A nondestructive testing system and method for structural uniformity of composite bulletproof helmets

Abstract

The present invention discloses a nondestructive testing system and method for the structural uniformity of a composite material bulletproof helmet, relating to the technical field of composite material mechanical manufacturing. The system and method comprise the following steps: extracting the path curvature in a topological diagram of the helmet material structure, performing polynomial fitting using the least squares method, generating an optimized scanning trajectory, converting the optimized scanning trajectory into a robotic arm control instruction, driving an ultrasonic probe to scan the helmet structure with multi-frequency focusing and incident angle deflection, and obtaining an original echo signal matrix; inputting the original echo signal matrix into an acoustic physical neural model, performing time-frequency feature analysis in a feature extraction layer to generate an acoustic wave propagation feature tensor, and performing sound velocity distortion correction in a physical constraint layer to obtain a calibrated echo signal matrix; and constructing a topological diagram of the helmet material structure and an acoustic physical neural model to achieve quantitative analysis of the internal microscopic defect distribution and connectivity characteristics of the composite material, thereby improving the overall accuracy and reliability of the nondestructive testing.

Description

Technical Field

[0001] The present invention relates to the technical field of composite material mechanical manufacturing, in particular to a nondestructive detection system and method for the structural uniformity of a composite material bulletproof helmet. Background Art

[0002] In recent years, the rapid development of composite materials technology, particularly the widespread application of high-performance composite materials in military equipment, has placed higher demands on the safety, reliability, and performance of protective equipment such as bulletproof helmets. Structural uniformity, as a key factor influencing the protective effectiveness of bulletproof helmets, has received increasing attention. Traditional nondestructive testing methods, such as X-ray imaging and ultrasonic testing, can, to a certain extent, assess the internal structural condition of composite materials.

[0003] However, existing technologies still have some shortcomings. First, existing optical transmission testing methods can only identify macroscopic defect morphology and fail to establish a mapping relationship between microscopic topological features such as pore distribution and connectivity paths and macroscopic mechanical parameters such as material stiffness and sound velocity gradient. This makes it impossible to predict the actual impact of defects on ballistic performance. Second, traditional ultrasonic testing uses a fixed scanning trajectory and static signal analysis model, failing to consider the dynamic interference of composite material microstructures on the sound wave propagation path. This leads to deviations in sound velocity distortion correction and makes it difficult to ensure detection accuracy. Summary of the Invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a non-destructive testing method for the structural uniformity of composite material bulletproof helmets, which solves the problems of insufficient detection accuracy and lack of micro-macro performance correlation in existing testing methods.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0007] In a first aspect, the present invention provides a nondestructive testing method for the structural uniformity of a composite bulletproof helmet, comprising: collecting a helmet microstructure dataset, extracting the coordinates of the force-bearing regions of the helmet pore positions, obtaining the pore space coordinates, and mapping the geometric properties of the pore space coordinates to form nodes of a topological structure; applying a geometric adjacency algorithm to connect adjacent pores in the helmet connectivity path to generate physical channel segments; mapping the transmission directions of the physical channel segments to obtain edges of the topological structure; and performing weighted fusion of the nodes and edges of the topological structure to generate a topological map of the helmet material structure.

[0008] The path curvature in the helmet material structure topology is extracted and polynomial fitting is performed using the least squares method to generate an optimized scanning trajectory. The optimized scanning trajectory is converted into a robotic arm control instruction, which drives the ultrasonic probe to scan the helmet structure with multi-frequency focusing and incident angle deflection to obtain the original echo signal matrix.

[0009] The original echo signal matrix is ​​input into the acoustic physics neural model. The feature extraction layer performs time-frequency feature analysis to generate the sound wave propagation feature tensor. The physical constraint layer performs sound velocity distortion correction to obtain the calibration echo signal matrix. The sound wave propagation feature tensor and the calibration echo signal matrix are subjected to multi-scale time-frequency analysis and nonlinear transformation to output the non-destructive testing solution.

[0010] Based on the non-destructive testing scheme, the improved incremental weighted graph algorithm is used to perform dynamic node clustering and edge strength redistribution on the helmet material structure topology graph, and the node weights are updated simultaneously and the edge connection strength is optimized.

[0011] As a preferred solution of the non-destructive testing method for the structural uniformity of the composite material bulletproof helmet described in the present invention, the helmet microstructure dataset includes helmet pore positions, helmet connection paths, helmet three-dimensional point clouds and composite material performance data.

[0012] As a preferred solution of the non-destructive testing method for the structural uniformity of the composite material bulletproof helmet of the present invention, wherein: the extraction of the path curvature in the topological diagram of the helmet material structure specifically includes the following steps:

[0013] B-spline interpolation is used to perform parameterized sampling on the topological map of the helmet material structure to obtain a discrete point set, which is then numerically differentiated to generate the path curvature.

[0014] As a preferred solution of the non-destructive testing method for the structural uniformity of the composite material bulletproof helmet of the present invention, wherein: the generating of the optimized scanning trajectory specifically includes the following steps:

[0015] Noise suppression and feature enhancement are performed on the path curvature to generate an optimized curvature distribution; polynomial fitting is performed on the optimized curvature distribution using the least squares method to form an optimized scanning trajectory.

[0016] As a preferred solution of the non-destructive testing method for the structural uniformity of the composite material bulletproof helmet of the present invention, the method of converting the optimized scanning trajectory into a robotic arm control instruction specifically includes the following steps:

[0017] Dynamic time warping is applied to perform timestamp discretization on the optimized scanning trajectory to obtain the angle instruction sequence, which is then encapsulated to form the robotic arm control instructions.

[0018] As a preferred solution of the non-destructive testing method for the structural uniformity of the composite material bulletproof helmet of the present invention, wherein: the obtaining of the original echo signal matrix specifically includes the following steps:

[0019] According to the control instructions of the robotic arm, the ultrasonic probe is driven to scan the helmet structure with multi-frequency focusing and incident angle deflection to obtain multi-channel echo signals;

[0020] The multi-channel echo signals are space-time aligned to generate the original echo signal matrix.

[0021] As a preferred solution of the non-destructive testing method for the structural uniformity of the composite material bulletproof helmet of the present invention, the specific construction process of the acoustic physical neural model is as follows:

[0022] Build and initialize the feature extraction layer and physical constraint layer, perform feature transfer and double stacking through residual connections, and construct an acoustic physical neural model.

[0023] As a preferred solution of the non-destructive testing method for the structural uniformity of the composite material bulletproof helmet of the present invention, the output non-destructive testing solution specifically includes the following operating steps:

[0024] The original echo signal matrix is ​​input into the acoustic physics neural model. The feature extraction layer performs multi-dimensional feature analysis through a three-dimensional convolution kernel to generate a sound wave propagation feature tensor.

[0025] The physical constraint layer is embedded in the elastic constitutive equation to perform sound velocity distortion correction on the acoustic wave propagation characteristic tensor to obtain the calibration echo signal matrix;

[0026] Apply wavelet packet energy spectrum to perform multi-scale time-frequency analysis and nonlinear transformation on the calibration echo signal matrix to obtain structural uniformity parameters and defect probability distribution;

[0027] Integrate structural uniformity parameters with defect probability distribution to output nondestructive testing solutions.

[0028] As a preferred solution of the non-destructive testing method for the structural uniformity of the composite material bulletproof helmet of the present invention, wherein: the node weights of the helmet material structure topology graph are updated and the edge connection strength is optimized synchronously, specifically comprising the following steps:

[0029] Use wavelet packet decomposition to extract defect-sensitive frequency band parameters in non-destructive testing schemes;

[0030] According to the defect-sensitive frequency band parameters, an improved incremental weighted graph algorithm is used to perform node dynamic clustering and edge strength redistribution on the helmet material structure topology graph to obtain the sound velocity gradient coefficient and stiffness distribution coefficient.

[0031] Based on the sound velocity gradient coefficient, the node weights are gradient updated, and the edge connection strength is coupled optimized according to the stiffness distribution coefficient.

[0032] In a second aspect, the present invention provides a nondestructive testing system for the structural uniformity of a composite bulletproof helmet, comprising a topological mapping module, a signal acquisition module, a solution generation module, and a dynamic optimization module;

[0033] The topological mapping module is used to collect the helmet microstructure data set, extract the force-bearing area coordinates of the helmet pore position, obtain the pore space coordinates, and map the geometric properties of the pore space coordinates to form the nodes of the topological structure. The geometric adjacency algorithm is applied to connect the adjacent pores of the helmet connection path to generate physical channel segments. The transmission direction of the physical channel segments is mapped to obtain the edges of the topological structure. The nodes and edges of the topological structure are weightedly fused to generate the topological map of the helmet material structure.

[0034] The signal acquisition module is used to extract the path curvature from the topological diagram of the helmet material structure and perform polynomial fitting using the least squares method to generate an optimized scanning trajectory. The optimized scanning trajectory is converted into a robotic arm control instruction to drive the ultrasonic probe to scan the helmet structure with multi-frequency focusing and incident angle deflection to obtain the original echo signal matrix;

[0035] The solution generation module is used to input the original echo signal matrix into the acoustic physics neural model. The feature extraction layer performs time-frequency feature analysis to generate the sound wave propagation feature tensor. The physical constraint layer performs sound velocity distortion correction to obtain the calibration echo signal matrix. The sound wave propagation feature tensor and the calibration echo signal matrix are subjected to multi-scale time-frequency analysis and nonlinear transformation to output the non-destructive testing solution.

[0036] The dynamic optimization module is used to perform dynamic node clustering and edge strength redistribution on the helmet material structure topology graph based on the non-destructive testing scheme through an improved incremental weight graph algorithm, simultaneously updating the node weights and optimizing the edge connection strength.

[0037] In a third aspect, the present invention provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program is executed by the processor, any step of the non-destructive testing method for the structural uniformity of a composite material bulletproof helmet as described in the first aspect of the present invention is implemented.

[0038] In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, any step of the non-destructive testing method for the structural uniformity of a composite material bulletproof helmet as described in the first aspect of the present invention is implemented.

[0039] The present invention achieves the following beneficial effects: by mapping helmet pore locations and connectivity paths to mechanical parameters using a topological map of the helmet material structure, this method enables quantitative analysis of the distribution of microscopic defects within the composite material and their connectivity characteristics, and establishes a correlation between microscopic and macroscopic performance. A collaborative detection mechanism, employing an acoustic physics neural model and dynamic scanning trajectory optimization, improves the accuracy of ultrasonic wave propagation modeling in complex microstructures, significantly improving correction errors caused by sound velocity distortion, and thus significantly enhancing the overall accuracy and reliability of nondestructive testing. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0041] Figure 1 Flowchart of the nondestructive testing method for the structural uniformity of composite bullet-proof helmets.

[0042] Figure 2 Schematic diagram of the nondestructive testing system for the structural uniformity of composite bullet-proof helmets.

[0043] Figure 3 Flowchart for generating a topological map of the helmet material structure.

[0044] Figure 4 Flowchart for outputting nondestructive testing plan. DETAILED DESCRIPTION

[0045] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0046] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0047] Secondly, the term "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in various places throughout this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive of other embodiments.

[0048] Reference Figures 1 to 4, is an embodiment of the present invention, which provides a non-destructive testing method for the structural uniformity of a composite material bulletproof helmet, comprising the following steps:

[0049] S1. Collect a helmet microstructure dataset, map the helmet pore positions into nodes of the topological structure, map the helmet connectivity paths into edges of the topological structure, and generate a helmet material structure topological map.

[0050] The specific steps are as follows:

[0051] S1.1. Collect a helmet microstructure dataset, which includes helmet pore locations, helmet connectivity paths, helmet 3D point clouds, and composite material performance data.

[0052] Deploy image sensors on the outer surface of the helmet to collect data on the location of helmet pores; deploy distributed fiber optic sensors in the helmet interlayer to monitor data on the helmet's connectivity paths; and deploy laser 3D scanners around the helmet to collect data on the helmet's 3D point cloud.

[0053] Composite material performance data includes mechanical parameters, thermal parameters, electrical parameters and acoustic parameters; mechanical parameters are collected by a universal testing machine, thermal parameters are collected by a differential scanning calorimeter, electrical parameters are obtained by a high resistance meter, and acoustic parameters are collected by an ultrasonic testing device;

[0054] The helmet microstructure dataset can not only accurately reflect the topological characteristics of the pore distribution and connectivity paths inside the helmet, but also provide a high-precision data basis for subsequent defect evolution analysis, thereby comprehensively improving the accuracy and reliability of the structural uniformity assessment of bulletproof helmets.

[0055] S1.2. Preprocess the helmet microstructure dataset. In the specific operation, the outlier detection method based on the 3σ criterion is used to perform outlier identification and filtering on the helmet microstructure dataset, and a sliding window is used for moving average filtering to eliminate random fluctuations in the data acquisition process; at the same time, the sym8 wavelet basis is used to perform a 5-layer decomposition of the helmet microstructure dataset to eliminate high-frequency noise and ensure the stability of the time-frequency characteristics in the helmet microstructure dataset; the B-spline interpolation is used to spatially resample the helmet microstructure dataset to eliminate the resolution difference between different sensors; the ICP algorithm is used to perform data registration and spatial alignment on the helmet microstructure dataset to achieve sub-pixel matching of cross-modal data, and dynamic time warping is used for timestamp alignment to eliminate acquisition timing deviation; then the min-max scaling is used to normalize the helmet microstructure dataset to avoid dimensionality differences, and the preprocessed helmet microstructure dataset is output.

[0056] S1.3. The coordinates of the stress area of ​​the helmet pore position are extracted by CAE topological analysis method to obtain the pore space coordinates, and the geometric properties of the pore space coordinates are mapped to form nodes of the topological structure. In the specific operation, the CAE topological analysis method is used to perform least squares surface fitting on the helmet pore position to obtain the initial pore distribution data; the initial pore distribution data is time-series aligned and spatially positioned by dynamic time warping to obtain the timestamp and spatial position; the timestamp of the initial pore distribution data is used as the x-axis in the three-dimensional space, and the spatial position is used as the y-axis to construct a local coordinate system, and the grid density is adjusted on the local coordinate system to generate a non-uniform sampling grid; the Gaussian integral method is used to perform stress numerical sampling on the non-uniform sampling grid to obtain the stress value of each node, and the stress value of each node is interpolated by cubic spline to generate the stress area distribution; DBSCAN density clustering is used to perform domain aggregation and core point extraction on the stress area distribution to obtain the pore space coordinates. The pore space coordinates are the mechanically sensitive positions that need to be focused on, which can identify the potential failure origin area of ​​the helmet under impact.

[0057] The local binary value method (LBP) is used to perform feature labeling on the pore space coordinates to generate a pore feature descriptor. The geometric parameter statistical algorithm is used to perform multi-dimensional geometric quantization on the pore feature descriptor to obtain the geometric properties of the pore space coordinates. The geometric properties are reduced in dimension through principal component analysis to obtain a core feature vector with mechanical-geometric composite characteristics. The core feature vector is converted into JSON format through Z-score standardization and mapped through the add_node function of the NetworkX library to generate nodes of the topological structure.

[0058] S1.4. Apply the geometric adjacency algorithm to connect the adjacent pores of the helmet connectivity path to generate physical channel segments; perform transmission direction mapping on the physical channel segments to obtain the edges of the topological structure. In the specific operation, the KD-Tree spatial index is used to quickly retrieve the k-nearest neighbors (k=5~8) of each helmet connectivity path, and triangulation is performed on the k-nearest neighbors to obtain the initial adjacency relationship; based on the initial adjacency relationship, the geometric adjacency algorithm is applied to perform spatial topological connection on the helmet connectivity path to generate initial connection lines, and the initial connection lines are connected to adjacent pores to obtain the initial pore network; the principal stress direction of the initial pore network is extracted through stress tensor analysis, and the path is fitted through B-spline interpolation to form physical channel segments.

[0059] The physical channel segments are filtered and smoothed through a sliding window, and stress gradient weights are assigned to generate weighted connection paths. The principal stress tracing algorithm is used to extract the directions of the weighted connection paths and generate standardized transfer directions. The standardized transfer directions are mapped through the GraphML (Graph Markup Language) protocol to obtain edges passing through the topological structure.

[0060] S1.5. Perform weighted fusion on the nodes and edges of the topological structure to generate a topological diagram of the helmet material structure. Specifically, DBSCAN clustering is used to expand the neighborhood of the nodes of the topological structure, and the nodes of adjacent topological structures are clustered and merged to obtain a fused node set. The least squares method is simultaneously applied to perform non-uniform surface fitting on the fused node set to generate a transition surface. The principal stress tracking algorithm is used to perform stress gradient integration on the edges of the topological structure to obtain a mechanical transfer path, and isogeometric analysis (IGA) is applied to the mechanical transfer path for stress field coupling to obtain a stress weighted path. The transition surface and the stress weighted path are weightedly fused by the gradient descent optimization method to generate a topological diagram of the helmet material structure.

[0061] S2. Extract the path curvature from the helmet material structure topology diagram and use the least squares method to perform polynomial fitting to generate an optimized scanning trajectory. The optimized scanning trajectory is converted into a robotic arm control instruction to drive the ultrasonic probe to perform multi-dimensional coupled scanning on the helmet structure and obtain the original echo signal matrix.

[0062] The specific steps are as follows:

[0063] S2.1. Use B-spline interpolation to perform parameterized sampling on the topological map of the helmet material structure to obtain a discrete point set, perform numerical differentiation on the discrete point set, and generate the path curvature. In the specific operation, use wavelet transform to perform multi-scale decomposition on the topological map of the helmet material structure to extract geometric features; use principal component analysis to perform dimensionality reduction processing on the geometric features to obtain a key feature point set; apply B-spline interpolation to perform parameterized sampling on the key feature point set to generate a parameterized geometric representation, and perform equidistant segmentation on the parameterized geometric representation to obtain a uniform parameter grid point set; use isoparametric transformation to perform coordinate mapping on the uniform parameter grid point set, and use Gaussian integral to perform weight distribution to form a discrete point set; then apply the central difference method to perform numerical differentiation on the discrete point set to calculate the first-order derivative and second-order derivative, and at the same time use the curvature formula in differential geometry to perform curvature field calculation on the first-order derivative and second-order derivative to generate the path curvature. The specific mathematical formula is as follows,

[0064] ;

[0065] in, represents the path curvature, represents the parameter index of the discrete point set, Representation parameters The first derivative of represents the second-order derivative of parameter u.

[0066] S2.2. Perform noise suppression and feature enhancement on the path curvature to generate an optimized curvature distribution; use the least squares method to perform polynomial fitting on the optimized curvature distribution to form an optimized scanning trajectory. In the specific operation, the path curvature is decomposed and frequency-segmented by selecting the Symlets wavelet basis to obtain high-frequency coefficients, and the high-frequency coefficients are suppressed by using a soft threshold to retain the main curvature features while eliminating high-frequency noise to obtain the denoised path curvature; the gradient change characteristics of the denoised path curvature are extracted, and the gradient change characteristics are divided into local intervals through a sliding window to obtain the gradient change intervals, and the differential operator is used to locate the extreme values ​​of the gradient change intervals to obtain the curvature extreme points; the curvature extreme points are weightedly aggregated using a Gaussian weighted function to enhance the feature significance of the curvature mutation area, and the amplitude is adjusted through Min-Max normalization to generate normalized curvature features, and the normalized curvature features are weighted integrated to obtain the optimized curvature distribution;

[0067] It should be noted that the soft threshold is defined based on the standard deviation of the high-frequency coefficient, and its value range is [0.5σ, 2σ] (σ is the standard deviation of the high-frequency coefficient); the difference operator is defined based on the central difference format of the gradient change interval, and its value range is [-1, 0,1].

[0068] The optimized curvature distribution is decomposed into an upper triangular matrix through QR decomposition, and the upper triangular matrix is ​​optimized by least squares optimization to obtain the optimal coefficients. According to the optimal coefficients, the optimized curvature distribution is fitted with a polynomial using the least squares method to obtain a parameterized trajectory. The parameterized trajectory is locally smoothed using a Savitzky-Golay filter to form a continuous scanning trajectory, which is then discretized and interpolated to optimize the continuity of the curvature and obtain the optimized scanning trajectory.

[0069] S2.3. Apply dynamic time warping to discretize the timestamps of the optimized scanning trajectory to obtain the angle instruction sequence, and encapsulate the angle instruction sequence to form the robot arm control instruction. In the specific operation, the curvature parameters of the optimized scanning trajectory are extracted, and the parameters are compressed through Min-Max scaling to obtain key frames; the time axis of the key frames is nonlinearly aligned with the standard time axis through the dynamic time warping algorithm to obtain the time offset of each key frame; then the time offset is uniformly resampled to form a discretized timestamp sequence; based on the discretized timestamp sequence, the optimized scanning trajectory is reconstructed through NURBS curve fitting to obtain the end pose, and the end pose is mapped to generate joint space coordinates; the joint space coordinates are smoothly interpolated using quintic polynomial interpolation to obtain the angle instruction sequence;

[0070] It should be noted that the standard timeline refers to the standard time template of the robot arm movement, which is defined based on the specific task cycle time.

[0071] According to the EtherCAT industrial communication protocol, the angle instruction sequence is packaged and encapsulated to obtain the instruction data block. At the same time, a safety check field and an emergency stop instruction slot are added to the instruction data block to form a robotic arm control instruction that can be directly issued.

[0072] S2.4. According to the control instructions of the robotic arm, the ultrasonic probe is driven to perform multi-frequency focusing and incident angle deflection scanning on the helmet structure, and multi-channel echo signals are obtained. The multi-channel echo signals are aligned in space and time to generate the original echo signal matrix. In the specific operation, at the scanning end, the control instructions of the robotic arm are sent to the EtherCAT master station to drive the ultrasonic probe of the robotic arm to perform multi-plane dynamic scanning on the helmet structure. At the same time, during the scanning process, the phased array transmitting unit of the ultrasonic probe is used to transmit focused ultrasonic waves in a multi-frequency mode to ensure that the sound beam is accurately focused and the phase delay is used. The controller deflects the beam of the ultrasonic probe to achieve a fixed range (such as ±30°) of incident angle deflection, improve detection coverage, and form a spatial scanning grid. At the receiving end, the spatial scanning grid is synchronously sampled to obtain multi-channel echo signals. At the same time, the multi-channel echo signals are bandpass filtered using a digital filter to generate frequency domain separation signals. The frequency domain separation signals are then digitized using a high-speed ADC to obtain a time domain signal sequence. Dynamic time warping is used to compensate for time delays in the time domain signal sequence, complete spatial alignment, and generate the original echo signal matrix.

[0073] S3. Input the original echo signal matrix into the acoustic physical neural model. The feature extraction layer performs time-frequency feature analysis. The physical constraint layer performs sound velocity distortion correction and outputs a nondestructive testing solution.

[0074] The specific steps are as follows:

[0075] S3.1. Construct and train an acoustic physics neural model. Specifically, in the TensorFlow framework, call a three-dimensional convolutional network through Keras parameters, set the size of the three-dimensional convolution kernel to 5×5×5, the step size to 1×1×1, and follow the three-dimensional convolution with BatchNormalization to normalize the features to improve training stability. Simultaneously, use the ReLU activation function for nonlinear transformation to complete the construction of the feature extraction layer. Call a three-layer fully connected network through the nn.Module parameter, embed the elastic constitutive equation pair into the three-layer fully connected network, set the number of hidden layers to 3, the number of neurons to 128, and follow the three-layer fully connected network with Dropout for regularization to prevent overfitting. At the same time, use L2 regularization for weight attenuation to complete the construction of the physical constraint layer.

[0076] The feature extraction layer and the physical constraint layer are multi-dimensionally fused using residual connections to obtain a fused feature tensor, which is then transferred to generate a high-level feature representation. The generated high-level feature representation is weighted using an attention mechanism to obtain weighted features. The weighted features are dynamically modulated using a gating mechanism to form optimized weights. Based on the optimized weights, the feature extraction layer and the physical constraint layer are stacked using cross-layer connections and feature compressed using average pooling to complete the construction of the acoustic physical neural model.

[0077] Next, the acoustic-physical neural model is trained. Furthermore, the original echo signal matrix is ​​divided into a sample set, a training set, and a validation set. On the sample set, data augmentation is performed using random flipping, and normalization is performed using Z-score standardization to generate enhanced sample data. On the training set, gradient updates are performed on the enhanced sample data using backpropagation, and the Adam optimizer is simultaneously applied to dynamically adjust the learning rate to obtain the training loss. On the validation set, the mean squared error loss function is used to quantify the training loss to obtain the validation error. Training terminates when the validation error exceeds the convergence threshold for five consecutive epochs, and the trained acoustic-physical neural model is output synchronously.

[0078] It should be noted that the convergence threshold is defined based on the statistical distribution of the verification error and its value range is [0.01-0.05].

[0079] S3.2. Input the original echo signal matrix into the acoustic physical neural model. The feature extraction layer performs multi-dimensional feature analysis through the three-dimensional convolution kernel to generate the sound wave propagation feature tensor. In the specific operation, the original echo signal matrix is ​​input into the acoustic physical neural model through the Input interface, and the original echo signal matrix is ​​subjected to multi-dimensional feature analysis through the three-dimensional convolution kernel. The time dimension convolution captures the sound wave propagation characteristics, the frequency dimension convolution extracts the multi-frequency component features, and the space dimension convolution extracts the spatial correlation features. The time-frequency feature analysis, multi-frequency component features and spatial correlation features are spliced ​​in the feature channel to obtain a multi-dimensional feature map. During the convolution process, SAME padding is used to fill the boundaries of the multi-dimensional feature map to ensure the consistency of the time and space dimensions of the multi-dimensional feature map. The filled multi-dimensional feature map is standardized by unit variance through batch normalization to eliminate the dimensional differences between channels. The ReLU activation function is then used for nonlinear transformation to suppress negative features and enhance the expression ability, and the standardized feature map is output. The principal component analysis is applied to perform dimensionality reduction processing on the standardized feature map to form the sound wave propagation feature tensor.

[0080] S3.3, the physical constraint layer embeds the elastic constitutive equation to perform sound velocity distortion correction on the acoustic wave propagation characteristic tensor and obtain the calibration echo signal matrix. In the specific operation, the acoustic wave propagation characteristic tensor is input into the three-layer fully connected network embedded with the elastic constitutive equation. The first layer of the network flattens the acoustic wave propagation characteristic tensor and maps it to the latent space through the fully connected function. At the same time, the composite material performance data is injected into the latent space as a physical constraint condition through 1×1 convolution, and the latent feature is output. The second layer of the network calculates the sound velocity distortion gradient through the elastic constitutive equation, and performs Hadamard product on the sound velocity distortion gradient and the latent feature to obtain the sound velocity distortion correction value. The specific mathematical formula is as follows:

[0081] ;

[0082] in, Indicates the sound velocity distortion correction value, represents the identity matrix, represents the flattened acoustic wave propagation characteristic tensor, represents the sound velocity distortion gradient, represents the Hadamard product;

[0083] It should be noted that the identity matrix refers to a square matrix whose main diagonal elements are all 1 and the rest of the elements are all 0, which is called by the numpy.eye function;

[0084] According to the sound velocity distortion correction value, the gating mechanism is applied to perform weighted correction on the latent features and output the corrected features. The third-layer network uses the ReLU activation function to perform nonlinear transformation on the corrected features and performs gradient clipping through gradient truncation to form a calibration echo signal matrix.

[0085] S3.4. Apply wavelet packet energy spectrum to perform multi-scale time-frequency analysis and nonlinear transformation on the calibration echo signal matrix to obtain structural uniformity parameters and defect probability distribution, and integrate the structural uniformity parameters with the defect probability distribution to output a non-destructive testing plan. In the specific operation, apply wavelet packet energy spectrum to perform multi-layer decomposition on the calibration echo signal matrix to obtain sub-band signals of different frequency bands; use wavelet packet transform algorithm to perform energy integration on sub-band signals of different frequency bands to obtain sub-band energy spectrum, and perform multi-scale time-frequency analysis on the sub-band energy spectrum to obtain time-frequency energy matrix. The time-frequency energy matrix can clearly present the energy change characteristics of the internal structure of the composite material; then use Sigmoid function to perform nonlinear transformation on the time-frequency energy matrix to obtain normalized energy value, and use probability density function to map the normalized energy value to the standard probability interval to generate a probability distribution matrix; use principal component analysis to perform dimension compression and parameter extraction on the probability distribution matrix to obtain structural uniformity parameters; use kernel density estimation to perform probability density fitting on the structural uniformity parameters to form a defect probability distribution;

[0086] Finally, the Bayesian fusion algorithm is used to perform weighted integration of structural uniformity parameters and defect probability distribution, and thermal map rendering is performed through the Unity rendering engine to generate comprehensive inspection results. The comprehensive inspection results are formatted and output according to the ISO standard template to form a non-destructive testing plan.

[0087] S4. Based on the nondestructive testing scheme, the node weights of the helmet material structure topology graph are updated through the improved incremental weight graph algorithm, and the connection strength of the edges is optimized simultaneously.

[0088] The specific steps are as follows:

[0089] S4.1. Use wavelet packet decomposition to extract defect-sensitive frequency band parameters in the non-destructive testing scheme. In the specific operation, use a sliding window to perform signal segmentation and windowing processing on the non-destructive testing scheme to obtain pre-processed signal segments, and use the db4 wavelet basis function to perform wavelet packet decomposition on the pre-processed signal segments to form multi-scale sub-band components; then use the peak detection method to extract features from the multi-scale sub-band components to obtain characteristic data of each frequency band; apply linear discriminant analysis to the characteristic data of each frequency band for feature selection to generate key frequency band feature vectors; next, use K-means clustering to perform weighted aggregation and parameter optimization on the key frequency band feature vectors to form a sensitive frequency band set, and randomly sample the sensitive frequency band set to obtain defect-sensitive frequency band parameters.

[0090] S4.2. According to the defect-sensitive frequency band parameters, the improved incremental weighted graph algorithm is used to perform node dynamic clustering and edge strength redistribution on the helmet material structure topology map to obtain the sound velocity gradient coefficient and stiffness distribution coefficient. In the specific operation, the defect-sensitive frequency band parameters are input into the helmet material structure topology map through feature projection, and the Gaussian kernel function is used for nonlinear transformation to enhance the defect feature representation and obtain a defect enhanced topology map; a sliding window is used to scan the defect enhanced topology map, and region division is performed within the window to form a local feature map sub-block; the improved incremental weighted graph algorithm is used to perform local clustering and parameter extraction on the local feature map sub-block to obtain Select candidate feature nodes; apply spectral clustering to dynamically aggregate the candidate feature nodes, merge similar nodes, and obtain node clustering results; simultaneously use the Sobel operator to perform gradient analysis on the node clustering results to obtain the defect probability gradient. Based on the defect probability gradient, use the gradient threshold to redistribute the weights of the edges. For example, define the gradient threshold based on the statistical distribution of the defect probability gradient and set the gradient threshold to 0.7. When the defect probability gradient is higher than the gradient threshold, it is defined as a high-gradient region, and the edge strength weight is set to 1.5. When the defect probability gradient is lower than the gradient threshold, it is defined as a low-gradient region, and the edge strength weight is set to 0.8.

[0091] It should be noted that the Sobel operator is directly called through the cv2.Sobel function of the OpenCV library.

[0092] Through breadth-first search, the defect enhancement topology graph that completes node clustering and edge strength redistribution should be subjected to neighborhood expansion and path search to obtain the defect propagation path; then the A* algorithm is used to quantify the cost of the defect propagation path to obtain the path cost matrix, and the path cost matrix is ​​normalized by Min-Max to obtain the sound velocity gradient coefficient, which reflects the degree of distortion of the sound wave propagation path; the Laplace matrix is ​​used to decompose the feature space of the defect enhancement topology graph that completes node clustering and edge strength redistribution, and the ratio of the first three non-zero eigenvalues ​​is selected and logarithmically transformed to obtain the stiffness distribution coefficient, which characterizes the anisotropy of material stiffness;

[0093] It should be noted that the Laplace matrix refers to the matrix that represents the topological structure in the defect enhancement topology graph, which is obtained by performing a difference operation between the adjacency relationship and the node degree on the defect enhancement topology graph;

[0094] It should be noted that the improved incremental weighted graph algorithm has made the following improvements based on the existing incremental weighted graph algorithm. On the one hand, a dynamic weight adjustment mechanism based on the defect probability gradient is used to improve the sensitivity of defect area features. On the other hand, the sliding window area division and spectral clustering dynamic aggregation are combined to achieve the coordinated optimization of node and edge strength.

[0095] S4.3. Based on the sound velocity gradient coefficient, the node weights are gradient updated. At the same time, the edge connection strength is coupled and optimized according to the stiffness distribution coefficient. In the specific operation, the gradient descent method is used with the sound velocity gradient coefficient as the learning rate adjustment factor to dynamically adjust the weights of each node in the defect enhancement topology to obtain a preliminary optimized node set; based on the preliminary optimized node set, the parameters are fine-tuned through back propagation, and the node weights are iteratively updated to obtain a stable node weight distribution; at the same time, linear interpolation is used to perform multi-scale sampling and feature fusion on the stiffness distribution coefficient, and the strength is normalized through Z-score standardization to generate edge strength optimization parameters; based on the edge strength optimization parameters, the edge connection strength is coupled and optimized, and Gaussian filtering is used for local smoothing to output the collaboratively optimized helmet material structure topology.

[0096] This embodiment also provides a nondestructive testing system for the structural uniformity of a composite bulletproof helmet, comprising: a topological mapping module, a signal acquisition module, a solution generation module, and a dynamic optimization module;

[0097] The topological mapping module is used to collect the helmet microstructure data set, extract the force-bearing area coordinates of the helmet pore position, obtain the pore space coordinates, and map the geometric properties of the pore space coordinates to form the nodes of the topological structure. The geometric adjacency algorithm is applied to connect the adjacent pores of the helmet connection path to generate physical channel segments. The transmission direction of the physical channel segments is mapped to obtain the edges of the topological structure. The nodes and edges of the topological structure are weightedly fused to generate the topological map of the helmet material structure.

[0098] The signal acquisition module is used to extract the path curvature from the topological diagram of the helmet material structure and perform polynomial fitting using the least squares method to generate an optimized scanning trajectory. The optimized scanning trajectory is converted into a robotic arm control instruction to drive the ultrasonic probe to scan the helmet structure with multi-frequency focusing and incident angle deflection to obtain the original echo signal matrix;

[0099] The solution generation module is used to input the original echo signal matrix into the acoustic physics neural model. The feature extraction layer performs time-frequency feature analysis to generate the sound wave propagation feature tensor. The physical constraint layer performs sound velocity distortion correction to obtain the calibration echo signal matrix. The sound wave propagation feature tensor and the calibration echo signal matrix are subjected to multi-scale time-frequency analysis and nonlinear transformation to output the non-destructive testing solution.

[0100] The dynamic optimization module is used to perform dynamic node clustering and edge strength redistribution on the helmet material structure topology graph based on the non-destructive testing scheme through an improved incremental weight graph algorithm, simultaneously updating the node weights and optimizing the edge connection strength.

[0101] This embodiment also provides a computer device suitable for the non-destructive testing method for the structural uniformity of a composite bullet-proof helmet, comprising: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the non-destructive testing method for the structural uniformity of a composite bullet-proof helmet as proposed in the above embodiment.

[0102] The computer device may be a terminal, comprising a processor, memory, a communication interface, a display, and an input device connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores an operating system and computer programs. The internal memory provides an environment for the operating system and computer programs stored in the non-volatile storage media. The communication interface of the computer device is used to communicate with external terminals via wired or wireless communication. Wireless communication may be achieved via Wi-Fi, a carrier network, NFC (near-field communication), or other technologies. The display of the computer device may be a liquid crystal display or an electronic ink display. The input device may be a touchscreen overlay on the display, buttons, a trackball, or a touchpad on the computer device housing, or an external keyboard, touchpad, or mouse.

[0103] This embodiment also provides a storage medium having a computer program stored thereon, which, when executed by a processor, implements the nondestructive testing method for the structural uniformity of a composite material bulletproof helmet as proposed in the above embodiment; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk or optical disk.

[0104] In summary, this invention establishes a mapping relationship between helmet pore locations and connectivity paths and mechanical parameters using a topological map of the helmet material structure. This allows for quantitative analysis of the distribution of microscopic defects within composite materials and their connectivity characteristics, thereby correlating microscopic and macroscopic performance. A collaborative detection mechanism, combining an acoustic physics neural model with dynamic scanning trajectory optimization, improves the accuracy of ultrasonic wave propagation modeling in complex microstructures, significantly mitigates correction errors caused by sound velocity distortion, and thus significantly enhances the overall accuracy and reliability of nondestructive testing.

[0105] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A nondestructive testing method for the structural uniformity of a composite bulletproof helmet, characterized by: include, Collect a helmet microstructure dataset, extract the force-bearing area coordinates of the helmet pore positions, obtain the pore space coordinates, and map the geometric properties of the pore space coordinates to form nodes of the topological structure. Apply a geometric adjacency algorithm to connect adjacent pores in the helmet connection path to generate physical channel segments. Map the transmission direction of the physical channel segments to obtain the edges of the topological structure. Perform weighted fusion of the topological structure nodes and edges to generate a topological map of the helmet material structure. The path curvature in the helmet material structure topology is extracted and polynomial fitting is performed using the least squares method to generate an optimized scanning trajectory. The optimized scanning trajectory is converted into a robotic arm control instruction, which drives the ultrasonic probe to scan the helmet structure with multi-frequency focusing and incident angle deflection to obtain the original echo signal matrix. The original echo signal matrix is ​​input into the acoustic physics neural model. The feature extraction layer performs time-frequency feature analysis to generate the sound wave propagation feature tensor. The physical constraint layer performs sound velocity distortion correction to obtain the calibration echo signal matrix. The sound wave propagation feature tensor and the calibration echo signal matrix are subjected to multi-scale time-frequency analysis and nonlinear transformation to output the non-destructive testing solution. Based on the non-destructive testing scheme, the improved incremental weighted graph algorithm is used to perform dynamic node clustering and edge strength redistribution on the helmet material structure topology graph, and the node weights are updated simultaneously and the edge connection strength is optimized.

2. The nondestructive testing method for the structural uniformity of a composite material bulletproof helmet according to claim 1, wherein: The helmet microstructure dataset includes helmet pore locations, helmet connection paths, helmet three-dimensional point clouds, and composite material performance data.

3. The nondestructive testing method for the structural uniformity of a composite material bulletproof helmet according to claim 1, wherein: The method of extracting the path curvature in the helmet material structure topology diagram specifically includes the following steps: B-spline interpolation is used to perform parameterized sampling on the topological map of the helmet material structure to obtain a discrete point set, which is then numerically differentiated to generate the path curvature.

4. The nondestructive testing method for the structural uniformity of a composite material bulletproof helmet according to claim 3, wherein: The generation of the optimized scanning trajectory specifically includes the following steps: Noise suppression and feature enhancement are performed on the path curvature to generate an optimized curvature distribution; polynomial fitting is performed on the optimized curvature distribution using the least squares method to form an optimized scanning trajectory.

5. The nondestructive testing method for the structural uniformity of a composite material bulletproof helmet according to claim 1, wherein: The process of converting the optimized scanning trajectory into a robot arm control instruction specifically includes the following steps: Dynamic time warping is applied to perform timestamp discretization on the optimized scanning trajectory to obtain the angle instruction sequence, which is then encapsulated to form the robotic arm control instructions.

6. The nondestructive testing method for structural uniformity of a composite material bulletproof helmet according to claim 5, characterized in that: The acquisition of the original echo signal matrix specifically includes the following steps: According to the control instructions of the robotic arm, the ultrasonic probe is driven to scan the helmet structure with multi-frequency focusing and incident angle deflection to obtain multi-channel echo signals; The multi-channel echo signals are space-time aligned to generate the original echo signal matrix.

7. The nondestructive testing method for structural uniformity of a composite material bulletproof helmet according to claim 1, wherein: The specific construction process of the acoustic physical neural model is as follows: Build and initialize the feature extraction layer and physical constraint layer, perform feature transfer and double stacking through residual connections, and construct an acoustic physical neural model.

8. The nondestructive testing method for structural uniformity of a composite material bulletproof helmet according to claim 1, wherein: The output nondestructive testing scheme specifically includes the following steps: The original echo signal matrix is ​​input into the acoustic physics neural model. The feature extraction layer performs multi-dimensional feature analysis through a three-dimensional convolution kernel to generate a sound wave propagation feature tensor. The physical constraint layer is embedded in the elastic constitutive equation to perform sound velocity distortion correction on the acoustic wave propagation characteristic tensor to obtain the calibration echo signal matrix; Apply wavelet packet energy spectrum to perform multi-scale time-frequency analysis and nonlinear transformation on the calibration echo signal matrix to obtain structural uniformity parameters and defect probability distribution; Integrate structural uniformity parameters with defect probability distribution to output nondestructive testing solutions.

9. The nondestructive testing method for the structural uniformity of a composite material bulletproof helmet according to claim 8, wherein: The updating of the node weights of the helmet material structure topology graph and the simultaneous optimization of the edge connection strengths specifically include the following steps: Use wavelet packet decomposition to extract defect-sensitive frequency band parameters in non-destructive testing schemes; According to the defect-sensitive frequency band parameters, an improved incremental weighted graph algorithm is used to perform node dynamic clustering and edge strength redistribution on the helmet material structure topology graph to obtain the sound velocity gradient coefficient and stiffness distribution coefficient. Based on the sound velocity gradient coefficient, the node weights are gradient updated, and the edge connection strength is coupled optimized according to the stiffness distribution coefficient.

10. A nondestructive testing system for the structural uniformity of a composite material bulletproof helmet, based on the nondestructive testing method for the structural uniformity of a composite material bulletproof helmet according to any one of claims 1 to 9, characterized in that: Including topology mapping module, signal acquisition module, solution generation module, dynamic optimization module; The topological mapping module is used to collect the helmet microstructure data set, extract the force-bearing area coordinates of the helmet pore position, obtain the pore space coordinates, and map the geometric properties of the pore space coordinates to form the nodes of the topological structure. The geometric adjacency algorithm is applied to connect the adjacent pores of the helmet connection path to generate physical channel segments. The transmission direction of the physical channel segments is mapped to obtain the edges of the topological structure. The nodes and edges of the topological structure are weightedly fused to generate the topological map of the helmet material structure. The signal acquisition module is used to extract the path curvature from the topological diagram of the helmet material structure and perform polynomial fitting using the least squares method to generate an optimized scanning trajectory. The optimized scanning trajectory is converted into a robotic arm control instruction to drive the ultrasonic probe to scan the helmet structure with multi-frequency focusing and incident angle deflection to obtain the original echo signal matrix; The solution generation module is used to input the original echo signal matrix into the acoustic physics neural model. The feature extraction layer performs time-frequency feature analysis to generate the sound wave propagation feature tensor. The physical constraint layer performs sound velocity distortion correction to obtain the calibration echo signal matrix. The sound wave propagation feature tensor and the calibration echo signal matrix are subjected to multi-scale time-frequency analysis and nonlinear transformation to output the non-destructive testing solution. The dynamic optimization module is used to perform dynamic node clustering and edge strength redistribution on the helmet material structure topology graph based on the non-destructive testing scheme through an improved incremental weight graph algorithm, simultaneously updating the node weights and optimizing the edge connection strength.

Images