A packaging control method integrating multi-variable dynamic analysis

Abstract

The present invention discloses a packaging control method integrating multi-variable dynamic analysis, which relates to the technical field of packaging dynamic control and covers the complete process from data collection to the implementation of protection strategies. First, by deploying multi-source sensors at packaging and transportation nodes, stress, acceleration and environmental parameters are collected in real time to form multi-dimensional raw data covering the entire process. Then, the data is subjected to time-domain alignment and feature extraction using denoising filtering, wavelet decomposition and environmental coupling methods to obtain structured data in a unified coordinate system. The packaging fatigue risk is accurately identified through the micro-crack jump function and the cumulative shock measurement model, and the crack propagation trend and failure prediction are generated. Finally, according to the section risk changes, a segmented protection strategy is executed and the protection plan is dynamically optimized to ensure efficient control. By means of non-linear modeling and real-time optimization, the limitations of traditional packaging control are overcome, and it has broad application prospects.

Description

Technical Field

[0001] The present invention relates to the technical field of packaging dynamic control, and specifically to a packaging control method integrating multi-variable dynamic analysis. Background Art

[0002] With the diversification of modern logistics models and the continuous improvement of the supply chain speed, the stability and safety of packaging during transportation and storage have increasingly become key factors affecting product quality and end-user experience. Different types of goods often need to go through multiple loading and unloading operations, sudden changes in temperature and humidity, high-frequency or low-frequency vibrations and other complex environments during long-distance transportation across regions and multiple stages. Especially in the distribution of goods that are vulnerable to impact or vibration-sensitive (such as precision instruments, electronic components or fragile items), if the cumulative stress on the packaging cannot be accurately identified and dynamically controlled, it is extremely easy to cause product damage, an increase in the return rate, and even potential safety hazards due to packaging failure. In the face of this demand, existing technologies usually rely on single-parameter monitoring or phased sampling inspection methods, and it is difficult to identify the subtle differences in each link of the transportation process with high timeliness and high precision. Coupled with the fact that the tolerance limits of different materials and different specifications of packaging under transportation and storage conditions are not unified, existing general packaging designs or static protection solutions are difficult to meet the requirements of complex and variable logistics scenarios.

[0003] In a Chinese invention patent with the authorization announcement number CN113065740B, a high-efficiency product packaging control system is disclosed, including: a set domain creation module, an information processing module, a physical address acquisition module, and a verification module; the set domain creation module is used to create a spatial coordinate set domain; the information processing module is used to obtain the layout distribution information and the first sequence information of the outer box, and obtain the first binding information according to the layout distribution information, the first sequence information and the spatial coordinate set domain; the physical address acquisition module is used to obtain the second sequence information of the product, and obtain the physical address of the outer box corresponding to the product based on the second sequence information and the first binding information; the verification module is used to perform verification measures according to the physical address to obtain a configuration signal, and perform product packaging measures according to the configuration signal; the present invention can accurately locate the position of the outer box required for the product, thereby ensuring the correctness of product packaging, reducing its control cost, and improving its control efficiency.

[0004] However, combining the existing technology and the above application scenarios;

[0005] In practice, simply relying on the limited records of vibration amplitude or environmental parameters provided by the existing technology cannot effectively quantify the superposition of multiple stresses suffered by the packaging during the entire circulation process, nor is it easy to capture the impact of high-frequency and intermittent small-amplitude shocks on the early microcracks or fatigue phenomena of packaging materials. When various dynamic factors are coupled with each other and superimposed at different logistics stages, potential damage to the packaging structure or sealing part often appears in advance, resulting in the rapid expansion of cracks that are difficult to detect with the naked eye in subsequent transportation or storage links. The existing technology lacks the ability to perform real-time fusion and in-depth analysis of multi-source data, and also lacks targeted protection measures for different sections, so that it is impossible to actively intervene in the packaging protection measures in special scenarios such as increased vibration intensity, sudden changes in temperature and humidity, or frequent loading and unloading, ultimately leading to a decline in product quality, an increase in the rework rate, and an increase in the enterprise's logistics costs. Therefore, a technical means that can monitor, analyze the multi-variable dynamics of packaging and flexibly adjust the protection plan according to the prediction results is urgently needed to achieve high-reliability and low-cost logistics control.

[0006] For this reason, the present invention provides a packaging control method integrating multi-variable dynamic analysis. Summary of the Invention

[0007] (I) Technical problems to be solved

[0008] Aiming at the deficiencies of the existing technology, the present invention provides a packaging control method integrating multi-variable dynamic analysis. By arranging multi-source sensors at packaging and transportation nodes, stress, acceleration and environmental parameters are collected in real time to form multi-dimensional raw data covering the entire process; then, time-domain alignment and feature extraction are performed on the data by using denoising filtering, wavelet decomposition and environmental coupling methods to obtain structured data in a unified coordinate system; the packaging fatigue risk is accurately identified through the microcrack jump function and the cumulative shock measurement model, and the crack propagation trend and failure prediction are generated. Finally, according to the change of section risk, a segmented protection strategy is implemented and the protection plan is dynamically optimized to ensure efficient control. The limitations of traditional packaging control are overcome through non-linear modeling and real-time optimization, thereby solving the technical problems recorded in the background art.

[0009] (II) Technical solutions

[0010] To achieve the above object, the present invention is realized through the following technical solutions: A packaging control method integrating multi-variable dynamic analysis, including, after the initial logistics transportation process is started and the monitoring requirement is confirmed, multi-source sensors are arranged at packaging and transportation nodes to collect environmental parameters and the like in real time, so as to obtain multi-dimensional raw data covering the entire process and establish a unified space-time mark;

[0011] After the initial storage of the multi-dimensional raw data stream, methods such as denoising filtering, wavelet decomposition, and environmental coupling are used to perform time-domain alignment and high-order feature extraction on the sensor output to obtain structured data with a unified coordinate system and covering multi-dimensional indicators;

[0012] When abnormal mutation signs are detected in the structured data, a micro-crack jump function and a high-order cumulative shock metric model are used to finely identify and quantitatively evaluate the packaging fatigue risk, and generate the crack propagation trend and potential failure time window of each sensor;

[0013] When the analysis results show that the risk indicators of a certain logistics section increase significantly, a segmented protection strategy is immediately executed and a dynamic optimization algorithm is triggered to differentially strengthen and real-time correct the packaging materials and buffer layout, so as to maintain a high safety margin and closed-loop control the entire process on the premise of minimizing resource input.

[0014] Preferably, determine the area to be monitored and the key parts where impacts or fatigue may concentrate to form a preliminary candidate coordinate set for sensor layout , where represents the spatial coordinates or installation scheme of the th candidate position; optimize the layout of the candidate coordinate set, and use the following formula to measure the coverage and comprehensive effectiveness of the deployment to construct the scheme coverage :

[0015]

[0016] where: is the number of candidate sensors; represents the optimal distance metric between the sensor position π and the packaging target monitoring area R; is a user-defined attenuation coefficient; is the set of sensor coordinates;

[0017] By maximizing the scheme coverage or meeting its threshold conditions, select the final deployment scheme, generate a list of the installation positions and installation methods of the formal sensor network, and generate a parameter set of sensor layout and installation information;

[0018] Preferably, after the initial layout of the sensor network, perform basic calibration on the accelerometer, gyroscope, and vibration sensor, record the calibration parameter set, denoted as ;

[0019] Deploy environmental sensing devices in the target area and record the parameter set , which represent temperature, humidity, air pressure, and the number of transfers / loading and unloading frequencies, etc.;

[0020] Register the calibrated sensor parameters together with the environmental sensor inventory into the calibration and environmental node configuration parameter set, and establish a corresponding relationship with the sensor layout and deployment information parameter set;

[0021] Preferably, according to the coordinates and calibration information, start real-time data collection at each transportation or storage stage, and each sensor samples at the set sampling frequency Output timing data , and each sensor synchronously outputs environmental data ;

[0022] Construct preliminary time stamps and geographical tags to form data objects , where and represent time and space / node tags;

[0023] Set a specific dynamic weight function to represent time and space / node tags;

[0024] Set a specific dynamic weight function Perform preliminary time decay weighting on the sensor data to extract the potential impact of recent vibration shocks on microcracks. An example of the time window decay weighting function is as follows:

[0025]

[0026] In the formula: represents the weighted result of sensor within the time window; is the decay factor of sensor ; is the upper limit of integration. Upload all real-time data together with the preliminary weighting results to the data management platform;

[0027] Preferably, map the timing signals of all sensors to a unified time scale , denoise and smooth the timing signal of each sensor , and obtain the preliminarily denoised signal ; Output the denoised and time-domain aligned data , and retain all time scales and sensor numbers in the corresponding metadata ;

[0028] Preferably, perform multi-scale decomposition on the output denoised and time-domain aligned data using wavelet transform, and define as the discrete wavelet transform operator:

[0029]

[0030] Wherein: represents the time-frequency coefficient under the scale and translation conditions; is the selected wavelet basis function, different and correspond to different frequency bands or time localizations;

[0031] is to quantify the distribution characteristics of vibration energy at different scales, and the following characteristic energy spectrum is defined , which is used to measure the cumulative impact intensity of the package within a specific frequency band :

[0032]

[0033] Wherein: is the time-frequency coefficient of the sensor at the scale and translation after wavelet transform;

[0034] , represents the set of selected frequency bands or scale intervals; represents the absolute value of the waveform amplitude, is a user-defined amplitude exponent; represents the difference or increment of the waveform amplitude in the translation dimension : ; is the local mutation weight coefficient, which is used to extract only the positive mutation part; is the outer-layer non-linear amplification exponent;

[0035] The output high-order characteristic data can be formalized as , and the corresponding relationship with the sensor number is retained; using wavelet transform to extract the characteristic energy spectrum in the high, medium, and low frequency intervals respectively, to target and locate the frequency band where the material is prone to fatigue, and generate several characteristic components for each sensor ;

[0036] Preferably, combining the environmental sensing data , perform coupling analysis on each vibration characteristic :

[0037]

[0038] Wherein: represents the fused environmental coupling index; It can be a linear or non - linear combined function of one or more environmental variables, is the adjustment coefficient;

[0039] After obtaining the environmental coupling index define the multi - dimensional index vector :

[0040]

[0041] Connect all sensors with all environmental nodes with all environmental nodes and finally output to a unified data management structure, retaining the corresponding relationship between sensors and environmental nodes;

[0042] Preferably, obtain the multi - dimensional index vector corresponding to each sensor , where is the unified calibrated time scale. According to the defined time window or logistics stage segmentation, all are concatenated in sequence into a feature trajectory sequence ;

[0043] Define the symptom function to compare the change rate of the multi - dimensional index at adjacent or neighboring times:

[0044]

[0045] In the formula: is the time difference used to measure the difference between adjacent times or adjacent logistics segments; represents the vector norm, is the non - linear exponent;

[0046] When exceeds the preset dynamic threshold, it is regarded as a possible micro - crack symptom, and record the time and the relevant sensor numbers in the short - term risk list; output the preliminary micro - crack symptom record and the corresponding change amplitude record ;

[0047] Preferably, for the time that has been marked as a suspected micro - crack symptom, further analyze the cumulative impact degree of its historical data to determine whether the crack may accelerate its expansion in the subsequent logistics stage. Define the following local fatigue impact measure :

[0048]

[0049] In the formula: is the jump variable degree of microcracks; is the partial energy index obtained by multi-scale wavelet transform, represents a specific frequency band or scale; is the index set of all selected scales or frequency bands of interest in multi-scale analysis; is the environmental variable function; is for the environmental variable is the function for deforming or mapping; is the experimentally calibrated material sensitivity coefficient; is the non-linear amplification index for microcrack symptoms; is the wavelet energy weight at each scale (or frequency band);

[0050] : the non-linear index for wavelet energy; is the environmental coupling adjustment coefficient; , is the time decay factor;

[0051] After obtaining the local fatigue impact measure Combined with the fatigue curve or physical mechanism model of the packaging material, the impact measure is converted into a risk factor of failure tendency :

[0052]

[0053] Among them: is the impact-failure mapping coefficient;

[0054] When the risk factor is close to 1, it means that the microcracks are very likely to further expand to visible or cause overall damage in the short term; when this value is low, it means that the packaging can still maintain its structural integrity, but continuous monitoring is required; finally, the sequence of packaging damage risk factors is output , and it is marked corresponding to the sensor number and the time scale ;

[0055] Preferably, the packaging damage risk factor is integrated into a time series matrix and compared or regressively corrected with the actual material failure cases to obtain the overall packaging failure probability function;

[0056] According to the accumulation of the microcrack risk factor to infer the potential damage probability at a subsequent moment :

[0057]

[0058] Among them: and , which is the model coefficient of the material expansion rate;

[0059] When the packaging damage risk factor is large, even if the remaining time is not long, failure is likely to occur; when the packaging damage risk factor is small, a longer time or additional impact is required to trigger damage;

[0060] When analyzing different transportation stages or storage conditions, sort the rising indicators of the microcrack risk factor and the potential damage probability in the multi-dimensional index vector to identify the vibration frequency bands, temperature and humidity levels, or loading and unloading operation links that require the most attention; output the identification results together with the predicted value of the potential damage probability , the potential damage probability within a given time range , the list of key influencing factors or rankings;

[0061] Preferably, according to the output potential damage probability and the packaging damage risk factor , combined with the main nodes and transportation methods in the logistics route, divide the overall logistics process into several sections and assign a section identifier to each section; map the time scale or interval to the corresponding logistics section s to form a section-time correspondence table;

[0062] Statistically analyze or predict the time distribution of packaging damage risk within each section s, mainly based on the packaging damage risk factor or the potential damage probability , and calculate the overall risk index of this section:

[0063]

[0064] where: is the time range corresponding to section s; Select the maximum risk factor of all sensors at the same moment in the same section; is a mapping function related to the environment or scenario, and finally output the overall risk index as a measure of the risk level of each section;

[0065] Preferably, based on the overall risk index of the section , combined with key influencing factors, formulate a protection plan corresponding to each section s, including:

[0066] In the section where the overall risk index is higher than expected, appropriately increase cushioning materials or strengthen the impact-resistant structure of the outer box;

[0067] Allocate more manpower or automatic equipment at frequently loaded and unloaded nodes to reduce impact; increase moisture-proof or thermal insulation layers in cold chain or high humidity sections, etc.;

[0068] Before the start of each section, appropriately increase the monitoring frequency of sensors and environmental monitoring equipment , focus on the high-risk sensor numbers , and compare the newly collected real-time data during this period with the calculated packaging breakage risk factor or potential breakage probability to observe whether it is consistent with the prediction or shows a deviation;

[0069] Refer to the following formula to preliminarily evaluate the protection effect of this section and generate a protection effect evaluation value :

[0070]

[0071] In the formula: is the previously predicted risk factor, is the newly observed risk factor in this section; represents taking the average over all time slices or multiple time points within the section ;

[0072] is the number of key sensors active within this section;

[0073] Preferably, when the evaluation result, the protection effect evaluation value is continuously lower than the effect threshold or is negative, it is determined that the current protection measures do not meet the expectations: If it is monitored that the predicted risk factor sharply rises within a short period of time, trigger the temporary optimization process;

[0074] Temporarily add a more robust outer box, add more anti-seismic cushions or strengthen auxiliary means during loading and unloading in high-risk sections, and define the incremental optimization intensity function :

[0075]

[0076] In the formula: and , and To optimize the sensitivity coefficient Used to capture the protection effect evaluation value When it is negative or less than zero, if it is positive, this item is recorded as zero;

[0077] Indicates the amplification of environmental excitation, Used to represent the comprehensive measurement function of environmental excitation;

[0078] After the optimization measures are executed, continue to collect and calculate the new predicted risk factors in real time , if it is found that the incremental optimization intensity function after optimization Is significantly improved or the new risk factors gradually tend to be stable;

[0079] Otherwise, continue to perform incremental optimization or switch to other protection schemes, and record the empirical parameters in the optimization process in the database.

[0080] (III) Beneficial effects

[0081] The present invention provides a packaging control method integrating multivariate dynamic analysis, with the following beneficial effects:

[0082] The present invention sets up multi-level sub-steps within the four major steps of monitoring - preprocessing - analysis - protection, each of which conducts refined processing on sensor layout, data denoising, feature extraction, micro-crack identification and risk prediction, and logistics section protection and optimization, thereby constructing a complete control system across dimensions and stages. Through the reasonable layout of multi-source sensors and high-order feature extraction based on wavelet transform, the scheme can deeply explore the potential hidden fatigue effects that may occur in complex logistics environments, and weaken the noise interference through non-linear filtering and outlier rejection. In this process, aiming at the potential damage of packaging materials under short-term high-amplitude impacts or long-term cumulative vibrations, the present invention defines a micro-crack jump and cumulative impact measurement model, and comprehensively uses formulas such as integral attenuation and environmental index amplification to achieve sensitive capture of early signs of fatigue evolution. For special scenarios such as repeated loading and unloading, high humidity or extreme temperature involved in multi-section transportation, the model can integrate historical and real-time monitoring in time series, accurately measure the occurrence probability of packaging cracks and the subsequent expansion speed. At the same time, unified time and space marker management ensures the consistency of data across steps, avoids parameter confusion and semantic disconnection, and enables the output of each sub-step to be seamlessly docked to the next link.

[0083] In the protection stage, the present invention allocates differentiated solutions according to the risk indices of different transportation or storage sections. It can either strengthen the outer box structure, add buffer pads or optimize the loading and unloading process in high-risk sections, or adopt appropriate streamlining strategies in low-risk areas to achieve a reasonable balance between resources and safety. Moreover, by comparing the continuous monitoring results with the prediction results, the evaluation value of the protection strategy effect is calculated. Once the observed risk is higher than expected, the dynamic optimization mechanism will be triggered, and the solution will be revised based on the incremental optimization intensity formula to achieve the closed-loop control of "data - prediction - execution - feedback". Benefiting from the high sensitivity of non-linear modeling methods to extreme environmental impacts and the high efficiency of real-time optimization in correcting strategy deviations, the present invention not only overcomes the limitations of the lack of multi-source integration and immediate adjustment in traditional packaging control, but also provides expandable technical support for recyclable packaging, material iterative design and cross-industry applications. BRIEF DESCRIPTION OF THE DRAWINGS

[0084] Figure 1 It is a schematic flow chart of the packaging control method integrating multi-variable dynamic analysis of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0085] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0086] Please refer to Figure 1 , the present invention provides a packaging control method integrating multi-variable dynamic analysis, including:

[0087] Step 1: After the initial logistics transportation process is started and the monitoring requirements are confirmed, multi-source sensors are arranged at packaging and transportation nodes to collect environmental parameters and other data in real time, so as to obtain multi-dimensional raw data covering the whole process and establish a unified space-time mark;

[0088] The content of Step 1 is as follows:

[0089] Step 101: Sensor layout planning and layout optimization

[0090] According to the known packaging size, material properties and general situation of the logistics route, determine the areas to be monitored and the key parts where impacts or fatigue may concentrate, and form a preliminary candidate coordinate set for sensor layout , where represents the spatial coordinates or installation scheme of the th candidate position; optimize the layout of the candidate coordinate set, use the following formula to measure the coverage and comprehensive effectiveness of the deployment, and construct the scheme coverage :

[0091]

[0092] Wherein: is the number of candidate sensors; represents the optimal distance metric between the sensor position π and the target monitoring area R of the package (or transportation carrier) (which can be customized according to the package shape and material distribution, such as surface fit, thickness of the protective layer, etc.); is a customized attenuation coefficient used to adjust the influence of distance on the deployment effect (which can be determined comprehensively according to material sensitivity and layout difficulty); is the set of sensor coordinates;

[0093] By maximizing the scheme coverage or meeting its threshold condition, select the final deployment scheme, so that the sensors can not only cover high-risk areas, but also avoid redundant or overly dense layout in logistics. After completion, generate a formal list of the installation positions and installation methods of the sensor network, including the coordinates or angles of each sensor on the package, and generate a parameter set of sensor layout and layout information;

[0094] When in use, by arranging at key positions significantly improves the capture probability of potential impact accumulation points, realizes effective coverage of high-risk parts; reduces redundant deployment: uses optimization functions such as scheme coverage to avoid reinstalling sensors at non-critical parts of the package, reduce costs and improve monitoring efficiency; can flexibly configure the sensor array according to different package materials, sizes and characteristics of logistics routes, which is conducive to large-scale application and improves adaptability.

[0095] Step 102, Basic calibration and environmental monitoring node setting

[0096] After the preliminary layout of the sensor network is completed, perform basic calibration on each accelerometer, gyroscope and vibration sensor, record calibration parameter sets such as their zero offsets, full-scale upper limits and sensitivity coefficients, denoted as and store them in the underlying firmware;

[0097] To obtain the environmental data of the logistics node, it is necessary to deploy environmental sensing devices at key transfer links or target areas with large temperature and humidity gradients, and record the parameter set which represent temperature, humidity, air pressure and transfer times / loading and unloading frequencies respectively; register the calibrated sensor parameters and the environmental sensor list together into the calibration and environmental node configuration parameter set, and establish a corresponding relationship with the sensor layout and layout information parameter set to ensure that the sensor network at the same position can accurately associate its environmental node;

[0098] During use, calibrate the zero bias and sensitivity coefficients of each sensor to ensure accurate comparison of the timing signals of the sensors in the subsequent steps and environmental data so that the data can be accurately compared and data consistency can be ensured; uniformly collect and correlate the variable links such as temperature, humidity, air pressure, and number of loading and unloading times to provide multi-dimensional environmental input for the subsequent algorithm; the unified calibration information can maintain the monitoring accuracy in different batches or multiple models of packaging, reducing the workload of repeated configuration.

[0099] Step 103, Real-time data collection and label management

[0100] According to the confirmed coordinates and calibration information, start real-time data collection at each transportation or storage stage. Each sensor outputs timing data at the set sampling frequency , and each sensor synchronously outputs environmental data ;

[0101] To prevent data from being difficult to manage due to simple accumulation, construct preliminary time stamps and geographical tags (such as GPS or logistics link ID) to form data objects , where and represent time and space / node tags, which are used for alignment and fusion in subsequent steps;

[0102] Set a specific dynamic weight function at this stage to perform preliminary time decay weighting on the sensor data to extract the potential impact of recent vibration shocks on microcracks. An example of this time window decay weighting function is as follows:

[0103]

[0104] In the formula: represents the weighted result of the sensor within the time window; is the attenuation factor of the sensor (which can be selected in combination with the sensitivity recorded during calibration), and the larger the value, the faster the attenuation of earlier data; is the upper limit of integration, representing the time length considered, which is used to focus on recent high-frequency shocks; upload all real-time data together with the preliminary weighting results to the data management platform;

[0105] During use, construct a complete data object through the time stamp and space tag to ensure accurate backtracking of historical events in subsequent steps; time window weighting can be used during the collection stage , highlighting the impact of recent significant stress on material fatigue and enhancing the recognition of critical impacts; meanwhile, it can automatically adapt to changes in the network environment to ensure the continuity and security of data in multi-stage transportation or warehousing scenarios.

[0106] Step 2: After the initial storage of the multi-dimensional raw data stream, use denoising filtering, wavelet decomposition, and environmental coupling to perform time-domain alignment and high-order feature extraction on the sensor output to obtain structured data with a unified coordinate system and covering multi-dimensional indicators.

[0107] The above Step 2 includes the following contents:

[0108] Step 201: Time-domain alignment and non-linear denoising filtering;

[0109] According to the data object generated in the first step , first perform time synchronization on the sensor outputs within the same logistics node or the same transportation section; map the time series signals of all sensors to a unified time scale so that the corresponding vibration stress readings and environmental readings can be found at any moment ; If there are missing or duplicate marks in some data, interpolation or clipping needs to be performed in combination with the basic calibration information of the sensors registered in the first step

[0110] to ensure a strict one-to-one correspondence relationship of the data object in the time dimension;

[0111] For each sensor time series signal , use morphological filtering or adaptive structure element method for denoising and smoothing to obtain the preliminarily denoised signal ; its core calculation can be expressed as: ;

[0112]

[0113] where is a non-linear filtering operator (such as a combination of adaptive morphological opening and closing operations), represents the set of filtering structure elements or kernel functions;

[0114] When performing this filtering, it can be adjusted according to different sensors and different installation positions , for example, select a stronger noise suppression strategy at positions with high vibration intensity, and use a more detail-preserving structure element in areas with weak amplitudes; the final output is a set of denoised and time-domain aligned data , and all time scales and sensor numbers are retained in the corresponding metadata;

[0115] During use, taking as the unified time scale to ensure that all can be fused and compared with each other, which can reduce the probability of subsequent misjudgment; non-linear filtering can significantly reduce the influence of high-frequency noise and retain relatively complete the waveform spikes that may cause microcracks; reduce the interference of outliers and random jitters, provide a more reliable data basis for subsequent index calculation and fatigue analysis model output, and reduce the probability of subsequent misjudgment.

[0116] Step 202, Multi-scale band decomposition and high-order feature extraction

[0117] Perform multi-scale decomposition on the output denoised and time-domain aligned data using wavelet transform, and define as the discrete wavelet transform operator:

[0118]

[0119] where: represents the time-frequency coefficient under the scale and translation conditions, is the selected wavelet basis function, different and correspond to different frequency bands or time localities, thus realizing multi-scale decomposition;

[0120] To quantify the distribution characteristics of vibration energy at different scales, define the following characteristic energy spectrum , which is used to measure the cumulative impact intensity of the package within a specific frequency band (or scale interval):

[0121]

[0122] where: is the time-frequency coefficient of the sensor at the scale and translation after wavelet transform;

[0123] , represents the set of selected frequency bands or scale intervals, which is consistent with the concept of in the previous text; represents the absolute value of the waveform amplitude, is a user-defined amplitude exponent ( paying more attention to large-amplitude impacts when represents the difference or increment of the waveform amplitude in the translation dimension : , used to capture the instantaneous mutation of the amplitude;

[0124] is the local mutation weight coefficient, used to amplify (or suppress) the influence of those suddenly rising amplitude peaks on the cumulative impact;

[0125] is used to extract only the positive mutation part (i.e., the amplitude growth part), and is recorded as zero when negative or without mutation; is the outer - layer non - linear amplification exponent (which can take different values from ), and can further enhance the sensitivity to large - amplitude or mutation signals;

[0126] When in use, by introducing the amplitude mutation term on the basis of the traditional absolute - value accumulation, the energy concentration phenomenon of the packaging material under instantaneous impact can be more precisely described, so as to provide a more sensitive metric for the subsequent fatigue analysis model to capture the early signs of micro - cracks. Different , , combinations can adapt to various packaging materials and different fatigue - sensitivity scenarios;

[0127] The finally output high - order feature data can be formalized as , and retain the corresponding relationship with the sensor number . This output will be integrated with the environmental information in the next step to form multi - dimensional metrics; Using wavelet transform to extract the characteristic energy spectra in the high, medium, and low - frequency intervals respectively, so as to target and locate the frequency bands where the material is prone to fatigue;

[0128] Generate several characteristic components to facilitate the subsequent algorithm to flexibly select the most discriminative metrics and reduce information loss.

[0129] Step 203, Environmental coupling and construction of multi - dimensional index system

[0130] Combine the environmental sensing data (such as temperature , , air pressure , number of loading and unloading times , etc.), and conduct coupling analysis on each vibration characteristic :

[0131]

[0132] Among them: represents the fused environmental coupling index; A linear or non - linear combination function of one or more environmental variables (for example, adding temperature and humidity with certain weights and then multiplying by the logarithm of air pressure), which can be adjusted according to material properties in specific implementations; is an adjustment coefficient, with a value greater than 0, used to control the amplification or attenuation effect of environmental variables on the vibration energy spectrum;

[0133] After obtaining the environmental coupling index a group of multi - dimensional index vectors can be further defined :

[0134]

[0135] Each multi - dimensional index vector combines the high - order vibration characteristics and the external environmental state, reflecting the comprehensive fatigue possibility of the package at time or within a section; Combining all sensors and all environmental nodes the generated multi - dimensional index vectors When finally output to a unified data management structure, the corresponding relationship between the sensors and environmental nodes needs to be retained;

[0136] In use, by coupling environmental parameters such as temperature and humidity with the shock energy spectrum and integrating multiple environmental variables, it can avoid the one - sidedness of asserting the fatigue degree based solely on temperature or humidity and reduce the single - parameter deviation.

[0137] Step 3: When abnormal mutation signs are detected in the structured data, use the micro - crack jump function and the high - order cumulative shock measurement model to finely identify and quantitatively evaluate the packaging fatigue risk, and generate the crack propagation trend and potential failure time window of each sensor;

[0138] The content of the said Step 3 is as follows:

[0139] Step 301: Feature pattern recognition and micro - crack sign detection

[0140] Obtain the multi - dimensional index vector corresponding to each sensor where is the calibrated unified time scale. According to the defined time window or logistics stage segmentation, all are sequentially spliced into a feature trajectory sequence ; ;

[0141] To highlight that micro - cracks often show sudden jumps or local anomalies in certain features in the early stage, the following sign function is defined Compare the change rate of multi - dimensional indices at adjacent or neighboring times:

[0142]

[0143] In the formula: is the time difference used to measure the difference between adjacent moments or adjacent logistics segments (which can be set according to the segment length or refresh frequency of the transportation process); represents the vector norm (for example norm), which is used to comprehensively measure the overall change amplitude of multi-dimensional indicators; is the non-linear exponent, which is used to amplify the larger changes, so as to highlight the abnormal indicators when potential cracks occur;

[0144] When exceeds the preset dynamic threshold (which can be obtained by combining the material fatigue curve, statistical prior or historical sample training), it is regarded as a possible sign of micro-cracks, and the time and relevant sensor numbers are recorded in the short-term risk list; output the preliminary record of micro-crack signs, such as , as well as the corresponding record of the change amplitude ;

[0145] When in use, using the sign function to characterize the jump of the feature vector, it is possible to detect the possible signs of micro-cracks in the packaging earlier, complete the early crack detection, and realize refined anomaly detection through the dynamic threshold or statistical distribution, filter the false mutations caused by short-term noise, and reduce false alarms; record the identified suspected micro-crack moments in the risk control list for key attention in the subsequent risk measurement model, and complete the systematic alarm.

[0146] Step 302, Damage risk measurement based on the fatigue analysis model

[0147] For the moments that have been marked as suspected micro-crack signs , further analyze the cumulative impact degree of its historical data to judge whether the crack may accelerate the expansion in the subsequent logistics stage, and define the following local fatigue impact measurement :

[0148]

[0149] In the formula: is the micro-crack jump measurement; is the partial energy index obtained through multi-scale wavelet transform, represents a specific frequency band or scale; is the index set of all interested scales or frequency bands selected in the multi-scale analysis; is the fusion result of the environmental variable function, such as temperature, humidity, air pressure, number of loading and unloading times, etc.; is the environmental variable Functions that perform deformation or mapping; To calibrate the material sensitivity coefficient by experiment; is the nonlinear amplification index for microcrack signs; is the wavelet energy weight at each scale (or frequency band), which can be set according to the sensitivity of the packaging material to vibrations of different frequencies;

[0150] : Nonlinear index of wavelet energy, if Then the large amplitude vibration energy can be amplified more strongly; is the environmental coupling adjustment coefficient, , used to control the reinforcing (or inhibiting) effects of external environmental variables; , is the time attenuation factor, indicating that the impact closer to the current moment is more influential; is the current time (or the time so far), and the integral upper limit represents the value between 0 and The accumulation of the whole process; : Integral variable, traversing the historical time series experienced by the package.

[0151] Traditional crack risk assessment methods often only consider a single dimension (such as the sum of acceleration amplitudes) or simple power weighting, which makes it difficult to simultaneously reflect the comprehensive impact of multi-scale vibration, environmental factors (temperature and humidity, loading and unloading times, etc.) and temporal cumulative effects on crack propagation. With the help of high-order nonlinear integration, the multi-scale wavelet energy spectrum, microcrack jump variable metric and environmental coupling index are integrated to solve the problem that single-dimensional or simple accumulation methods cannot effectively distinguish the microcrack generation and propagation states.

[0152] Packaging materials often do not break immediately when microcracks appear in the early stage, but if they are subsequently subjected to frequent local impacts or extreme environments (such as high humidity, repeated loading and unloading), the cracks will rapidly deteriorate. This parameter targets the mechanism where recent impacts are more influential than long-term impacts, and uses the time attenuation factor to And the jump amplitude The nonlinear amplification can highlight the most active crack development period and identify potential deterioration risks in advance;

[0153] Obtaining local fatigue impact measurements Finally, the impact measurement is converted into a risk factor for failure tendency by combining the fatigue curve or physical mechanism model of the packaging material. :

[0154]

[0155] in: is the impact-failure mapping coefficient, the larger the value, the more vulnerable the material is to the cumulative impact;

[0156] When risk factors A value close to 1 indicates that microcracks are very likely to further expand to visible cracks or cause overall damage in the short term; when the value is low, it means that the package can still maintain its structural integrity, but monitoring needs to be continued.

[0157] Finally, output the sequence of package damage risk factors and mark it corresponding to the sensor number and the time scale The larger the value, the more vulnerable the material is to cumulative shocks Emphasize the dominance of recent shocks and achieve precise quantification of material fatigue; the risk factor within the range of intuitively reflects the probability level of crack expansion or damage, facilitating rapid decision-making at the business end and completing the risk factor mapping.

[0158] Step 303: Prediction of package structure failure probability and identification of key factors

[0159] Integrate the package damage risk factors into a time series matrix and compare or perform regression correction with the actual failure cases of the material (if recorded historically or obtained through accelerated tests) to obtain the overall package failure probability function; infer the potential damage probability at a subsequent moment based on the accumulation of microcrack risk factors :

[0160]

[0161] Where: and are the model coefficients of the material expansion rate, indicating the possibility of crack development over time based on the cumulative shock degree;

[0162] When the package damage risk factor is large, even if the remaining time is not long, failure is likely to occur; when the package damage risk factor is small, a longer time or additional shocks are required to trigger damage;

[0163] When analyzing different transportation stages or storage conditions, interpretable methods of machine learning (such as feature importance analysis, weighted regression coefficients, etc.) can be used to sort the increase in indicators in the multi-dimensional index vector that can trigger the microcrack risk factor and the potential damage probability to identify the vibration frequency bands, temperature and humidity levels, or loading and unloading operation links that require the most attention; the identification results together with the potential damage probability The predicted value, the potential breakage probability within a given time range and the potential breakage probability within a given time range , a list of key influencing factors or a ranked output;

[0164] When in use, the potential breakage probability can break through the current time limit and give a quantitative result for the potential breakage in the subsequent transportation or storage stage; by using the feature importance analysis of machine learning or regression models, accurately locate the high-frequency or high-humidity intervals that are most likely to induce cracks, increasing interpretability; after identifying the key causes, more efficient differential protection can be implemented in the fourth step, reducing resource investment while enhancing reliability and achieving targeted protection recommendations.

[0165] Step Four: When the analysis results show that the risk indicators of a certain logistics section increase significantly, immediately execute a segmented protection strategy and trigger a dynamic optimization algorithm to differentially strengthen and real-time correct the packaging materials and buffer layout, so as to maintain a high safety margin and closed-loop control the entire process on the premise of minimizing resource investment;

[0166] The above Step Four includes the following content:

[0167] Step 401: Logistics section division and risk matching

[0168] Based on the output potential breakage probability and the packaging breakage risk factors , combined with the main nodes and transportation modes (such as land transportation, sea transportation, air transportation, cold chain, etc.) in the logistics route, divide the overall logistics process into several sections (for example ), and assign a section identifier to each section; map the time scale or interval to the corresponding logistics section s to form a section-time correspondence table for distinguishing different environmental characteristics or impact modes;

[0169] Statistically analyze or predict the time distribution of packaging breakage risks within each section s, mainly based on the packaging breakage risk factors or the potential breakage probability , and calculate the overall risk index of this section:

[0170]

[0171] where: is the time range corresponding to section s; Select the maximum risk factor of all sensors within the same section time to capture the most vulnerable part; A mapping function related to the environment or scenario (which can be determined according to the transportation mode) for amplifying or reducing the failure probability and the comprehensive risk of the section; finally output the overall risk index as a measure of the risk level of each section;

[0172] When in use, the constructed risk index measures the overall risk of each section, making the packaging protection more targeted, avoiding one-size-fits-all overprotection or underprotection, achieving precise resource allocation, and unifying time and space mapping; when crossing multiple transportation modes and multiple stations, it can be integrated with the generated packaging damage risk factor and the potential damage probability data for seamless connection; for low-risk sections, the protection measures can be simplified, while for high-risk sections, the padding can be strengthened or the outer box can be upgraded, thus improving the overall efficiency and balancing efficiency and cost.

[0173] Step 402, Execution and Real-time Monitoring of Protection Strategies

[0174] Based on the overall risk index of the section , combined with key influencing factors (vibration frequency band, temperature and humidity level, number of loading and unloading operations, etc.), formulate the protection plan corresponding to each section, including:

[0175] In sections where the overall risk index is higher than expected, appropriately increase the padding material or strengthen the impact-resistant structure of the outer box;

[0176] Allocate more manpower or automatic equipment at nodes with frequent loading and unloading to reduce the impact; increase moisture-proof or heat-insulating layers in cold chain or high-humidity sections, etc.;

[0177] Before the start of each section, appropriately increase the monitoring frequency of sensors and environmental monitoring equipment , focus on the high-risk sensor numbers , upload the new real-time data (including acceleration, temperature and humidity, air pressure, etc.) collected during this period to the data management platform, and compare it with the already calculated packaging damage risk factor or the potential damage probability to observe whether it is consistent with the prediction or shows a deviation;

[0178] Refer to the following formula to conduct a preliminary evaluation of the protection effect of this section and generate a protection effect evaluation value :

[0179]

[0180] In the formula: is the previously predicted risk factor, The new risk factors actually observed in this section (if obtained through short-term retraining or online analysis); Indicates that within the section Take the average over all time slices or multiple time points; The number of key sensors active within this section;

[0181] If Displays a significant positive value, indicating that the actual risk factor is lower than the predicted value and the protection strategy is effective: if a negative value or too low value appears, it means that the actual risk may be higher than the prediction and subsequent dynamic optimization is required.

[0182] When in use, generate an evaluation value of the protection effect Can measure the deviation between the actual observation and the previous prediction in a timely manner, providing an early signal for dealing with sudden anomalies; the sensors increase the monitoring frequency within the section, which is linked to the implementation effect of the strategy. Once a crack amplification trend appears, an immediate response can be made, and real-time monitoring is in a closed loop: data is continuously accumulated: the execution records and evaluation results are stored synchronously, which can be used as an experience library for the next batch or new material upgrade.

[0183] Step 403, Dynamic Optimization and Adaptive Feedback

[0184] When the evaluation result, the evaluation value of the protection effect Is continuously lower than the effect threshold Or a negative value appears, it is determined that the current protection measure does not meet the expectation: if the predicted risk factor Rises sharply within a short period of time, triggering a temporary optimization process;

[0185] To improve the protection effect, a stronger outer box can be temporarily added in high-risk sections, more seismic pads can be added, or auxiliary means can be strengthened during the loading and unloading process. Define an incremental optimization intensity function :

[0186]

[0187] In the formula: And , And Are optimization sensitivity coefficients Used to capture the situation when the evaluation value of the protection effect Is negative or less than zero. If it is positive, this item is recorded as zero; Represents the amplification of environmental excitation, Used to represent the comprehensive measurement function of environmental excitation, which can be obtained by weighting variables such as temperature, humidity, or the frequency of loading and unloading; The larger the value, the greater the degree of optimization or additional protection resources required.

[0188] After implementing the optimization measures (such as replacing the outer box material with a higher grade, increasing the thickness of the cushioning pad, improving the handling operation, etc.), continue to collect and calculate the new predicted risk factors in real time. , if it is found that the incremental optimization intensity function is significantly improved or the new risk factors gradually tend to be stable, it indicates that the optimization has achieved results;

[0189] Otherwise, continue to perform incremental optimization or switch to other protection schemes, and record the empirical parameters (such as the corresponding protection effects at different time periods ) during the optimization process in the database to provide reference for subsequent logistics route planning and packaging selection.

[0190] When in use, the incremental optimization intensity allows the system to quickly add protection resources or change the scheme when the actual risk emerges, quickly respond to risks beyond expectations; continuous monitoring and cyclic correction can effectively prevent microcracks from evolving from early local small defects to global damage, improve the overall safety margin; incorporate the optimized results into the database, complete R & D iteration and multiple cyclic feedback, and provide more empirical basis for future packaging design or logistics schemes.

[0191] I. Ways to obtain the material fatigue curve

[0192] Standardized fatigue test methods:

[0193] Stress-life (S-N) test: For common packaging materials such as paper packaging materials, plastic films or composite films, the S-N curve can be measured on a constant load testing machine or a repeated bending instrument, that is, record the fatigue cycle times when the material reaches failure (appears obvious cracks or fractures) at different stress / strain levels.

[0194] According to the test results, plot the S-N curve, and obtain the average life or distribution range at different stress levels through interpolation or regression algorithms. Accelerated fatigue test: For some polymer packaging or composite structures (such as multilayer films with strong barrier properties, aluminum-plastic composite materials, etc.), vibration accumulation or repeated heat-sealing opening and closing tests can be carried out in a constant or controllable temperature and humidity environment, so as to observe the fatigue evolution speed of the material at a specific step rate. Combining the real logistics situation, evaluate the effective life of the material with the acceleration effect.

[0195] Material failure mechanism and microscopic detection:

[0196] For the outer packaging of precision instruments or fragile items, the microscopic structural changes of the material during the crack initiation and propagation stages can be further observed through a microscope or a scanning electron microscope (SEM) to provide microscopic mechanical evidence for the fatigue model. In addition, data on interface delamination or fiber fracture after the material is stressed under different temperature and humidity conditions can be collected, and the weakening coefficient of the extreme environment on the life in the subsequent mathematical model or perform calibration;

[0197] Map the data to the fatigue analysis model of this solution: The obtained S-N data or the results of accelerated fatigue tests can be converted into a material fatigue curve function through regression or curve fitting and be referenced in the high-order cumulative shock metric. If the jump variable degree and microcrack index in this solution need to be closer to the actual situation, these experimental data need to be combined to correct the model parameters (including material sensitivity coefficient K, attenuation coefficient n, mapping coefficient s, etc.).

[0198] II. Specific Hardware Implementation

[0199] Selection and Installation of Multi-source Sensors

[0200] Vibration and acceleration sensors: It is recommended to use triaxial accelerometers (such as MEMS sensors) and medium-high frequency vibration pickups (such as piezoelectric ones) to cover a wide vibration frequency band (1Hz to several kHz). For the packaging of electronic precision instruments, the sensitivity and sampling frequency of the sensors can be appropriately increased;

[0201] Environmental sensors: Temperature and humidity sensors should have certain waterproof and anti-interference capabilities: For the pressure sensor, a high-precision MEMS barometer can be selected. If the detection of loading and unloading frequency is involved, a transit counter (such as RFID access control counting, Bluetooth counter, etc.) can be added to the handling appliance or box and uploaded synchronously with the timestamp;

[0202] Installation method: The sensors can be attached to the key stress concentration areas of the packaging lining or placed on the outer surface of the outer box; For high-value products, multiple sensors can also be deployed inside and outside the box to form redundancy and backup for comparison and cross-verification. All sensors are connected to a wireless gateway or a data recording module and have the function of segmented or full-time sampling;

[0203] 2 Data Acquisition and Preliminary Transmission Equipment

[0204] Data recording module: A low-power microcontroller (MCU) or a small single-board computer can be selected, with built-in Flash storage or an SD card for caching when the network is interrupted midway.

[0205] Communication method: If during long-distance transportation, the data needs to be relayed to the cloud platform through 4G / 5G, LoRa or satellite communication; at the warehousing node or port, it can be switched to Wi-Fi or a wired local area network for batch uploading.

[0206] Gateway and Back-end Data Management: It is recommended to use an edge gateway for basic data cleaning, segmentation and packaging, and attach time and geographical tags. If the logistics vehicle is equipped with a GPS module, the location information can be incorporated to further enrich the multi-dimensional analysis of this solution. The back-end data management platform can adopt a cloud database or distributed storage to carry large-scale data from different sensors and different logistics routes, and interface with each algorithm module described in the second, third, and fourth steps of this solution.

[0207] Those of ordinary skill in the art can realize that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are executed in hardware or software depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of this application.

[0208] Those skilled in the art can clearly understand that for the convenience and simplicity of description, the specific working processes of the systems, devices, and units described above can refer to the corresponding processes in the foregoing method embodiments and will not be elaborated here.

[0209] In the several embodiments provided in this application, it should be understood that the disclosed systems, devices, and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units is only for some logical function divisions. In actual implementation, there can be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed mutual coupling, direct coupling, or communication connection can be through some interfaces. The indirect coupling or communication connection of the device or unit can be in electrical, mechanical, or other forms.

[0210] The units described as separate components may or may not be physically separated. The components displayed as units may or may not be physical units, that is, they can be located in one place, or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0211] As described above, this is only the specific implementation manner of this application, but the protection scope of this application is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed in this application and should be covered by the protection scope of this application. Therefore, the protection scope of this application should be subject to the protection scope of the claims.

Claims

1. A packaging control method integrating multivariate dynamic analysis, characterized in that: including, After the initial logistics transportation process is started and the monitoring requirements are confirmed, multi-source sensors are deployed at packaging and transportation nodes to collect environmental parameters, transfer times, and loading and unloading frequencies in real time, so as to obtain multi-dimensional raw data covering the entire process and establish a unified spatio-temporal marker; After the multi-dimensional raw data stream is initially stored, denoising filtering, wavelet decomposition, and environmental coupling are used to perform time-domain alignment and high-order feature extraction on the sensor output, resulting in structured data with a unified coordinate system and covering multi-dimensional indicators; When abnormal mutation signs are detected in the structured data, a micro-crack jump function and a high-order cumulative shock metric model are used to finely identify and quantitatively evaluate the packaging fatigue risk, generating the crack propagation trend and potential failure time window of each sensor; When the analysis results show that the risk indicators of a certain logistics section increase significantly, a segmented protection strategy is immediately implemented and a dynamic optimization algorithm is triggered to differentially strengthen and real-time correct the packaging materials and buffer layout, so as to maintain a high safety margin and closed-loop control the entire process on the premise of minimizing resource input; Perform multi-scale decomposition on the data after wavelet transform denoising and time-domain alignment; To quantify the distribution characteristics of vibration energy at different scales, a characteristic energy spectrum is defined to measure the cumulative shock intensity of the package within a specific frequency band ; The output high-order characteristic data can be formalized as , and the corresponding relationship with the sensor number is retained, where is the sensor number; Couple and analyze the characteristic energy spectrum in combination with environmental sensing data. After obtaining the environmental coupling index, generate a multi-dimensional index vector ; Output the multi-dimensional index vectors generated by all sensors and all environmental nodes to the unified data management structure for final output; ​ Obtain the multi-dimensional index vectors corresponding to each sensor , and all the multi-dimensional index vectors , are sequentially concatenated into a feature trajectory sequence ; Define the symptom function Compare the change rates of multi-dimensional indicators at adjacent or nearby times. When the symptom function exceeds the preset dynamic threshold, it is regarded as a possible symptom of microcracks, and record this moment and the relevant sensor numbers in the short-term risk list; Output preliminary micro-crack sign records and corresponding change amplitude records; Among them, for the moments that have been marked as signs of suspected microcracks , analyze their historical cumulative impact levels and define the local fatigue impact metric : After obtaining the local fatigue impact metric and combining with the fatigue curve or physical mechanism model of the packaging material, the impact metric is converted into a risk factor of failure tendency , and after summarization, a risk factor sequence is output , and it is marked corresponding to the sensor number and time .

2. A packaging control method integrating multi-variable dynamic analysis according to claim 1, characterized in that: Determine the area to be monitored and the key parts where impacts or fatigue may concentrate, form a preliminary candidate coordinate set for sensor deployment, optimize the layout of the candidate coordinate set, and construct a coverage measure and comprehensive effectiveness measure for the deployment of the plan; Select the final deployment plan by maximizing the plan coverage or meeting its threshold conditions, and generate a list of the installation locations and installation methods of the formal sensor network.

3. A packaging control method integrating multi-variable dynamic analysis according to claim 2, characterized in that: Perform basic calibration on accelerometers, gyroscopes, and vibration sensors, and record the calibration parameter set; Deploy environmental sensing devices in the target area to collect and record the parameter set, register the calibrated sensor parameters and the environmental sensor list together, and establish a corresponding relationship.

4. A packaging control method integrating multi-variable dynamic analysis according to claim 3, characterized in that: According to the coordinates and calibration information, start real-time data collection at each transportation or storage stage, collect the time-series data output by each sensor at the set sampling frequency, and each sensor synchronously outputs environmental data; Construct preliminary time markers and geographical markers, and set a specific dynamic weight function to perform preliminary time decay weighting on the sensor data to extract the potential impact of recent vibration shocks on micro-cracks.

5. A packaging control method integrating multi-variable dynamic analysis according to claim 4, characterized in that: After mapping all sensed timing signals to a unified time scale ， Denoise and smooth the timing signals of each sensor to obtain the preliminarily denoised signals; Output the data after denoising and time-domain alignment, and retain all time stamps and sensor numbers in the corresponding metadata.

6. A packaging control method integrating multi-variable dynamic analysis according to claim 5, characterized in that: Integrate risk factors into a time series matrix and compare or perform regression correction with the actual material failure cases to obtain the overall packaging failure probability function; Infer the potential breakage probability within a given time range based on the accumulation of risk factors ; ​ For multi-dimensional index vectors in which risk factors can be triggered and the potential breakage probability Sort the increasing indicators to identify the vibration frequency bands, temperature and humidity levels, or handling operation links that require the most attention; Output the recognition result together with the predicted value of the potential damage probability within a given time range of the potential damage probability and the list of key influencing factors.

7. A packaging control method integrating multi-variable dynamic analysis according to claim 6, characterized in that: Divide the overall logistics process into several sections and assign section identifiers to each section; for the moment map it to the corresponding logistics section to form a section-time correspondence table; for each section statistically analyze or predict the time distribution of the packaging damage risk within it, and based on the risk factor or the potential damage probability , calculate the overall risk index of this section as a measure of the risk level of each section.

8. A packaging control method integrating multivariate dynamic analysis according to claim 7, characterized in that: Based on sections Overall risk index , combined with key influencing factors, formulate the corresponding protection plan for each section ​ Including: Appropriately increasing cushioning materials or strengthening the impact-resistant structure of the outer box in sections where the overall risk index is higher than expected; Allocating more manpower or automated equipment at nodes with frequent loading and unloading to reduce impact; Adding moisture-proof or thermal insulation layers in cold chain or high-humidity sections; Before the start of each section, increase the monitoring frequency of sensors and environmental monitoring equipment and focus on high-risk sensor numbers . Compare the new real-time data collected during this period with the risk factors or the potential probability of damage .

9. A packaging control method integrating multivariate dynamic analysis according to claim 8, characterized in that: Generate a protection effect evaluation value to preliminarily evaluate the protection effect of this section; when the evaluation result, that is, the protection effect evaluation value, is continuously lower than the effect threshold or a negative value appears, it is determined that the current protection measure fails to meet the expectation: if it is monitored that the predicted risk factor rises sharply in a short period of time, trigger a temporary optimization process; Temporarily add stronger outer boxes, add more anti-seismic pads in high-risk sections or strengthen auxiliary means in the loading and unloading links, and define an incremental optimization intensity function; after the optimization measures are executed, continue to collect and calculate the new predicted risk factors in real time. If it is not found that the incremental optimization intensity function is significantly improved after optimization or the new risk factors gradually tend to be stable, continue to perform incremental optimization or switch to other protection schemes.

Images