A customized packaging design and management system for home appliances

Abstract

The present invention relates to the field of data processing technology, and in particular to a customized packaging design and management system for household appliances, comprising a dynamic feature genetic encoding module, a cross-domain parameter coupling optimization engine, a two-level genetic optimization architecture, a dynamic case feature reconstruction module, a co-evolutionary verification interface, and an adaptive processing instruction generator. The geometric features of household appliances are converted into a hybrid-encoded chromosome data stream; the cross-domain parameter coupling optimization engine integrates the logistics vibration spectrum and the material impact resistance parameters, constructs a dynamic constraint matrix, and generates a multi-objective non-dominated solution set; the co-evolutionary verification interface corrects the optimization weight coefficients through physical simulation feedback, forming a closed-loop verification link. The present invention solves the problems of static template adaptation lag, insufficient multi-constraint coupling, and missing historical case feedback, thereby improving packaging design efficiency and structural reliability.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a customized packaging design and management system for household appliances. Background Art

[0002] The customized packaging design and management system for home appliances integrates user demand data, product 3D model parameters, and logistics environment constraints to build an intelligent design architecture based on parametric templates. The system employs a modular design approach, analyzing product appearance characteristics through geometric topology optimization algorithms. Combining material mechanics simulation with drop impact dynamics models, it generates a minimized packaging solution that meets the structural strength requirements of specific transportation scenarios. A lightweight iteration engine is embedded in the design process, using finite element analysis to verify the packaging's pressure-bearing performance. A historical case database is also used for multi-objective optimization, achieving a dynamic balance between material cost and protective effectiveness. The system features a built-in collaborative design interface that supports multi-terminal data synchronization and version control. It also integrates a virtual reality visualization module to assist designers in human-computer interaction verification, ultimately outputting a CNC machining instruction set suitable for automated production lines, thereby improving packaging development efficiency and reducing resource consumption.

[0003] Existing home appliance packaging management systems suffer from technical pain points: the parametric template update mechanism relies on a static database architecture and lacks real-time adjustment capabilities based on dynamic identification of product features, resulting in delayed adaptation of templates to the geometric features of new home appliances. The system is unable to autonomously analyze non-standard surface parameters through topology optimization algorithms, requiring manual intervention in template sizing and structural layout, increasing repetitive design costs. Insufficient dynamic coupling analysis of logistics constraints and material performance parameters prevents automated dimensional convergence during packaging structure iterations. The retrieval logic of the historical case library fails to form a closed-loop feedback loop with real-time design requirements, limiting the efficiency of template self-optimization in multi-constraint scenarios. These shortcomings stem from the lack of a dynamic feature extraction engine, an imperfect real-time data stream fusion mechanism, and a lack of deep integration of adaptive generation algorithms. There is an urgent need to build a dynamic parameter coupling model and an automated template update framework to improve system responsiveness. Summary of the Invention

[0004] In response to the shortcomings of the existing technology, the present invention provides a customized packaging design and management system for household appliances, which is used to solve the problems of template update lag and low design efficiency caused by the static parametric template architecture in the existing household appliance packaging design system, the inability to adapt to product geometric characteristics in real time, insufficient dynamic coupling of logistics and material parameters, and the lack of closed-loop feedback of historical cases.

[0005] In order to solve the above technical problems, the specific technical solutions of the present invention are as follows:

[0006] The present invention provides a customized packaging design and management system for household appliances, comprising a dynamic feature genetic encoding module, a cross-domain parameter coupling optimization engine, a two-level genetic optimization architecture, a dynamic case feature reconstruction module, a co-evolution verification interface, and an adaptive processing instruction generator;

[0007] The dynamic feature genetic encoding module is used to convert the geometric features of the home appliance into chromosome encoding data that can be processed by the genetic algorithm, and synchronously output the chromosome encoding data to the two-level genetic optimization architecture;

[0008] The cross-domain parameter coupling optimization engine is used to integrate logistics environment parameters and material performance constraints, establish a multi-objective optimization model and send optimization instructions to the two-level genetic optimization architecture;

[0009] The dual-level genetic optimization architecture includes a topology optimization layer and a case self-optimization layer, which is used to receive the chromosome encoding data and optimization instructions, generate packaging structure parameters and case retrieval strategies through genetic algorithm crossover mutation, and write the optimization results into the dynamic case feature reconstruction module;

[0010] The dynamic case feature reconstruction module is used to update the feature clustering center of the historical case database according to the optimization results, and feed back the updated feature matching rules to the case self-optimization layer;

[0011] The co-evolution verification interface is used to import the candidate solutions output by the topology optimization layer into the physical simulation module, and reversely adjust the weight coefficients of the cross-domain parameter coupling optimization engine according to the simulation results;

[0012] The adaptive machining instruction generator is used to match the optimized structural parameters with the production line equipment constraints to generate a numerical control machining instruction set including a version identifier.

[0013] Furthermore, in the customized packaging design and management system for household appliances of the present invention, the dynamic feature genetic coding module includes:

[0014] A 3D scanning preprocessing unit is used to filter out noise and reconstruct the surface of the appliance surface point cloud data, and transmit the reconstructed mesh model to the parameterized encoding unit;

[0015] A parameterized encoding unit is used to receive the mesh model, segment it into characteristic surface patches according to curvature mutation boundaries, and perform parameterized expression using non-uniform rational B-splines (NURBS) to generate surface control point data;

[0016] A chromosome generation unit is used to generate a binary gene sequence according to the support point density distribution in the surface control point data, encode the NURBS continuity parameters into real-number gene segments, and transmit the mixed-encoded chromosome data stream to the topology optimization layer and the case self-optimization layer of the two-level genetic optimization architecture through a data bus.

[0017] Furthermore, in the customized packaging design and management system for household appliances described in the present invention, the cross-domain parameter coupling optimization engine includes:

[0018] A constraint fusion unit is used to perform fast Fourier transform on the logistics vibration spectrum to obtain frequency domain energy distribution, generate a dynamic constraint matrix based on the material impact resistance curve, and input the dynamic constraint matrix into the Pareto solver unit;

[0019] The Pareto solver unit adopts the NSGA-II non-dominated sorting algorithm with a congestion comparison operator to generate a non-dominated solution set of packaging structure topology parameters and case retrieval priorities based on the dynamic constraint matrix, and distributes the optimization instructions to the topology optimization layer and case self-optimization layer of the two-level genetic optimization architecture through the data bus.

[0020] Furthermore, in the customized packaging design and management system for household appliances of the present invention, the two-level genetic optimization architecture further includes:

[0021] Cross-layer elite sharing unit is used to regularly extract high-fitness individual gene fragments with finite element stress concentration coefficients below a threshold from the topology optimization layer, and extract historical case parameters with Mahalanobis distance similarity above a threshold from the case self-optimization layer. Gene fragments are exchanged in the shared storage area through a crossover rule based on cosine similarity, triggering adaptive crossover probability adjustment to drive the co-evolution of the two-layer genetic population;

[0022] The fitness calculation of the topology optimization layer is based on the maximum principal stress and material usage of the finite element analysis, and the similarity matching of the case self-optimization layer adopts the t-SNE manifold projection and improved Hausdorff distance measurement algorithm.

[0023] Furthermore, in the customized packaging design and management system for household appliances according to the present invention, the dynamic case feature reconstruction module includes:

[0024] The incremental clustering unit uses an online K-means++ algorithm with an exponentially decaying forgetting factor to update the feature vector cluster centers based on the optimized parameter set output by the genetic algorithm. It also dynamically adjusts the weights of historical cases through a sliding time window mechanism and transmits the updated cluster centers to the time series alignment unit.

[0025] The timing alignment unit is used to receive the cluster center, perform a sliding window Fourier transform on the time-frequency characteristics of the logistics environment parameters, align the timing differences between historical cases and the current design scenario through a dynamic time warping algorithm, and feed back the aligned feature vector to the case self-optimization layer.

[0026] Furthermore, in the customized packaging design and management system for household appliances of the present invention, the collaborative evolution verification interface includes:

[0027] The physical simulation agent unit is used to generate the stress propagation path inside the package based on the particle collision force calculation model of the discrete element method, and output the energy absorption distribution thermal map to the weight correction unit;

[0028] The weight correction unit is used to dynamically adjust the lightweight and strength weight coefficients of the multi-objective Pareto model in the cross-domain parameter coupling optimization engine according to the deviation value between the maximum deformation variable in the thermal map and the preset threshold value using a proportional and integral control strategy, and feed back the corrected weight coefficients to the Pareto solution unit through the data bus.

[0029] Furthermore, in the customized packaging design and management system for household appliances of the present invention, the adaptive processing instruction generator includes:

[0030] The equipment constraint mapping unit is used to analyze the interpolation accuracy parameters of the production line equipment, establish the geometric inclusion relationship between the machining path and the tool radius through the Delaunay triangulation algorithm, generate the minimum machining clearance constraint rules, and input the constraint rules into the version tree management unit;

[0031] The version tree management unit uses the hash chain technology of the Merkle tree structure to maintain the derivation relationship between CNC instruction versions, binds the version identifier corresponding to the constraint rule with the clustering model version number of the dynamic case feature reconstruction module, and records the version mapping relationship through the blockchain distributed ledger to realize full-link data traceability.

[0032] Beneficial effects of the present invention:

[0033] The present invention uses a dynamic feature genetic encoding module to convert the geometric features of household appliances into chromosome data that can be genetically optimized, achieving real-time adaptation of product form and parametric templates; a cross-domain parameter coupling optimization engine integrates logistics vibration spectrum and material impact resistance parameters to construct a multi-objective dynamic constraint matrix, driving the packaging structure to automatically balance between lightweight and strength goals; a two-level genetic optimization architecture uses the co-evolution mechanism of the topology optimization layer and the case self-optimization layer, combined with finite element stress analysis and historical case similarity matching, to simultaneously optimize material distribution and case retrieval strategies, thereby improving the design convergence efficiency in multi-constraint scenarios; a dynamic case feature reconstruction module eliminates the timing differences between historical data and the current scenario based on incremental clustering and dynamic time warping algorithms, enhancing the real-time matching accuracy of the case library; a co-evolution verification interface dynamically corrects and optimizes weight coefficients through physical simulation feedback, forming a design-verification closed-loop iterative link; an adaptive processing instruction generator uses geometric constraint rules and blockchain version traceability technology to achieve mapping and binding of processing parameters and case library versions, ensuring the traceability and consistency of the entire production data chain, thereby significantly shortening the packaging design cycle, reducing material waste and improving structural reliability. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] In order to more clearly illustrate the technical solution of the present invention, the following is a brief introduction to the drawings required for use in the embodiments. Obviously, for ordinary technicians in this field, other drawings can be obtained based on the drawings without paying any creative labor.

[0035] Figure 1 This is a system architecture diagram of a customized packaging design and management system for home appliances provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0036] In order to make the purpose, technical solutions and advantages of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below in conjunction with the specific embodiments of the present invention and the corresponding drawings. Obviously, the embodiments described are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention. The technical solutions provided by each embodiment of the present invention are described in detail below in conjunction with the drawings. In order to better understand the purpose of the present invention, the present invention is further described in detail below.

[0037] See also Figure 1 The present invention provides a customized packaging design and management system for home appliances, including a dynamic feature genetic encoding module, a cross-domain parameter coupling optimization engine, a two-level genetic optimization architecture, a dynamic case feature reconstruction module, a co-evolution verification interface, and an adaptive processing instruction generator;

[0038] The dynamic feature genetic encoding module is used to convert the geometric features of the home appliance into chromosome encoding data that can be processed by the genetic algorithm, and synchronously output the chromosome encoding data to the two-level genetic optimization architecture;

[0039] The dynamic feature genetic encoding module transforms the geometric features of home appliances into a data structure that can be optimized by a genetic algorithm through a multi-level data processing process. The 3D scanning preprocessing unit first uses a laser scanner to acquire high-density point cloud data on the appliance surface. A statistical filtering algorithm is used to remove outlier noise points based on the local density distribution. The point cloud is smoothed using the moving least squares method. A Poisson surface reconstruction algorithm is then used to generate a closed triangular mesh model. The reconstructed mesh model is then transferred to the parametric encoding unit. After receiving the mesh model, the parametric encoding unit identifies surface mutation boundaries using a curvature extreme value detection algorithm, segments the model into multiple feature surface patches, and parametrically represents each patch using a non-uniform rational B-spline (NURBS) control point optimization algorithm. This generates a surface feature dataset containing control point coordinates, weights, and node vectors. Based on the support point density distribution in the dataset, the chromosome generation unit quantifies the presence of support points per square millimeter into a binary gene sequence. It also encodes the geometric constraints within the NURBS surface continuity parameters into real-number gene segments, forming a hybrid-encoded chromosome data structure. This chromosome data stream is then synchronously transmitted via a data bus to the topology optimization layer and case-based self-optimization layer of the dual-level genetic optimization architecture. This process, through a progressive process of geometric feature extraction, parametric mapping, and hybrid encoding, transforms complex surface features into standardized input data that can be iteratively optimized by the genetic algorithm, providing fundamental data support for the dynamic adaptation of packaging structures.

[0040] The cross-domain parameter coupling optimization engine is used to integrate logistics environment parameters and material performance constraints, establish a multi-objective optimization model and send optimization instructions to the two-level genetic optimization architecture;

[0041] The cross-domain parameter coupling optimization engine dynamically optimizes packaging parameters through multi-source data integration and algorithm collaboration. The constraint fusion unit receives the time-domain signal from the logistics vibration spectrum and uses a fast Fourier transform to decompose the vibration signal into a frequency-domain energy distribution, extracting the energy peaks in each frequency band as dynamic load characteristics. Simultaneously, it analyzes the yield strength and damping parameters in the material's impact resistance curve, matrix-fusing the frequency-domain energy distribution with the material's mechanical properties to generate a dynamic constraint matrix containing equivalent dynamic loads, static boundary conditions, and impact resistance thresholds. This matrix is ​​then transmitted to the Pareto solver via a data interface. Based on the dynamic constraint matrix, the Pareto solver constructs a multi-objective optimization model using the NSGA-II non-dominated sorting algorithm with a congestion comparison operator. This non-dominated sorting strategy screens candidate solutions that simultaneously meet lightweighting and structural strength objectives. The congestion comparison operator maintains the diversity of the solution set's distribution in the objective space, generating a non-dominated solution set with a prioritized list of historical case retrievals and optimization schemes for the packaging structure's topological parameters. This solution set is distributed via a data bus to the topology optimization layer and case self-optimization layer of the dual-level genetic optimization architecture. The topology optimization layer receives structural parameters for iterative finite element optimization, while the case self-optimization layer uses historical case features for similarity matching based on retrieval priorities. This process, through the synergy of frequency-domain feature extraction, multi-constraint fusion, and multi-objective optimization algorithms, enables dynamic coupled analysis of logistics environment parameters and material properties, providing cross-domain optimization decision-making instructions for packaging design.

[0042] The dual-level genetic optimization architecture includes a topology optimization layer and a case self-optimization layer, which is used to receive the chromosome encoding data and optimization instructions, generate packaging structure parameters and case retrieval strategies through genetic algorithm crossover mutation, and write the optimization results into the dynamic case feature reconstruction module;

[0043] A two-level genetic optimization architecture achieves iterative improvement of packaging parameters through hierarchical optimization and cross-layer interaction. After receiving chromosome-encoded data, the topology optimization layer calculates the maximum principal stress distribution of the packaging structure under drop conditions based on finite element analysis. It then generates a fitness assessment based on the material volume fraction and selects high-fitness individual gene fragments with stress concentration coefficients below a preset threshold. The optimized structural parameters are then written into the dynamic case feature reconstruction module via a data interface. The case self-optimization layer parses the case retrieval priority list in the optimization instructions and uses the t-SNE manifold projection algorithm to map historical case feature vectors into a low-dimensional space. Using an improved Hausdorff distance to measure the geometric differences between the current design constraints and historical cases, it extracts a set of historical case parameter sets with similarities above a threshold. This strategy is then generated and fed back to the optimization instruction generation chain. The cross-layer elite sharing unit regularly extracts binary gene sequences from low-stress concentration areas within high-fitness individuals in the topology optimization layer. These sequences are then cross-recombined with real-valued gene segments from highly similar cases in the case self-optimization layer based on cosine similarity. This triggers an adaptive crossover probability adjustment mechanism, generating new population data that integrates structural strength and historical experience. This data is then synchronously injected into the finite element iteration process of the topology optimization layer and the similarity matching engine of the case self-optimization layer via a shared storage area. This process, through cross-layer interaction between stress distribution features and historical case features, enables the coordinated optimization of packaging structural parameters across both physical performance and empirical data. This drives the continuous writing of optimization results into the dynamic case feature reconstruction module to update feature cluster centers.

[0044] The dynamic case feature reconstruction module is used to update the feature clustering center of the historical case database according to the optimization results, and feed back the updated feature matching rules to the case self-optimization layer;

[0045] The dynamic case feature reconstruction module achieves real-time updating of historical case features and optimization of matching rules through dynamic clustering and time series alignment. After receiving the optimized parameter set output by the two-level genetic optimization architecture, the incremental clustering unit uses an online K-means++ algorithm with an exponentially decaying forgetting factor to update the feature vector cluster centers. It then partitions the time series data of historical cases using a sliding time window mechanism, dynamically adjusts the weight coefficients of cases within the window, and reduces the impact of outdated cases outside the time window on the cluster centers. The updated cluster centers are then transmitted to the time series alignment unit. The time series alignment unit performs a sliding window Fourier transform on the vibration acceleration time series data of the logistics environment parameters to extract the energy distribution pattern of the time-frequency features. It calculates the difference in time series features between historical cases and the current design scenario using a dynamic time warping algorithm and aligns the time series fluctuation patterns of the two using a dynamic curved path, eliminating feature offsets caused by inconsistent transport phase divisions. The aligned feature vectors are fed back to the case self-optimization layer via a data bus to update the retrieval rules of the similarity matching engine, driving the case self-optimization layer to call historical case parameters that match the spatiotemporal characteristics of the current design scenario in subsequent iterations. The above processing enables the feature space of the historical case database to evolve synchronously with real-time design requirements through incremental updates of cluster centers and dynamic calibration of time series, supporting precise adaptation and closed-loop feedback of case retrieval strategies.

[0046] The co-evolution verification interface is used to import the candidate solutions output by the topology optimization layer into the physical simulation module, and reversely adjust the weight coefficients of the cross-domain parameter coupling optimization engine according to the simulation results;

[0047] The co-evolutionary verification interface achieves closed-loop verification and optimization of design solutions through physical simulation and parameter feedback. After receiving the candidate packaging structure parameters output by the topology optimization layer, the physical simulation agent unit establishes a kinematic model of the filling particles using the discrete element method. It calculates the collision forces and energy transfer paths between particles using Hertzian contact theory, simulates the internal stress distribution of the packaging during transportation, and generates thermal map data that includes energy absorption efficiency and deformation characteristics. This thermal map is then transmitted to the weight correction unit. The weight correction unit analyzes the maximum deformation data in the thermal map, calculates its deviation from the preset safety threshold, and adopts the proportional-integral control strategy to dynamically adjust the weight coefficient distribution of the multi-objective Pareto model in the cross-domain parameter coupling optimization engine. The proportional term responds to the real-time deformation deviation, and the integral term accumulates the historical deviation trend. The new weight ratio of the lightweight target and the structural strength target is generated through weighted fusion; the corrected weight coefficient is fed back to the Pareto solver unit through the data bus, triggering the multi-objective optimization model to recalculate the non-dominated solution set, update the material distribution scheme of the topology optimization layer and the retrieval strategy priority of the case self-optimization layer, and form a closed-loop interactive link of simulation verification, parameter correction, and model iteration, so that the packaging design parameters can be co-evolved between physical performance verification and mathematical optimization.

[0048] The adaptive machining instruction generator is used to match the optimized structural parameters with the production line equipment constraints to generate a numerical control machining instruction set including a version identifier.

[0049] The adaptive machining instruction generator achieves precise generation and data traceability of CNC instructions through device constraint matching and version management. The device constraint mapping unit analyzes the interpolation accuracy parameters of the production line's CNC machine tools to obtain the minimum resolution data for the tool motion trajectory. Based on the Delaunay triangulation algorithm, it constructs a geometric topological network of the machining path point set, calculates the geometric inclusion relationship between the tool radius and the path spacing, generates a constraint rule set that meets the minimum machining clearance requirements, and transmits this rule set to the version tree management unit via a data interface. After receiving the constraint rule set, the version tree management unit uses the hash chain technology of the Merkle tree structure to construct a CNC instruction version derivation relationship tree, chain-binding the root hash value of each machining instruction version to its parent version hash value. At the same time, the version identifier corresponding to the constraint rule is mapped and associated with the clustering model version number of the dynamic case feature reconstruction module. The binding relationship between the version identifier, clustering model version number, and timestamp is recorded in the distributed ledger through the distributed consensus mechanism of the blockchain nodes, forming an unalterable full-lifecycle data chain. The CNC machining instruction set adapts to the communication protocols of different production line equipment through protocol conversion middleware, generates an instruction version tree containing evolutionary algebraic identifiers, supports rapid location of historical version differences and triggers parameter rollback operations during production backtracking, and realizes dynamic matching of optimized structural parameters and production line processing capabilities and full-link data consistency maintenance.

[0050] The present invention provides a customized packaging design and management system for household appliances. This system achieves dynamic optimization and automated generation of packaging designs through multi-module collaboration. The system first digitizes the geometric features of household appliances using a dynamic feature genetic encoding module. A 3D scanning preprocessing unit performs noise filtering and surface reconstruction on surface point cloud data to generate a high-precision mesh model. A parameterized encoding unit segments the mesh model into characteristic surface patches based on curvature mutation boundaries and extracts surface control point data using a non-uniform rational B-spline (NURBS) parameterization method. Finally, a chromosome generation unit maps the support point density distribution into a binary gene sequence and encodes surface continuity parameters into real-number gene segments, forming a hybrid-encoded chromosome data stream. This data stream is synchronously transmitted via a data bus to the topology optimization layer and case-based self-optimization layer of a two-level genetic optimization architecture.

[0051] The cross-domain parameter coupling optimization engine receives the logistics vibration spectrum and material performance parameters. The constraint fusion unit converts the vibration spectrum into a frequency-domain energy distribution through fast Fourier transform (FFT) and combines it with the material's impact resistance curve to generate a dynamic constraint matrix. The Pareto solver uses the NSGA-II non-dominated sorting algorithm with a crowding comparison operator to generate a non-dominated solution set of packaging structure topology parameters and case retrieval priorities based on the dynamic constraint matrix. The optimization instructions are then distributed to the two-level genetic optimization architecture via a data bus. The topology optimization layer of the two-level genetic optimization architecture calculates maximum principal stress and material usage based on finite element analysis and generates fitness evaluation results. The case self-optimization layer maps historical case feature vectors to a low-dimensional space through t-SNE manifold projection and selects highly similar cases using an improved Hausdorff distance metric. The cross-level elite sharing unit periodically extracts gene fragments with low stress concentration coefficients from the topology optimization layer and performs cosine similarity-based crossover recombination with the parameters of highly similar cases in the case self-optimization layer, driving the coevolution of the two-level genetic populations.

[0052] The incremental clustering unit of the dynamic case feature reconstruction module uses an online K-means++ algorithm with an exponentially decaying forgetting factor to update the feature vector cluster centers based on the genetic optimization results and dynamically adjust the weights of historical cases through a sliding time window. The time alignment unit performs a sliding window Fourier transform on the time-frequency characteristics of the logistics environment parameters, uses a dynamic time warping algorithm to align the timing differences between historical cases and the current design scenario, and feeds the updated feature matching rules back to the case self-optimization layer. The physical simulation agent unit of the co-evolution verification interface establishes a particle collision force model based on the discrete element method, simulates the stress propagation path within the package, and generates an energy absorption thermogram. The weight correction unit uses a proportional-integral control strategy to dynamically adjust the lightweight and strength weight coefficients of the multi-objective Pareto model based on the deviation between the maximum deformation variable in the thermogram and the preset threshold, and feeds the corrected parameters back to the cross-domain parameter coupling optimization engine via the data bus.

[0053] The device constraint mapping unit of the adaptive machining instruction generator analyzes the interpolation accuracy parameters of production line equipment, establishes the geometric inclusion relationship between the machining path and the tool radius using the Delaunay triangulation algorithm, and generates the minimum machining clearance constraint rule. The version tree management unit uses a Merkle tree-structured hash chain to maintain the derivation relationship between CNC instruction versions, binding the version identifier corresponding to the constraint rule to the clustering model version number of the dynamic case feature reconstruction module. This version mapping relationship is recorded in a blockchain distributed ledger, enabling full-link data traceability. These modules establish a closed-loop feedback link through a data bus and shared storage area, enabling dynamic iteration and self-optimization of packaging structure parameters, case matching rules, and machining instruction versions.

[0054] Specifically, in the customized packaging design and management system for household appliances according to the present invention, the dynamic feature genetic coding module includes:

[0055] A 3D scanning preprocessing unit is used to filter out noise and reconstruct the surface of the appliance surface point cloud data, and transmit the reconstructed mesh model to the parameterized encoding unit;

[0056] A parameterized encoding unit is used to receive the mesh model, segment it into characteristic surface patches according to curvature mutation boundaries, and perform parameterized expression using non-uniform rational B-splines (NURBS) to generate surface control point data;

[0057] A chromosome generation unit is used to generate a binary gene sequence according to the support point density distribution in the surface control point data, encode the NURBS continuity parameters into real-number gene segments, and transmit the mixed-encoded chromosome data stream to the topology optimization layer and the case self-optimization layer of the two-level genetic optimization architecture through a data bus.

[0058] The present invention provides a customized packaging design and management system for household appliances. This system achieves dynamic optimization and automated generation of packaging designs through multi-module collaboration. The system first digitizes the geometric features of household appliances using a dynamic feature genetic encoding module. A 3D scanning preprocessing unit performs noise filtering and surface reconstruction on surface point cloud data to generate a high-precision mesh model. A parameterized encoding unit segments the mesh model into characteristic surface patches based on curvature mutation boundaries and extracts surface control point data using a non-uniform rational B-spline (NURBS) parameterization method. Finally, a chromosome generation unit maps the support point density distribution into a binary gene sequence and encodes surface continuity parameters into real-number gene segments, forming a hybrid-encoded chromosome data stream. This data stream is synchronously transmitted via a data bus to the topology optimization layer and case-based self-optimization layer of a two-level genetic optimization architecture.

[0059] The cross-domain parameter coupling optimization engine receives the logistics vibration spectrum and material performance parameters. The constraint fusion unit converts the vibration spectrum into a frequency-domain energy distribution through fast Fourier transform (FFT) and combines it with the material's impact resistance curve to generate a dynamic constraint matrix. The Pareto solver uses the NSGA-II non-dominated sorting algorithm with a crowding comparison operator to generate a non-dominated solution set of packaging structure topology parameters and case retrieval priorities based on the dynamic constraint matrix. The optimization instructions are then distributed to the two-level genetic optimization architecture via a data bus. The topology optimization layer of the two-level genetic optimization architecture calculates maximum principal stress and material usage based on finite element analysis and generates fitness evaluation results. The case self-optimization layer maps historical case feature vectors to a low-dimensional space through t-SNE manifold projection and selects highly similar cases using an improved Hausdorff distance metric. The cross-level elite sharing unit periodically extracts gene fragments with low stress concentration coefficients from the topology optimization layer and performs cosine similarity-based crossover recombination with the parameters of highly similar cases in the case self-optimization layer, driving the coevolution of the two-level genetic populations.

[0060] The incremental clustering unit of the dynamic case feature reconstruction module uses an online K-means++ algorithm with an exponentially decaying forgetting factor to update the feature vector cluster centers based on the genetic optimization results and dynamically adjust the weights of historical cases through a sliding time window. The time alignment unit performs a sliding window Fourier transform on the time-frequency characteristics of the logistics environment parameters, uses a dynamic time warping algorithm to align the timing differences between historical cases and the current design scenario, and feeds the updated feature matching rules back to the case self-optimization layer. The physical simulation agent unit of the co-evolution verification interface establishes a particle collision force model based on the discrete element method, simulates the stress propagation path within the package, and generates an energy absorption thermogram. The weight correction unit uses a proportional-integral control strategy to dynamically adjust the lightweight and strength weight coefficients of the multi-objective Pareto model based on the deviation between the maximum deformation variable in the thermogram and the preset threshold, and feeds the corrected parameters back to the cross-domain parameter coupling optimization engine via the data bus.

[0061] The device constraint mapping unit of the adaptive machining instruction generator analyzes the interpolation accuracy parameters of production line equipment, establishes the geometric inclusion relationship between the machining path and the tool radius using the Delaunay triangulation algorithm, and generates the minimum machining clearance constraint rule. The version tree management unit uses a Merkle tree-structured hash chain to maintain the derivation relationship between CNC instruction versions, binding the version identifier corresponding to the constraint rule to the clustering model version number of the dynamic case feature reconstruction module. This version mapping relationship is recorded in a blockchain distributed ledger, enabling full-link data traceability. These modules establish a closed-loop feedback link through a data bus and shared storage area, enabling dynamic iteration and self-optimization of packaging structure parameters, case matching rules, and machining instruction versions.

[0062] Specifically, the customized packaging design and management system for home appliances of the present invention includes the following:

[0063] A constraint fusion unit is used to perform fast Fourier transform on the logistics vibration spectrum to obtain frequency domain energy distribution, generate a dynamic constraint matrix based on the material impact resistance curve, and input the dynamic constraint matrix into the Pareto solver unit;

[0064] The Pareto solver unit adopts the NSGA-II non-dominated sorting algorithm with a congestion comparison operator to generate a non-dominated solution set of packaging structure topology parameters and case retrieval priorities based on the dynamic constraint matrix, and distributes the optimization instructions to the topology optimization layer and case self-optimization layer of the two-level genetic optimization architecture through the data bus.

[0065] The cross-domain parameter coupling optimization engine dynamically optimizes packaging parameters through multi-source data processing and optimization algorithms. The constraint fusion unit receives the time-domain signal from the logistics vibration spectrum and uses a fast Fourier transform to decompose the vibration signal into a frequency-domain energy distribution. The energy peaks in each frequency band are extracted as dynamic load characteristics. Combined with the yield strength and damping characteristic parameters from the material's impact resistance curve, the frequency-domain energy distribution is matrix-fused with the material parameters to generate a dynamic constraint matrix containing equivalent dynamic loads, static boundary conditions, and impact resistance thresholds. This matrix is ​​then transmitted to the Pareto solver via a data interface.

[0066] Based on the dynamic constraint matrix, the Pareto solver constructs a multi-objective optimization model using the NSGA-II non-dominated sorting algorithm with a congestion comparison operator. This non-dominated sorting strategy screens candidate solutions that simultaneously meet material lightweighting and structural strength requirements. The congestion comparison operator maintains the diversity of the solution set's distribution in the objective space, generating a non-dominated solution set that includes optimized topological parameters for the packaging structure and a prioritized list of historical case retrieval. This solution set is then distributed via a data bus to the topology optimization layer and case self-optimization layer of a two-level genetic optimization architecture. The topology optimization layer receives the structural parameters for iterative finite element optimization, while the case self-optimization layer uses historical case features for similarity matching based on retrieval priorities. This process, through the synergy of frequency domain feature extraction, multi-constraint fusion, and a multi-objective optimization algorithm, enables dynamic coupled analysis of logistics environment parameters and material properties, providing cross-domain optimization decision-making guidance for packaging design.

[0067] Specifically, the customized packaging design and management system for home appliances of the present invention further includes:

[0068] Cross-layer elite sharing unit is used to regularly extract high-fitness individual gene fragments with finite element stress concentration coefficients below a threshold from the topology optimization layer, and extract historical case parameters with Mahalanobis distance similarity above a threshold from the case self-optimization layer. Gene fragments are exchanged in the shared storage area through a crossover rule based on cosine similarity, triggering adaptive crossover probability adjustment to drive the co-evolution of the two-layer genetic population;

[0069] The fitness calculation of the topology optimization layer is based on the maximum principal stress and material usage of the finite element analysis, and the similarity matching of the case self-optimization layer adopts the t-SNE manifold projection and improved Hausdorff distance measurement algorithm.

[0070] A two-level genetic optimization architecture achieves iterative improvement of packaging parameters through hierarchical optimization and cross-layer interaction. The topology optimization layer calculates the maximum principal stress distribution of the packaging structure under drop conditions based on finite element analysis results. This calculation, combined with the material volume fraction, generates a fitness assessment value and selects high-fitness individual gene fragments whose stress concentration coefficients fall below a preset threshold. The case self-optimization layer uses the t-SNE manifold projection algorithm to map high-dimensional historical case feature vectors to a low-dimensional space. Using a modified Hausdorff distance to measure the geometric differences between the current design constraints and historical cases, the system extracts parameter sets from historical cases with similarities exceeding a threshold.

[0071] The cross-layer elite sharing unit regularly extracts binary gene sequences from high-fitness individuals in the topology optimization layer, capturing regions of low stress concentration. It also extracts real-valued gene segments associated with constraint parameters from high-similarity cases in the case self-optimization layer. In a shared storage area, these gene segments are paired and combined based on cosine similarity matching rules. This triggers a gene recombination operation, in which the crossover probability adaptively adjusts over the evolutionary generations, generating a new population that integrates structural strength and historical experience. The recombined gene data is then fed into the finite element iteration process in the topology optimization layer and the similarity matching engine in the case self-optimization layer, driving the simultaneous evolution of material distribution optimization and case retrieval strategies. This process, through cross-layer interaction between stress distribution characteristics and historical case features, enables the coordinated optimization of packaging structural parameters across both physical performance and empirical data.

[0072] Specifically, in the customized packaging design and management system for home appliances according to the present invention, the dynamic case feature reconstruction module includes:

[0073] The incremental clustering unit uses an online K-means++ algorithm with an exponentially decaying forgetting factor to update the feature vector cluster centers based on the optimized parameter set output by the genetic algorithm. It also dynamically adjusts the weights of historical cases through a sliding time window mechanism and transmits the updated cluster centers to the time series alignment unit.

[0074] The timing alignment unit is used to receive the cluster center, perform a sliding window Fourier transform on the time-frequency characteristics of the logistics environment parameters, align the timing differences between historical cases and the current design scenario through a dynamic time warping algorithm, and feed back the aligned feature vector to the case self-optimization layer.

[0075] The dynamic case feature reconstruction module continuously optimizes historical case features through dynamic clustering and time series alignment. After receiving the optimized parameter set output by the genetic algorithm, the incremental clustering unit uses an online K-means++ algorithm with an exponentially decaying forgetting factor to update the feature vector cluster centers. It then partitions the time series data of historical cases using a sliding time window, dynamically adjusting the weight coefficients of cases within the window to reduce the impact of outdated cases outside the time window on the cluster centers. The updated cluster centers are then transmitted to the time series alignment unit.

[0076] The time series alignment unit performs a sliding window Fourier transform on the vibration acceleration and temperature and humidity data in the logistics environment parameters to extract the energy distribution pattern of the time-frequency characteristics. Based on the dynamic time warping algorithm, the time series feature differences between the historical case and the current design scenario are calculated. Dynamic curved paths are used to align the time series fluctuation patterns of the two, eliminating feature offsets caused by inconsistent transport phase divisions. The aligned feature vectors are fed back to the case self-optimization layer via the data bus to update the retrieval rules of the similarity matching engine. This process, through incremental updates of cluster centers and dynamic calibration of time series, ensures that the feature space of the historical case database evolves synchronously with real-time design requirements, supporting precise adaptation of case retrieval strategies.

[0077] Specifically, in the customized packaging design and management system for home appliances according to the present invention, the collaborative evolution verification interface includes:

[0078] The physical simulation agent unit is used to generate the stress propagation path inside the package based on the particle collision force calculation model of the discrete element method, and output the energy absorption distribution thermal map to the weight correction unit;

[0079] The weight correction unit is used to dynamically adjust the lightweight and strength weight coefficients of the multi-objective Pareto model in the cross-domain parameter coupling optimization engine according to the deviation value between the maximum deformation variable in the thermal map and the preset threshold value using a proportional and integral control strategy, and feed back the corrected weight coefficients to the Pareto solution unit through the data bus.

[0080] The weight correction unit analyzes the maximum deformation data in the thermal map, calculates its deviation from the preset safety threshold, and uses a proportional-integral control strategy to dynamically adjust the weight coefficients of the multi-objective Pareto model in the cross-domain parameter coupling optimization engine. The proportional term responds to the real-time deviation, and the integral term accumulates historical deviation trends. Through weighted fusion, a new weight ratio between the lightweighting and structural strength objectives is generated. The revised weight coefficients are fed back to the Pareto solver via the data bus, triggering the multi-objective optimization model to recalculate the non-dominated solution set and update the material distribution scheme of the topology optimization layer and the retrieval strategy of the case self-optimization layer. This process, through the closed-loop interaction between simulation data and the optimization algorithm, forms a co-evolutionary mechanism between physical verification and mathematical optimization for packaging design parameters.

[0081] Specifically, in the customized packaging design and management system for household appliances according to the present invention, the adaptive processing instruction generator includes:

[0082] The equipment constraint mapping unit is used to analyze the interpolation accuracy parameters of the production line equipment, establish the geometric inclusion relationship between the machining path and the tool radius through the Delaunay triangulation algorithm, generate the minimum machining clearance constraint rules, and input the constraint rules into the version tree management unit;

[0083] The version tree management unit uses the hash chain technology of the Merkle tree structure to maintain the derivation relationship between CNC instruction versions, binds the version identifier corresponding to the constraint rule with the clustering model version number of the dynamic case feature reconstruction module, and records the version mapping relationship through the blockchain distributed ledger to realize full-link data traceability.

[0084] The technical features of the present invention are explained as follows:

[0085] Dynamic feature genetic coding module: Converts the geometric features of household appliances (such as surface shape and support point distribution) into a standardized data format that can be processed by genetic algorithms. Through three-dimensional scanning, parametric coding and chromosome generation steps, the physical features are mapped into genetic sequences that can be genetically optimized.

[0086] 3D scanning preprocessing unit: denoises and reconstructs the surface point cloud data of household appliances obtained by laser scanning, removes abnormal points through statistical filtering, and uses the Poisson surface reconstruction algorithm to generate a closed triangular mesh model.

[0087] Parametric encoding unit: Based on curvature mutation detection, the mesh model is segmented into feature surface patches, and the non-uniform rational B-spline (NURBS) method is used to optimize the control point weights to generate parametric data describing the surface geometric characteristics.

[0088] NURBS parametric expression: A mathematical modeling method that accurately describes the shape of complex surfaces through control points, weights, and node vectors, and supports local modification and continuity control of surfaces (such as G1 / G2 continuity).

[0089] Chromosome generation unit: Encode the support point density distribution into a binary gene sequence and the surface continuity parameter into a real-number gene segment to form a mixed data type for the genetic algorithm to perform crossover and mutation operations.

[0090] Cross-domain parameter coupling optimization engine: Integrates cross-domain parameters such as logistics vibration spectrum and material impact resistance to build a multi-objective optimization model to drive the coordinated optimization of packaging structure and case retrieval strategy.

[0091] Constraint fusion unit: Perform frequency domain analysis on logistics vibration signals, extract energy distribution characteristics, and generate a dynamic constraint matrix based on the mechanical properties of the material to provide input conditions for multi-objective optimization.

[0092] Pareto solver unit: uses the NSGA-II algorithm to screen non-dominated solution sets, balances lightweight and structural strength goals, maintains solution set diversity through crowding comparison, and outputs optimization instructions to the genetic architecture.

[0093] NSGA-II non-dominated sorting algorithm: A multi-objective optimization algorithm that uses a non-dominated sorting hierarchical solution set and combines a crowding operator to avoid local convergence. It is suitable for trade-off decisions on conflicting objectives in packaging design.

[0094] Two-level genetic optimization architecture: This includes a topology optimization layer (optimizing material distribution based on finite element analysis) and a case self-optimization layer (based on historical case similarity matching), achieving global optimization through cross-layer interaction.

[0095] Topology optimization layer: Based on the results of finite element stress analysis, the fitness value of the material distribution scheme is calculated to screen gene fragments with low stress concentration and low material consumption.

[0096] Case self-optimization layer: Projects historical case features into a low-dimensional space, and drives the evolution of case retrieval strategies by improving the Hausdorff distance metric and the similarity with current design requirements.

[0097] Dynamic case feature reconstruction module: Dynamically updates the case library feature cluster center based on the optimization results, and eliminates the transportation environment differences between historical data and the current scenario through time series alignment.

[0098] Incremental clustering unit: Use the online K-means++ algorithm to incrementally update the cluster center, introduce an exponential decay factor to reduce the weight of outdated cases, and maintain the timeliness of the case library.

[0099] Time series alignment unit: Extracts time-frequency features of logistics parameters (such as vibration, temperature and humidity), aligns historical and current time series data through dynamic time warping (DTW), and modifies case applicability scores.

[0100] Co-evolution verification interface: The physical simulation results are fed back into the optimization model, and closed-loop feedback between design parameters and simulation verification is achieved through dynamic weight adjustment.

[0101] Physical simulation agent unit: Based on the discrete element method (DEM), it simulates the collision process of packaging filling particles, outputs stress propagation path and energy absorption thermal map, and verifies structural reliability.

[0102] Weight correction unit: Based on the deviation of the simulated deformation, the proportional-integral (PI) control strategy is used to adjust the multi-objective weight coefficients to balance the priority of lightweight and strength.

[0103] Adaptive machining instruction generator: Analyzes the machining constraints of production line equipment, generates an adaptive CNC instruction set, and implements full-link data traceability through version management.

[0104] Equipment constraint mapping unit: Establishes the geometric inclusion relationship between tool path and machining accuracy through Delaunay triangulation, generates minimum machining clearance rules and avoids interference.

[0105] Version tree management unit: Use Merkle tree hash chain to record the instruction version derivation relationship, bind the case library version number, and use blockchain to store evidence to ensure that data cannot be tampered with.

[0106] Merkle tree structure hash chain technology: Verifies data integrity through tree hash structure, ensures traceability of version changes, and supports fast difference comparison and version backtracking.

[0107] Blockchain distributed ledger: Distributed storage version mapping relationship and timestamp, multi-node data synchronization through consensus mechanism, to prevent single point tampering risk.

[0108] Dynamic Time Warping Algorithm: Flexible alignment of time series data of different lengths to eliminate the impact of differences in transportation stage divisions on case matching.

[0109] Improved Hausdorff distance metric algorithm: Enhanced geometric shape similarity evaluation, reduced noise interference through segmented matching, and improved historical case retrieval accuracy.

[0110] Proportional-integral control strategy: Dynamically adjust parameter weights based on real-time deviation and historical accumulated error to achieve a smooth transition of the optimization target.

[0111] Delaunay triangulation algorithm: Generates a triangular mesh that maximizes the minimum angle, which is used to construct a geometric constraint model for the machining path to ensure machining feasibility.

[0112] Sliding window Fourier transform: Extracts the local frequency domain features of the time series signal and captures the transient change pattern of logistics vibration energy.

[0113] Genetic algorithm crossover mutation: Generate new individuals through gene segment exchange and random perturbation, drive population diversity evolution, and avoid premature convergence of optimization.

[0114] Finite element analysis: A numerical simulation method that calculates the stress distribution of packaging structures under conditions such as dropping and stacking, and evaluates structural reliability.

[0115] The adaptive machining instruction generator achieves precise machining instruction generation and data traceability through device constraint mapping and version control. The device constraint mapping unit analyzes the interpolation accuracy parameters of the production line's CNC machine tools to obtain the minimum resolution data for the tool motion trajectory. Using the Delaunay triangulation algorithm, it constructs a geometric topological network of the machining path point set, calculates the inclusion relationship between the tool radius and the path spacing, and generates a constraint rule set that meets the minimum machining clearance requirements. This rule set is then transmitted to the version tree management unit via a data interface.

[0116] After receiving the constraint rule set, the version tree management unit uses Merkle tree-structured hash chain technology to construct a CNC instruction version derivation tree. The hash value of each machining instruction version is chained to the hash value of its parent version. The version identifier corresponding to the constraint rule is mapped and associated with the clustering model version number of the dynamic case feature reconstruction module. The binding relationship between the version identifier, clustering model version number, and timestamp is recorded in the distributed ledger through the distributed consensus mechanism of the blockchain nodes. This process, through the combination of geometric constraint generation and version traceability technology, makes the iterative process of machining instructions verifiable and traceable, supporting the integrity management of production data throughout its lifecycle.

[0117] During the closed-loop feedback and production traceability phase, the incremental clustering unit of the dynamic case feature reconstruction module uses an online K-means++ algorithm with an exponentially decaying forgetting factor to update the feature vector cluster centers based on genetic optimization results and adjust the weights of historical cases using a sliding time window. The time series alignment unit performs a sliding window Fourier transform on the vibration acceleration time series data of logistics environment parameters and uses a dynamic time warping algorithm to eliminate time series differences between historical cases and the current transportation scenario. The physical simulation agent unit of the co-evolution verification interface simulates the collision process of packaging particles using the discrete element method to generate energy absorption thermograms. The weight correction unit uses a proportional-integral control strategy to adjust the lightweight and strength weight coefficients of the Pareto model based on the deformation deviation value. The adaptive machining instruction generator establishes tool path geometric constraints using the Delaunay triangulation algorithm. The version tree management unit uses a Merkle tree hash chain and a blockchain distributed ledger to record the mapping between instruction versions and case library versions, enabling machining parameter traceability and closed-loop design verification, supporting improved design efficiency throughout the entire process.

[0118] This invention addresses the shortcomings of existing technologies through a dynamic parameterized architecture and a multi-module collaborative mechanism. The dynamic feature genetic encoding module converts the geometric features of home appliances into chromosome-encoded data that can be processed by a genetic algorithm. The 3D scanning preprocessing unit extracts point cloud data and reconstructs the mesh model. The parameterized encoding unit segments surface patches based on curvature mutation boundaries and generates NURBS control point data. The chromosome generation unit encodes support point density and continuity parameters into a hybrid gene sequence, which is synchronously transmitted to the two-level genetic optimization architecture via a data bus, achieving dynamic adaptation of product geometric features and real-time updating of template parameters.

[0119] The cross-domain parameter coupling optimization engine integrates the logistics vibration spectrum and material impact resistance parameters. The constraint fusion unit generates frequency domain dynamic load characteristics through fast Fourier transform and constructs a dynamic constraint matrix based on the material performance curve. The Pareto solver unit uses the NSGA-II algorithm to generate the non-dominated solution set of packaging structure topology parameters and case retrieval strategy. Through the collaborative iteration of the topology optimization layer and the case self-optimization layer of the two-level genetic optimization architecture, it drives the dynamic coupling analysis of logistics parameters and material properties, and solves the problem of automatic dimensional convergence under multiple constraints.

[0120] The dynamic case feature reconstruction module updates the historical case cluster centers through an incremental clustering unit. The timing alignment unit eliminates timing differences between historical cases and the current scenario using a sliding window Fourier transform and a dynamic time warping algorithm. The co-evolutionary verification interface feeds physical simulation results back to the parameter coupling engine, and the adaptive machining instruction generator establishes a closed-loop feedback loop through device constraint mapping and blockchain version traceability. These modules, through genetic algorithm-driven data interaction and dynamic weight adjustment, achieve precise matching of historical case features with real-time design requirements, forming a self-optimizing closed-loop design system.

Claims

1. A customized packaging design and management system for home appliances, characterized in that: It includes a dynamic feature genetic encoding module, a cross-domain parameter coupling optimization engine, a two-level genetic optimization architecture, a dynamic case feature reconstruction module, a co-evolution verification interface, and an adaptive processing instruction generator; The dynamic feature genetic encoding module is used to convert the geometric features of the home appliance into chromosome encoding data that can be processed by the genetic algorithm, and synchronously output the chromosome encoding data to the two-level genetic optimization architecture; The cross-domain parameter coupling optimization engine is used to integrate logistics environment parameters and material performance constraints, establish a multi-objective optimization model and send optimization instructions to the two-level genetic optimization architecture; The dual-level genetic optimization architecture includes a topology optimization layer and a case self-optimization layer, which is used to receive the chromosome encoding data and optimization instructions, generate packaging structure parameters and case retrieval strategies through genetic algorithm crossover mutation, and write the optimization results into the dynamic case feature reconstruction module; The dynamic case feature reconstruction module is used to update the feature clustering center of the historical case database according to the optimization results, and feed back the updated feature matching rules to the case self-optimization layer; The co-evolution verification interface is used to import the candidate solutions output by the topology optimization layer into the physical simulation module, and reversely adjust the weight coefficients of the cross-domain parameter coupling optimization engine according to the simulation results; The adaptive machining instruction generator is used to match the optimized structural parameters with the production line equipment constraints to generate a numerical control machining instruction set including a version identifier.

2. The customized packaging design and management system for home appliances according to claim 1, characterized in that: The dynamic feature genetic encoding module includes: A 3D scanning preprocessing unit is used to filter out noise and reconstruct the surface of the appliance surface point cloud data, and transmit the reconstructed mesh model to the parameterized encoding unit; A parameterized encoding unit is used to receive the mesh model, segment it into characteristic surface patches through curvature mutation boundaries, and perform non-uniform rational B-spline parameterization expression to generate surface control point data; A chromosome generation unit is used to generate a binary gene sequence according to the support point density distribution in the surface control point data, encode the NURBS continuity parameters into real-number gene segments, and transmit the mixed-encoded chromosome data stream to the topology optimization layer and the case self-optimization layer of the two-level genetic optimization architecture through a data bus.

3. The customized packaging design and management system for home appliances according to claim 1, characterized in that: The cross-domain parameter coupling optimization engine includes: A constraint fusion unit is used to perform fast Fourier transform on the logistics vibration spectrum to obtain frequency domain energy distribution, generate a dynamic constraint matrix based on the material impact resistance curve, and input the dynamic constraint matrix into the Pareto solver unit; The Pareto solver unit adopts the NSGA-II non-dominated sorting algorithm with a congestion comparison operator to generate a non-dominated solution set of packaging structure topology parameters and case retrieval priorities based on the dynamic constraint matrix, and distributes the optimization instructions to the topology optimization layer and case self-optimization layer of the two-level genetic optimization architecture through the data bus.

4. The customized packaging design and management system for home appliances according to claim 1, characterized in that: The two-level genetic optimization architecture also includes: Cross-layer elite sharing unit is used to regularly extract high-fitness individual gene fragments with finite element stress concentration coefficients below a threshold from the topology optimization layer, and extract historical case parameters with Mahalanobis distance similarity above a threshold from the case self-optimization layer. Gene fragments are exchanged in the shared storage area through a crossover rule based on cosine similarity, triggering adaptive crossover probability adjustment to drive the co-evolution of the two-layer genetic population; The fitness calculation of the topology optimization layer is based on the maximum principal stress and material usage of the finite element analysis, and the similarity matching of the case self-optimization layer adopts the t-SNE manifold projection and improved Hausdorff distance measurement algorithm.

5. The customized packaging design and management system for home appliances according to claim 1, characterized in that: The dynamic case feature reconstruction module includes: The incremental clustering unit uses an online K-means++ algorithm with an exponentially decaying forgetting factor to update the feature vector cluster centers based on the optimized parameter set output by the genetic algorithm. It also dynamically adjusts the weights of historical cases through a sliding time window mechanism and transmits the updated cluster centers to the time series alignment unit. The timing alignment unit is used to receive the cluster center, perform a sliding window Fourier transform on the time-frequency characteristics of the logistics environment parameters, align the timing differences between historical cases and the current design scenario through a dynamic time warping algorithm, and feed back the aligned feature vector to the case self-optimization layer.

6. The customized packaging design and management system for home appliances according to claim 1, characterized in that: The collaborative evolution verification interface includes: The physical simulation agent unit is used to generate the stress propagation path inside the package based on the particle collision force calculation model of the discrete element method, and output the energy absorption distribution thermal map to the weight correction unit; The weight correction unit is used to dynamically adjust the lightweight and strength weight coefficients of the multi-objective Pareto model in the cross-domain parameter coupling optimization engine according to the deviation value between the maximum deformation variable in the thermal map and the preset threshold value using a proportional and integral control strategy, and feed back the corrected weight coefficients to the Pareto solution unit through the data bus.

7. The customized packaging design and management system for home appliances according to claim 1, characterized in that: The adaptive processing instruction generator includes: The equipment constraint mapping unit is used to analyze the interpolation accuracy parameters of the production line equipment, establish the geometric inclusion relationship between the machining path and the tool radius through the Delaunay triangulation algorithm, generate the minimum machining clearance constraint rules, and input the constraint rules into the version tree management unit; The version tree management unit uses the hash chain technology of the Merkle tree structure to maintain the derivation relationship between CNC instruction versions, binds the version identifier corresponding to the constraint rule with the clustering model version number of the dynamic case feature reconstruction module, and records the version mapping relationship through the blockchain distributed ledger to realize full-link data traceability.

Images