A method for online automatic testing of the compressive strength of corrugated boxes

Abstract

The present invention provides a method for online automatic testing of the compressive capacity of corrugated cardboard boxes, belonging to the technical field of corrugated cardboard box production. The present invention constructs a micro-deformation loading system to apply a test load that does not exceed 10% of the expected maximum load-bearing capacity, uses a laser displacement sensor array to measure the micro-deformation displacement data of the six surfaces of the cardboard box, and simultaneously collects acoustic emission signals and extracts spectral characteristic parameters. Based on the physical mechanism model of the cardboard fiber structure, the micro-deformation data is analyzed, the stress field distribution is calculated, and the analysis is combined with the hierarchical attention cardboard strength assessment model. The stiffness coefficient is calculated and compared with the benchmark threshold. The strain energy distribution diagram is used to identify structural weak areas and quantify the damage risk index. The predicted value of the compressive strength of the corrugated cardboard box and its confidence interval are comprehensively evaluated, and the prediction result is output, thereby achieving an accurate assessment of the compressive capacity of the corrugated cardboard box under non-destructive conditions.

Description

Technical Field

[0001] The invention belongs to the technical field of corrugated paper box production, and in particular relates to a method for online automatic testing of the compression resistance of a corrugated paper box. Background Art

[0002] Corrugated cardboard is one of the most widely used packaging materials in modern logistics and packaging. Its compressive performance is directly related to the safety of packaged goods. Traditionally, the compressive resistance of corrugated cardboard has been evaluated primarily through destructive testing. This involves applying gradually increasing pressure to the cardboard using specialized pressure testing equipment until the structure fails, and recording the maximum load as an indicator of compressive strength. Alternatively, there are empirical calculation methods, such as the Mackey formula, which estimate the compressive strength of cardboard using basic cardboard parameters, but these methods have limited accuracy.

[0003] However, traditional destructive testing not only wastes samples but also cannot be applied to real-time quality control on production lines. Empirical methods, on the other hand, fail to fully consider the material's microstructural properties and local deformation behavior, making it difficult to accurately assess actual compressive performance. This is particularly true when dealing with cartons of varying sizes and materials, leading to significant prediction errors. For packaging high-end precision products, such errors can lead to significant financial losses.

[0004] Furthermore, existing technologies struggle to effectively identify weak areas within the carton structure, preventing early warning of potential damage risks, which is crucial for production line quality control. Accurately predicting the compressive strength of corrugated cartons and identifying structural weaknesses without damaging them has become a key technical challenge urgently needed to be addressed in the industry. Summary of the Invention

[0005] In view of this, the present invention provides a method for online automatic testing of the compressive strength of a corrugated box, which can solve the technical problem in the prior art that it is impossible to accurately predict the compressive strength of the corrugated box without destroying it.

[0006] The present invention is implemented as follows: The present invention provides a method for online automatic testing of the compressive resistance of corrugated cardboard boxes, including: constructing a micro-deformation loading system to apply a test load to the top of the corrugated cardboard box; using a high-precision laser displacement sensor array to measure micro-deformation displacement data; synchronously collecting acoustic emission signals and extracting spectral characteristic parameters; analyzing micro-deformation displacement data based on a physical mechanism model of the cardboard fiber structure, and applying a stress distribution calculation equation to calculate the stress field distribution of the corrugated cardboard box wall; inputting the micro-deformation displacement data and spectral characteristic parameters into a pre-trained hierarchical attention cardboard box strength assessment model for analysis; calculating the stiffness coefficient and comparing it with a benchmark threshold; identifying structural weak areas and quantifying the damage risk index; using a multi-source data fusion algorithm to comprehensively evaluate the compressive strength prediction value and confidence interval; and outputting the test results.

[0007] The test load refers to the pressure applied to the top of the corrugated box by the micro-deformation loading system, and its magnitude is within 10% of the expected maximum load-bearing capacity.

[0008] Among them, the micro-deformation loading system refers to a pressure-applying device driven by a high-precision servo motor, which ensures a smooth and precise loading process through closed-loop feedback control. The loading force range is 50N to 2000N, and the accuracy is not less than 0.5%.

[0009] Among them, the high-precision laser displacement sensor array refers to a measurement network composed of 24 triangulation principle laser displacement sensors with a measurement accuracy of 1 micron and a sampling frequency of 1000Hz, which is used to capture tiny deformations on the surface of the carton.

[0010] The micro-deformation displacement data refers to a set of numerical values ​​of deformation displacement of six surfaces of the corrugated box under the action of a test load, collected by a high-precision laser displacement sensor array.

[0011] The acoustic emission signal refers to the high-frequency sound wave signal generated by the internal fibers of the corrugated box under the action of the test load, with a frequency range of 20kHz to 200kHz, which is collected by a piezoelectric sensor array.

[0012] The spectrum characteristic parameters refer to a set of characteristic parameters extracted after spectrum analysis of the acoustic emission signal, including energy distribution of each frequency band, peak frequency, center frequency and bandwidth.

[0013] Among them, the physical mechanism model of the carton fiber structure refers to a stress-strain analysis model based on the anisotropic mechanical properties of paper materials, taking into account the interlayer structural characteristics and fiber arrangement direction of corrugated paper.

[0014] Among them, the stress distribution calculation equation is based on Hooke's law and is used to calculate the stress distribution state of the corrugated box wall under the conditions of micro-deformation displacement data. The input includes micro-deformation displacement data, material elastic modulus tensor, Poisson's ratio parameter, corrugated structure geometric parameters and test load size, and the output is three-dimensional stress field distribution and strain energy density distribution.

[0015] Among them, the hierarchical attention carton strength assessment model refers to a deep learning model that combines spatial hierarchical feature extraction and attention mechanism, which is used to predict the compressive strength of corrugated boxes from micro-deformation displacement data and spectral feature parameters; its hierarchical structure takes into account the multi-layer composite structure characteristics of corrugated boxes, and forms a corresponding relationship with the corrugated wave distance and corrugated wall thickness parameters in the corrugated structure geometric parameters through the feature extraction capabilities of different scales.

[0016] This invention combines a high-precision micro-deformation loading system with a laser displacement sensor array to accurately capture the micro-deformation characteristics of cartons under low load conditions (within 10% of the maximum load capacity). Combined with acoustic emission signal analysis, this system establishes a comprehensive non-destructive assessment system, ensuring the structural integrity of the cartons during testing.

[0017] This method overcomes the limitations of traditional destructive testing. By integrating micro-deformation displacement data with acoustic spectrum characteristics through a hierarchical attention carton strength assessment model, it achieves highly accurate predictions of carton compressive strength. Furthermore, based on stress field distribution and strain energy analysis, the method accurately identifies structural weaknesses and quantifies the risk of damage, providing an early warning mechanism for production line quality control and successfully resolving the technical issue of accurately predicting compressive strength without damaging the carton. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 is a flow chart of the method of the present invention.

[0019] Figure 2 Schematic diagram of the composition of the online automatic testing system for the compressive strength of corrugated boxes in Example 2.

[0020] Figure 3 Schematic diagram of the micro-deformation loading system structure in Example 2.

[0021] Figure 4 Schematic diagram of the structure of the acoustic emission signal acquisition system in Example 2.

[0022] Figure 5 These are the acoustic emission spectrum characteristics of different types of corrugated boxes in Example 2.

[0023] Figure 6 This is the three-dimensional model of stress-strain distribution of the corrugated box in Example 2. DETAILED DESCRIPTION

[0024] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0025] like Figure 1 FIG. 1 is a flow chart of a method for online automatic testing of the compressive strength of a corrugated box provided by the present invention. The method comprises the following steps:

[0026] A method for online automatic testing of the compressive strength of a corrugated box comprises the following steps:

[0027] S01. Construct a micro-deformation loading system and apply a test load not exceeding 10% of the expected maximum load capacity to the top of the corrugated box;

[0028] S02. Using a high-precision laser displacement sensor array to measure micro-deformation displacement data of six surfaces of the corrugated box under the test load;

[0029] S03, synchronously collecting the acoustic emission signal generated during the test load application process and extracting spectrum characteristic parameters;

[0030] S04. Analyze the micro-deformation displacement data based on the physical mechanism model of the carton fiber structure, and calculate the stress field distribution of the corrugated carton wall using the stress distribution calculation equation;

[0031] S05, inputting the micro-deformation displacement data and the spectral feature parameters into a pre-trained hierarchical attention carton strength assessment model for analysis;

[0032] S06. Calculating the stiffness coefficient of the corrugated box based on the slope of the elastic deformation curve calculated based on the micro-deformation displacement data and comparing the result with a reference threshold;

[0033] S07. Generating a strain energy distribution map based on the stress field distribution to identify weak areas of the corrugated box structure and quantify a potential damage risk index;

[0034] S08. Comprehensively evaluate the predicted value and confidence interval of the compressive strength of corrugated boxes using a multi-source data fusion algorithm;

[0035] S09. Outputting the predicted compressive strength value, the confidence interval, the stress field distribution, and the damage risk index as test results.

[0036] Among them, the micro-deformation loading system refers to a pressure-applying device driven by a high-precision servo motor. Closed-loop feedback control ensures a smooth and precise loading process. The loading force range is 50N to 2000N, with an accuracy of not less than 0.5%.

[0037] The test load refers to the pressure applied to the top of the corrugated box by the micro-deformation loading system, and its magnitude is within 10% of the expected maximum load-bearing capacity.

[0038] Among them, the high-precision laser displacement sensor array refers to a measurement network composed of 24 triangulation principle laser displacement sensors with a measurement accuracy of 1 micron and a sampling frequency of 1000Hz, which is used to capture tiny deformations on the surface of the carton.

[0039] The micro-deformation displacement data refers to a set of numerical values ​​of deformation displacement of six surfaces of the corrugated box under the action of the test load, collected by the high-precision laser displacement sensor array.

[0040] Among them, the acoustic emission signal refers to the high-frequency sound wave signal generated by the internal fibers of the corrugated box under the action of the test load, with a frequency range of 20kHz to 200kHz, which is collected by a piezoelectric sensor array.

[0041] The spectrum characteristic parameters refer to a set of characteristic parameters extracted after spectrum analysis of the acoustic emission signal, including energy distribution of each frequency band, peak frequency, center frequency and bandwidth.

[0042] Among them, the physical mechanism model of carton fiber structure refers to a stress-strain analysis model based on the anisotropic mechanical properties of paper materials, taking into account the interlayer structural characteristics and fiber arrangement direction of corrugated paper.

[0043] Among them, the stress distribution calculation equation is based on Hooke's law and is used to calculate the stress distribution state of the corrugated box wall under the conditions of the micro-deformation displacement data. The input includes the micro-deformation displacement data, the material elastic modulus tensor, the Poisson's ratio parameter, the corrugated structure geometric parameters and the test load size, and the output is the three-dimensional stress field distribution and the strain energy density distribution; the material elastic modulus tensor refers to the second-order tensor that describes the elastic properties of the corrugated box material, which is obtained through experimental measurement; the Poisson's ratio parameter refers to the absolute value of the ratio of the lateral strain to the axial strain of the material under uniaxial stress; the corrugated structure geometric parameters include corrugated height, corrugated wave pitch, corrugated wall thickness and corrugated shape coefficient.

[0044] The stress field distribution refers to the three-dimensional stress distribution state of the corrugated box wall obtained by calculating the stress distribution calculation equation, which is used for subsequent analysis and evaluation.

[0045] Among them, the hierarchical attention carton strength assessment model refers to a deep learning model that combines spatial hierarchical feature extraction and attention mechanism, which is used to predict the compressive strength of corrugated boxes from the micro-deformation displacement data and the spectral feature parameters.

[0046] The elastic deformation curve refers to a curve formed by the micro-deformation displacement data changing with the load under the action of the test load.

[0047] The stiffness coefficient refers to the inverse of the displacement change rate caused by a unit load increment in the elastic deformation stage, with the unit being N / mm, and is obtained by calculating the slope of the elastic deformation curve.

[0048] The reference threshold refers to a standard value of the stiffness coefficient preset according to the specifications and uses of the corrugated box, which is used for comparison with the stiffness coefficient.

[0049] The strain energy distribution diagram refers to a visual representation of the internal strain energy density distribution obtained by calculating the stress field distribution.

[0050] The structurally weak area refers to an area with a higher strain energy density identified from the strain energy distribution diagram, indicating a location in the corrugated box structure that is more susceptible to damage.

[0051] The damage risk index refers to a dimensionless parameter obtained based on the analysis of the weak areas of the structure, ranging from 0 to 100, indicating the probability of damage to the corrugated box under a standard load.

[0052] Among them, the multi-source data fusion algorithm refers to a calculation method that integrates the micro-deformation displacement data, the spectral characteristic parameters, the stiffness coefficient and the damage risk index using a Bayesian reasoning framework to improve prediction accuracy and reliability.

[0053] The predicted compressive strength value refers to the estimated value of the final compressive strength of the corrugated box calculated by the multi-source data fusion algorithm, and the unit is N.

[0054] The confidence interval refers to the uncertainty range of the predicted compressive strength value, which is calculated by statistical methods and expressed as a percentage range of the predicted value.

[0055] The specific structure of the hierarchical attention carton strength assessment model is a fusion architecture of a multi-level spatial feature extraction network and a temporal information processing network, wherein the spatial feature extraction network uses six convolutional layers to encode the micro-deformation displacement data of the six sides of the corrugated carton, the number of convolution kernels in each layer is 64, 128, 256, 256, 512, 512, the convolution kernel size is 3×3, the step size is 1, and the ReLU activation function is used; the temporal information processing network uses a bidirectional long short-term memory network to process the temporal features of the spectral feature parameters, and the number of hidden layer neurons is 512; the hierarchical attention carton strength assessment model introduces a multi-head cross attention mechanism in the feature fusion layer, the number of attention heads is 8, and the embedding dimension is 64, which is used to capture the micro-deformation displacement data and the spectral feature parameters. The correlation between the numbers; the output layer uses a fully connected layer to map the fused features into the compressive strength prediction value and its uncertainty estimate, and the number of neurons in the fully connected layer is 256, 128, 64, and 2, respectively; the multi-head cross-attention mechanism can adaptively focus on the deformation mode and acoustic characteristics most relevant to the weak area of ​​the structure by learning the correlation weights between the micro-deformation displacement data and the spectral feature parameters. The multi-head cross-attention mechanism is closely related to the process of identifying the weak area of ​​the structure in step S07; the hierarchical structure design of the hierarchical attention carton strength assessment model specially considers the multi-layer composite structure characteristics of the corrugated carton, and forms a corresponding relationship with the corrugated wave distance and corrugated wall thickness parameters in the corrugated structure geometric parameters through the feature extraction capabilities of different scales.

[0056] The steps for establishing the training data set of the hierarchical attention carton strength assessment model specifically include first extracting 20,000 corrugated carton samples of different specifications, sizes, corrugation types, and raw material characteristics from the production line; then performing a micro-deformation test on each sample and recording the complete micro-deformation displacement data and the acoustic emission signal; then sending each sample to a standard compression tester for destructive testing and recording its final destructive strength value as label data; then dividing the data set into a training set, a validation set, and a test set in a ratio of 8:1:1; finally, normalizing all feature data to a mean of 0 and a standard deviation of 1, and using the principal component analysis method to reduce the dimensionality of high-dimensional features, retaining the principal components with an explained variance ratio of 99%.

[0057] The steps of training the hierarchical attention carton strength evaluation model specifically include first initializing the model parameters using a Gaussian distribution with a mean of 0 and a standard deviation of 0.01; then using the Adam optimizer for model training, with the initial learning rate set to 0.001, and using a cosine annealing learning rate scheduling strategy with a minimum learning rate of 0.00001; then setting the batch size to 64 and the number of training rounds to 200; evaluating the model performance on the validation set after each round of training, using the mean absolute percentage error as the evaluation metric; activating the early stopping mechanism to terminate the training when the performance of the validation set does not improve for 10 consecutive rounds; and finally The best model performance is evaluated on the test set to ensure that the average prediction error is less than 5%. A mixed precision training strategy is adopted throughout the training process to improve computational efficiency, and an L2 regularization method with a weight decay value of 0.0001 is used to prevent overfitting. During the training process, the performance of the model on different types of corrugated boxes is simultaneously monitored to ensure that the model has good generalization ability for various corrugated structures. The training process pays special attention to the model's sensitivity to the high-value areas of the stress field distribution calculated in step S04, and improves the model's ability to identify weak areas of the structure by designing a special loss function weighting method.

[0058] The specific implementation of the above steps is described in detail below.

[0059] The specific implementation method of constructing the micro-deformation loading system in step S01 is to first install a high-precision servo motor and its drive controller. The controller uses a 32-bit microprocessor to achieve closed-loop control with a control frequency of 1000Hz. Then configure the loading force sensor with a range of 0 to 2000N, a sensitivity of 0.01% of full scale, and a signal-to-noise ratio of not less than 85dB. Then install a balanced loading platform, and control the platform flatness error within 0.01mm to ensure uniform force distribution during loading. Subsequently, a closed-loop force feedback control algorithm is designed. The algorithm adopts a proportional integral differential (PID) control structure, in which the proportional gain is set to 0.85, the integral time constant is 0.05s, the differential time constant is 0.02s, and the integral saturation threshold is set to ±15% of the control output to prevent control overshoot caused by integral saturation. Finally, a loading curve planning module is established, and a quintic polynomial curve is used to achieve smooth loading to ensure that the loading rate is always controlled within the range of 10N / s to 50N / s, and the acceleration is limited to 20N / The loading system is connected to the main control computer via industrial Ethernet, with a data transmission bandwidth of 100Mbps and a transmission delay of less than 1ms. The purpose of step S01 is to provide a stable and precise micro-load loading environment to ensure that the micro-deformation characteristics of the carton are obtained without damaging the carton structure. The closed-loop control technology in this step can accurately control the load size to avoid the risk of irreversible deformation of the carton due to excessive force.

[0060] The specific implementation method of using a high-precision laser displacement sensor array to measure micro-deformation displacement data in step S02 is to first build a laser triangulation sensor bracket system, which is made of carbon fiber composite material with a thermal expansion coefficient of less than 5× / °C to ensure stability during measurement. Twenty-four laser displacement sensors were then strategically arranged: six on the top, four on the bottom, and three to four on each of the four side surfaces. The spacing between sensors was dynamically adjusted based on the carton size, but never exceeded 150 mm. The measurement points were distributed across representative areas of each carton surface, particularly at corner seams and in the center. A high-speed synchronous data acquisition system was then configured, utilizing time-division multiplexing technology to maintain clock synchronization error within 1 μs. The data acquisition card had a 16-bit sampling accuracy and an input range of ±10 V. A data processing algorithm was then designed, employing a sliding window median filter with a window length of 5 to remove measurement noise. A bicubic spline interpolation algorithm was used to construct a complete deformation field of the carton surface, with an interpolation grid density of 4 nodes per square centimeter. Finally, a three-dimensional deformation field visualization module was implemented, generating a pseudo-color cloud map of the carton deformation, with a color mapping ranging from blue (minimum deformation) to red (maximum deformation). The purpose of this step is to obtain the deformation distribution data of each surface of the corrugated box under the action of small loads. This data is the basis for subsequent analysis. Through high-precision measurement, the subtle deformation characteristics of the box are captured to reflect its structural integrity and mechanical properties.

[0061] The specific implementation for synchronously collecting acoustic emission signals and extracting spectral characteristic parameters in step S03 involves first attaching eight high-sensitivity piezoelectric sensors to the carton surface. The sensors have a sensitivity of 75dB (referring to 0dB = 1V / μbar), a frequency response range of 20kHz to 200kHz, and a spacing of no more than 200mm. Next, an acoustic emission signal amplification and filtering circuit is configured, with a preamplifier gain of 40dB, a bandpass filter with a passband of 20kHz to 200kHz, and a roll-off slope of 48dB / octave. Next, the signal triggering and acquisition logic is designed, using a floating threshold adaptive triggering scheme. The initial trigger threshold is set to three times the RMS value of the background noise, and the sampling rate is set to 500kHz to meet the Nyquist sampling theorem. Short-time Fourier transform (STFT) analysis is then performed on the collected acoustic signals for time-frequency analysis. A time window length of 2ms and a window overlap of 50% are used, and a Hanning window is used to reduce spectral leakage. Finally, spectral characteristic parameters are extracted, including energy parameters (energy distribution in each frequency band, energy contribution in the three frequency bands of 20-50kHz, 50-100kHz, and 100-200kHz), frequency parameters (peak frequency, center frequency, weighted frequency, and bandwidth), and statistical parameters (spectral entropy, spectral skewness, and spectral kurtosis). The acoustic emission parameter extraction algorithm uses wavelet packet decomposition with a decomposition level of 5 and the db4 wavelet as the basis function. This step is used to obtain the acoustic signal characteristics generated by the internal fibers of the carton under a small load. These acoustic characteristics can reflect the microscopic changes in the fiber structure within the carton and are an important basis for predicting macroscopic compressive performance.

[0062] The specific implementation method of analyzing the micro-deformation displacement data based on the physical mechanism model of the carton fiber structure in step S04 is to first establish an anisotropic constitutive model of the carton material, regard the corrugated carton material as an orthotropic material, use the generalized Hooke's law to describe the stress-strain relationship, and the material properties are determined by 9 independent elastic constants (three elastic moduli 、 、 , three Poisson ratios 、 、 , three shear moduli 、 、 ). Then, a corrugated structure geometric model is established, and the corrugation is described as a sinusoidal waveform. Its shape is defined by wave height, wave pitch and wall thickness parameters. The bonding characteristics between the face paper and the corrugated paper are taken into account in the model. The carton is then discretized into a finite element grid, and the grid density is set to no less than 8 nodes per corrugated wave pitch to accurately capture the deformation characteristics of the corrugated structure. The micro-deformation displacement data obtained in step S02 is then input into the finite element model as a boundary condition, and the full-field stress distribution is calculated using the displacement method. Finally, the material parameters are optimized through iterative calculation so that the root mean square error between the deformation field predicted by the model and the measured deformation field is less than 2%. The calculation process adopts a nonlinear finite element method, considering the large deformation effect, and the unit type is selected as a 20-node hexahedral unit, and the integration point is 3×3×3 Gaussian integration points. The role of this step is based on the principles of physical mechanics, and the stress field distribution of the carton wall is inferred through micro-deformation displacement data, providing a theoretical basis for the subsequent evaluation of the carton structure strength.

[0063] In step S05, the micro-deformation displacement data and spectral feature parameters are input into the hierarchical attention carton strength assessment model. The specific implementation method is to first preprocess the micro-deformation displacement data, including removing outliers (using a modified Z-score method with a threshold of 3.5), normalizing (using the standard score method to normalize the data to a mean of 0 and a standard deviation of 1), and spatial downsampling (using an adaptive grid method to maintain high resolution in key areas and appropriately reduce resolution in other areas). The spectral feature parameters are then preprocessed, including feature normalization (using the minimum-maximum normalization method to scale features to the range [0, 1]) and feature selection (using recursive feature elimination to retain the top 80% of the features ranked by importance). The processed micro-deformation displacement data is then input into a six-layer convolutional neural network, whose convolution kernel parameters are set according to the aforementioned requirements, and the output is a 512-dimensional spatial feature vector. The processed spectral feature parameters are then input into a bidirectional long short-term memory network, which consists of two layers, each with 512 neurons, and the output is a 512-dimensional temporal feature vector. Finally, the spatial and temporal feature vectors are fused using a multi-head cross-attention mechanism. This attention mechanism uses the scaled dot-product attention algorithm with a temperature parameter set to 0.1. The fused features are then mapped into compressive strength predictions and confidence intervals via a fully connected layer. This step utilizes a deep learning model to extract high-level features from multi-source feature data and uses the attention mechanism to focus on the most relevant feature combinations for accurate strength assessment.

[0064] The specific implementation method for calculating the stiffness coefficient based on the slope of the elastic deformation curve calculated from the micro-deformation displacement data in step S06 is to first extract the corresponding data points of load and displacement from the micro-deformation displacement data. For each measurement point, the displacement values ​​at different load levels are recorded to form a load-displacement dataset. The load-displacement data is then subjected to noise filtering using a Savitzky-Golay filter with a window length of 7 and a polynomial order of 3, preserving the main trends in the data while removing high-frequency noise. The load-displacement curve is then fitted using a weighted least squares method. The fitting function uses a linear polynomial (straight line) model, with weights set to the inverse of the data point displacement values ​​to enhance the fitting accuracy in the small deformation area. The slope of the fitted line is then calculated. The inverse of this slope is the stiffness coefficient, expressed in N / mm. Finally, the calculated stiffness coefficient is compared against a preset benchmark threshold. This threshold is determined based on the carton specifications and is typically 90% of the standard carton stiffness. For example, for a 5-ply corrugated box, the benchmark threshold is typically set between 800N / mm and 1200N / mm; for a 3-ply corrugated box, the benchmark threshold is typically set between 400N / mm and 700N / mm. Comparison results are categorized as "pass" (no less than 95% of the benchmark threshold), "warning" (between 85% and 95% of the benchmark threshold), and "fail" (less than 85% of the benchmark threshold). This step quantitatively assesses the carton's elastic stiffness. The stiffness coefficient is a key indicator of the carton's structural stability and is crucial for predicting its compressive performance.

[0065] The specific implementation method for generating a strain energy distribution map from the stress field distribution in step S07 is to first calculate the strain energy density at each unit node based on the stress field distribution calculated in step S04. The calculation formula is: strain energy density equals half the inner product of the stress component and the strain component. This unit node strain energy density information is then mapped to the surface of the three-dimensional carton model, and the node averaging method is used to handle the transition of strain energy density between units. An adaptive color mapping algorithm is then applied to generate a strain energy distribution cloud map. The color range is from blue (low strain energy) to red (high strain energy). The number of color levels is set to 10, and the grading method uses percentile segmentation to ensure that each color level covers the same number of data points. Structural weaknesses are then identified based on the strain energy distribution. Areas with strain energy density above the 90th percentile are marked as high-risk areas, areas between the 75th and 90th percentiles are marked as medium-risk areas, and areas below the 75th percentile are marked as low-risk areas. Finally, the damage risk index is calculated using a weighted summation method, with the weights assigned being the proportion of high-risk areas × 100, the proportion of medium-risk areas × 60, and the proportion of low-risk areas × 20. The concentration of high-risk areas is also taken into account; the higher the concentration, the greater the risk index. The risk index is normalized to a range of 0 to 100. This step uses an energy-based approach to identify weak links in the carton structure, predict possible damage locations, and quantify the structural damage risk, providing a basis for improving carton design.

[0066] The specific implementation method for comprehensively evaluating the predicted compressive strength value using a multi-source data fusion algorithm in step S08 is to first establish a Bayesian network model. The network structure consists of four input nodes (micro-deformation displacement characteristics, spectral characteristic parameters, stiffness coefficient, and failure risk index) and one output node (the predicted compressive strength value). The conditional probability distribution of each node is then defined. A Gaussian mixture model is used to describe the probability distribution of continuous variables, with the number of mixture components set to 3. The model parameters are estimated using the expectation-maximization algorithm. Bayesian inference is then performed using the Monte Carlo Markov Chain (MCMC) method. The Metropolis-Hastings algorithm is used for sampling, with the number of samples set to 10,000. The first 2,000 samples are discarded as a warm-up period. The posterior distribution of the predicted compressive strength value is then calculated based on the sampling results. The median of the posterior distribution is used as the final predicted value, and the 2.5% and 97.5% quantiles are used as the lower and upper limits of the 95% confidence interval. Finally, the prediction reliability index is calculated. The mean absolute percentage error (MAPE) of the prediction model is calculated based on historical data. This error is typically kept below 5%. If this threshold is exceeded, the system will issue a calibration prompt. This step integrates various test indicators and uses probabilistic statistical methods to obtain reliable compressive strength predictions. It also provides a range of uncertainty for the prediction, providing a scientific basis for decision-making in practical applications.

[0067] The specific implementation for outputting the test results in step S09 involves first generating a compressive strength prediction report, including the predicted value (in N), the relative prediction error (based on historical data statistics, typically ±3% to ±5%), the 95% confidence interval (expressed as a percentage of the predicted value, typically within the range of ±5% to ±10%), and test condition parameters (such as ambient temperature and humidity). Next, a stress distribution visualization report is generated, including a 3D color stress cloud map, a principal stress vector map, and stress concentration area identification. The image resolution must be at least 1024×768 pixels, using a rainbow color scale and a numerical scale. Next, a damage risk assessment report is generated, including a risk index value (a dimensionless value ranging from 0 to 100), a risk level classification (0 to 30 for low risk, 30 to 70 for medium risk, and 70 to 100 for high risk), and a description of the weak area location (identified using the carton surface partition number). Finally, the test results are exported in PDF, Excel, and JSON formats, and a historical data comparison function is provided for comparative analysis with historical test results of similar cartons. The purpose of this step is to present complex test analysis results to users in an intuitive and easy-to-understand manner, allowing users to quickly make judgments and decisions. At the same time, it provides a standardized data interface to support subsequent data analysis and management.

[0068] Specifically, the core of this invention lies in combining micro-deformation testing technology with a multi-source data fusion algorithm, based on material mechanics theory and deep learning methods, to establish a mapping relationship between the microscopic deformation behavior of corrugated boxes and their macroscopic compressive strength. When a corrugated box is subjected to low loads, its structure undergoes minute but measurable elastic deformation. While this deformation is imperceptible to the naked eye, it contains rich material and structural information.

[0069] First, the invention utilizes a high-precision laser displacement sensor array to precisely capture micro-deformation data on the carton's six surfaces, achieving micron-level measurement accuracy. This data can reveal local stress concentration points and deformation patterns in the material. Simultaneously, it collects acoustic emission signals generated by the carton's internal fiber structure under stress. These high-frequency acoustic signals (20kHz to 200kHz) can reveal changes in the material's internal microstructure and potential precursors to damage.

[0070] Secondly, based on a physical model of the carton fiber structure, the present invention applies stress distribution equations to convert micro-deformation displacement data into a three-dimensional stress field distribution, and then calculates the strain energy distribution. This process adheres to Hooke's law and the principle of energy conservation in material mechanics. The unevenness of the strain energy density distribution directly reflects the weak areas of the structure, i.e., the locations most likely to fail under actual loading conditions.

[0071] The hierarchical attention carton strength assessment model designed by this invention utilizes a fusion architecture of a multi-level spatial feature extraction network and a temporal information processing network, combined with a multi-head cross-attention mechanism. This model can adaptively identify patterns in micro-deformation data and acoustic signal features that are most relevant to compressive strength. The model's hierarchical structure specifically considers the multi-layered composite structure of corrugated carton boxes, effectively capturing deformation characteristics at different scales.

[0072] Finally, a multi-source data fusion algorithm based on a Bayesian inference framework integrates micro-deformation displacement data, spectral characteristic parameters, stiffness coefficient, and damage risk index to comprehensively evaluate the predicted compressive strength of corrugated boxes and its confidence interval. The entire prediction process has a solid physical and statistical basis and has demonstrated high accuracy and reliability in extensive experimental verification.

[0073] A specific embodiment 1 of the present invention is provided below. The specific implementation of each step in this embodiment 1 is described in detail as follows.

[0074] The specific implementation method of constructing the micro-deformation loading system in step S01 is to first install a high-precision servo motor and its drive controller. The controller uses a 32-bit microprocessor to achieve closed-loop control with a control frequency of 1000Hz. Then configure the loading force sensor with a range of 0 to 2000N, a sensitivity of 0.01% of full scale, and a signal-to-noise ratio of not less than 85dB. Then install a balanced loading platform, and control the platform flatness error within 0.01mm to ensure uniform force distribution during loading. Subsequently, a closed-loop force feedback control algorithm is designed. The algorithm adopts a proportional integral differential (PID) control structure, in which the proportional gain is set to 0.85, the integral time constant is 0.05s, the differential time constant is 0.02s, and the integral saturation threshold is set to ±15% of the control output to prevent control overshoot caused by integral saturation. Finally, a loading curve planning module is established, and a quintic polynomial curve is used to achieve smooth loading to ensure that the loading rate is always controlled within the range of 10N / s to 50N / s, and the acceleration is limited to 20N / The loading system is connected to the main control computer via industrial Ethernet, with a data transmission bandwidth of 100Mbps and a transmission delay of less than 1ms. The purpose of step S01 is to provide a stable and precise micro-load loading environment to ensure that the micro-deformation characteristics of the carton are obtained without damaging the carton structure. The closed-loop control technology in this step can accurately control the load size to avoid the risk of irreversible deformation of the carton due to excessive force.

[0075] The specific implementation method of using a high-precision laser displacement sensor array to measure micro-deformation displacement data in step S02 is to first build a laser triangulation sensor bracket system, which is made of carbon fiber composite material with a thermal expansion coefficient of less than 5× / °C to ensure stability during measurement. Twenty-four laser displacement sensors were then strategically arranged: six on the top, four on the bottom, and three to four on each of the four side surfaces. The spacing between sensors was dynamically adjusted based on the carton size, but never exceeded 150 mm. The measurement points were distributed across representative areas of each carton surface, particularly at corner seams and in the center. A high-speed synchronous data acquisition system was then configured, utilizing time-division multiplexing technology to maintain clock synchronization error within 1 μs. The data acquisition card had a 16-bit sampling accuracy and an input range of ±10 V. A data processing algorithm was then designed, employing a sliding window median filter with a window length of 5 to remove measurement noise. A bicubic spline interpolation algorithm was used to construct a complete deformation field of the carton surface, with an interpolation grid density of 4 nodes per square centimeter. Finally, a three-dimensional deformation field visualization module was implemented, generating a pseudo-color cloud map of the carton deformation, with a color mapping ranging from blue (minimum deformation) to red (maximum deformation). The purpose of this step is to obtain the deformation distribution data of each surface of the corrugated box under the action of small loads. This data is the basis for subsequent analysis. Through high-precision measurement, the subtle deformation characteristics of the box are captured to reflect its structural integrity and mechanical properties.

[0076] The specific implementation for synchronously collecting acoustic emission signals and extracting spectral characteristic parameters in step S03 involves first attaching eight high-sensitivity piezoelectric sensors to the carton surface. The sensors have a sensitivity of 75dB (referring to 0dB = 1V / μbar), a frequency response range of 20kHz to 200kHz, and a spacing of no more than 200mm. Next, an acoustic emission signal amplification and filtering circuit is configured, with a preamplifier gain of 40dB, a bandpass filter with a passband of 20kHz to 200kHz, and a roll-off slope of 48dB / octave. Next, the signal triggering and acquisition logic is designed, using a floating threshold adaptive triggering scheme. The initial trigger threshold is set to three times the RMS value of the background noise, and the sampling rate is set to 500kHz to meet the Nyquist sampling theorem. Short-time Fourier transform (STFT) analysis is then performed on the collected acoustic signals for time-frequency analysis. A time window length of 2ms and a window overlap of 50% are used, and a Hanning window is used to reduce spectral leakage. Finally, spectral characteristic parameters are extracted, including energy parameters (energy distribution in each frequency band, energy contribution in the three frequency bands of 20-50kHz, 50-100kHz, and 100-200kHz), frequency parameters (peak frequency, center frequency, weighted frequency, and bandwidth), and statistical parameters (spectral entropy, spectral skewness, and spectral kurtosis). The acoustic emission parameter extraction algorithm uses wavelet packet decomposition with a decomposition level of 5 and the db4 wavelet as the basis function. This step is used to obtain the acoustic signal characteristics generated by the internal fibers of the carton under a small load. These acoustic characteristics can reflect the microscopic changes in the fiber structure within the carton and are an important basis for predicting macroscopic compressive performance.

[0077] The specific implementation method of analyzing the micro-deformation displacement data based on the physical mechanism model of the carton fiber structure in step S04 is to first establish an anisotropic constitutive model of the carton material, regard the corrugated carton material as an orthotropic material, use the generalized Hooke's law to describe the stress-strain relationship, and the material properties are determined by 9 independent elastic constants (three elastic moduli 、 、 , three Poisson ratios 、 、 , three shear moduli 、 、 ). Then, a corrugated structure geometric model is established, and the corrugation is described as a sinusoidal waveform. Its shape is defined by wave height, wave pitch and wall thickness parameters. The bonding characteristics between the face paper and the corrugated paper are taken into account in the model. The carton is then discretized into a finite element grid, and the grid density is set to no less than 8 nodes per corrugated wave pitch to accurately capture the deformation characteristics of the corrugated structure. The micro-deformation displacement data obtained in step S02 is then input into the finite element model as a boundary condition, and the full-field stress distribution is calculated using the displacement method. Finally, the material parameters are optimized through iterative calculation so that the root mean square error between the deformation field predicted by the model and the measured deformation field is less than 2%. The calculation process adopts a nonlinear finite element method, considering the large deformation effect, and the unit type is selected as a 20-node hexahedral unit, and the integration point is 3×3×3 Gaussian integration points. The role of this step is based on the principles of physical mechanics, and the stress field distribution of the carton wall is inferred through micro-deformation displacement data, providing a theoretical basis for the subsequent evaluation of the carton structure strength.

[0078] The stress distribution calculation equation is based on the generalized Hooke's law for orthotropic materials, and its matrix form can be expressed as: ,

[0079] The flexibility matrix The components of are:

[0080] , , ,

[0081] , , ,

[0082] , , ,

[0083] , , .

[0084] Inverting the above flexibility matrix, we can get the stiffness matrix , which represents the relationship between strain and stress:

[0085] .

[0086] For the corrugated structure, the laminated composite material theory is used, and the equivalent elastic moduli along the corrugated direction (taken as direction 1) and perpendicular to the corrugated direction (taken as direction 2) are:

[0087] ,

[0088] ,

[0089] in, and Represent the elastic modulus of facial tissue and corrugated paper respectively, and Respectively represent the thickness of facial tissue and corrugated paper.

[0090] Based on the acquired micro-deformation displacement data and the geometric parameters of the corrugated box, the finite element method is used to solve the stress equilibrium equation:

[0091] ,

[0092] in, represents the divergence operator, represents the stress tensor, Represents the body force. Using the displacement method, the displacement field Expressed as node displacement and shape functions Combination of:

[0093] ,

[0094] in, Indicates the number of element nodes (for a 20-node hexahedral element, ). Substituting the displacement field into the strain-displacement relationship yields the strain field:

[0095] ,

[0096] in, represents the strain-displacement matrix. Substituting the strain field into the constitutive relation yields the stress field:

[0097] ,

[0098] in, Represents the above stiffness matrix. Applying the principle of minimum potential energy, for any virtual displacement , should meet the following requirements:

[0099] ,

[0100] in, represents the surface force. The virtual displacement field is expressed as , substitute into the above formula and consider The arbitrariness of , we get the discrete form of the equilibrium equation:

[0101] ,

[0102] in, represents the overall stiffness matrix, Denotes the overall load vector, which are:

[0103] ,

[0104] .

[0105] Solving the above equations gives the node displacement , and then obtain the strain field and stress field For thin-walled corrugated boxes, the membrane stress is further calculated. and bending moment , used to assess buckling stability:

[0106] ,

[0107] ,

[0108] in, Indicates the wall thickness, represents the coordinate along the wall thickness. The accuracy of the calculated final stress field distribution is verified by comparing it with the measured deformation field, and the material parameters are iteratively optimized until the root mean square error is less than 2%.

[0109] In step S05, the micro-deformation displacement data and spectral feature parameters are input into the hierarchical attention carton strength assessment model. The specific implementation method is to first preprocess the micro-deformation displacement data, including removing outliers (using a modified Z-score method with a threshold of 3.5), normalizing (using the standard score method to normalize the data to a mean of 0 and a standard deviation of 1), and spatial downsampling (using an adaptive grid method to maintain high resolution in key areas and appropriately reduce resolution in other areas). The spectral feature parameters are then preprocessed, including feature normalization (using the minimum-maximum normalization method to scale features to the range [0, 1]) and feature selection (using recursive feature elimination to retain the top 80% of the features ranked by importance). The processed micro-deformation displacement data is then input into a six-layer convolutional neural network, whose convolution kernel parameters are set according to the aforementioned requirements, and the output is a 512-dimensional spatial feature vector. The processed spectral feature parameters are then input into a bidirectional long short-term memory network, which consists of two layers, each with 512 neurons, and the output is a 512-dimensional temporal feature vector. Finally, the spatial and temporal feature vectors are fused using a multi-head cross-attention mechanism. This attention mechanism uses the scaled dot-product attention algorithm with a temperature parameter set to 0.1. The fused features are then mapped into compressive strength predictions and confidence intervals via a fully connected layer. This step utilizes a deep learning model to extract high-level features from multi-source feature data and uses the attention mechanism to focus on the most relevant feature combinations for accurate strength assessment.

[0110] The calculation process of the multi-head cross attention mechanism can be expressed as:

[0111] ,

[0112] in, represents the query matrix (from the spatial eigenvectors), represents the key matrix (from the time series eigenvectors), represents the value matrix (also from the time series eigenvectors), represents the dimension of the key vector, Represents the temperature parameter (set to 0.1). The multi-head attention mechanism calculates 8 parallel attention heads, then concatenates the results and undergoes a linear transformation:

[0113] ,

[0114] in, , and is the parameter matrix to be learned.

[0115] The specific implementation method for calculating the stiffness coefficient based on the slope of the elastic deformation curve calculated from the micro-deformation displacement data in step S06 is to first extract the corresponding data points of load and displacement from the micro-deformation displacement data. For each measurement point, the displacement values ​​at different load levels are recorded to form a load-displacement dataset. The load-displacement data is then subjected to noise filtering using a Savitzky-Golay filter with a window length of 7 and a polynomial order of 3, preserving the main trends in the data while removing high-frequency noise. The load-displacement curve is then fitted using a weighted least squares method. The fitting function uses a linear polynomial (straight line) model, with weights set to the inverse of the data point displacement values ​​to enhance the fitting accuracy in the small deformation area. The slope of the fitted line is then calculated. The inverse of this slope is the stiffness coefficient, expressed in N / mm. Finally, the calculated stiffness coefficient is compared with a preset benchmark threshold. This threshold is determined based on the carton specifications and is typically 90% of the standard carton stiffness. For example, for a 5-ply corrugated carton, the benchmark threshold is typically set between 800N / mm and 1200N / mm; for a 3-ply corrugated carton, the benchmark threshold is typically set between 400N / mm and 700N / mm. The comparison results are categorized as pass (no less than 95% of the benchmark threshold), warning (between 85% and 95% of the benchmark threshold), and fail (less than 85% of the benchmark threshold). This step quantitatively assesses the carton's elastic stiffness. The stiffness coefficient is a key indicator of the carton's structural stability and is crucial for predicting its compressive performance.

[0116] The calculation process of the weighted least squares method for fitting the load-displacement curve can be expressed as:

[0117] ,in, Indicates the The load value of the data point, Indicates the The displacement value of each data point, Indicates the The weight of the data point, represents the total number of data points, Indicates the slope of the fitted straight line. Stiffness coefficient Calculated as:

[0118] ,

[0119] The unit is N / mm.

[0120] The specific implementation method for generating a strain energy distribution map from the stress field distribution in step S07 is to first calculate the strain energy density at each unit node based on the stress field distribution calculated in step S04. The calculation formula is: strain energy density equals half the inner product of the stress component and the strain component. This unit node strain energy density information is then mapped to the surface of the three-dimensional carton model, and the node averaging method is used to handle the transition of strain energy density between units. An adaptive color mapping algorithm is then applied to generate a strain energy distribution cloud map. The color range is from blue (low strain energy) to red (high strain energy). The number of color levels is set to 10, and the grading method uses percentile segmentation to ensure that each color level covers the same number of data points. Structural weaknesses are then identified based on the strain energy distribution. Areas with strain energy density above the 90th percentile are marked as high-risk areas, areas between the 75th and 90th percentiles are marked as medium-risk areas, and areas below the 75th percentile are marked as low-risk areas. Finally, the damage risk index is calculated using a weighted summation method, with the weights assigned being the proportion of high-risk areas × 100, the proportion of medium-risk areas × 60, and the proportion of low-risk areas × 20. The concentration of high-risk areas is also taken into account; the higher the concentration, the greater the risk index. The risk index is normalized to a range of 0 to 100. This step uses an energy-based approach to identify weak links in the carton structure, predict possible damage locations, and quantify the structural damage risk, providing a basis for improving carton design.

[0121] The calculation formula of strain energy density can be expressed as:

[0122] ,

[0123] in, represents the strain energy density, and denote the components of the stress and strain tensors, respectively, 、 、 represents the normal stress component, 、 、 represents the shear stress component, 、 、 represents the normal strain component, 、 、 represents the shear strain component.

[0124] The calculation formula of the damage risk index can be expressed as:

[0125] ,

[0126] in, represents the damage risk index, 、 、 Represent the weight coefficients of high, medium and low risk areas respectively, 、 、 Represents the proportion of high, medium and low risk areas to the total area, The concentration factor of high-risk areas is calculated as:

[0127] ,

[0128] in, represents the number of connected components in high-risk areas, represents the maximum possible number of connected components (set to 10% of the total number of units), represents the concentration influence coefficient (set to 0.5).

[0129] The specific implementation method for comprehensively evaluating the predicted compressive strength value using a multi-source data fusion algorithm in step S08 is to first establish a Bayesian network model. The network structure consists of four input nodes (micro-deformation displacement characteristics, spectral characteristic parameters, stiffness coefficient, and failure risk index) and one output node (the predicted compressive strength value). The conditional probability distribution of each node is then defined. A Gaussian mixture model is used to describe the probability distribution of continuous variables, with the number of mixture components set to 3. The model parameters are estimated using the expectation-maximization algorithm. Bayesian inference is then performed using the Monte Carlo Markov Chain (MCMC) method. The Metropolis-Hastings algorithm is used for sampling, with the number of samples set to 10,000. The first 2,000 samples are discarded as a warm-up period. The posterior distribution of the predicted compressive strength value is then calculated based on the sampling results. The median of the posterior distribution is used as the final predicted value, and the 2.5% and 97.5% quantiles are used as the lower and upper limits of the 95% confidence interval. Finally, the prediction reliability index is calculated. The mean absolute percentage error (MAPE) of the prediction model is calculated based on historical data. This error is typically kept below 5%. If this threshold is exceeded, the system will issue a calibration prompt. This step integrates various test indicators and uses probabilistic statistical methods to obtain reliable compressive strength predictions. It also provides a range of uncertainty for the prediction, providing a scientific basis for decision-making in practical applications.

[0130] The probability density function of the Gaussian mixture model can be expressed as:

[0131] ,in, represents the observation data, represents the model parameters, Indicates the number of mixed ingredients, Indicates the The weights of the mixture components (satisfying ), Indicates the mean , the covariance matrix is Multivariate Gaussian distribution of :

[0132] ,

[0133] in, Indicates the data dimension, represents the determinant of the covariance matrix.

[0134] The specific implementation for outputting the test results in step S09 involves first generating a compressive strength prediction report, including the predicted value (in N), the relative prediction error (based on historical data statistics, typically ±3% to ±5%), the 95% confidence interval (expressed as a percentage of the predicted value, typically within the range of ±5% to ±10%), and test condition parameters (such as ambient temperature and humidity). Next, a stress distribution visualization report is generated, including a 3D color stress cloud map, a principal stress vector map, and stress concentration area identification. The image resolution must be at least 1024×768 pixels, using a rainbow color scale and a numerical scale. Next, a damage risk assessment report is generated, including a risk index value (a dimensionless value ranging from 0 to 100), a risk level classification (0 to 30 for low risk, 30 to 70 for medium risk, and 70 to 100 for high risk), and a description of the weak area location (identified using the carton surface partition number). Finally, the test results are exported in PDF, Excel, and JSON formats, and a historical data comparison function is provided for comparative analysis with historical test results of similar cartons. The purpose of this step is to present complex test analysis results to users in an intuitive and easy-to-understand manner, allowing users to quickly make judgments and decisions. At the same time, it provides a standardized data interface to support subsequent data analysis and management.

[0135] The loss function in the hierarchical attention box strength evaluation model is a weighted combination of mean square error loss and weighted region-sensitive loss:

[0136] ,

[0137] in, Represents the mean squared error loss, which is used to measure the difference between the predicted strength and the actual strength:

[0138] ,

[0139] in, Indicates the The actual compressive strength of the samples, represents the compressive strength predicted by the model, Indicates the sample size. Represents weighted region-sensitive loss, which is used to enhance the model's ability to identify structurally weak areas:

[0140] ,

[0141] in, Indicates the Sample No. The actual strain energy density in the region is represents the strain energy density predicted by the model, Indicates the number of regional divisions, Indicates the The weight of each region is proportional to the frequency of damage in the historical damage data:

[0142] ,

[0143] in, Indicates the The frequency of damage in a region in historical data. and are the weight coefficients of the two loss terms, which are set to 0.7 and 0.3 respectively.

[0144] Through the above implementation, the online automatic testing method for the compressive strength of corrugated cardboard boxes realizes the non-destructive prediction of the compressive strength of the cardboard boxes, has the advantages of fast testing speed, high prediction accuracy, and simple operation, and can be effectively applied to the quality control process of corrugated cardboard box production lines.

[0145] In order to better understand and implement the present invention, Example 2 of a specific application scenario of the present invention is provided below: In response to the problem of damage to corrugated boxes during the transportation of electronic products, researchers used the method of the present invention to conduct online compressive capacity testing on corrugated boxes of different specifications. Three common types of corrugated boxes were selected for this test: Type A (3-layer single corrugated box), Type B (5-layer double corrugated box) and Type C (7-layer triple corrugated box), and 100 samples of each type were selected for testing. The test environment temperature was controlled at 23±1°C and the relative humidity was controlled at 50±2% to ensure the consistency and reliability of the test results. As Figure 2 The figure shows the schematic diagram of the online automatic testing system for the compression resistance of corrugated boxes.

[0146] First, if Figure 3As shown in Figure 1, a micro-deformation loading system that meets the requirements of step S01 was constructed. The system uses a servo motor with a rated power of 2.5kW and a maximum torque of 12N·m. The controller uses a 32-bit microprocessor with a control frequency set to 1000Hz. The loading force sensor has a range of 0 to 2000N, a sensitivity of 0.01% of full scale, and a signal-to-noise ratio of 87dB. The flatness error of the balanced loading platform is controlled within 0.008mm. The PID parameters of the closed-loop force feedback control algorithm are shown in Table 1:

[0147] Table 1 PID control parameters of micro-deformation loading system

[0148]

[0149] Then, according to the requirements of step S02, a high-precision laser displacement sensor array is arranged. The sensor bracket is made of carbon fiber composite material with a thermal expansion coefficient of 4.2× / ℃. The specific arrangement of the 24 laser displacement sensors is shown in Table 2:

[0150] Table 2 Laser displacement sensor layout plan

[0151]

[0152] Next, according to the requirements of step S03, the acoustic emission signal is collected and the spectral characteristic parameters are extracted. Eight piezoelectric sensors are attached to the surface of the carton. The sensor sensitivity is 75dB, the frequency response range is 20kHz to 200kHz, and the sensor spacing is controlled within 180mm. The trigger threshold of the acoustic emission signal is set to 3.2 times the root mean square value of the background noise, and the sampling rate is 500kHz. The collected acoustic signal is analyzed using short-time Fourier transform. The main spectral characteristic parameters of the three carton types are shown in Table 3:

[0153] Table 3 Acoustic emission spectrum characteristic parameters of different types of cartons

[0154]

[0155] Figure 5 The horizontal axis represents frequency, and the vertical axis represents normalized amplitude. The graph plots the frequency spectra of three different types of cartons. This chart clearly shows that as the number of corrugated carton layers increases, the peak frequency of the acoustic emission signal gradually shifts toward higher frequencies, and the energy share in the high-frequency band (100-200kHz) gradually increases. Based on the requirements of step S04, an anisotropic constitutive model for the carton material was established. The material elastic constants for type A carton are shown in Table 4:

[0156] Table 4 Material elastic constants of type A cartons

[0157]

[0158] After inputting the micro-deformation displacement data into the finite element model, the calculated stress field distribution shows that when the A-type carton is subjected to a test load of 200N, the maximum stress occurs at the connection between the side and the bottom, reaching 1.25MPa, and the stress concentration coefficient is 2.8.

[0159] Following the requirements of step S05, the hierarchical attention cardboard strength assessment model is used to process micro-deformation displacement data and spectral feature parameters. The model structure includes six convolutional layers and a bidirectional long short-term memory network, and a multi-head cross-attention mechanism is used to fuse features. The model configuration and training parameters are shown in Table 5:

[0160] Table 5 Hierarchical attention model configuration parameters

[0161]

[0162] According to the requirements of step S06, the stiffness coefficients of the three cartons were calculated and compared with the benchmark thresholds. The results are shown in Table 6:

[0163] Table 6 Comparison of stiffness coefficients of different types of cartons

[0164]

[0165] According to the requirements of step S07, the strain energy distribution map is generated through the stress field distribution, and the structural weak areas are identified. The damage risk assessment results of the three types of cartons are shown in Table 7:

[0166] Table 7 Damage risk assessment results for different types of cartons

[0167]

[0168] Figure 6A three-dimensional surface graph displays the stress distribution on the carton surface. The horizontal and vertical axes represent the relative position coordinates of the carton surface, and the vertical axis represents the stress value (MPa). A color gradient is used to represent stress magnitude, from blue (low stress) to yellow to red (high stress). The graph also identifies areas of varying risk: red dots indicate high-risk areas (stress > 1.5 MPa), orange dots indicate medium-risk areas (1.0 MPa < ≤ 1.5 MPa), and green dots indicate low-risk areas (stress ≤ 1.0 MPa). Stress concentration areas at the corners (risk index 76.4) and the central compression area (risk index 58.2) are specifically noted. According to the requirements of step S08, a multi-source data fusion algorithm is used to perform a comprehensive assessment of compressive strength. The Bayesian network model uses a Gaussian mixture model to describe the conditional probability distribution of each node, with three components. Using the MCMC method, 10,000 samples were sampled, and the prediction results are shown in Table 8.

[0169] Table 8 Prediction results of compressive strength of different types of cartons

[0170]

[0171] Finally, according to the requirements of step S09, a test result report is generated, which includes three parts: compressive strength prediction, stress analysis, and risk assessment. For the weak areas of the C-type carton, the recommended optimized design scheme is shown in Table 9:

[0172] Table 9 C-type carton structure optimization suggestions

[0173]

[0174] Traditional methods for testing the compressive strength of corrugated cardboard primarily rely on destructive testing, where a press applies increasing pressure to the cardboard until it breaks, and the maximum pressure at failure is recorded as the compressive strength. This method has the following problems: First, the testing process completely destroys the sample, resulting in a waste of resources; second, each cardboard can only be tested once, making full inspection on the production line impossible; third, it is impossible to predict the cardboard's weak areas and possible failure modes in advance, making it difficult to improve the design in a targeted manner; and fourth, the test results are highly discrete and significantly affected by operating conditions. The proposed online automatic testing method for the compressive strength of corrugated cardboard offers the following significant advantages over traditional methods: first, it utilizes the principle of micro-deformation testing, which is non-destructive to the cardboard; second, it acquires multi-dimensional information through a high-precision sensor array and acoustic emission technology, which can comprehensively reflect the structural characteristics of the cardboard; third, it utilizes a method that combines physical models and deep learning to achieve high-precision prediction of compressive strength, with an average prediction error of less than 5%; finally, it can identify structural weaknesses and quantify the risk of failure, providing a direct basis for optimizing cardboard structure.

[0175] It should be noted that the variables involved in the present invention are explained in detail as shown in Tables 10 and 11 below.

[0176] Table 10 Variable Explanation Table (Part 1)

[0177]

[0178] Table 11 Variable Explanation Table (Part 2)

[0179]

[0180] The above description is only a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any technician familiar with this technical field can easily think of changes or replacements within the technical scope disclosed by the present invention, which should be covered by the scope of protection of the present invention.

Claims

1. A method for online automatic testing of the compressive strength of corrugated boxes, characterized in that: include: Construct a micro-deformation loading system to apply the test load to the top of the corrugated box; Use high-precision laser displacement sensor array to measure micro-deformation displacement data; Acoustic emission signals are collected synchronously and spectral characteristic parameters are extracted; micro-deformation displacement data is analyzed based on the physical mechanism model of the carton fiber structure, and the stress distribution calculation equation is applied to calculate the stress field distribution of the corrugated carton wall; The micro-deformation displacement data and spectral feature parameters are input into the pre-trained hierarchical attention carton strength assessment model for analysis to obtain the predicted value and confidence interval of the corrugated carton compressive strength; Calculate the stiffness coefficient and compare it with the benchmark threshold; identify weak areas of the structure and quantify the damage risk index; use a multi-source data fusion algorithm to comprehensively evaluate the compressive strength prediction value and confidence interval, specifically: establish a Bayesian network model, which includes four input nodes, namely micro-deformation displacement characteristics, spectral characteristic parameters, stiffness coefficient and damage risk index, and an output node. The output node is the compressive strength prediction value, and a reliable compressive strength prediction result is obtained through probabilistic statistical methods; the compressive strength prediction value, the confidence interval, the stress field distribution and the damage risk index are output as test results.

2. The method for online automatic testing of the compressive strength of corrugated boxes according to claim 1, characterized in that: The test load refers to the pressure applied to the top of the corrugated box by the micro-deformation loading system, which is within 10% of the expected maximum load-bearing capacity.

3. The method for online automatic testing of the compressive strength of corrugated boxes according to claim 2, characterized in that: The micro-deformation loading system refers to a pressure-applying device driven by a high-precision servo motor. Closed-loop feedback control ensures a smooth and precise loading process. The loading force range is 50N to 2000N, with an accuracy of no less than 0.5%.

4. The method for online automatic testing of the compressive strength of corrugated boxes according to claim 3, characterized in that: The high-precision laser displacement sensor array refers to a measurement network composed of 24 triangulation principle laser displacement sensors with a measurement accuracy of 1 micron and a sampling frequency of 1000 Hz, which is used to capture tiny deformations on the surface of the carton.

5. The method for online automatic testing of the compressive strength of corrugated boxes according to claim 4, characterized in that: The micro-deformation displacement data refers to a set of numerical values ​​of deformation displacement of six surfaces of the corrugated box under the action of a test load, collected by a high-precision laser displacement sensor array.

6. The method for online automatic testing of the compressive strength of corrugated boxes according to claim 5, characterized in that: The acoustic emission signal refers to the high-frequency sound wave signal generated by the fibers inside the corrugated box under the action of the test load, with a frequency range of 20kHz to 200kHz, which is collected by a piezoelectric sensor array.

7. The method for online automatic testing of the compressive strength of corrugated boxes according to claim 6, characterized in that: The spectrum characteristic parameters refer to a set of characteristic parameters extracted after spectrum analysis of the acoustic emission signal, including energy distribution of each frequency band, peak frequency, center frequency and bandwidth.

8. The method for online automatic testing of the compressive strength of corrugated boxes according to claim 7, characterized in that: The physical mechanism model of the carton fiber structure refers to a stress-strain analysis model based on the anisotropic mechanical properties of paper materials, taking into account the interlayer structural characteristics and fiber arrangement direction of corrugated paper.

9. The method for online automatic testing of the compressive strength of corrugated boxes according to claim 8, characterized in that: The stress distribution calculation equation is based on Hooke's law and is used to calculate the stress distribution state of the corrugated box wall under the conditions of micro-deformation displacement data. The input includes micro-deformation displacement data, material elastic modulus tensor, Poisson's ratio parameter, corrugated structure geometric parameters and test load size. The output is three-dimensional stress field distribution and strain energy density distribution.

Images