-based program for detecting internal bubble defects in engineering plastic products is configured to implement the steps of the laser scattering tomography-based method for detecting internal bubble defects in engineering plastic products as described above.

[0098] Other embodiments or specific implementations of the laser scattering tomography-based equipment for detecting internal bubble defects in engineering plastic products described in this invention can be referred to the above-described method embodiments, and will not be repeated here.

[0099] Furthermore, to achieve the above objectives, the present invention also provides a medium storing a laser scattering tomography-based internal bubble defect detection program for engineering plastic products. When the laser scattering tomography-based internal bubble defect detection program for engineering plastic products is executed by a processor, it implements the steps of the laser scattering tomography-based internal bubble defect detection method for engineering plastic products as described above.

[0100] The above are merely preferred embodiments of the present invention and do not limit the scope of the patent. Any equivalent structural or procedural transformations made based on the description and drawings of the present invention, or direct or indirect applications in other related technical fields, are similarly included within the scope of patent protection of the present invention.

Claims

1. A method for detecting internal bubble defects in engineering plastic products based on laser scattering tomography, characterized in that, The method comprises: S1, using a laser emitting device to emit a modulated laser beam to the engineering plastic product to be detected, and simultaneously using a multi-angle scattered light receiving array to synchronously collect a multi-angle time-domain scattered light signal sequence, the multi-angle time-domain scattered light signal sequence representing intensity attenuation characteristics and time distribution characteristics of laser pulses scattered inside the engineering plastic product at different incident angles; S2, obtaining a material thickness parameter and a real-time environmental temperature parameter of the engineering plastic product, and performing dynamic compensation processing on the multi-angle time-domain scattered light signal sequence according to the material thickness parameter and the real-time environmental temperature parameter, so as to eliminate interference effects of refractive index fluctuations caused by temperature changes on scattered signals, and generate a compensated scattered characteristic parameter set; S3, based on the compensated scattered characteristic parameter set, applying a tomographic reconstruction algorithm to construct a three-dimensional scattering coefficient distribution map inside the engineering plastic product, the three-dimensional scattering coefficient distribution map reflecting spatial variation characteristics of the material internal density gradient, wherein the internal bubble defect area presents a local high value area with a scattering coefficient value significantly higher than that of the engineering plastic matrix material; S4, performing an adaptive threshold segmentation algorithm on the three-dimensional scattering coefficient distribution map to obtain a bubble defect candidate region set, and then performing true and false discrimination on the bubble defect candidate region set according to bubble morphological feature screening rules to determine a real bubble defect area; S5, for the real bubble defect area, using a hyperboloid fitting model to calculate volume radius parameters, geometric center coordinates and spatial distribution density parameters of each bubble defect, and when the volume radius parameter of any bubble defect exceeds a preset structural safety threshold, determining that the engineering plastic product has an internal bubble defect affecting the structural integrity.

2. The method for detecting internal bubble defects of an engineering plastic product based on laser scattering tomography according to claim 1, wherein Step S2 specifically comprises: S21, collecting surface temperature distribution data of the engineering plastic product in real time through an infrared thermal imaging device, querying a pre-constructed temperature-refractive index disturbance mapping relationship table according to the surface temperature distribution data, and obtaining refractive index compensation coefficients corresponding to the temperature of each region; S22, for the scattered light signal corresponding to each incident angle, applying the refractive index compensation coefficient to the scattered light intensity normalization processing process, and combining the material thickness parameter of the engineering plastic product to calculate an effective optical path difference compensation amount; S23, data fusion of the normalized scattered light signal and the effective optical path difference compensation amount to generate a compensated scattered characteristic parameter set containing scattering phase function parameters and attenuation length parameters.

3. The method for detecting internal bubble defects of an engineering plastic product based on laser scattering tomography according to claim 2, wherein Step S3 specifically comprises: S31, inputting the compensated scattered characteristic parameter set into an iterative tomographic reconstruction model, the iterative tomographic reconstruction model being built-in with a priori constraint condition based on Monte Carlo scattering simulation; S32, iteratively inverting and solving a spatial distribution matrix of the internal scattering coefficient of the engineering plastic product by minimizing the mean square error between the actual measured scattered signal and the simulated scattered signal, to generate a high-resolution three-dimensional scattering coefficient distribution map; S33, performing anisotropic diffusion filtering processing on the three-dimensional scattering coefficient distribution map to effectively suppress image noise generated in the tomographic reconstruction process while retaining the bubble edge definition.

4. The method for detecting internal bubble defects of an engineering plastic product based on laser scattering tomography according to claim 3, wherein Step S4 specifically comprises: S41, determine a dynamic segmentation threshold based on the statistical distribution characteristics of the scattering coefficients of the engineering plastic product substrate region, and the dynamic segmentation threshold is adaptively adjusted according to the standard deviation of the local region scattering coefficient; S42, perform a connected region labeling algorithm on the three-dimensional scattering coefficient distribution map to obtain an initial bubble defect candidate region set; S43, for each candidate region in the initial bubble defect candidate region set, calculate its sphericity feature parameter and surface roughness feature parameter, and compare the calculation results with the pre-constructed bubble morphological feature database, and when the sphericity feature parameter is in a predetermined reasonable range and the surface roughness feature parameter is lower than a threshold, the candidate region is confirmed as a real bubble defect region.

5. The method for detecting internal bubble defects in an engineered plastic article based on laser scatter tomography according to claim 4, wherein, The construction process of the bubble morphological feature database includes: B1, collect bubble defect samples of known size and morphology in standard engineering plastic samples, and accurately measure the sphericity feature parameter and surface roughness feature parameter of each bubble defect; B2, establish a hierarchical mapping table of bubble morphological feature parameters based on the bubble volume radius parameter, wherein small size bubbles correspond to a first sphericity range and large size bubbles correspond to a second sphericity range; B3, establish a correlation between the bubble morphological feature parameters in the hierarchical mapping table and the critical influence threshold of the mechanical properties of the engineering plastic product to form a bubble morphological feature database for bubble defect true-false discrimination.

6. The method for detecting internal bubble defects in an engineered plastic article based on laser scatter tomography according to claim 1, wherein, Further comprising the following steps performed after step S5: S6, generate a defect spatial distribution thermodynamic map according to the spatial distribution density of the real bubble defect region, and evaluate the residual strength coefficient of the engineering plastic product based on the defect spatial distribution thermodynamic map; S7, when the residual strength coefficient is lower than a predetermined safety threshold, generate a defect analysis report containing bubble defect spatial distribution feature parameters and corresponding injection molding process parameter adjustment suggestions.

7. The laser scatter tomography based method of detecting internal bubble defects in an engineered plastic article of claim 1, wherein, Step S1 specifically includes: S11, obtain the material type parameter of the engineering plastic product to be detected, query the pre-constructed material type-laser wavelength mapping relationship table according to the material type parameter, and determine the target wavelength range of the modulated laser beam suitable for the engineering plastic product; S12, control the laser emitting device to emit a modulated laser beam in the target wavelength range, and synchronously capture a multi-angle time-domain scattered light signal sequence through time-correlated single-photon counting technology.

8. The method for detecting internal bubble defects in an engineered plastic article based on laser scatter tomography according to claim 7, wherein, The pulse width of the modulated laser beam is set to a parameter value matched with the typical scattering time scale of the bubbles inside the engineering plastic product, so as to ensure that the scattering signal has sufficient time resolution capability.

9. A laser scattering tomography-based apparatus for detecting internal bubble defects in an engineering plastic product, characterized by, The device includes a memory, a processor, and a laser scattering tomography-based engineering plastic product internal bubble defect detection program stored on the memory and executable on the processor, and the laser scattering tomography-based engineering plastic product internal bubble defect detection program is configured to implement the steps of the laser scattering tomography-based engineering plastic product internal bubble defect detection method according to any one of claims 1 to 8.

10. A medium characterized by, The medium stores a program for detecting internal bubble defects of an engineering plastic product based on laser scatter tomography, and the program, when executed by a processor, implements the steps of the method for detecting internal bubble defects of an engineering plastic product based on laser scatter tomography according to any one of claims 1 to 8.

Images