Package management method and system combined with image processing technology

Abstract

The invention relates to the technical field of circulating packaging, in particular to a packaging management method and system combined with an image processing technology. The invention discloses a package management method combined with an image processing technology. The method comprises the following steps: S1, collecting damage condition data and turnover times of each module in a cycle package in a use period; s2, acquiring circulating packaging information, processing the damage condition data based on the circulating packaging information, and generating a damage state sequence of each circulating packaging module; s3, based on the damage state sequence of each module, obtaining a residual turnover frequency prediction result of each module through a residual turnover frequency prediction model for circularly packaging each module; and S4, generating a combination scheme according to the prediction result of the remaining turnover times and the combination of modules of the package spare part library optimization cycle package. According to the invention, damage intelligent identification, residual turnover frequency prediction and module combination optimization of circulating packaging can be realized, so that intelligent and high-efficiency circulating packaging management under a supply chain closed loop is realized.

Description

Technical Field

[0001] This invention relates to the field of recyclable packaging technology, and in particular to a packaging management method and system that combines image processing technology. Background Technology

[0002] In current closed-loop supply chain reusable packaging management practices, a common problem is performance degradation caused by the long-term repeated use of reusable packaging modules. Specifically, packaging boxes and pallet modules exhibit surface wear, structural deformation, or color fading after multiple uses. This not only affects the appearance integrity of the packaging but, more importantly, reduces its protective capabilities and lifespan. Traditional management methods rely primarily on manual experience or fixed-cycle replacement strategies. When a reusable packaging module is damaged, the entire reusable packaging is scrapped, making it difficult to accurately assess the actual damage status of each module. This results in many usable modules being prematurely scrapped, while some modules that have reached their critical state continue to circulate without being detected in time, posing a risk of further damage. Existing technologies for reusable packaging management often employ lifespan prediction methods based on fixed turnover counts, such as recording the number of times packaging is used via RFID or QR codes, and replacing it when a preset threshold is reached. Therefore, there is an urgent need for a packaging management method and system that combines image processing technology to achieve intelligent damage identification, prediction of remaining turnover counts, and optimization of module combinations for reusable packaging, thereby realizing intelligent and efficient reusable packaging management within a closed-loop supply chain. Summary of the Invention

[0003] To overcome the drawback of low utilization efficiency of reusable packaging, this invention provides a packaging management method and system that combines image processing technology.

[0004] The technical solution of this invention is: a packaging management method combining image processing technology, comprising the following steps: S1: Collect damage data and turnover number of each module in the cyclic packaging during its service life; S2: Obtain the cyclic packaging information, process the damage status data based on the cyclic packaging information, and generate a damage status sequence of each module of the cyclic packaging; S3: Based on the damage state sequence of each module, the remaining turnover prediction result of each module is obtained by using the remaining turnover prediction model of each module in a cyclic packaging. S4: Based on the predicted remaining turnover times and the optimized combination of each module of the cyclic packaging in the packaging parts library, generate a combination scheme; S5: Recombine the modules according to the combination scheme to generate a new cyclic package.

[0005] Preferably, the step of collecting damage data and turnover count of each module in the reusable packaging during its service life includes: collecting damage data of each module in each reusable packaging through sensors deployed in the environment, wherein the damage data of each module includes surface texture map, structural deformation map and color degradation map; obtaining the unique identifier and turnover count of each reusable packaging through an identification reading device; and establishing a unique module number for each module of the reusable packaging based on the unique identifier of the reusable packaging.

[0006] Preferably, the step of acquiring reusable packaging information and processing the damage status data based on the reusable packaging information to generate a damage status sequence for each module of the reusable packaging includes: acquiring reusable packaging information from a database, wherein the reusable packaging information includes the module composition requirements of the reusable packaging, the material coefficients of each module, the maximum number of turnovers of each module, the initial edge contours of each module, and the base color of each module; performing multi-scale texture feature analysis on the surface texture map using local binary mode variance analysis and gradient magnitude statistics techniques; and obtaining texture roughness by constructing a gray-level co-occurrence matrix and calculating the contrast features and the dispersion of the local gradient distribution of the gray-level co-occurrence matrix. The geometric deformation analysis of the structural deformation map is performed using contour shape context matching and dynamic time warping techniques. The edge deformation degree is obtained by calculating the Hausdorff distance and curvature change correlation between the actual edge contour and the initial edge contour. The aging color difference quantification of the color degradation map is performed using color space transformation and spectral reflectance analysis techniques. The color offset degree is obtained by calculating the Euclidean distance and hue saturation change between the current color feature and the reference color in the CIELAB uniform color space. The loss state sequence is generated by standardizing and combining the module unique number, texture roughness, edge deformation degree and color offset.

[0007] Preferably, the step of obtaining the remaining turnover prediction result of each module based on the damage state sequence of each module through the remaining turnover prediction model of the cyclic packaging of each module includes: normalizing the texture roughness, edge deformation degree, color offset degree, turnover number, material coefficient of each module and maximum turnover number of each module, calculating the remaining turnover prediction result of each module through the remaining turnover prediction model, sorting the modules according to the remaining turnover prediction result and the module's unique number, and generating a module remaining turnover ranking table.

[0008] Preferably, the step of calculating the remaining turnover prediction results for each module using the remaining turnover prediction model includes: the formula for the remaining turnover prediction model is: ; In the formula, For the first The remaining number of turnovers for a class module. For the first Material coefficient of class module, For the first Maximum number of times a class module can be turned around. For the first The number of times a class module is turned over. For the first Texture roughness of class modules, For the first Edge deformation of class modules For the first Color offset of the class module.

[0009] Preferably, the step of generating a combination scheme by optimizing the combination of various modules of the cyclic packaging based on the prediction result of the remaining turnover times and the packaging parts library includes: based on the module remaining turnover times sorting table, marking modules with remaining turnover times lower than a preset threshold as scrap and recyclable, counting the number of each type of module marked as scrap and recyclable, calling the corresponding number of new modules from the packaging parts library, performing k-means clustering on modules with remaining turnover times greater than or equal to the preset threshold and the new modules to obtain the number of clusters, counting the number of each type of module in each cluster, calculating the number of complete packages within each cluster and the number of remaining modules of each type according to the module composition requirements of a single cyclic packaging, calculating the number of complete packages between clusters based on the number of remaining modules of each type, and using a genetic algorithm to solve the combination scheme of each type of module based on the number of complete packages within each cluster and the number of complete packages between clusters.

[0010] Preferably, the step of using a genetic algorithm to solve the combination scheme of various modules based on the number of complete packages within the group and the number of complete packages between the groups includes: summing the number of complete packages within the group and the number of complete packages between the groups to obtain the total number of packages; adopting an encoding strategy with cyclic package combination schemes as the basic unit, where a chromosome represents a complete module allocation scheme, a chromosome is composed of gene segments connected by the total number of packages, each gene segment corresponds to a package to be constructed, each gene segment contains multiple genes, each gene represents a unique number of a module allocated to a package, an initial population is randomly generated, a tournament selection method is used to randomly select chromosomes, fitness is calculated through a fitness function, the chromosome with the highest fitness is selected to enter the next generation, crossover and mutation operations are performed to update the chromosomes, and when the maximum number of iterations is reached, a combination scheme containing the unique number of the module is output.

[0011] Preferably, the random selection of chromosomes, the calculation of fitness using a fitness function, and the selection of the chromosome with the highest fitness to enter the next generation include: randomly selecting chromosomes, calculating the average turnover number and the square of the standard deviation of the turnover number for the combination schemes represented by the chromosomes; if the package to be constructed corresponding to the gene segment meets the module composition requirements of the single cyclic package, the integrity factor is 1; otherwise, the integrity factor is 0; and calculating the fitness based on the average turnover number, the square of the standard deviation of the turnover number, and the integrity factor using a fitness formula, wherein the fitness formula is: ; In the formula, For fitness, As an integrity factor, This represents the average number of turnovers. This is the square of the standard deviation of the number of turnovers.

[0012] Preferably, the step of recombining the modules according to the combination scheme to generate a new cyclic package includes: obtaining a unique module number according to the combination scheme, combining the modules into a new cyclic package using the unique module identifier, and assigning a unique identifier to the new cyclic package.

[0013] A packaging management system incorporating image processing technology includes: Data acquisition module: Collects damage data and turnover number of each module in the cyclic packaging during its service life; Sequence generation module: acquires cyclic packaging information, processes the damage status data based on the cyclic packaging information, and generates a damage status sequence for each module of the cyclic packaging; Prediction result generation module: Based on the damage state sequence of each module, the remaining turnover prediction result of each module is obtained by cyclically packaging the remaining turnover prediction model of each module; Combination scheme generation module: Based on the predicted remaining turnover times and the optimized combination of various modules of the packaging parts library, a combination scheme is generated; Packaging assembly module: Reassemble the modules according to the assembly scheme to generate new reusable packaging. Beneficial effects

[0014] This invention integrates multi-source damage data and module attribute data, and uses an established prediction model for remaining turnover times to perform comprehensive calculations. This overcomes the limitations of existing technologies that rely solely on turnover times. It can reflect the individualized damage status and remaining lifespan of each module in actual use, reducing the two extreme cases of "over-scrapping" or "using with defects". It maximizes the utilization potential of each module while ensuring packaging safety. This invention utilizes image processing technology to perform automated and quantitative analysis of damage status data, generating standardized damage status sequences. This replaces the traditional subjective judgment that relies on human experience, reduces human error, and transforms management decisions from experience-driven to data-driven, thereby improving the objectivity and efficiency of management. This invention, based on module sorting and clustering analysis of remaining turnover counts, and combined with a genetic algorithm for global optimization, can intelligently generate optimal combination schemes for module reuse and new module supplementation. This method can group modules with similar remaining turnover counts together to form packaging units with synchronized turnover cycles, reducing the frequency of package disassembly due to premature scrapping of some modules, while also reducing the quantity and frequency of new module procurement. This effectively lowers the overall operating costs of packaging and spare parts inventory pressure. Attached Figure Description

[0015] Figure 1 This is a flowchart of the packaging management method using the combined image processing technology of the present invention; Figure 2 This is a structural diagram of the packaging management system based on the combined image processing technology of the present invention. Detailed Implementation

[0016] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0017] Example 1: A packaging management method combining image processing technology, such as... Figure 1 As shown, it includes the following steps: S1: Collect damage data and turnover number of each module in the cyclic packaging during its service life; Damage data of each module in each reusable package is collected by sensors deployed in the environment. The damage data of each module includes surface texture map, structural deformation map and color degradation map. The unique identifier and turnover number of each reusable package are obtained by an identification reading device. Based on the unique identifier of the reusable package, a unique module number is established for each module of the reusable package.

[0018] It should be further explained that hardware is deployed at key operational nodes of the closed-loop supply chain network, with these nodes extensively covering warehouse storage areas and cargo loading and unloading platforms. Multimodal sensor networks are deployed here.

[0019] The data acquisition units constituting this sensor network mainly include three types of specialized equipment: a high-resolution industrial camera for optical imaging, a 3D structured light scanner for geometric shape recording, and a high-precision colorimeter for color characteristic analysis. The industrial camera operates under strictly controlled lighting conditions, acquiring surface texture maps that clearly reveal microstructural changes on the material surface, including fine scratches, wear patches, and corrosion marks. The structured light scanner constructs a 3D point cloud model representing the object's shape by emitting specific gratings and analyzing their deformation patterns, thereby accurately identifying geometric anomalies such as bending deformation, local dents, or edge warping in the packaging module. The colorimeter uses spectral analysis technology to convert the material's apparent color into standardized chromaticity values, generating a color degradation map that objectively reflects color decay caused by environmental exposure.

[0020] Simultaneously, the process of acquiring identification information is carried out. Key management information is extracted from electronic tags or printed codes attached to the packaging using UHF RFID readers or QR code identification devices. The unique identifier serves as the identity credential for each package, typically encoded with a unique numerical sequence, establishing a permanent correspondence with a specific packaging instance. The turnover rate serves as a dynamically updated indicator of usage intensity, essentially recording the cumulative number of complete logistics cycles completed by the reusable packaging unit and its modules. Each "turnover" represents a complete lifecycle loop from outbound, put into use, transported, returned for inspection, to final return for standby. This parameter quantifies the historical workload of the reusable packaging unit, and its counting logic follows the complete logistics cycle: after the packaging unit undergoes the complete process of outbound, transport, and return, an automatic accumulation operation is performed.

[0021] Based on the received unique identifier of the reusable packaging, a unique identification code is generated for each independent component of the packaging by querying a pre-defined module relationship mapping table. This mapping table essentially establishes the correspondence between the overall packaging and its constituent parts. For example, when a package with the code CP20240001 is identified, its constituent modules are assigned hierarchical identification numbers according to the mapping relationship. For instance, the box unit corresponds to CP20240001-A01, the cover unit to CP20240001-B01, and the base unit to CP20240001-C01. This hierarchical identification system ensures that each module has an independent traceable identity throughout its circulation cycle, establishing the necessary technical foundation for subsequent damage assessment and lifespan prediction based on individual modules.

[0022] S2: Obtain the cyclic packaging information, process the damage status data based on the cyclic packaging information, and generate a damage status sequence of each module of the cyclic packaging; Retrieve recyclable packaging information from the database. This information includes the module composition requirements, material coefficients of each module, maximum turnover times of each module, initial edge contours of each module, and reference colors of each module. Perform multi-scale texture feature analysis on the surface texture map using local binary mode variance analysis and gradient magnitude statistics. Obtain texture roughness by constructing a gray-level co-occurrence matrix and calculating its contrast characteristics and the discreteness of local gradient distribution. Perform geometric deformation analysis on the structural deformation map using contour shape context matching and dynamic time warping techniques. Obtain edge deformation degree by calculating the Hausdorff distance and curvature change correlation between the actual edge contour and the initial edge contour. Perform aging color difference quantification on the color degradation map using colorimetric space transformation and spectral reflectance analysis techniques. Obtain color shift degree by calculating the Euclidean distance and hue saturation change between the current color feature and the reference color in the CIELAB uniform color space. Standardize and combine the module unique number, texture roughness, edge deformation degree, and color shift degree to generate a loss state sequence.

[0023] It should be further explained that pre-stored cyclic packaging information is retrieved from the database. This information constitutes the baseline parameters for subsequent analysis. Specifically, it includes the required module composition type and quantity for each cyclic package, the material property coefficients corresponding to each module, the theoretically maximum allowable number of turnovers threshold for each module, the initial edge contour geometric data of each module in its intact state, and the standard color reference value calibrated at the factory for each module. In the surface texture analysis stage, a multi-scale texture feature analysis is performed on the acquired surface texture map using a collaborative approach of Local Binary Mode ANOVA and gradient magnitude statistics. Local Binary Mode ANOVA focuses on characterizing the local pattern changes and stability of the texture image, while gradient magnitude statistics quantifies the severity of intensity changes in the image. Furthermore, by constructing a gray-level co-occurrence matrix and accurately calculating its contrast feature parameters, while simultaneously statistically analyzing the discrete variation coefficient of the local gradient distribution, these two dimensions of indicators are fused and weighted to finally output a comprehensive texture roughness value, which objectively reflects the microscopic wear condition of the material surface. In the geometric deformation assessment phase, contour shape context matching combined with dynamic time warping is used to perform detailed geometric deformation analysis on the acquired edge deformation map. Contour shape context matching establishes the correspondence between the actual contour and the initial contour by describing the spatial distribution characteristics of contour points, while dynamic time warping is used to align and compare these two contour sequences that differ in sampling density and local morphology. Based on these matching results, the Hausdorff distance between the actual edge contour and the initial edge contour is calculated to capture the maximum deformation deviation, and the correlation of the curvature changes of the two contour lines is analyzed to assess the consistency of the deformation pattern. In the color aging quantification process, colorimetric space transformation and spectral reflectance analysis techniques are applied to analyze the color degradation map. First, the device-related RGB color data is converted to the perceptually uniform CIELAB color space, while examining the changes in the spectral reflectance characteristics of the material surface. The overall color difference amplitude is assessed by calculating the Euclidean distance between the current color feature and the reference color in CIELAB space, and the degree of change in hue angle and saturation component is quantified separately. These color difference indicators are weighted and integrated to generate a color shift value that comprehensively characterizes the degree of color aging.

[0024] Each module's unique identifier is standardized and integrated with its three corresponding feature metrics—texture roughness, edge deformation, and color offset. Standardization ensures all feature values ​​are within a uniform dimension and numerical range. These standardized features are then combined with the module identifier according to a predefined data structure to form a standardized damage state sequence. This sequence characterizes the overall damage status of each module at the end of its current service life, providing a precise data foundation for subsequent remaining lifetime prediction and optimal combination decisions.

[0025] S3: Based on the damage state sequence of each module, the remaining turnover prediction result of each module is obtained by using the remaining turnover prediction model of each module in a cyclic packaging. After normalizing the texture roughness, edge deformation, color offset, turnover count, material coefficient of each module, and maximum turnover count of each module, the remaining turnover count prediction result of each module is calculated by the remaining turnover count prediction model. The modules are sorted according to the remaining turnover count prediction result and the module's unique number to generate a module remaining turnover count ranking table.

[0026] The formula for the remaining turnover prediction model is: ; In the formula, For the first The remaining number of turnovers for a class module. For the first Material coefficient of class module, For the first Maximum number of times a class module can be turned around. For the first The number of times a class module is turned over. For the first Texture roughness of class modules, For the first Edge deformation of class modules For the first Color offset of the class module.

[0027] It should be further explained that before entering the prediction model, multiple input parameters are first preprocessed using normalization. These parameters include texture roughness, edge deformation, and color shift obtained from the damage state sequence, as well as the current turnover count, material coefficients of each module, and the maximum turnover count of each module. The normalization process transforms parameters with different dimensions and numerical ranges into a standard range of zero to one, ensuring the stability of the prediction model and the accuracy of the calculation results. After data preprocessing, the remaining turnover count prediction model for each module is called to perform calculations, quantifying each module to obtain the remaining turnover count prediction results. Subsequently, based on the remaining turnover count prediction results and the corresponding unique module number, they are arranged in descending order. This sorting method can intuitively show the distribution of the remaining value of all modules, and the resulting module remaining turnover count sorting table reveals the health status level of each module, providing direct data support for subsequent optimization and combination decisions.

[0028] S4: Based on the predicted remaining turnover times and the optimized combination of each module of the cyclic packaging in the packaging parts library, generate a combination scheme; Based on the module remaining turnover sorting table, modules with remaining turnover counts below a preset threshold are marked as scrapped and recycled. The number of each type of module marked as scrapped and recycled is counted. The corresponding number of new modules of the same type are called from the packaging parts library. Modules with remaining turnover counts greater than or equal to the preset threshold and the new modules are subjected to k-means clustering to obtain the number of clusters. The number of each type of module in each cluster is counted. According to the module composition requirements of a single cyclic package, the number of complete packages within each cluster and the number of remaining modules of each type can be calculated. The number of complete packages between groups is calculated based on the number of remaining modules of each type. Based on the number of complete packages within each group and the number of complete packages between groups, a genetic algorithm is used to solve the combination scheme of each type of module.

[0029] The total number of packages is obtained by summing the number of complete packages within the group and the number of complete packages between the groups. An encoding strategy based on cyclic package combination schemes is adopted. A chromosome represents a complete module allocation scheme. A chromosome is composed of gene segments connected by the total number of packages. Each gene segment corresponds to a package to be constructed. Each gene segment contains multiple genes. Each gene represents a unique number of a module allocated to a package. An initial population is randomly generated. A tournament selection method is used to randomly select chromosomes. Fitness is calculated by a fitness function. The chromosome with the highest fitness is selected to enter the next generation. Crossover and mutation operations are performed to update the chromosomes. When the maximum number of iterations is reached, the combination scheme containing the unique number of the module is output.

[0030] Chromosomes are randomly selected, and the average turnover rate and the square of the standard deviation of the turnover rate for the combination schemes represented by the chromosomes are calculated. If the package to be constructed corresponding to the gene segment meets the module composition requirements of the single cyclic package, the integrity factor is 1; otherwise, the integrity factor is 0. The fitness is calculated based on the average turnover rate, the square of the standard deviation of the turnover rate, and the integrity factor using the fitness formula, where the fitness formula is: ; In the formula, For fitness, As an integrity factor, This represents the average number of turnovers. This is the square of the standard deviation of the number of turnovers.

[0031] It should be further explained that, based on the module remaining turnover sorting table, modules with remaining turnover counts below a preset threshold are uniformly marked as awaiting scrapping and recycling. The preset threshold is set based on the experience of personnel in this field and engineering practices. Subsequently, the detailed types and corresponding quantities of these marked modules are statistically analyzed, and a request is automatically initiated to the packaging parts library based on the statistical results to obtain new modules of the corresponding type and quantity to replenish the inventory. After the module update is completed, the in-use modules with remaining turnover counts greater than or equal to the preset threshold and the newly added modules are included in the combination optimization process. The k-means clustering algorithm is used to group these modules, and the optimal number of clusters is automatically determined based on the remaining turnover count characteristics of the modules. After clustering, the specific quantity of each type of module in each cluster is statistically analyzed in detail to provide a data foundation for subsequent packaging combination. According to the module composition requirements of a single cyclic package, the packaging composition capacity at two levels is calculated separately. First, within each cluster, the number of complete packages that can be formed within the group is calculated based on the number of each type of module in each cluster, and the number of remaining modules of each type that cannot be paired within each group is recorded. Then, these remaining modules are integrated across groups to calculate the number of complete packages that can be formed between groups. Finally, the number of complete packages within a group and the number of complete packages between groups are added together to obtain the theoretical total number of packages.

[0032] Based on the completed packaging quantity statistics, the number of complete packages within a group and the number of complete packages between groups are arithmetically summed to determine the theoretical total number of packages that can be formed by each type of module. Based on the determined total number of packages, a genetic algorithm model is constructed using an encoding strategy with complete cyclic packaging combination schemes as the basic unit. In this model, each chromosome completely represents a specific module allocation scheme. The structure of a chromosome consists of a series of consecutive gene segments of the total number of packages, where each independent gene segment corresponds to a complete packaging unit to be assembled. Within each gene segment, there are multiple ordered genes, each gene precisely corresponding to a specific module allocated to that package, and using the module's unique number as its gene value. The initialization phase creates an initial population containing multiple chromosomes through a random generation mechanism. The specific method for constructing each chromosome is as follows: all current modules are randomly shuffled, and then, according to the required number of gene segments, the unique numbers of these modules are sequentially filled into the gene segments corresponding to each packaging unit until all modules are allocated, forming the initial module allocation scheme.

[0033] During population evolution, a tournament selection mechanism is used to screen for high-quality chromosomes. Specifically, a specified number of chromosomes are randomly selected from the current population to form a competition group. The fitness value of each chromosome within each group is precisely calculated using a fitness function, and the chromosome with the highest fitness is selected to enter the next generation. After selection, crossover and mutation operations are performed on the selected chromosome population. Crossover involves exchanging gene segments from different chromosomes to recombine solutions, while mutation introduces new solutions by randomly changing the unique module number of specific genes. These genetic operations work together to continuously update and optimize the chromosome population.

[0034] When the evolutionary process reaches the preset maximum number of iterations, it terminates and outputs the optimal solution. This optimal solution is a complete combination scheme containing unique numbers for all modules. This scheme ensures the integrity of the packaging composition and achieves the optimal configuration of the remaining lifetime of the modules, providing precise guidance for subsequent physical assembly operations.

[0035] The fitness function calculation process comprehensively considers both the quality and completeness of the combination scheme. First, the average turnover rate of all modules in the combination scheme represented by the chromosome is calculated, along with the squared standard deviation of these turnover rates. For each gene segment in the chromosome corresponding to the package to be constructed, it is verified whether the module composition meets the requirements of a single cyclic package. If it fully meets the requirements, the completeness factor corresponding to that package is set to 1; otherwise, it is set to 0. Based on the calculated average turnover rate, the squared standard deviation of the turnover rate, and the completeness factor, the overall fitness value of the chromosome is calculated using the fitness formula.

[0036] S5: Recombine the modules according to the combination scheme to generate a new cyclic package.

[0037] The module's unique number is obtained according to the combination scheme. The modules are combined into a new circular package using the module's unique identifier, and a unique identifier is assigned to the new circular package.

[0038] It should be further explained that the unique serial number of each module listed in the assembly scheme is parsed and obtained, and these serial numbers serve as the key basis for module identification. In the packaging and assembly area, the corresponding physical module is accurately located and extracted from the module temporary storage area by scanning or entering the module's unique identifier, as instructed.

[0039] The next step is module assembly. Strictly following the provided assembly plan, modules with different unique numbers are physically assembled in a prescribed structural order. For example, a specific numbered housing module is mechanically connected and secured to its corresponding numbered cover and base modules to ensure a stable overall structure. Throughout this process, the unique identifier of each module is verified in real time to ensure the assembly operation matches the predetermined plan.

[0040] After the physical assembly of the modules is completed, new unique identifiers need to be assigned to these newly formed reusable packages. This assignment process is implemented through identifier management, which allocates an unused unique code from the identifier pool and establishes a permanent association between this code and the unique numbers of all modules constituting the new package. This newly assigned unique identifier will become the identity credential for the reusable package in subsequent circulation cycles, and its associated information will be updated in the packaging master record in the database in real time.

[0041] The reusable packaging, reorganized and given new identifiers, undergoes quality inspection and is then reintegrated into the closed-loop supply chain. Throughout the process, the unique identifier of the new packaging and its constituent modules are fully recorded, and the current status information of all modules involved in the reorganization is updated to ensure data consistency in the entire packaging management process.

[0042] Example 2: Based on Example 1, a packaging management system combining image processing technology, such as... Figure 2 As shown, it includes: Data acquisition module: Collects damage data and turnover number of each module in the cyclic packaging during its service life; Sequence generation module: acquires cyclic packaging information, processes the damage status data based on the cyclic packaging information, and generates a damage status sequence for each module of the cyclic packaging; Prediction result generation module: Based on the damage state sequence of each module, the remaining turnover prediction result of each module is obtained by cyclically packaging the remaining turnover prediction model of each module; Combination scheme generation module: Based on the predicted remaining turnover times and the optimized combination of various modules of the packaging parts library, a combination scheme is generated; Packaging assembly module: Reassemble the modules according to the assembly scheme to generate new reusable packaging.

[0043] The embodiments of the present invention have been described in detail above with reference to the accompanying drawings. However, the present invention is not limited to the above embodiments. Within the scope of knowledge possessed by those skilled in the art, various changes can be made without departing from the spirit of the present invention.

Claims

1. A package management method of a joint image processing technique, characterized by, The method comprises the following steps: S1: collecting damage condition data and turnover times of each module of the circulating packaging during a use cycle; S2: obtaining circulating packaging information, processing the damage condition data based on the circulating packaging information, and generating a damage state sequence of each module of the circulating packaging; S3: based on the damage state sequence of each module, obtaining a residual turnover time prediction result of each module through a residual turnover time prediction model of each module of the circulating packaging; S4: optimizing the combination of each module of the circulating packaging according to the residual turnover time prediction result and the packaging equipment library, and generating a combination scheme; S5: recombining each module according to the combination scheme to generate new circulating packaging.

2. The packaging management method of claim 1, wherein, The collection of damage condition data and turnover times of each module of the circulating packaging during a use cycle comprises: collecting damage condition data of each module in each circulating packaging through sensors deployed in the environment, the damage condition data of each module comprising a surface texture graph, a structural deformation graph, and a color degradation graph; obtaining a unique identifier and turnover times of each circulating packaging through an identifier reading device; and establishing a module unique number for each module of the circulating packaging based on the unique identifier of the circulating packaging.

3. The packaging management method of claim 2, wherein, The obtaining of the circulating packaging information and the processing of the damage condition data based on the circulating packaging information to generate a damage state sequence of each module of the circulating packaging comprises: obtaining circulating packaging information from a database, the circulating packaging information comprising module composition requirements of the circulating packaging, module material coefficients, maximum turnover times of each module, initial edge profiles of each module, and reference colors of each module; performing multi-scale texture feature analysis on the surface texture graph using local binary pattern variance analysis and gradient amplitude statistical techniques; obtaining texture roughness by constructing a gray level co-occurrence matrix and calculating contrast features of the gray level co-occurrence matrix and a discrete degree of local gradient distribution; performing geometric deformation analysis on the structural deformation graph using contour shape context matching and dynamic time warping techniques; obtaining edge deformation degree by calculating Hausdorff distance between an actual edge profile and an initial edge profile and curvature change correlation; performing aging color difference quantification on the color degradation graph using colorimetric space conversion and spectral reflectance analysis techniques; obtaining color shift degree by calculating Euclidean distance and hue saturation change of current color features and reference colors in a CIELAB uniform color space; and generating a loss state sequence by standardizing and combining the module unique number, texture roughness, edge deformation degree, and color shift degree.

4. The packaging management method of claim 3, wherein, The obtaining of a residual turnover time prediction result of each module based on the damage state sequence of each module through a residual turnover time prediction model of each module of the circulating packaging comprises: normalizing the texture roughness, edge deformation degree, color shift degree, turnover times, module material coefficients, and maximum turnover times of each module, and then calculating the residual turnover time prediction result of each module through the residual turnover time prediction model; and generating a module residual turnover time ranking table by sorting the modules according to the residual turnover time prediction result and the module unique number.

5. The packaging management method of claim 4, wherein, The remaining turnover times prediction result of each module is calculated by the remaining turnover times prediction model, including that a remaining turnover times prediction model formula is: ； In the formula, is the number of remaining turns of the first class module, is the material coefficient of the first class module, is the maximum number of turns of the first class module, is the number of turns of the first class module, is the texture roughness of the first class module, is the edge deformation degree of the first class module, is the color offset degree of the first class module.

6. The packaging management method of claim 4, wherein, The combination scheme is generated by optimizing the combination of each module based on the remaining turnover times prediction result and the packaging equipment library, including that based on the module remaining turnover times sorting table, modules with a remaining turnover time lower than a preset threshold are marked as scrap recycling, the number of each type of module marked as scrap recycling is counted, a corresponding number of new modules of each type are called from the packaging equipment library, modules with a remaining turnover time greater than or equal to the preset threshold and the new modules are subjected to k-means mean clustering to obtain a cluster group number, the number of each type of module in each cluster group is counted, the number of complete packages that can be formed in each cluster group is calculated according to the module composition requirement of a single cycle package, the number of complete packages between cluster groups is calculated according to the number of each type of remaining module, and the combination scheme of each type of module is solved based on the number of complete packages in the cluster group and the number of complete packages between cluster groups using a genetic algorithm.

7. The packaging management method of claim 6, wherein, The combination scheme of each type of module is solved based on the number of complete packages in the cluster group and the number of complete packages between cluster groups using a genetic algorithm, including that the number of complete packages in the cluster group and the number of complete packages between cluster groups are summed to obtain a total number of packages, an encoding strategy taking a cycle package combination scheme as a basic unit is adopted, one chromosome represents one complete module distribution scheme, the chromosome is connected by a total number of package genes, each gene corresponds to one to-be-formed package, each gene contains a plurality of genes, and each gene represents a unique number of a module allocated to the package, an initial population is randomly generated, a tournament selection method is adopted, chromosomes are randomly selected, the fitness is calculated through a fitness function, the chromosome with the highest fitness is selected to enter the next generation, the chromosome is updated through crossover and mutation operations, and the combination scheme containing the unique number of the module is output when the maximum number of iterations is reached.

8. The packaging management method of claim 7, wherein, The chromosomes are randomly selected, the fitness is calculated through the fitness function, and the chromosome with the highest fitness is selected to enter the next generation, including that the chromosomes are randomly selected, the average turnover times and the square of the turnover times standard deviation of the combination scheme represented by the chromosome are calculated, if the to-be-formed package corresponding to the gene segment meets the module composition requirement of the single cycle package, the integrity factor is 1, if not, the integrity factor is 0, and the fitness is calculated based on the average turnover times, the square of the turnover times standard deviation, and the integrity factor through a fitness formula, wherein the fitness formula is: ； wherein is the fitness, is the integrity factor, is the average number of turns, is the square of the standard deviation of the number of turns.

9. The packaging management method of claim 7, wherein, The modules are recombined according to the combination scheme to generate a new cycle package, including that the unique number of the module is obtained according to the combination scheme, the modules are combined into a new cycle package through the unique identification of the module, and the new cycle package is assigned a unique identification.

10. A package management system of a joint image processing technology for implementing the package management method of any one of claims 1 to 9, characterized by It includes: A data acquisition module acquires damage condition data and turnover times of each module of a cycle package in a use cycle; A sequence generation module obtains cycle package information, processes the damage condition data based on the cycle package information, and generates a damage state sequence of each module of the cycle package. The prediction result generation module: based on the damage state sequence of each module, the remaining turnover number prediction result of each module is obtained by recycling and packaging the remaining turnover number prediction model of each module; The combination scheme generation module: according to the remaining turnover number prediction result and the packaging equipment library optimization, the combination of recycling and packaging each module is generated, and the combination scheme is generated; The packaging combination module: according to the combination scheme, each module is recombined to generate a new recycling and packaging.

Images