Order management method and system based on big data

Abstract

The present application relates to the field of order management, and in particular to a method and system for order management based on big data. The method includes: obtaining a returned flower parameter set, based on the returned flower parameter set, collecting and analyzing flower status data in real time through a multimodal data fusion strategy to determine a flower freshness data set; extracting flower types based on the returned flower parameter set, structurally splitting the returned flowers based on the flower freshness data set and the flower types, and generating a modular bouquet parameter set; obtaining a user's real-time bouquet customization requirement set, and dynamically generating a flower pre-sale data set based on the user's real-time bouquet customization requirement set and the modular bouquet parameter set; pushing the flower pre-sale data set to the user, receiving user feedback information, generating a flower secondary sales order data set, and outputting a flower processing log. The present application reasonably monitors the freshness of flowers during the flower sales process, thereby realizing the reuse of flower resources.

Description

Technical Field

[0001] The present application relates to the field of order management, and in particular to an order management method and system based on big data. Background Art

[0002] In the digital economy era, the flower order management system has become a core hub connecting planting bases, logistics networks and consumer terminals, supporting the trillion-level gift consumption market and daily emotional expression needs, and directly affecting the quality of urban life and the economic benefits of the flower industry.

[0003] However, the existing flower order management system is unable to match the corresponding resale channels according to the dynamic changes in the status of flowers when handling returned flowers, which can easily lead to irreversible loss of the freshness of flowers, not only increasing product losses, but also causing waste of flowers and rising costs. Summary of the Invention

[0004] This application provides an order management method and system based on big data to solve the above technical problems.

[0005] In a first aspect, the present application provides an order management method based on big data, the method comprising:

[0006] A returned flower parameter set is obtained. Based on the returned flower parameter set, flower status data is collected and analyzed in real time using a multimodal data fusion strategy to determine a flower freshness dataset. Flower types are extracted based on the returned flower parameter set. The returned flowers are structurally split based on the flower freshness dataset and the flower types to generate a modular bouquet parameter set. A user's real-time bouquet customization requirement set is obtained. Based on the user's real-time bouquet customization requirement set and the modular bouquet parameter set, a flower pre-sale dataset is dynamically generated. The flower pre-sale dataset is pushed to the user, user feedback is received, a flower secondary sales order dataset is generated, and a flower processing log is output.

[0007] This solution first uses a multimodal data fusion strategy to analyze and quantify the freshness of flowers in real time, replacing subjective manual assessments. Then, based on the flower type and freshness data, returned bouquets are modularly broken down into three categories: good quality (for resale), moderately defective (for public welfare matching), and severely damaged (for fertilizer recycling), generating a standardized inventory list. Then, based on real-time user customization needs, a dynamic pricing engine is used to accurately match module inventory and generate pre-sale plans. Finally, a distributed recording system establishes a full-link trusted record, completing the transaction loop and outputting the flower processing log to the merchant. This returned flower processing process uses multimodal data fusion to accurately quantify freshness, significantly reducing the misjudgment rate and preventing high-quality returned flowers from being wasted. It then conducts efficient structural sorting and modular reorganization based on type and freshness, significantly shortening processing time and reducing waste during the shelf life. At the same time, it dynamically matches module inventory based on user needs and combines freshness-based intelligent pricing to effectively activate secondary sales and greatly improve conversion rates. Ultimately, it forms a digital closed-loop management of the entire process (sorting → sales → log), transforming returned flowers, which were traditionally regarded as losses, into reconfigurable resources that can accurately match multi-level needs (resale, charity, recycling), systematically achieving zero resource waste and maximizing value regeneration.

[0008] Optionally, based on the returned flower parameter set, collecting and analyzing flower status data in real time through a multimodal data fusion strategy to determine a flower freshness data set includes:

[0009] The returned flower parameter set includes returned flower image data, environmental temperature and humidity data, packaging status information, and transportation time data;

[0010] Extracting visual corruption features based on the returned flower image data using a deep vision method;

[0011] The visual corruption features include the proportion of browning pixels at the edge of petals and the deviation coefficient of petal stretch;

[0012] Based on the environmental temperature and humidity data, predicting the water loss rate through a time series prediction method;

[0013] generating a spoilage acceleration factor through regression analysis based on the packaging status information and the transportation time data, wherein the transportation time data is negatively correlated with the flower freshness grade, i.e., the longer the transportation time, the lower the flower freshness grade;

[0014] Based on a multimodal data fusion strategy, the flower freshness is subjected to correlation analysis and comprehensive evaluation according to the visual corruption characteristics, the water loss rate, the corruption acceleration factor, and the transportation time data to determine the flower freshness dataset.

[0015] Through this solution, a flower freshness assessment method based on a multimodal data fusion strategy significantly improves the comprehensiveness and accuracy of corruption feature identification by integrating four-dimensional data sources: vision, environment, packaging, and transportation. In particular, it breaks through the blind spot of traditional single-modal methods in detecting hidden damage. It uses a deep vision model to extract microscopic morphological features, a time series prediction model to correlate temperature and humidity with the rate of water loss, and regression analysis to decouple the synergistic effects of packaging and transportation, effectively eliminating empirical misjudgments.

[0016] Optionally, the multimodal data fusion strategy includes:

[0017] Performing normalization and quantification processing based on the visual corruption characteristics, the water loss rate, and the corruption acceleration factor to generate a standardized corruption index set;

[0018] Based on a preset corruption association model, dynamically assigning weight coefficients to the indicators in the standardized corruption indicator set according to the types of flowers;

[0019] Performing a weighted fusion calculation on the standardized corruption indicator set according to the assigned weight coefficient to generate an initial freshness score;

[0020] Generating a dynamic adjustment factor based on the transportation time data and the flower type, correcting the initial freshness score through a nonlinear mapping relationship, and classifying the freshness level of each type of flower;

[0021] The flower freshness dataset is generated according to the flower freshness level of each type.

[0022] This solution utilizes a multimodal fusion mechanism to address three core pain points. First, it cross-validates visual features, moisture loss rates, and spoilage acceleration factors to cover both explicit and implicit spoilage defects. Second, it dynamically assigns weights based on flower types to achieve precise "one product, one policy" assessment (for example, increasing the weight of roses' transportation time). Finally, it introduces a nonlinear dynamic adjustment factor to compensate for the accelerated attenuation effect of long transportation. This systematically overcomes the inaccurate assessment problems of traditional methods under complex coupling factors, and avoids secondary sales disputes and resource mismatch losses caused by misjudgment of product appearance.

[0023] Optionally, the flower freshness grade includes good quality, moderately defective, and severely damaged;

[0024] The "good condition" category refers to flowers in good condition, whose freshness level meets the preset resale threshold. Flowers of this level are clearly marked and used for discount promotions;

[0025] The medium defect category refers to flowers that have shown some signs of decay. The freshness level of the flowers is lower than the preset resale threshold but meets the preset reuse threshold. Flowers of this level are donated to public welfare.

[0026] The severely damaged category refers to flowers that are severely rotten or damaged, and the freshness level of the flowers is lower than a preset reusable threshold. Flowers of this level are used to be converted into organic fertilizers.

[0027] Through this plan, a three-level classification mechanism is used to achieve a coordinated improvement in resource recycling and operational efficiency. Specifically, targeted discount sales of flowers with good appearance increase the resale rate of returned flowers and reduce the cost of purchasing new products. Public donations of flowers with moderate defects create social value. Fertilizer conversion of flowers with severe damage reduces the amount of solid waste handled. At the same time, classified processing increases warehouse turnover efficiency by several times, and automated grading greatly reduces the error rate of manual sorting, creating a dual benefit of maximizing resource value and optimizing operating costs.

[0028] Optionally, extracting flower types according to the returned flower parameter set, structurally splitting the returned flowers based on the flower freshness dataset and the flower types, and generating a modular bouquet parameter set includes:

[0029] Extracting the flower type and the flower freshness level of each flower based on the flower freshness dataset;

[0030] Performing cluster analysis on the returned flowers according to the flower species to generate a number of independent flower species sets;

[0031] In each of the independent flower seed sets, the flower seeds are secondary grouped according to the flower freshness level to determine a good quality subset, a moderately defective subset, and a severely damaged subset within the current independent flower seed set, thereby generating an independent flower seed status set;

[0032] According to the different flower types and the flower freshness levels, searching a preset flower seed preservation database to determine the preservation time interval of each subset of the current independent flower seed status set;

[0033] According to the fresh-keeping time interval, performing corresponding independent operations on each subset of the independent flower seed status set;

[0034] The independent operation includes:

[0035] For the good-looking subset, retain the original flower branches and generate a complete flower material module;

[0036] For the moderate defect subset, the rotten parts are removed and the usable parts are retained to generate a downgraded flower material module;

[0037] For the severely damaged subset, petals and leaves are peeled off to generate a fragment material module;

[0038] The number of available flowers in each subset of the independent flower species status set is counted to generate the modular bouquet parameter set including the flower species, the flower freshness level, the available flower quantity, and the freshness preservation time interval.

[0039] Through this solution, the good-quality subset, moderately defective subset, and severely damaged subset within the current independent flower seed set are determined, and an independent flower seed status set is generated. Through precise stratification of the three-level subsets, the vast majority of returned flowers are effectively utilized (sale of intact flowers, donation of downgraded flowers, and fertilizer conversion of fragmented materials), reducing the waste rate compared to traditional methods. By binding operational decisions on freshness-keeping time, the response speed of selling flowers with good quality is increased several times, greatly improving the punctuality rate of public welfare donations and significantly improving timeliness.

[0040] Optionally, dynamically generating a flower pre-sale data set based on the user's real-time bouquet customization requirement set and the modular bouquet parameter set includes:

[0041] Analyze the user's real-time bouquet customization requirement set and extract the flower types and corresponding quantity requirements requested by the user;

[0042] Based on the modular bouquet parameter set, a subset of available returned flower inventory that matches the flower type is retrieved in real time, the subset of available returned flower inventory including the good-quality subset of the corresponding type, the available flower quantity, and the freshness-keeping time interval;

[0043] determining the total available quantity of all flowers based on the available returned flower inventory subset;

[0044] According to the total available quantity of all flowers, determine whether the total available quantity meets the quantity requirement:

[0045] If all conditions are met, a full refund pre-sale plan for flowers will be generated, including the specific quantity of flowers and discounted prices;

[0046] If the requirements are partially met, a combination plan of new and old flowers will be generated, including the ratio of new and old flowers, combined discount price and the maximum number of flowers that can be met.

[0047] The flower pre-sale data set is dynamically generated according to the full return flower pre-sale plan and the new and old flower combination plan.

[0048] Through this solution, available returned flowers marked as "good condition" are accurately matched to user customization needs, which significantly improves the reuse rate of such returned resources and avoids idle resources and waste. At the same time, the system dynamically generates a pre-sale plan for all returned flowers or a combination plan of new and old flowers according to inventory satisfaction, and pushes the proposed plan to users, effectively reducing order loss caused by insufficient inventory, thereby optimizing the overall order conversion rate. In addition, the generation of the combined discount price in the plan is strictly based on the base price of flowers and the discount coefficient dynamically determined according to the freshness level, ensuring the transparency of the pricing process and clear rules, significantly enhancing the credibility of the final quotation to users, and further consolidating the conversion effect.

[0049] Optionally, if the requirements are partially met, a new and old flower material combination plan including the ratio of new and old flower materials, combined discount price and maximum quantity that can be met is generated, including:

[0050] Determining the maximum quantity of returned flowers that can be satisfied and the ratio of new and old flowers based on the available returned flower inventory subset and the user's real-time bouquet customization requirements;

[0051] According to the types of unsold flowers and the types of returned flowers, the proportion of unsold flowers and the proportion of returned flowers in the new and old flower material combination plan, a preset flower price database is searched to extract the flower benchmark price corresponding to each flower type in the new and old flower material combination plan;

[0052] Dynamically determining a price discount coefficient for the returned flowers based on the flower freshness level corresponding to the good-quality subset;

[0053] The flower freshness grade is positively correlated with the price discount coefficient, that is, the higher the flower freshness grade, the greater the price discount coefficient and the smaller the price discount;

[0054] Based on the proportion of unsold flowers required to be used in the new and old flower material combination plan and on the basis of the flower benchmark price, an actual unit price of unsold flowers is generated;

[0055] Based on the proportion of returned flowers required to be used in the new and old flower material combination plan, the product of the base price of flowers and the price discount coefficient is used as the actual unit price of the returned flowers;

[0056] Determine the final combined selling price of the corresponding type of flowers based on the number of available flowers allocated for each type of flowers in the new and old flower combination plan, the actual unit price of the unsold flowers, and the actual unit price of the returned flowers;

[0057] Summarizing the final combined selling prices of all the flower types in the new and old flower material combination scheme to generate a new and old flower material combination price, and using the new and old flower material combination price as the combined discounted price;

[0058] The new and old flower material combination plan is constructed based on the ratio of the new and old flower materials, the combined discount price and the maximum satisfying quantity.

[0059] Through this solution, by dynamically generating a combination of new and old flower materials, the ratio of new and old flower materials is calculated in real time based on the maximum number of returned flowers, and the price discount coefficient is dynamically set based on the freshness level of the flowers. Finally, a stepped combination discount price is generated in combination with the benchmark price of flowers. On the one hand, by flexibly adjusting the ratio of new and old flower materials, the time window of limited inventory is maximized to avoid the corruption and loss of high-value flowers. On the other hand, scientific graded pricing is achieved through the price discount coefficient. While ensuring reasonable profits for high-freshness flowers, the turnover of medium and low-freshness inventory is accelerated, thereby solving the core pain points of inefficient supply and demand matching and the difficulty in balancing corporate profits and user price sensitivity.

[0060] Optionally, the method further includes:

[0061] Obtain the user's address and the geographical location of the shipping location, and set an initial search radius based on the freshness-limiting time interval of the good-appearance subset with the shipping location as the center;

[0062] dynamically adjusting the initial search radius according to the flower freshness level of each of the flower types in the good-quality subset to generate a dynamic search radius, wherein the lower the flower freshness level, the smaller the dynamic search radius;

[0063] Generate a priority user push set based on the user's address within the range corresponding to the dynamic search radius;

[0064] Based on the freshness aging interval and the estimated transportation time, the priority user push set is screened, wherein the estimated transportation time must be less than or equal to the lower limit of the freshness aging interval. If a matching user exists after screening, a matching user push set is generated;

[0065] If the flower freshness level drops to the moderate defect category, searching the preset public welfare unit database within the range corresponding to the dynamic search radius to generate a set of candidate public welfare units;

[0066] Screening the candidate public welfare units based on the freshness-keeping time interval and the estimated transportation time, wherein the estimated transportation time must be less than or equal to the lower limit of the freshness-keeping time interval;

[0067] If there is a matching public welfare organization after screening, a public welfare donation push set will be generated.

[0068] Through this solution, a dynamic search radius mechanism is used to achieve spatial adaptation of resource allocation, ensuring that high-grade flowers expand the scope of commercial conversion, and low-grade flowers lock adjacent areas to control risks; and establish automatic triggering rules for public welfare channels, that is, when the freshness drops to the donation threshold, it switches to searching the public welfare database, and combines hard time verification to eliminate invalid donations.

[0069] Optionally, the step of pushing the flower pre-sale data set to a user, receiving user feedback information, generating a flower secondary sale order data set, and outputting a flower processing log includes:

[0070] Pushing the flower pre-sale dataset to target users via a preset electronic communication channel;

[0071] Receive the user feedback information, analyze the user feedback information, determine whether the user accepts the corresponding plan in the flower pre-sale data set, and if the user accepts, extract the corresponding full return flower pre-sale plan and the new and old flower combination plan;

[0072] Generate a flower secondary sales order dataset based on the fully refunded flower pre-sale plan and the new and old flower combination plan accepted by the user, and based on the new and old flower ratio, the combined discount price, and the maximum available quantity in the flower pre-sale dataset;

[0073] The flower processing log is generated and outputted according to the flower secondary sales order data set.

[0074] Through this solution, electronic communication channels are used to accurately reach target users, so that returned flowers can be quickly matched with demand within the freshness window, improving user conversion rate and satisfaction, and significantly reducing the loss rate of high-value flowers. Structured order data is automatically generated to avoid order recording errors and compliance risks caused by manual operations. At the same time, the full-process operation log ensures the auditability and abnormal warning capabilities of the business closed loop. Ultimately, the utilization rate of returned flower resources is maximized, the dynamic response efficiency of the supply chain is improved, and the coordinated optimization of secondary sales compliance management is achieved.

[0075] In a second aspect, the present application provides an order management system based on big data, the system comprising:

[0076] A flower analysis module is used to obtain a parameter set of returned flowers, and based on the parameter set of returned flowers, collect and analyze flower status data in real time through a multimodal data fusion strategy to determine the freshness level of the flowers;

[0077] a structure splitting module, configured to extract flower types according to the returned flower parameter set, perform structure splitting on the returned flowers based on the flower freshness dataset and the flower types, and generate a modular bouquet parameter set;

[0078] The customization and pre-sale module is used to obtain the user's real-time bouquet customization requirement set and dynamically generate a flower pre-sale data set based on the user's real-time bouquet customization requirement set and the modular bouquet parameter set;

[0079] The interaction and order module is used to push the flower pre-sale data set to users, receive user feedback information, generate a flower secondary sale order data set, and output a flower processing log. BRIEF DESCRIPTION OF THE DRAWINGS

[0080] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, a brief introduction will be given below to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0081] Figure 1 A schematic diagram of an application scenario provided in one embodiment of the present application;

[0082] Figure 2 A flowchart of an order management method based on big data provided in one embodiment of the present application;

[0083] Figure 3 A schematic diagram of the structure of an order management system based on big data provided in one embodiment of the present application; DETAILED DESCRIPTION

[0084] To make the purpose, technical solutions, and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0085] In this document, the term "and / or" simply describes a relationship between related objects, indicating that three possible relationships exist. For example, "A and / or B" can represent: A exists alone, A and B exist simultaneously, or B exists alone. Furthermore, the character " / " in this document, unless otherwise specified, generally indicates an "or" relationship between the related objects.

[0086] The embodiments of the present application are described in further detail below with reference to the accompanying drawings.

[0087] When handling returned flowers, the existing flower order management system is unable to match the corresponding resale channels according to the dynamic changes in the flowers' status, which can easily lead to irreversible loss of the flowers' freshness, increasing product losses, wasting flowers and increasing costs.

[0088] Based on this, the present application provides a big data-based order management method and system. First, a multimodal data fusion strategy is used to analyze and quantify the freshness of flowers in real time, replacing subjective manual evaluation. Then, based on the flower type and freshness data, the returned bouquets are modularly disassembled into three categories: good appearance (re-sale), moderately defective (public welfare matching), and severely damaged (fertilizer recycling), generating a standardized inventory list. Then, based on the user's real-time customization needs, a dynamic pricing engine is used to accurately match module inventory and generate pre-sale plans. Finally, a distributed recording system is used to establish a full-link trusted record, complete the transaction loop, and output the flower processing log to the merchant. This returned flower processing process uses multimodal data fusion to accurately quantify freshness, significantly reducing the misjudgment rate and preventing high-quality returned flowers from being wasted. It then conducts efficient structural sorting and modular reorganization based on type and freshness, significantly shortening processing time and reducing waste during the shelf life. At the same time, it dynamically matches module inventory based on user needs and combines freshness-based intelligent pricing to effectively activate secondary sales and greatly improve conversion rates. Ultimately, it forms a digital closed-loop management of the entire process (sorting → sales → log), transforming returned flowers, which were traditionally regarded as losses, into reconfigurable resources that can accurately match multi-level needs (resale, charity, recycling), systematically achieving zero resource waste and maximizing value regeneration.

[0089] Figure 1 This is a schematic diagram of an application scenario provided by this application. During the resale process of returned flowers, the method provided by this application can be used to match resale channels based on the dynamic state changes of the returned flowers during their processing, thereby avoiding unnecessary waste of flowers and preventing cost increases.

[0090] Specifically, the method of the present application is applied to any server, which is connected to a logistics management system and a flower e-commerce platform. The server obtains the returned flower parameter set provided by the logistics management system and the user's real-time bouquet customization requirement set provided by the flower e-commerce platform through the server. First, a multimodal data fusion strategy is used to analyze and quantify the freshness of the flowers in real time, replacing subjective manual evaluation; then, based on the flower type and freshness data, the returned bouquets are modularly disassembled into three categories of components: good appearance (resale), moderate defects (public welfare matching), and severe damage (fertilizer recycling), and a standardized inventory list is generated; then, combined with the user's real-time customization requirements, a dynamic pricing engine is used to accurately match the module inventory and generate a pre-sale plan; finally, a full-link trusted record is established through a distributed recording system, the transaction loop is completed, and the flower processing log is output to the merchant.

[0091] For specific implementation methods, please refer to the following embodiments.

[0092] Figure 2 This is a flowchart of an order management method based on big data provided by an embodiment of the present application. The method of this embodiment can be applied to the server in the above scenario. Figure 2 As shown, the method includes:

[0093] S201. Obtain a returned flower parameter set, and based on the returned flower parameter set, collect and analyze flower status data in real time through a multimodal data fusion strategy to determine a flower freshness data set.

[0094] The returned flower parameter set may refer to a data set of flowers returned to the warehouse due to customer rejection, including returned flower image data, ambient temperature and humidity data, packaging status information, and transportation time data. The data comes from the logistics management system.

[0095] The multimodal data fusion strategy can refer to an intelligent analysis strategy of multi-dimensional sensor data based on visual spoilage characteristics, water loss rate, spoilage acceleration factor and transportation time data.

[0096] The flower freshness dataset can be the freshness of returned bouquets quantified through a multimodal data fusion strategy.

[0097] Specifically, traditional flower return processing relies heavily on manual experience to determine freshness, which presents significant limitations. Subjective assessment methods struggle to accurately quantify key indicators like petal moisture content or micro-deterioration in the stem, leading to some reusable flowers being mistakenly rejected and wasting valuable resources. Furthermore, the manual sorting process is time-consuming, and delayed disposal causes flowers to continue to decay, missing the optimal reuse window. Furthermore, image, environmental, and transportation data are dispersed and independent, making it impossible to correlate causal relationships between packaging damage and localized deterioration, making it difficult to generate a scientific grading basis. This step deploys a multimodal sensor network to capture real-time data on physical form, air temperature and humidity, and transportation duration. A fusion analysis mechanism is then built to generate an objective freshness metric. This real-time freshness metric is then used to dynamically classify returned flowers. This multi-source data fusion reduces the risk of freshness misjudgment, eliminates the risk of subjective misjudgment, and shortens disposal timelines, laying a crucial foundation for scientific decision-making in the precise reuse of flower resources.

[0098] S202: Extract flower types according to the returned flower parameter set, perform structural decomposition on the returned flowers based on the flower freshness dataset and the flower types, and generate a modular bouquet parameter set.

[0099] Structural splitting can be a technical operation that groups flowers of the same type according to their freshness level, removes the rotten parts and retains the usable parts.

[0100] The modular bouquet parameter set may be a structured list recording the flower types, flower freshness levels, available flower quantities, and preservation time intervals.

[0101] Specifically, traditional flower return processing suffers from two core pain points: First, crude classification, based solely on appearance, fails to fully consider the differences in characteristics of different flower species and the specific details of their usable parts (e.g., discarding partially rotten but otherwise intact flowers based solely on overall appearance). This results in the mistaken discarding of high-value, reusable flowers, resulting in resource waste. Second, a lack of timeliness. Processing strategies fail to effectively correlate with the inherent shelf life of flowers and lack a dynamic consideration of their decay over time. As a result, some returned flowers, though not yet tampering, are not promptly and appropriately handled (e.g., sold or donated) within their limited shelf life, ultimately leading to their value being lost. This step involves physical disassembly, precisely separating returned flowers into three atomic components: Good-quality flowers are connected to the resale chain based on digital parameterized profiles; moderately defective flowers are targeted for charitable purposes; and severely damaged flowers are sent to the fertilizer recycling system. This structured deconstruction strategy based on objective data not only breaks through the value constraints of traditional whole flower forms on high-quality flower materials (such as intact flower heads being freed from the forced downgrading of withered branches and leaves), but also builds a "Lego-style" supply network based on a standardized component library. That is, through the cross-mapping of freshness parameters and species labels, the good-looking categories are accurately matched with high-end customization, the moderately defective categories and the severely damaged categories to activate social welfare and ecological cycles, and ultimately systematically resolve the structural mismatch contradiction between non-standard product resources and multi-level demands.

[0102] S203: Obtain a user's real-time bouquet customization requirement set, and dynamically generate a flower pre-sale data set based on the user's real-time bouquet customization requirement set and the modular bouquet parameter set.

[0103] The user's real-time bouquet customization requirement set can be the dynamic requirements of flower type, color theme, and freshness requirements submitted by the user, and the data comes from the flower e-commerce platform.

[0104] The flower pre-sale dataset can be a bouquet solution for sale generated by matching demand with module inventory.

[0105] Specifically, traditional resale of returned flowers faces the dual bottlenecks of a supply-demand impasse and inflexible pricing. Warehouses only statically categorize inventory by freshness, failing to correlate it with user-defined requests (e.g., "20 red roses") in real time, leading to lost orders for highly matched products. Furthermore, fixed discount schemes fail to dynamically adjust based on freshness levels, making them less attractive to price-sensitive users. This innovative approach implements dynamic inventory precision matching, using real-time retrieval of modular bouquet parameter sets to identify a subset of available returned inventory, addressing the "inventory without demand" backlog. A pricing strategy for combining new and used flowers is developed, dynamically generating a discount factor based on the freshness level of returned flowers (with smaller discounts for flowers in good condition) and re-selling unsold flowers at the base price, balancing price attractiveness with profit margins. A demand fulfillment grading mechanism is also introduced, generating a "maximum available quantity" notification when inventory is low, preventing customer churn due to unmet needs. Without this step, returned roses may become unavailable due to lack of immediate demand, and static pricing fails to engage potential consumers, reducing resale conversion rates by over 30%. This not only opens up the channel for dynamic matching of supply and demand, but also continuously approaches the optimal pricing point through intelligent algorithms, ultimately maximizing the value of returned resources.

[0106] S204: Push the flower pre-sale data set to the user, receive user feedback information, generate a flower secondary sale order data set, and output a flower processing log.

[0107] The flower secondary sales order dataset can be the order record of the user's confirmed purchase, including module code combination, delivery time, etc.

[0108] Flower processing logs can record the complete operational flow and quality data from return to resale.

[0109] Specifically, traditional order systems are unable to accurately identify potential users and appropriate delivery areas, and they also lack the ability to collect user feedback, which can lead to users abandoning purchases due to not receiving push notifications in a timely manner. This step addresses the issue of information transmission failure through multi-channel instant reach (simultaneous push notifications via app / SMS / WeChat). It also establishes an automated order generation closed loop, directly parsing the pre-sale plan to generate structured order data after user confirmation, eliminating manual entry errors. Furthermore, through system log collection technology, a full-process operation log tracking system is established, recording the complete path from sorting and assembly to sales, generating corresponding flower processing logs, and providing them to merchants through human-computer interaction devices such as high-definition display screens. This not only generates tamper-proof quality credit certificates at all stages of the supply chain but also provides the flower industry with a reusable digital management paradigm for non-standard products.

[0110] This solution first uses a multimodal data fusion strategy to analyze and quantify the freshness of flowers in real time, replacing subjective manual assessments. Then, based on the flower type and freshness data, returned bouquets are modularly broken down into three categories: good quality (for resale), moderately defective (for public welfare matching), and severely damaged (for fertilizer recycling), generating a standardized inventory list. Then, based on real-time user customization needs, a dynamic pricing engine is used to accurately match module inventory and generate pre-sale plans. Finally, a distributed recording system establishes a full-link trusted record, completing the transaction loop and outputting the flower processing log to the merchant. This returned flower processing process uses multimodal data fusion to accurately quantify freshness, significantly reducing the misjudgment rate and preventing high-quality returned flowers from being wasted. It then conducts efficient structural sorting and modular reorganization based on type and freshness, significantly shortening processing time and reducing waste during the shelf life. At the same time, it dynamically matches module inventory based on user needs and combines freshness-based intelligent pricing to effectively activate secondary sales and greatly improve conversion rates. Ultimately, it forms a digital closed-loop management of the entire process (sorting → sales → log), transforming returned flowers, which were traditionally regarded as losses, into reconfigurable resources that can accurately match multi-level needs (resale, charity, recycling), systematically achieving zero resource waste and maximizing value regeneration.

[0111] In some embodiments, the returned flower parameter set includes returned flower image data, ambient temperature and humidity data, packaging status information, and transportation time data; based on the returned flower image data, visual corruption features are extracted through deep vision methods; visual corruption features include the proportion of browning pixels on the petal edges and the petal stretch deviation coefficient; based on the ambient temperature and humidity data, the water loss rate is predicted through a time series prediction method; based on the packaging status information and transportation time data, a corruption acceleration factor is generated through regression analysis, wherein the transportation time data is negatively correlated with the flower freshness level, that is, the longer the transportation time, the lower the flower freshness level; based on the multimodal data fusion strategy, the flower freshness is correlated and comprehensively evaluated according to the visual corruption features, water loss rate, corruption acceleration factor and transportation time data to determine the flower freshness data set.

[0112] The returned flower parameter set may be a set including images, environment, packaging, and shipping information of the returned flowers.

[0113] The returned flower image data may be visual data reflecting the appearance of the flower, including petal shape, color changes, and the like.

[0114] The ambient temperature and humidity data may be the real-time temperature and humidity values ​​of the environment in which the flowers are located.

[0115] The packaging status information can be a quantitative description of the packaging integrity (such as damage, sealing).

[0116] The transportation time data can be the total time from shipment to return receipt, and then from return receipt to shipment back to the warehouse.

[0117] The ratio of browning pixels at the edge of petals may be the ratio of the browning area at the edge of petals to the total pixels of the image, reflecting the degree of oxidation and corruption.

[0118] The petal expansion deviation coefficient may be a deviation value between the actual expansion of the flower and the standard expansion, and may represent the degree of dehydration.

[0119] The water loss rate can be used to predict the water loss of flowers per unit time.

[0120] The spoilage acceleration factor can be a coefficient that quantifies the acceleration of the spoilage process caused by packaging defects and transportation time.

[0121] Multimodal data fusion strategy can be a comprehensive analysis framework that integrates visual, environmental, packaging and transportation data.

[0122] The flower freshness dataset may be a structured data set containing the freshness level of each flower.

[0123] Specifically, traditional methods that rely on manual visual inspection or single sensor data (such as images) have inherent flaws. Visual data makes it difficult to quantify the hidden damage caused by moisture loss due to temperature and humidity fluctuations, and environmental data cannot identify the accelerated microbial growth caused by damaged packaging. Flower spoilage is essentially a dynamic process that is a nonlinear superposition of multiple factors, including transportation time (negatively correlated with freshness), packaging status, and environmental parameters. Its complexity requires the construction of a collaborative analysis framework covering the entire spoilage chain. At the same time, the differences in spoilage characteristics among different flower species (such as roses' sensitivity to petal browning and lilies' prone to stem softening) require that the evaluation system dynamically adjust weights based on species characteristics to achieve accurate grading. This is the basis for determining the subsequent flow of resources (promotion of good-looking flowers, donation of moderately defective flowers, and conversion of severely damaged flowers into fertilizer). In addition, the rapid decline in freshness over time of returned flowers necessitates replacing inefficient manual evaluation with automated multimodal acquisition (imagery / environment / packaging / transportation) and fusion computing to avoid the loss of time-sensitive reusable resources. To address the above issues, this step first extracts visual corruption features: inputs the returned flower image into the pre-trained deep vision model, outputs the percentage of browning pixels on the edge of the petals (segments the edge area of ​​the petals and counts the percentage of browning pixels) and the petal stretch deviation coefficient (compares the deviation between the actual petal expansion angle and the standard template); then predicts the water loss rate: takes the ambient temperature and humidity time series data as input, uses the time series prediction model, and uses the high temperature and low humidity sequence to trigger a high water loss warning, outputting the hourly water loss rate curve for the next 24 hours; then generates a corruption acceleration factor: the independent variables are the packaging damage level (0-10 points) and the transportation time (hours), and the dependent variable is the corruption rate baseline value (through historical data The output is the spoilage acceleration factor under the combined effects of packaging and transportation time. Finally, multimodal fusion is used to assess freshness: visual features, water loss rate, and spoilage acceleration factor are converted into standardized indicators in the range [0, 1]. Preset weight templates are called according to the flower type (for example, roses focus on browning percentage, and lilies focus on stretching). A comprehensive spoilage index is synthesized according to the weights: comprehensive spoilage index = browning percentage + stretching deviation + water loss + spoilage acceleration factor. Based on the negative correlation between transportation time and freshness, the freshness level of orders with long transportation time is reduced. The output is a structured dataset containing flower ID, type, and freshness level (good / moderately defective / severely damaged), which is used as the flower freshness dataset.

[0124] Through this solution, a flower freshness assessment method based on a multimodal data fusion strategy significantly improves the comprehensiveness and accuracy of corruption feature identification by integrating four-dimensional data sources: vision, environment, packaging, and transportation. In particular, it breaks through the blind spot of traditional single-modal methods in detecting hidden damage. It uses a deep vision model to extract microscopic morphological features, a time series prediction model to correlate temperature and humidity with the rate of water loss, and regression analysis to decouple the synergistic effects of packaging and transportation, effectively eliminating empirical misjudgments.

[0125] In some embodiments, normalization and quantification processing is performed based on visual corruption characteristics, water loss rate and corruption acceleration factor to generate a standardized corruption indicator set; based on a preset corruption association model, weight coefficients of each indicator in the standardized corruption indicator set are dynamically assigned according to the type of flowers; a weighted fusion calculation is performed on the standardized corruption indicator set according to the assigned weight coefficients to generate an initial freshness score; a dynamic adjustment factor is generated based on transportation time data and flower type, the initial freshness score is corrected through a nonlinear mapping relationship, and the freshness level of each type of flower is divided; and a flower freshness data set is generated based on the freshness level of each type of flower.

[0126] Normalized quantification processing can be the process of converting three types of heterogeneous data, namely visual corruption characteristics (the proportion of browning pixels on the edge of petals, the deviation coefficient of petal stretch), water loss rate, and corruption acceleration factor, into standardized numerical values ​​with a unified dimension.

[0127] The standardized corruption indicator set can be a set of numerical values ​​generated after normalization and quantification processing, including three types of standardized data: visual corruption indicators, water loss indicators, and corruption factor indicators.

[0128] The corruption association model may be a knowledge base model that stores the weighted relationships between different flower types and corruption indicators.

[0129] The initial freshness score may be a preliminary score generated by weighted fusion of a standardized set of corruption indicators, reflecting the theoretical freshness without considering the transportation time.

[0130] The dynamic adjustment factor can be a correction coefficient generated based on the transportation time and the type of flowers to compensate for the accelerated spoilage caused by long transportation.

[0131] The nonlinear mapping relationship may be a non-proportional conversion rule describing the relationship between the revised score and the final flower freshness grade.

[0132] The flower freshness grade may be a classification label that identifies the availability status of returned flowers, including three grades: good quality, moderately defective, and severely damaged.

[0133] Specifically, traditional flower return management relies on a single visual assessment, which has significant limitations. It cannot detect hidden corruption caused by the transportation environment (such as water loss caused by local high temperature), and the three types of heterogeneous data, visual, environmental, and transportation (representing spatial corruption, hidden decay, and process loss, respectively) are difficult to directly coordinate due to differences in dimensions and reliability. Moreover, different types of flowers have different sensitivities to corruption factors and transportation time (for example, the water loss weight of lilies needs to reach 70%, while the visual browning weight of carnations only needs to be 30%). If a fixed weight model is used, the misjudgment rate of high-value flowers will surge. To address the above issues, this step first performs linear normalization on the proportion of browning pixels on the edge of the petals and maps it to the interval [0,1]. At the same time, the water loss rate is compressed to the same dimension through the Sigmoid function, and the corruption acceleration factor is normalized after taking the inverse to form a unified and comparable standardized corruption indicator set. Then, the preset corruption association model is queried according to the type of flowers to obtain the basic weight coefficient (for example, the rose is set with a visual weight of 0.4, a water weight of 0.4, and a corruption factor weight of 0.2). If the transportation time exceeds the threshold, the corruption factor weight is dynamically increased by 10%; and then the weighted The average value generates an initial score (for example, the weighted values ​​of 0.7 / 0.5 / 0.6 for rose indicators are calculated to be 0.62). Finally, a dynamic adjustment factor is adjusted based on the transportation time and flower type (for example, a factor of 0.85 is corresponding to a 36-hour rose). The initial score is multiplied by this factor to obtain a revised score (0.62 × 0.85 ≈ 0.53). This score is then mapped to a grade range using a nonlinear piecewise function (0.53 is classified as a moderate defect). This completes the full-link conversion from raw data to freshness grades, and generates a flower freshness dataset based on the freshness grade of each flower type.

[0134] This solution utilizes a multimodal fusion mechanism to address three core pain points. First, it cross-validates visual features, moisture loss rates, and spoilage acceleration factors to cover both explicit and implicit spoilage defects. Second, it dynamically assigns weights based on flower types to achieve precise "one product, one policy" assessment (for example, increasing the weight of roses' transportation time). Finally, it introduces a nonlinear dynamic adjustment factor to compensate for the accelerated attenuation effect of long transportation. This systematically overcomes the inaccurate assessment problems of traditional methods under complex coupling factors, and avoids secondary sales disputes and resource mismatch losses caused by misjudgment of product appearance.

[0135] In some embodiments, the flower freshness levels include good quality, moderate defect, and severely damaged; the good quality category refers to flowers in good condition, and the flower freshness level meets the preset resale threshold. Flowers of this level are clearly marked and used for discount promotions; the moderate defect category refers to flowers in a certain state of decay, and the flower freshness level is lower than the preset resale threshold but meets the preset reuse threshold. Flowers of this level are donated for public welfare; the severely damaged category refers to flowers in a severely corrupt or damaged state, and the flower freshness level is lower than the preset reuse threshold. Flowers of this level are used to be converted into organic fertilizer.

[0136] Good condition flowers can be returned flowers that meet resale quality standards.

[0137] Moderately defective flowers may be flowers that have some reusable value but cannot be sold directly.

[0138] Severe damage may include flowers that have lost their ornamental value and cannot be repaired.

[0139] The preset resaleable threshold may be a minimum quality score for flowers to be resaleable.

[0140] The preset reusability threshold may be a minimum quality score at which flowers may be used for non-sale purposes.

[0141] Specifically, the existing technology only performs a simple binary classification of returned flowers into those that can be resold and those that cannot be resold, resulting in some flowers that could have been reused not receiving reasonable resource processing. In addition, the flower industry has long faced the problem of resource depletion caused by mixed storage. Traditional processing methods mix flowers of different freshness, resulting in high-quality flowers being contaminated by low-quality ones and depreciating in value. In addition, charitable donations require whole and intact flowers, while organic fertilizers only require fragments. Ungraded flowers cannot adapt to differentiated needs. This step first uses the preset threshold comparison module to determine the grade based on the freshness score. If the score is ≥80, the flowers are marked as good and the following steps are performed: affixing a green identification label (including a discount QR code), storing them in a constant temperature fresh-keeping warehouse, and synchronizing them to the "discount flowers" section of the e-commerce platform. If the score is between 50 and 80, the flowers are marked as moderately defective and the following steps are performed: removing corrupted parts (such as yellowed petals / softened stems), spraying them with an antibacterial solution, packaging them in a special charity donation box, and automatically matching them with the recipient through the charity organization's demand database. If the score is <50, the flowers are marked as severely damaged and the following steps are performed: removing the petals / leaves, sending them to a flower shredder, mixing them with organic auxiliary materials for composting and fermentation, and generating a fertilizer composition test report. Finally, dynamic monitoring is implemented, namely scanning real-time corruption characteristics of good-quality inventory within each cycle, tracking the temperature and humidity of moderately defective flowers during transportation with GPS, and monitoring the temperature and pH value of severely damaged flowers during the fermentation process in real time.

[0142] Through this plan, a three-level classification mechanism is used to achieve a coordinated improvement in resource recycling and operational efficiency. Specifically, targeted discount sales of flowers with good appearance increase the resale rate of returned flowers and reduce the cost of purchasing new products. Public donations of flowers with moderate defects create social value. Fertilizer conversion of flowers with severe damage reduces the amount of solid waste handled. At the same time, classified processing increases warehouse turnover efficiency by several times, and automated grading greatly reduces the error rate of manual sorting, creating a dual benefit of maximizing resource value and optimizing operating costs.

[0143] In some embodiments, based on a flower freshness dataset, the flower species and flower freshness level of each flower are extracted; based on the flower species, a cluster analysis is performed on the returned flowers to generate a number of independent flower seed sets; within each independent flower seed set, the flowers are secondary grouped based on the flower freshness level to determine a good quality subset, a moderately defective subset, and a severely damaged subset within the current independent flower seed set, thereby generating an independent flower seed status set; based on different flower species and flower freshness levels, a preset flower seed preservation database is searched to determine the status of each subset in the current independent flower seed status set. Freshness-keeping time interval; according to the freshness-keeping time interval, perform corresponding independent operations on each subset in the independent flower species state set; independent operations include: for the good-quality subset, retain the original flower branches and generate a complete flower material module; for the moderately defective subset, remove the rotten parts and retain the usable parts to generate a degraded flower material module; for the severely damaged subset, peel off the petals and leaves to generate a fragmented material module; count the number of available flowers in each subset in the independent flower species state set, and generate a modular bouquet parameter set including flower type, flower freshness level, available flower quantity, and freshness-keeping time interval.

[0144] The independent flower seed set can be a group of flowers of the same type formed by clustering the returned flowers according to variety similarity.

[0145] The good-looking subset can be flowers with intact petals and no decay.

[0146] The moderate blemish subset can be flowers with localized decay (e.g. browning of petal edges).

[0147] The severe damage subset can be flowers with overall decay or structural damage.

[0148] The shelf life range can be the effective length of time that different flower varieties can maintain their sales or use value at a specific freshness level (such as "roses in good condition: 24-48 hours").

[0149] The complete flower material module can be a standardized flower material unit directly generated from a good-looking subset.

[0150] The downgraded flower material module can be a secondary flower material unit formed by removing the rotten parts of the moderately defective subset.

[0151] The debris material module can be recyclable materials such as petals and leaves peeled from the severely damaged subset.

[0152] Specifically, the traditional disposal model for returned flowers lacks classification accuracy. It simply relies on apparent characteristics for general grading and fails to distinguish between the characteristics of flower varieties and the details of repairable parts. As a result, flowers with reuse value are mistakenly judged and discarded, resulting in resource loss. In addition, there is a lack of time management, and the disposal strategy is not dynamically linked to the shelf life of flowers. There is a lack of adaptive mechanism for the decay of freshness over time, resulting in some returned flowers not reaching the corruption threshold. This step first uses a deep learning model to analyze flower images based on precise image recognition classification, identifying their variety, degree of openness, freshness, and specific defects (such as the location and degree of petal damage and root decay), achieving a more refined classification than traditional extensive grading. Secondly, a multi-dimensional value assessment model is constructed, integrating factors such as the market value of the variety, the state of flower openness, the proportion of usable parts (such as the proportion of healthy parts remaining after removing local decay), and the cost of defect repair to calculate the remaining economic value of each returned flower. Next, a dynamic freshness period is calculated, combining the identified variety characteristics, the current freshness level, and environmental parameters (such as temperature and humidity) to predict the remaining effective freshness window for each flower under specific storage conditions. Finally, a disposal strategy is intelligently generated. Based on the value assessment results, the remaining freshness period, and real-time market demand (such as specific variety promotions and donation channel matching), preset rules or decision tree algorithms are used to calculate the number of available flowers in each subset of the independent flower state set. This then generates a modular bouquet parameter set, including flower variety, flower freshness level, available flower quantity, and freshness period.

[0153] Through this solution, the good-quality subset, moderately defective subset, and severely damaged subset within the current independent flower seed set are determined, and an independent flower seed status set is generated. Through precise stratification of the three-level subsets, the vast majority of returned flowers are effectively utilized (sale of intact flowers, donation of downgraded flowers, and fertilizer conversion of fragmented materials), reducing the waste rate compared to traditional methods. By binding operational decisions on freshness-keeping time, the response speed of selling flowers with good quality is increased several times, greatly improving the punctuality rate of public welfare donations and significantly improving timeliness.

[0154] In some embodiments, the user's real-time bouquet customization demand set is analyzed to extract the flower types and corresponding quantity requirements requested by the user; based on the modular bouquet parameter set, a subset of available returned flower inventory that matches the flower types is retrieved in real time, and the available returned flower inventory subset includes a good-quality subset of the corresponding type and its available flower quantity and shelf life range; based on the available returned flower inventory subset, the total available quantity of all flower materials is determined; based on the total available quantity of all flower materials, it is judged whether the total available quantity meets the quantity requirements: if all requirements are met, a full return flower material pre-sale plan including the specific flower material quantity and discount price is generated; if part of the requirements are met, a new and old flower material combination plan including the ratio of new and old flower materials, the combined discount price and the maximum satisfactory quantity is generated; based on the full return flower material pre-sale plan and the new and old flower material combination plan, a flower pre-sale data set is dynamically generated.

[0155] The user's real-time bouquet customization requirement set may be a real-time bouquet requirement set submitted by the user through an online flower customization platform, including personalized requirements such as the required flower type, quantity, and color combination.

[0156] The available returned flower inventory subset may be an available inventory subset that is screened in real time from the modular bouquet parameter set and matches the flower types required by the user, and only includes flowers of "good appearance category".

[0157] The total available quantity may be the sum of the quantities of all matching flowers in the available returned flower inventory subset.

[0158] The fully-returnable flower pre-sale plan can be a sales plan that fully utilizes returned flowers to meet user needs, including the specific quantity of flowers and discounted prices.

[0159] The ratio of new and old flower materials can be the ratio of the number of returned flowers (old flower materials) to the number of unsold flowers (new flower materials) in the combination plan.

[0160] The combined discount price can be the total selling price of a mixed package of new and old flowers.

[0161] The maximum satisfyable quantity may be the maximum quantity of flowers that can satisfy the user's demand when only the currently available returned flower inventory is used.

[0162] A combination plan of new and old flowers can be a sales plan that mixes returned flowers with unsold flowers (new flowers), including the ratio of new and old flowers, the combined discount price, and the maximum quantity that can be met.

[0163] Specifically, the traditional flower return processing system has significant flaws. The problem lies in the lack of inventory dynamics: since the freshness of returned flowers decays rapidly over time, and the system fails to establish a real-time linkage mechanism between preservation time and inventory status, the sales plan pushed to users is out of touch with the actual available resources. This disconnect directly leads to rigid demand matching: when the returned inventory cannot fully meet user needs, the system can usually only rigidly suggest that users modify their needs or wait for replenishment, lacking a flexible mechanism that can partially meet user needs. This step first analyzes the user's real-time bouquet customization requirements to extract the flower types and corresponding quantity requirements. It then performs real-time inventory matching based on the modular bouquet parameter set. By filtering flowers in the "good condition" category and aggregating their quantities, a subset of available return flower inventory and the total available quantity are generated. Dynamic proposal generation is then performed based on the comparison between the total available quantity and the user's requested quantity. If the total available quantity is greater than or equal to the user's requested quantity, a fully returnable flower pre-sale proposal is directly generated, including a specific flower list, discount price, and delivery time limit. If the total available quantity is insufficient, the maximum available quantity and the ratio of new and old flowers are calculated. Combined with the flower benchmark price and a dynamic discount factor based on freshness level, a new and old flower combination proposal is generated, including the new and old flower ratio, combined discount price, and replenishment suggestions. Finally, the generated proposal is written into the flower pre-sale dataset for accurate push notification to users and, after receiving user feedback, can be invoked by downstream order processing processes.

[0164] Through this solution, available returned flowers marked as "good condition" are accurately matched to user customization needs, which significantly improves the reuse rate of such returned resources and avoids idle resources and waste. At the same time, the system dynamically generates a pre-sale plan for all returned flowers or a combination plan of new and old flowers according to inventory satisfaction, and pushes the proposed plan to users, effectively reducing order loss caused by insufficient inventory, thereby optimizing the overall order conversion rate. In addition, the generation of the combined discount price in the plan is strictly based on the base price of flowers and the discount coefficient dynamically determined according to the freshness level, ensuring the transparency of the pricing process and clear rules, significantly enhancing the credibility of the final quotation to users, and further consolidating the conversion effect.

[0165] In some embodiments, the maximum quantity of returned flowers and the ratio of new and old flowers are determined based on the available returned flower inventory subset and the user's real-time bouquet customization demand set; based on the types of unsold flowers and the types of returned flowers, the proportion of unsold flowers and the proportion of returned flowers in the new and old flower combination plan, the preset flower price database is retrieved to extract the flower benchmark price corresponding to each flower type in the new and old flower combination plan; based on the freshness level of flowers corresponding to the good-quality subset, the price discount coefficient of returned flowers is dynamically determined; the flower freshness level is positively correlated with the price discount coefficient, that is, the higher the flower freshness level, the larger the price discount coefficient and the smaller the price discount; based on the unsold flowers required to be used in the new and old flower combination plan, the price discount coefficient is determined dynamically. Based on the percentage of flowers sold and the benchmark price of flowers, the actual unit price of unsold flowers is generated; based on the percentage of returned flowers required to be used in the new and old flower material combination plan, the product of the benchmark price of flowers and the price discount coefficient is used as the actual unit price of returned flowers; based on the number of available flowers allocated to each flower type in the new and old flower material combination plan, the actual unit price of unsold flowers and the actual unit price of returned flowers, the final combined selling price of the corresponding type of flowers is determined; the final combined selling prices of all flower types in the new and old flower material combination plan are summarized to generate the combined price of new and old flowers, and the combined price of new and old flowers is used as the combined discount price; based on the proportion of new and old flowers, the combined discount price and the maximum quantity that can be satisfied, a new and old flower material combination plan is constructed.

[0166] The price discount factor may be a price discount rate set to reflect differences in freshness levels of returned flowers.

[0167] The flower benchmark price may be a standard selling price of unsold flowers in a preset price database.

[0168] The actual unit price of unsold flowers may be the real-time unit price of brand new flowers (non-returned flowers) in a user-customized bouquet, which is a base price without discounts.

[0169] The actual unit price of the returned flowers may be the discounted unit price of the returned flowers after secondary processing in the combination plan, which is lower than the price of new flowers.

[0170] The combined price of the new and old flowers can be the total price of the mixed bouquet that the user ultimately pays.

[0171] Specifically, in the secondary sales scenario of fresh flowers, traditional processing methods face severe challenges due to the three core contradictions of the extremely short freshness window of returned flowers (usually ≤72 hours), high price sensitivity of users, and dynamic two-way fluctuations in inventory and demand. If only relying on the full return plan, when inventory is insufficient, it will directly lead to loss of sales opportunities and waste of resources; and if a unified discount strategy is adopted, it will be impossible to balance the value protection of high-freshness flowers and the rapid turnover demand of low-freshness flowers. This step is based on a comparative analysis of the user's real-time bouquet customization demand set and the available subset of returned flower inventory. First, the maximum sufficiency quantity of returned flowers is calculated, and the ratio of new and old flowers is dynamically set based on the gap between the demand quantity and the maximum sufficiency quantity. The preset flower price database is then retrieved to obtain the benchmark price of each type of flower. A stepped price discount factor is generated based on the freshness level of the flowers in the good-quality subset (the higher the freshness level, the larger the discount factor). The actual unit price of unsold flowers (base price × 100%) and the actual unit price of returned flowers (base price × price discount factor) are then calculated. Finally, the quantity of each flower variety is allocated based on the new-to-old ratio, and the final combined selling price of each variety is generated based on the corresponding actual unit price. This is then aggregated to form a combined discounted price. The new-to-old flower ratio, combined discounted price, and maximum sufficiency quantity are then combined to create a complete new and old flower combination plan.

[0172] Through this solution, by dynamically generating a combination of new and old flower materials, the ratio of new and old flower materials is calculated in real time based on the maximum number of returned flowers, and the price discount coefficient is dynamically set based on the freshness level of the flowers. Finally, a stepped combination discount price is generated in combination with the benchmark price of flowers. On the one hand, by flexibly adjusting the ratio of new and old flower materials, the time window of limited inventory is maximized to avoid the corruption and loss of high-value flowers. On the other hand, scientific graded pricing is achieved through the price discount coefficient. While ensuring reasonable profits for high-freshness flowers, the turnover of medium and low-freshness inventory is accelerated, thereby solving the core pain points of inefficient supply and demand matching and the difficulty in balancing corporate profits and user price sensitivity.

[0173] In some embodiments, the user's address and the geographic location of the shipping location are obtained. With the shipping location as the center, an initial search radius is set based on the freshness interval of the good-quality subset. The initial search radius is dynamically adjusted based on the flower freshness level of each flower type in the good-quality subset to generate a dynamic search radius, wherein the lower the flower freshness level, the smaller the dynamic search radius. Within the range corresponding to the dynamic search radius, a priority user push set is generated based on the user's address. The priority user push set is filtered based on the freshness interval and estimated transportation time, wherein the estimated transportation time must be less than or equal to the lower limit of the freshness interval. If a matching user is found after the filtering, a matching user push set is generated. If the flower freshness level drops to a moderately defective category, a preset public welfare institution database is searched within the range corresponding to the dynamic search radius to generate a candidate public welfare institution set. The candidate public welfare institution set is filtered based on the freshness interval and estimated transportation time, wherein the estimated transportation time must be less than or equal to the lower limit of the freshness interval. If a matching public welfare institution is found after the filtering, a public welfare donation push set is generated.

[0174] The user address location can be the geographic coordinates (such as longitude and latitude) corresponding to the delivery address provided by the user when placing an order, which comes from the order management system.

[0175] The geographical location of the shipping place can be the geographical coordinates of the flower warehouse or processing center, which comes from the logistics management system.

[0176] The initial search radius may be an initial geographical range centered on the place of shipment and calculated based on the maximum transportation distance allowed by the freshness preservation period of the flowers.

[0177] The dynamic search radius may be a geographical search range that is adjusted in real time according to the freshness level of the flowers, where the lower the freshness, the smaller the radius.

[0178] The priority user push set may be a set of users within the dynamic search radius whose historical consumption preferences match the currently returned flowers.

[0179] Estimated shipping time can be the estimated logistics time from the shipping point to the target location.

[0180] The matching user push set can be a set of target users that can actually be reached after being screened for freshness.

[0181] The public welfare unit database can be a digital warehouse that stores the basic information of public welfare organizations in a structured manner, including key fields such as organizational attributes, receiving capacity, geographic coordinates, etc., to support the precise matching of public welfare donations.

[0182] The public welfare institution set may be an operational institution subset dynamically selected from the public welfare institution database based on real-time business needs.

[0183] The charity donation push set may be a set of charity organizations that meet the conditions for receiving flowers with moderate defects.

[0184] Specifically, there is an inherent conflict between the timeliness of flower spoilage and the efficiency of geographical matching. The traditional fixed-radius push mechanism ignores the freshness decay gradient, resulting in two major systemic failures: high-freshness flowers miss out on potential customers far away due to insufficient coverage radius, while low-freshness flowers suffer a surge in spoilage rate during transportation due to out-of-range delivery; at the same time, although moderately defective flowers have charitable donation value, it is difficult to achieve effective matching before spoilage due to the lag in manual coordination and the dispersion of institutional locations. This step first extracts the user's address from the order system and obtains the shipping location from the logistics system. With the shipping location as the center, the initial search radius (e.g., 2440m) is calculated based on the lower limit of the freshness range for the good-looking subset (e.g., 24 hours) and a preset average transportation speed (e.g., 60km / h). The freshness rating of each type of flower in the good-looking subset is then queried (e.g., 85 for roses, 70 for lilies). The radius is dynamically adjusted according to the preset mapping rule (lower freshness ratings result in smaller scaling). (e.g., rose radius = 2440m × 85% = 2074m, lily radius = 2440m × 70% = 1708m). , taking the minimum value as the dynamic search radius (1708m); then, within the dynamic search radius, search for users who have historically purchased similar flowers and are marked as "accepting discounted products" or "high repurchase rate" to generate a priority user push set; then, call the logistics interface to calculate the estimated transportation time from the place of shipment to each user, and only retain users with transportation time ≤ the lower limit of the freshness preservation period to form a matching user push set; finally, monitor the freshness of the flowers in real time. When the grade drops to moderate defects (such as <60 points), search the public welfare database within the dynamic search radius and calculate the transportation time. Then, filter out organizations with transportation time ≤ the lower limit of the freshness preservation period to generate a public welfare donation push set.

[0185] Through this solution, a dynamic search radius mechanism is used to achieve spatial adaptation of resource allocation, ensuring that high-grade flowers expand the scope of commercial conversion, and low-grade flowers lock adjacent areas to control risks; and establish automatic triggering rules for public welfare channels, that is, when the freshness drops to the donation threshold, it switches to searching the public welfare database, and combines hard time verification to eliminate invalid donations.

[0186] In some embodiments, a flower pre-sale data set is pushed to target users through a preset electronic communication channel; user feedback information is received, analyzed, and it is determined whether the user accepts the corresponding plan in the flower pre-sale data set. If accepted, the corresponding full refund flower pre-sale plan and new and old flower combination plan are extracted; based on the full refund flower pre-sale plan and new and old flower combination plan accepted by the user, a flower secondary sales order data set is generated based on the ratio of new and old flowers in the flower pre-sale data set, the combined discount price and the maximum satisfyable quantity; based on the flower secondary sales order data set, a flower processing log is generated and output.

[0187] The preset electronic communication channel may be a system-preset digital information push channel (such as APP messages, SMS, emails), which is used to deliver flower pre-sale plans to users in a targeted manner.

[0188] Target users can be specific user groups that receive push notifications, including priority user push notification sets (matched users) and public welfare donation push notification sets (matched public welfare organizations).

[0189] User feedback information can be the user's operational response to the push plan (such as acceptance / rejection), including the plan selection identifier and additional requirements.

[0190] The flower secondary sales order dataset can be standardized order data generated after the user confirms the plan, including the plan type, flower material details, quantity, price, and timeliness mark.

[0191] Flower processing logs can be operational audit files that record the entire order execution process, including timestamps, user IDs, status changes, and exception events.

[0192] Specifically, returning flowers is subject to strict time constraints (short shelf life), and traditional manual delivery methods make it difficult to accurately match demand within a limited time, resulting in high-value flowers being wasted due to expiration. At the same time, secondary sales involve complex rules such as discount strategies and mixing of new and old flowers, and manual operations are prone to order record errors and compliance risks. This step first pushes sales proposals from the flower pre-sale dataset to target users based on priority (priority users receive priority) through pre-set electronic communication channels (such as SMS gateway APIs). Users then submit feedback, including an action identifier and additional requirements, by replying to a link or button. When a user accepts the proposal, the system dynamically generates order data based on the proposal type. For a fully refundable flower pre-sale proposal, the system extracts the corresponding flower's shelf life and available quantity from the modular bouquet parameter set, combining this with the discount price to generate an order entry. For a combination of new and old flowers, a mixed order entry is generated based on the ratio of new and old flowers, the combined discount price, and the actual quantity used. This is ultimately packaged into a flower secondary sales order dataset with a unique order ID. Simultaneously, the system automatically records key events such as the push time, user ID, proposal acceptance type, flower outbound operations, and logistics node status, generating a structured flower processing log that is output to the merchant. If the user rejects the proposal or the shelf life is approaching, a rematching process is triggered or an early warning event is recorded to notify manual intervention.

[0193] Through this solution, electronic communication channels are used to accurately reach target users, so that returned flowers can be quickly matched with demand within the freshness window, improving user conversion rate and satisfaction, and significantly reducing the loss rate of high-value flowers. Structured order data is automatically generated to avoid order recording errors and compliance risks caused by manual operations. At the same time, the full-process operation log ensures the auditability and abnormal warning capabilities of the business closed loop. Ultimately, the utilization rate of returned flower resources is maximized, the dynamic response efficiency of the supply chain is improved, and the coordinated optimization of secondary sales compliance management is achieved.

[0194] Figure 3 A schematic diagram of a structure of an order management system based on big data provided by an embodiment of the present application is shown as follows: Figure 3 As shown, an order management system 300 based on big data in this embodiment includes: a flower analysis module 301, a structure splitting module 302, a customization and pre-sale module 303, and an interaction and order module 304;

[0195] Flower analysis module 301 is used to obtain a returned flower parameter set, and based on the returned flower parameter set, collect and analyze flower status data in real time through a multimodal data fusion strategy to determine the flower freshness level;

[0196] a structure splitting module 302 for extracting flower types according to the returned flower parameter set, performing structure splitting on the returned flowers based on the flower freshness dataset and the flower types, and generating a modular bouquet parameter set;

[0197] The customization and pre-sale module 303 is used to obtain a user's real-time bouquet customization requirement set and dynamically generate a flower pre-sale data set based on the user's real-time bouquet customization requirement set and the modular bouquet parameter set;

[0198] The interaction and order module 304 is used to push the flower pre-sale data set to the user, receive user feedback information, generate a flower secondary sale order data set, and output a flower processing log.

[0199] Optionally, the flower analysis module 301 collects and analyzes flower status data in real time based on the returned flower parameter set through a multimodal data fusion strategy to determine a flower freshness data set, specifically for:

[0200] The returned flower parameter set includes returned flower image data, environmental temperature and humidity data, packaging status information, and transportation time data;

[0201] Extracting visual corruption features based on the returned flower image data using a deep vision method;

[0202] The visual corruption features include the proportion of browning pixels at the edge of petals and the deviation coefficient of petal stretch;

[0203] Based on the environmental temperature and humidity data, predicting the water loss rate through a time series prediction method;

[0204] generating a spoilage acceleration factor through regression analysis based on the packaging status information and the transportation time data, wherein the transportation time data is negatively correlated with the flower freshness grade, i.e., the longer the transportation time, the lower the flower freshness grade;

[0205] Based on a multimodal data fusion strategy, the flower freshness is subjected to correlation analysis and comprehensive evaluation according to the visual corruption characteristics, the water loss rate, the corruption acceleration factor, and the transportation time data to determine the flower freshness dataset.

[0206] Optionally, the flower analysis module 301, when based on the multimodal data fusion strategy, is specifically configured to:

[0207] Performing normalization and quantification processing based on the visual corruption characteristics, the water loss rate, and the corruption acceleration factor to generate a standardized corruption index set;

[0208] Based on a preset corruption association model, dynamically assigning weight coefficients to the indicators in the standardized corruption indicator set according to the types of flowers;

[0209] Performing a weighted fusion calculation on the standardized corruption indicator set according to the assigned weight coefficient to generate an initial freshness score;

[0210] Generating a dynamic adjustment factor based on the transportation time data and the flower type, correcting the initial freshness score through a nonlinear mapping relationship, and classifying the freshness level of each type of flower;

[0211] The flower freshness dataset is generated according to the flower freshness level of each type.

[0212] Optionally, the structure splitting module 302 is specifically configured to:

[0213] The "good condition" category refers to flowers in good condition, whose freshness level meets the preset resale threshold. Flowers of this level are clearly marked and used for discount promotions;

[0214] The medium defect category refers to flowers that have shown some signs of decay. The freshness level of the flowers is lower than the preset resale threshold but meets the preset reuse threshold. Flowers of this level are donated to public welfare.

[0215] The severely damaged category refers to flowers that are severely rotten or damaged, and the freshness level of the flowers is lower than a preset reusable threshold. Flowers of this level are used to be converted into organic fertilizers.

[0216] Optionally, when extracting flower types based on the returned flower parameter set, performing structural splitting on the returned flowers based on the flower freshness dataset and the flower types, and generating a modular bouquet parameter set, the structure splitting module 302 is specifically configured to:

[0217] Extracting the flower type and the flower freshness level of each flower based on the flower freshness dataset;

[0218] Performing cluster analysis on the returned flowers according to the flower species to generate a number of independent flower species sets;

[0219] In each of the independent flower seed sets, the flower seeds are secondary grouped according to the flower freshness level to determine a good quality subset, a moderately defective subset, and a severely damaged subset within the current independent flower seed set, thereby generating an independent flower seed status set;

[0220] According to the different flower types and the flower freshness levels, searching a preset flower seed preservation database to determine the preservation time interval of each subset of the current independent flower seed status set;

[0221] According to the fresh-keeping time interval, performing corresponding independent operations on each subset of the independent flower seed status set;

[0222] The independent operation includes:

[0223] For the good-looking subset, retain the original flower branches and generate a complete flower material module;

[0224] For the moderate defect subset, the rotten parts are removed and the usable parts are retained to generate a downgraded flower material module;

[0225] For the severely damaged subset, petals and leaves are peeled off to generate a fragment material module;

[0226] The number of available flowers in each subset of the independent flower species status set is counted to generate the modular bouquet parameter set including the flower species, the flower freshness level, the available flower quantity, and the freshness preservation time interval.

[0227] Optionally, the customization and pre-sale module 303 dynamically generates a flower pre-sale data set based on the user's real-time bouquet customization requirement set and the modular bouquet parameter set, specifically for:

[0228] Analyze the user's real-time bouquet customization requirement set and extract the flower types and corresponding quantity requirements requested by the user;

[0229] Based on the modular bouquet parameter set, a subset of available returned flower inventory that matches the flower type is retrieved in real time, the subset of available returned flower inventory including the good-quality subset of the corresponding type, the available flower quantity, and the freshness-keeping time interval;

[0230] determining the total available quantity of all flowers based on the available returned flower inventory subset;

[0231] According to the total available quantity of all flowers, determine whether the total available quantity meets the quantity requirement:

[0232] If all conditions are met, a full refund pre-sale plan for flowers will be generated, including the specific quantity of flowers and discounted prices;

[0233] If the requirements are partially met, a combination plan of new and old flowers will be generated, including the ratio of new and old flowers, combined discount price and the maximum number of flowers that can be met.

[0234] The flower pre-sale data set is dynamically generated according to the full return flower pre-sale plan and the new and old flower combination plan.

[0235] Optionally, when generating a new and old flower material combination plan including a ratio of new and old flower materials, a combined discount price, and a maximum quantity that can be satisfied based on the partial satisfaction, the customization and pre-sale module 303 is specifically configured to:

[0236] Determining the maximum quantity of returned flowers that can be satisfied and the ratio of new and old flowers based on the available returned flower inventory subset and the user's real-time bouquet customization requirements;

[0237] According to the types of unsold flowers and the types of returned flowers, the proportion of unsold flowers and the proportion of returned flowers in the new and old flower material combination plan, a preset flower price database is searched to extract the flower benchmark price corresponding to each flower type in the new and old flower material combination plan;

[0238] Dynamically determining a price discount coefficient for the returned flowers based on the flower freshness level corresponding to the good-quality subset;

[0239] The flower freshness grade is positively correlated with the price discount coefficient, that is, the higher the flower freshness grade, the greater the price discount coefficient and the smaller the price discount;

[0240] Based on the proportion of unsold flowers required to be used in the new and old flower material combination plan and on the basis of the flower benchmark price, an actual unit price of unsold flowers is generated;

[0241] Based on the proportion of returned flowers required to be used in the new and old flower material combination plan, the product of the base price of flowers and the price discount coefficient is used as the actual unit price of the returned flowers;

[0242] Determine the final combined selling price of the corresponding type of flowers based on the number of available flowers allocated for each type of flowers in the new and old flower combination plan, the actual unit price of the unsold flowers, and the actual unit price of the returned flowers;

[0243] Summarizing the final combined selling prices of all the flower types in the new and old flower material combination scheme to generate a new and old flower material combination price, and using the new and old flower material combination price as the combined discounted price;

[0244] The new and old flower material combination plan is constructed based on the ratio of the new and old flower materials, the combined discount price and the maximum satisfying quantity.

[0245] Optionally, the interaction and order module 304 is specifically configured to:

[0246] Pushing the flower pre-sale dataset to target users via a preset electronic communication channel;

[0247] Receive the user feedback information, analyze the user feedback information, determine whether the user accepts the corresponding plan in the flower pre-sale data set, and if the user accepts, extract the corresponding full return flower pre-sale plan and the new and old flower combination plan;

[0248] Generate a flower secondary sales order dataset based on the fully refunded flower pre-sale plan and the new and old flower combination plan accepted by the user, and based on the new and old flower ratio, the combined discount price, and the maximum available quantity in the flower pre-sale dataset;

[0249] The flower processing log is generated and outputted according to the flower secondary sales order data set.

[0250] Optionally, the system 300 further includes a range selection module 305, specifically configured to:

[0251] Obtain the user's address and the geographical location of the shipping location, and set an initial search radius based on the freshness-limiting time interval of the good-appearance subset with the shipping location as the center;

[0252] dynamically adjusting the initial search radius according to the flower freshness level of each of the flower types in the good-quality subset to generate a dynamic search radius, wherein the lower the flower freshness level, the smaller the dynamic search radius;

[0253] Generate a priority user push set based on the user's address within the range corresponding to the dynamic search radius;

[0254] Based on the freshness aging interval and the estimated transportation time, the priority user push set is screened, wherein the estimated transportation time must be less than or equal to the lower limit of the freshness aging interval. If a matching user exists after screening, a matching user push set is generated;

[0255] If the flower freshness level drops to the moderate defect category, searching the preset public welfare unit database within the range corresponding to the dynamic search radius to generate a set of candidate public welfare units;

[0256] Screening the candidate public welfare units based on the freshness-keeping time interval and the estimated transportation time, wherein the estimated transportation time must be less than or equal to the lower limit of the freshness-keeping time interval;

[0257] If there is a matching public welfare organization after screening, a public welfare donation push set will be generated.

Claims

1. An order management method based on big data, characterized in that: include: Obtaining a returned flower parameter set, and based on the returned flower parameter set, collecting and analyzing flower status data in real time through a multimodal data fusion strategy to determine a flower freshness dataset; Extracting flower types according to the returned flower parameter set, structurally splitting the returned flowers based on the flower freshness dataset and the flower types, and generating a modular bouquet parameter set; Obtaining a user's real-time bouquet customization requirement set, and dynamically generating a flower pre-sale data set based on the user's real-time bouquet customization requirement set and the modular bouquet parameter set; Push the flower pre-sale data set to the user, receive user feedback information, generate a flower secondary sales order data set, and output a flower processing log; The method of collecting and analyzing flower status data in real time based on the returned flower parameter set through a multimodal data fusion strategy to determine a flower freshness data set includes: The returned flower parameter set includes returned flower image data, environmental temperature and humidity data, packaging status information, and transportation time data; Extracting visual corruption features based on the returned flower image data using a deep vision method; The visual corruption features include the proportion of browning pixels at the edge of petals and the deviation coefficient of petal stretch; Based on the environmental temperature and humidity data, predicting the water loss rate through a time series prediction method; generating a spoilage acceleration factor through regression analysis based on the packaging status information and the transportation time data, wherein the transportation time data is negatively correlated with the flower freshness grade, i.e., the longer the transportation time, the lower the flower freshness grade; Based on a multimodal data fusion strategy, correlation analysis and comprehensive evaluation of flower freshness are performed according to the visual spoilage characteristics, the water loss rate, the spoilage acceleration factor, and the transportation time data to determine the flower freshness dataset; The multimodal data fusion strategy includes: Performing normalization and quantification processing based on the visual corruption characteristics, the water loss rate, and the corruption acceleration factor to generate a standardized corruption index set; Based on a preset corruption association model, dynamically assigning weight coefficients to the indicators in the standardized corruption indicator set according to the types of flowers; Performing a weighted fusion calculation on the standardized corruption indicator set according to the assigned weight coefficient to generate an initial freshness score; Generating a dynamic adjustment factor based on the transportation time data and the flower type, correcting the initial freshness score through a nonlinear mapping relationship, and classifying the flower freshness level of each type; The flower freshness dataset is generated according to the flower freshness level of each type.

2. The method according to claim 1, characterized in that The flower freshness grades include good quality, moderately defective, and severely damaged; The "good condition" category refers to flowers in good condition, whose freshness level meets the preset resale threshold. Flowers of this level are clearly marked and used for discount promotions; The medium defect category refers to flowers that have shown some signs of decay. The freshness level of the flowers is lower than the preset resale threshold but meets the preset reuse threshold. Flowers of this level are donated to public welfare. The severely damaged category refers to flowers that are severely rotten or damaged, and the freshness level of the flowers is lower than a preset reusable threshold. Flowers of this level are used to be converted into organic fertilizers.

3. The method according to claim 1, characterized in that Extracting flower types according to the returned flower parameter set, structurally splitting the returned flowers based on the flower freshness dataset and the flower types, and generating a modular bouquet parameter set, including: Extracting the flower type and the flower freshness level of each flower based on the flower freshness dataset; Performing cluster analysis on the returned flowers according to the flower species to generate a number of independent flower species sets; In each of the independent flower seed sets, the flower seeds are secondary grouped according to the flower freshness level to determine a good quality subset, a moderately defective subset, and a severely damaged subset within the current independent flower seed set, thereby generating an independent flower seed status set; According to the different flower types and the flower freshness levels, searching a preset flower seed preservation database to determine the preservation time interval of each subset of the current independent flower seed status set; According to the fresh-keeping time interval, performing corresponding independent operations on each subset of the independent flower seed status set; The independent operation includes: For the good-looking subset, retain the original flower branches and generate a complete flower material module; For the moderate defect subset, the rotten parts are removed and the usable parts are retained to generate a downgraded flower material module; For the severely damaged subset, petals and leaves are peeled off to generate a fragment material module; The number of available flowers in each subset of the independent flower species status set is counted to generate the modular bouquet parameter set including the flower species, the flower freshness level, the available flower quantity, and the freshness preservation time interval.

4. The method according to claim 3, characterized in that The step of dynamically generating a flower pre-sale data set based on the user's real-time bouquet customization requirement set and the modular bouquet parameter set includes: Analyze the user's real-time bouquet customization requirement set and extract the flower types and corresponding quantity requirements requested by the user; Based on the modular bouquet parameter set, a subset of available returned flower inventory that matches the flower type is retrieved in real time, the subset of available returned flower inventory including the good-quality subset of the corresponding type, the available flower quantity, and the freshness-keeping time interval; determining the total available quantity of all flowers based on the available returned flower inventory subset; According to the total available quantity of all flowers, determine whether the total available quantity meets the quantity requirement: If all conditions are met, a full refund pre-sale plan for flowers will be generated, including the specific quantity of flowers and discounted prices; If the requirements are partially met, a combination plan of new and old flowers will be generated, including the ratio of new and old flowers, combined discount price and the maximum number of flowers that can be met. The flower pre-sale data set is dynamically generated according to the full return flower pre-sale plan and the new and old flower combination plan.

5. The method according to claim 4, characterized in that If the requirements are partially met, a new and old flower material combination plan is generated, including the ratio of new and old flower materials, the combined discount price and the maximum quantity that can be met, including: Determining the maximum quantity of returned flowers that can be satisfied and the ratio of new and old flowers based on the available returned flower inventory subset and the user's real-time bouquet customization requirements; According to the types of unsold flowers and the types of returned flowers, the proportion of unsold flowers and the proportion of returned flowers in the new and old flower material combination plan, a preset flower price database is searched to extract the flower benchmark price corresponding to each flower type in the new and old flower material combination plan; Dynamically determining a price discount coefficient for the returned flowers based on the flower freshness level corresponding to the good-quality subset; The flower freshness grade is positively correlated with the price discount coefficient, that is, the higher the flower freshness grade, the greater the price discount coefficient and the smaller the price discount; Based on the proportion of unsold flowers required to be used in the new and old flower material combination plan and on the basis of the flower benchmark price, an actual unit price of unsold flowers is generated; Based on the proportion of returned flowers required to be used in the new and old flower material combination plan, the product of the base price of flowers and the price discount coefficient is used as the actual unit price of the returned flowers; Determine the final combined selling price of the corresponding type of flowers based on the number of available flowers allocated for each type of flowers in the new and old flower combination plan, the actual unit price of the unsold flowers, and the actual unit price of the returned flowers; Summarizing the final combined selling prices of all the flower types in the new and old flower material combination scheme to generate a new and old flower material combination price, and using the new and old flower material combination price as the combined discounted price; The new and old flower material combination plan is constructed based on the ratio of the new and old flower materials, the combined discount price and the maximum satisfying quantity.

6. The method according to claim 3, characterized in that The method further comprises: Obtain the user's address and the geographical location of the shipping location, and set an initial search radius based on the freshness-limiting time interval of the good-appearance subset with the shipping location as the center; dynamically adjusting the initial search radius according to the flower freshness level of each of the flower types in the good-quality subset to generate a dynamic search radius, wherein the lower the flower freshness level, the smaller the dynamic search radius; Generate a priority user push set based on the user's address within the range corresponding to the dynamic search radius; Based on the freshness aging interval and the estimated transportation time, the priority user push set is screened, wherein the estimated transportation time must be less than or equal to the lower limit of the freshness aging interval. If a matching user exists after screening, a matching user push set is generated; If the flower freshness level drops to the moderate defect category, searching the preset public welfare unit database within the range corresponding to the dynamic search radius to generate a set of candidate public welfare units; Screening the candidate public welfare units based on the freshness-keeping time interval and the estimated transportation time, wherein the estimated transportation time must be less than or equal to the lower limit of the freshness-keeping time interval; If there is a matching public welfare organization after screening, a public welfare donation push set will be generated.

7. The method according to claim 4, characterized in that The step of pushing the flower pre-sale data set to the user, receiving user feedback information, generating a flower secondary sale order data set, and outputting a flower processing log includes: Pushing the flower pre-sale dataset to target users via a preset electronic communication channel; Receive the user feedback information, analyze the user feedback information, determine whether the user accepts the corresponding plan in the flower pre-sale data set, and if the user accepts, extract the corresponding full return flower pre-sale plan and the new and old flower combination plan; Generate a flower secondary sales order dataset based on the fully refunded flower pre-sale plan and the new and old flower combination plan accepted by the user, and based on the new and old flower ratio, the combined discount price, and the maximum available quantity in the flower pre-sale dataset; The flower processing log is generated and outputted according to the flower secondary sales order data set.

8. An order management system based on big data, characterized in that: include: A flower analysis module is used to obtain a returned flower parameter set, and based on the returned flower parameter set, collect and analyze flower status data in real time through a multimodal data fusion strategy to determine a flower freshness data set; a structure splitting module, configured to extract flower types according to the returned flower parameter set, perform structure splitting on the returned flowers based on the flower freshness dataset and the flower types, and generate a modular bouquet parameter set; The customization and pre-sale module is used to obtain the user's real-time bouquet customization requirement set and dynamically generate a flower pre-sale data set based on the user's real-time bouquet customization requirement set and the modular bouquet parameter set; An interaction and order module, configured to push the flower pre-sale data set to users, receive user feedback, generate a flower secondary sale order data set, and output a flower processing log; The flower analysis module is specifically used for: The returned flower parameter set includes returned flower image data, environmental temperature and humidity data, packaging status information, and transportation time data; Extracting visual corruption features based on the returned flower image data using a deep vision method; The visual corruption features include the proportion of browning pixels at the edge of petals and the deviation coefficient of petal stretch; Based on the environmental temperature and humidity data, predicting the water loss rate through a time series prediction method; generating a spoilage acceleration factor through regression analysis based on the packaging status information and the transportation time data, wherein the transportation time data is negatively correlated with the flower freshness grade, i.e., the longer the transportation time, the lower the flower freshness grade; Based on a multimodal data fusion strategy, correlation analysis and comprehensive evaluation of flower freshness are performed according to the visual spoilage characteristics, the water loss rate, the spoilage acceleration factor, and the transportation time data to determine the flower freshness dataset; The multimodal data fusion strategy in the flower analysis module is specifically used to: Performing normalization and quantification processing based on the visual corruption characteristics, the water loss rate, and the corruption acceleration factor to generate a standardized corruption index set; Based on a preset corruption association model, dynamically assigning weight coefficients to the indicators in the standardized corruption indicator set according to the types of flowers; Performing a weighted fusion calculation on the standardized corruption indicator set according to the assigned weight coefficient to generate an initial freshness score; Generating a dynamic adjustment factor based on the transportation time data and the flower type, correcting the initial freshness score through a nonlinear mapping relationship, and classifying the flower freshness level of each type; The flower freshness dataset is generated according to the flower freshness level of each type.

Images