Fabric evaluation decision-making method, device and equipment and storage medium

Abstract

The invention relates to the technical field of intelligent fabric decision, in particular to a fabric evaluation decision method, device and equipment and a storage medium. Performing attenuation analysis on the market performance data and the clothing demand data to obtain attenuation factors; calculating a preset reference time, an attenuation factor, the current sales data and the current time to obtain a demand attenuation coefficient; predicting the social media data and the e-commerce platform key event according to the demand attenuation coefficient, a preset demand prediction model and a preset association rule mining algorithm to obtain a stock quantity decision and an optimal stock quantity; generating a fabric evaluation decision according to the stock quantity decision and the optimal stock quantity; through generation of stock and fabric decisions, stock optimization, stockout and overstock reduction and agile response of a supply chain are realized, and the competitiveness and profit level of a clothing enterprise are remarkably improved.

Description

Technical Field

[0001] The present invention relates to the field of intelligent fabric decision-making technology, and in particular to a fabric evaluation and decision-making method, device, equipment and storage medium. Background Art

[0002] In existing technology, apparel companies generally formulate procurement plans based on static historical sales and fixed seasonal coefficients. The system summarizes POS sales on a monthly basis and applies a linear extrapolation model to derive demand for the next four weeks. The results are then directly mapped to fabric purchase orders. This method does not quantify the natural decline in sales over time, nor does it have a mechanism to immediately incorporate external events such as social media popularity, live broadcast schedules, and platform promotions into the forecasting process, resulting in a lag in the model's response to real market demand. The inventory strategy uses a simple safety stock multiple, which cannot distinguish between popular items and long-tail categories. Problems such as delayed replenishment of popular items and backlogs of slow-moving items are common. The entire process relies on manual Excel processing with inconsistent data calibers, and fabric evaluation is based solely on historical unit consumption. As a result, the accuracy of fabric procurement is not high enough, becoming a core bottleneck restricting the efficiency of the quick response supply chain. Summary of the Invention

[0003] In order to overcome the deficiencies of the prior art, the present invention aims to provide a fabric evaluation decision method, system, device and storage medium.

[0004] A fabric evaluation decision method comprises: obtaining market performance data and clothing demand data; performing attenuation analysis on the market performance data and clothing demand data to obtain an attenuation factor; obtaining current sales data and current time; calculating a preset benchmark time, an attenuation factor, current sales data, and current time to obtain a demand attenuation coefficient; obtaining social media data and key events on an e-commerce platform; predicting the social media data and key events on the e-commerce platform based on the demand attenuation coefficient, a preset demand forecasting model, and a preset association rule mining algorithm to obtain a stocking quantity decision and an optimal inventory quantity; and generating a fabric evaluation decision based on the stocking quantity decision and the optimal inventory quantity.

[0005] Furthermore, the prediction of social media data and key events on the e-commerce platform based on the demand attenuation coefficient, the preset demand forecasting model, and the preset association rule mining algorithm to obtain stocking quantity decisions and optimal inventory levels includes:

[0006] Social media data is analyzed for popularity based on a preset pruning exact linear time algorithm to obtain popular events. An optimized event association network is constructed based on an association rule mining algorithm, key events on the e-commerce platform, and popular events. The optimized event association network is predicted based on a demand forecasting model, a hierarchical analysis method, and a demand attenuation coefficient to determine stocking quantity decisions and optimal inventory levels.

[0007] Furthermore, the method of performing heat analysis on social media data according to a preset pruned exact linear time algorithm to obtain heat events includes: performing heat testing on social media data according to the pruned exact linear time algorithm and a preset standard deviation difference condition and a preset standard deviation range condition to obtain a heat testing result; when the heat testing result is that the volatility of the sequence after the backtracking change point is greater than the volatility of the sequence before the backtracking change point and meets the standard deviation range condition, a heat surge event is generated; a heat event is generated according to the heat surge event; when the heat testing result is that the volatility of the sequence after the backtracking change point is less than the volatility of the sequence before the backtracking change point and meets the standard deviation range condition, a heat drop event is generated; and a heat event is generated according to the heat drop event.

[0008] Furthermore, the heat test of social media data is performed according to the pruned exact linear time algorithm and preset standard deviation difference conditions and preset standard deviation range conditions to obtain a heat test result, including: iteratively analyzing the social media data according to a preset time series starting point and a preset time window to obtain an optimal backtracking point; performing backtracking analysis on the social media data according to the pruned exact linear time algorithm and the optimal backtracking point to obtain a backtracking change point sequence; performing heat test on the backtracking change point sequence according to the standard deviation difference condition and the standard deviation range condition to obtain a heat test result.

[0009] Furthermore, the optimized event association network is constructed based on the association rule mining algorithm, key events of the e-commerce platform and hot events, including: classifying the key events of the e-commerce platform to obtain multiple classified events; performing association analysis on the multiple classified events and hot events according to the association rule mining algorithm to obtain association analysis results; constructing an event association network based on the association analysis results and preset association rules; performing false verification on the event association network according to preset independence test conditions to obtain false verification results; and filtering the event association network according to the false verification results to obtain an optimized event association network.

[0010] Furthermore, the method predicts the optimized event association network according to the demand forecasting model, the hierarchical analysis method and the demand attenuation coefficient to obtain the stocking quantity decision and the optimal inventory quantity, including: performing a hierarchical analysis on the optimized event association network according to the hierarchical analysis method and the demand attenuation coefficient to obtain the clothing sentiment tendency score weight, the cross-platform propagation speed score weight and the coverage score weight; predicting the clothing sentiment tendency score weight, the cross-platform propagation speed score weight and the coverage score weight according to the demand forecasting model and the preset stocking cycle to obtain the predicted sales volume and stocking quantity decision; calculating the predicted sales volume, the preset safety factor, the preset in-transit inventory and the preset replenishment capacity to obtain the optimal inventory quantity.

[0011] Furthermore, the method of generating a fabric evaluation decision based on the stocking quantity decision and the optimal inventory quantity includes: generating fabric demand data and clothing style data based on the optimal inventory quantity and the stocking quantity decision; performing feature extraction on the fabric demand data and clothing style data to obtain fabric texture features and clothing area features; performing style migration on the clothing area features based on a preset style migration model and fabric texture features to obtain a target predicted style; rendering the target predicted style based on a preset texture alignment error to obtain an initial rendering; performing pre-inference on the initial rendering based on a preset virtual model posture library to obtain clothing fit; adjusting the initial rendering based on the clothing fit to obtain a final rendering; and generating a fabric evaluation decision based on the final rendering.

[0012] Furthermore, a fabric evaluation and decision-making device includes: a first data acquisition module for acquiring market performance data and clothing demand data; a decay analysis module for performing decay analysis on the market performance data and clothing demand data to obtain a decay factor; a second data acquisition module for acquiring current sales data and current time; a demand decay calculation module for calculating a preset benchmark time, a decay factor, current sales data and current time to obtain a demand decay coefficient; a third data acquisition module for acquiring social media data and key events of an e-commerce platform; a prediction module for predicting social media data and key events of an e-commerce platform based on the demand decay coefficient, a preset demand prediction model and a preset association rule mining algorithm to obtain a stocking quantity decision and an optimal inventory quantity; and a fabric evaluation and decision module for generating a fabric evaluation decision based on the stocking quantity decision and the optimal inventory quantity.

[0013] Furthermore, a fabric evaluation and decision-making device is provided, comprising: a memory and at least one processor, wherein the memory stores instructions; at least one of the processors calls the instructions in the memory to enable the computer device to execute each step of any one of the above-mentioned fabric evaluation and decision-making methods.

[0014] Furthermore, a computer-readable storage medium is provided, wherein instructions are stored on the computer-readable storage medium, and when the instructions are executed by a processor, each step of any one of the above-mentioned fabric evaluation and decision-making methods is implemented.

[0015] In the technical solution of the present invention, by acquiring market performance data and clothing demand data, a panoramic insight into external dynamics and internal characteristics is achieved; through attenuation analysis, an attenuation factor reflecting the natural attenuation rate of sales is obtained, and the demand attenuation coefficient is calculated by combining the benchmark time, current sales data and time, and the real demand intensity at different stages is dynamically quantified; based on this coefficient, a time series forecasting model and an association rule mining algorithm are used to integrate social media data and key e-commerce events for prediction, generate stocking quantity decisions and optimal inventory quantities, and finally output fabric evaluation decisions; this solution realizes full-link data connectivity, enhances the timeliness and accuracy of demand perception, effectively balances the relationship between supply and demand, reduces inventory backlogs and out-of-stock risks, provides reliable data-driven support for supply chain optimization of clothing companies, and improves market response efficiency and competitiveness. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] The above and / or additional aspects and advantages of the present invention will become apparent and readily understood from the following description of the embodiments with reference to the accompanying drawings, in which:

[0017] Figure 1 A first flow chart of a fabric evaluation and decision-making method provided by an embodiment of the present invention;

[0018] Figure 2 A second flow chart of a fabric evaluation and decision-making method provided by an embodiment of the present invention;

[0019] Figure 3 A third flow chart of a fabric evaluation and decision-making method provided by an embodiment of the present invention;

[0020] Figure 4 A fourth flow chart of a fabric evaluation and decision-making method provided by an embodiment of the present invention;

[0021] Figure 5 A fifth flow chart of a fabric evaluation and decision-making method provided by an embodiment of the present invention;

[0022] Figure 6 A sixth flow chart of a fabric evaluation and decision-making method provided by an embodiment of the present invention;

[0023] Figure 7 A seventh flow chart of a fabric evaluation and decision-making method provided by an embodiment of the present invention;

[0024] Figure 8 A schematic structural diagram of a fabric evaluation and decision-making device provided by an embodiment of the present invention;

[0025] Figure 9 A schematic diagram of the structure of a fabric evaluation and decision-making device provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0026] The present invention provides a fabric assessment decision-making method, system, device and storage medium. By integrating operator behavior data and equipment status data, and combining segmentation analysis and multi-level verification, it achieves a comprehensive and accurate assessment of equipment operation compliance, improves the compliance analysis effect of equipment operation, and reduces the operational risks and safety hazards caused by traditional compliance analysis methods.

[0027] The terms "first," "second," "third," "fourth," and so on (if any) in the description and claims of the present invention and in the accompanying drawings are used to distinguish similar objects and are not necessarily used to describe a particular order or precedence. It should be understood that the terms used in this manner are interchangeable where appropriate, so that the embodiments described herein can be implemented in an order other than that shown or described herein. In addition, the terms "including" or "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product, or apparatus that includes a series of steps or elements is not necessarily limited to those steps or elements expressly listed, but may include other steps or elements not expressly listed or inherent to such process, method, product, or apparatus.

[0028] For ease of understanding, the specific process of the embodiment of the present invention is described below. Figure 1 In one embodiment of the present invention, a fabric evaluation and decision-making method includes:

[0029] 101. Obtain market performance data and clothing demand data;

[0030] In this embodiment, market performance data and apparel demand data are obtained through the Zhixiaobu app, connected to the enterprise ERP (Enterprise Resource Planning), POS (Point of Sale) and e-commerce backend systems. Market performance data is used to reflect external market dynamics and apparel competitiveness, and includes sales performance data, market competition data, and user feedback data. Apparel demand data is used to focus on the demand characteristics and life cycle of apparel itself. Apparel demand data includes basic attribute data (product information: SKU code, category (tops / pants / dresses), style (casual / commuting / sports), fabric material, price range; life cycle stage: new product launch period, growth period, maturity period, decline period).

[0031] 102. Conduct attenuation analysis on market performance data and clothing demand data to obtain attenuation factors;

[0032] In this example, the decay factor reflects the rate at which clothing sales naturally decay over time;

[0033] 103. Get current sales data and current time;

[0034] 104. Calculate the preset base time, attenuation factor, current sales data, and current time to obtain a demand attenuation coefficient;

[0035] In this embodiment, the calculation formula of the demand attenuation coefficient is as follows:

[0036] , Dynamically quantify the real demand intensity at different sales stages through the demand attenuation coefficient;

[0037] 105. Obtain social media data and key events of e-commerce platforms;

[0038] In this embodiment, social media data (such as popularity of outfit notes on Xiaohongshu, discussion volume of Douyin topics, etc.) and key events of e-commerce platforms (such as live broadcast schedules, 618 promotion nodes, etc.);

[0039] 106. Predict key events on social media data and e-commerce platforms based on the demand attenuation coefficient, the preset demand forecasting model, and the preset association rule mining algorithm to determine stocking quantity decisions and optimal inventory levels;

[0040] In this embodiment, the demand forecasting model uses a time series forecasting algorithm (such as LSTM);

[0041] 107. Generate fabric evaluation decision based on stocking quantity decision and optimal inventory quantity;

[0042] In this embodiment, relying on the Zhixiaobu APP, enterprise ERP, POS and e-commerce backend system, comprehensive market performance data and clothing demand data are obtained to achieve a panoramic insight into external dynamics and internal characteristics; through attenuation analysis, the attenuation factor reflecting the natural attenuation rate of sales is obtained, and the demand attenuation coefficient is calculated by combining the benchmark time, current sales data and time, and the real demand intensity at different stages is dynamically quantified; based on this coefficient, the time series forecasting model and association rule mining algorithm are used to integrate social media data and key e-commerce events for prediction, generate stocking quantity decisions and optimal inventory quantities, and finally output fabric evaluation decisions; this solution realizes full-link data connection, enhances the timeliness and accuracy of demand perception, effectively balances the supply and demand relationship, reduces inventory backlogs and out-of-stock risks, provides data-driven reliable support for supply chain optimization of apparel companies, and improves market response efficiency and competitiveness.

[0043] See also Figure 2 A second embodiment of a fabric evaluation and decision-making method according to an embodiment of the present invention includes:

[0044] 201. Performing heat analysis on social media data according to a preset pruning exact linear time algorithm to obtain hot events;

[0045] In this embodiment, the pruning exact linear time algorithm is used to quickly locate backtracking change points in linear time by minimizing the objective function of the piecewise fitting error and the change point penalty term. Its "pruning" mechanism can eliminate redundant candidate points, avoid ineffective calculations, and adapt to the efficient processing of massive social media data.

[0046] 202. An optimized event association network is constructed based on the association rule mining algorithm, key events and hot events of the e-commerce platform;

[0047] In this embodiment, based on association rule mining algorithms (such as Apriori and FP-Growth), popular events and key events of the e-commerce platform (such as the 618 promotion, live broadcast scheduling, and platform traffic support) are integrated to build an optimized event association network;

[0048] 203. Based on the demand forecasting model, hierarchical analysis method and demand decay coefficient, the optimized event association network is predicted to obtain the stocking quantity decision and the optimal inventory level;

[0049] In this embodiment, core indicators are extracted from the network and weighted using AHP, such as clothing sentiment, cross-platform dissemination speed, and coverage. The weights are then adjusted based on the demand decay coefficient to adapt to the product lifecycle. A demand forecasting model (such as LSTM) inputs the weighted indicator scores and historical data to generate sales forecasts. The stocking quantity (450 pieces) is generated based on the gap in the stocking cycle (for example, sales of 600 pieces are predicted in the next 12 days, while the current inventory is 150 pieces), and event-driven factors are added. The optimal inventory level balances out-of-stock risks and inventory backlogs.

[0050] In this embodiment, a pruned precise linear time algorithm is used to locate social media popularity change points with linear complexity, eliminate redundant calculations, and generate structured popularity events in real time. Based on the association rule mining algorithm, popularity events and key e-commerce events (such as promotions and live broadcast scheduling) are integrated to build an optimized event association network and clearly present the event transmission link. The core indicators of the network are weighted through the hierarchical analysis method, and the demand attenuation coefficient is combined to adapt to the product life cycle. The demand forecasting model inputs weighted data to generate sales forecasts, and the stocking quantity is generated by combining the stocking cycle gap and event factors. The optimal inventory quantity is calculated to balance the risk of out-of-stock and backlog. This solution helps the supply chain respond to market trends efficiently.

[0051] See also Figure 3 A third embodiment of a fabric evaluation and decision-making method according to an embodiment of the present invention includes:

[0052] 301. Performing a popularity test on the social media data according to the pruning exact linear time algorithm and a preset standard deviation difference condition and a preset standard deviation range condition to obtain a popularity test result;

[0053] In this embodiment, by minimizing the objective function of the segmented fitting error and the change point penalty term, the backtracking change point is quickly located with linear time complexity while ensuring accuracy;

[0054] 302. When the heat test result shows that the volatility of the sequence after the retrospective change point is greater than the volatility of the sequence before the retrospective change point and meets the standard deviation range condition, a heat surge event is generated;

[0055] 303. Generate a heat event based on the heat surge event;

[0056] In this embodiment, a sudden increase in popularity event includes core information such as the change point time, standard deviation difference value, and related topics (e.g., "The standard deviation of a sweater topic increased from 300 to 750 on October 5, triggering a sudden increase in popularity");

[0057] 304. When the heat test result shows that the volatility of the sequence after the retrospective change point is less than the volatility of the sequence before the retrospective change point and meets the standard deviation range condition, a heat drop event is generated;

[0058] 305. Generate a heat event based on a heat drop event;

[0059] In this embodiment, the sudden drop in popularity events and the sudden rise in popularity events are classified as hot events, providing clear market signals for downstream business modules such as fabric demand forecasting and inventory adjustment;

[0060] In this embodiment, the accuracy and efficiency of capturing market dynamics are effectively improved through the social media heat detection mechanism, providing reliable support for business decision-making; the pruning precise linear time algorithm is adopted, with the goal of minimizing the segmented fitting error and the change point penalty term, and the linear time complexity is used to quickly locate the backtracking change point to ensure efficient data analysis; combined with the preset standard deviation difference and range condition double verification, if the volatility after the change point is greater than the previous sequence and reaches the threshold, a heat surge event is generated, otherwise a heat drop event is generated. The event contains core information such as the change point time, standard deviation difference, and related topics; the two types of events are uniformly classified as heat events, and market trends are clearly fed back; this solution provides real-time market signals for downstream fabric demand forecasts, inventory adjustments, etc., helping companies quickly adapt to heat changes and improve supply chain adjustment efficiency and decision-making accuracy.

[0061] See also Figure 4 A fourth embodiment of a fabric evaluation and decision-making method according to an embodiment of the present invention includes:

[0062] 401. Iteratively analyze the social media data based on a preset time series starting point and a preset time window to obtain an optimal backtracking point;

[0063] In this embodiment, social media data (such as topic discussion volume and likes time series curves) is iteratively analyzed based on the starting point of the time series and a preset window size (e.g., 24 hours / window). The trend stability index of the data in each window is calculated, and the end point of the window with the highest trend stability (minimum error) is selected as the optimal backtracking point. This means that the backtracking starting point that best captures the characteristics of popularity changes is selected. For example, if a topic's trend is stable before the third day and increases in volatility after the third day, the optimal backtracking point is set as the end of the third day, providing a benchmark reference for subsequent change point detection. This step avoids redundant calculations caused by unbounded backtracking, ensuring that subsequent analysis focuses on the time coordinates that have the greatest impact on popularity changes, thereby improving analysis efficiency.

[0064] 402. Perform backtracking analysis on social media data based on the pruning exact linear time algorithm and the optimal backtracking point to obtain a backtracking change point sequence;

[0065] In this embodiment, the Pruned Exact Linear Time (PELT) algorithm is an efficient time series change point detection algorithm. Its objective function is a piecewise fitting error and a change point penalty term. By minimizing this function, a retrospective analysis of social media data is performed. At the same time, pruning rules are used to eliminate candidate time points that are unlikely to be change points, thereby finding a retrospective change point sequence.

[0066] 403. Perform a heat test on the retrospective change point sequence according to the standard deviation difference condition and the standard deviation range condition to obtain a heat test result;

[0067] In this embodiment, the standard deviation difference condition is the volatility of the series after the change point (standard deviation ) is greater than the volatility of the series before the change point (standard deviation ), and the standard deviation range conditions are met (the difference exceeds the preset threshold (such as }), it is judged as a sudden increase in heat; the standard deviation difference condition is the volatility of the sequence after the change point (standard deviation ) is less than the volatility of the series before the change point (standard deviation ), and the standard deviation range conditions are met (the difference exceeds the preset threshold (such as }, determined to be a sudden drop in heat);

[0068] In this embodiment, the starting point of the time series is used as a benchmark, and the data is iteratively analyzed according to the time window. The trend stability index is calculated, and the optimal backtracking point is screened out to avoid redundant calculations of unbounded backtracking, ensuring that subsequent analysis focuses on key time coordinates and improves efficiency; the pruning exact linear time algorithm (PELT) is then used to backtrack and analyze the data to find the backtracking change point sequence; finally, based on the standard deviation difference condition and the standard deviation range condition test, the sudden increase and decrease in popularity are determined, thereby improving the accuracy of identifying sudden changes in popularity and reducing misjudgments. This provides a reliable basis for companies to timely capture changes in market hotspots and formulate targeted fabric strategies, thereby improving their response speed to market dynamics.

[0069] See also Figure 5 A fifth embodiment of a fabric evaluation and decision-making method according to an embodiment of the present invention includes:

[0070] 501. Classify key events of the e-commerce platform to obtain multiple classified events;

[0071] In this embodiment, key events on the e-commerce platform are classified according to event nature (promotional, operational, external), impact scope (global / category / single product), and time period (short-term sudden / medium-term continuous / long-term planned), resulting in multiple classified events.

[0072] 502. Perform association analysis on multiple classified events and hot events according to an association rule mining algorithm to obtain an association analysis result;

[0073] In this embodiment, the association analysis results include support (frequency of event co-occurrence), confidence (probability of event B occurring after event A occurs), and lift (intensity of the impact of event A on event B) to quantify the strength of the association;

[0074] 503. Construct an event association network based on the association analysis results and preset association rules;

[0075] In this embodiment, classified events and hot events are used as nodes, and node attributes include event type, occurrence time, and impact weight. The confidence level of the association rule is used as the edge weight. The higher the weight, the greater the visual intensity of the edge. In the example network, "618 Big Sale (Node A)" is connected to "Category Hotness Surge (Node B)" through an edge with a weight of 0.85, and "New Product Launch (Node C)" is connected to "Single Product Hotness Surge (Node D)" through an edge with a weight of 0.75, intuitively presenting the event transmission chain.

[0076] 504. Performing false verification on the event correlation network according to the preset independence verification conditions to obtain a false verification result;

[0077] In this embodiment, the independence test condition is to calculate the significance p-value of the association rule through the chi-square test method. If the p-value is ≤ 0.05 (the preset significance level), the association is considered statistically significant; if the p-value is > 0.05, it is determined to be a false association (possibly caused by random coincidence). For example, the "store decoration event" and the "sweatshirt popularity surge event" have a high co-occurrence frequency, but the independence test shows a p-value of 0.12 (> 0.05), which is determined to be a false association (no actual causal relationship). The chi-square test is a non-parametric test method used to analyze the association of categorical variables. The core principle is to determine whether there is a significant association between variables by measuring the deviation between the actual observed data and the theoretical expected data.

[0078] 505. Filter the event correlation network according to the false verification result to obtain an optimized event correlation network;

[0079] In this embodiment, based on the false verification results, false correlation edges with p-values ​​greater than 0.05 in the event correlation network are removed, and statistically significant true correlations are retained, ultimately obtaining an optimized event correlation network.

[0080] In this embodiment, through a systematic event correlation analysis and optimization process, the reliability and decision-making value of the correlation between e-commerce platform events and popularity are improved; the association rule mining algorithm is used to quantify the correlation strength between events and popularity, and then a correlation network is constructed to intuitively present the event transmission link; through verified false correlations, statistically significant real correlations are retained, and the redundant correlations in the optimized network are reduced, thereby improving the accuracy of correlation authenticity, providing clear logical support for fabric demand forecasting and inventory adjustments, helping enterprises capture event-driven market dynamics, reduce decision-making errors, and improve supply chain response efficiency.

[0081] See also Figure 6 A sixth embodiment of a fabric evaluation and decision-making method in an embodiment of the present invention includes:

[0082] 601. Perform hierarchical analysis on the optimized event association network based on the hierarchical analysis method and the demand decay coefficient to obtain the clothing sentiment tendency score weight, cross-platform communication speed score weight, and coverage score weight;

[0083] In this embodiment, the core influencing indicators are scientifically weighted using the analytic hierarchy process (AHP) and the demand attenuation coefficient. The clothing sentiment tendency score is used to quantify the polarity of users' evaluation of clothing on social media (such as the rate of positive comments and the proportion of negative keywords), which is directly related to consumption intention; the cross-platform propagation speed score is used to measure the diffusion efficiency of hot events in multiple channels (TikTok, Xiaohongshu, and e-commerce platforms) (such as the time from the first release to full platform coverage), reflecting the speed of market penetration; the coverage score is used to calculate the scale of users reached by the hot event (such as the number of topic discussions and the number of interactive users), reflecting the market influence;

[0084] 602. Predict the clothing sentiment tendency score weight, cross-platform communication speed score weight, and coverage score weight based on the demand forecast model and the preset stocking cycle to obtain predicted sales and stocking quantity decisions;

[0085] In this embodiment, the demand forecasting model maps weighted indicator scores to sales forecasts. Stocking quantity decisions include generating a stocking quantity (450 units) based on the predicted sales and the gap between the stocking cycle (e.g., predicted sales of 600 units in the next 12 days, with 150 units currently in stock), while also overlaying driving factors from the event correlation network (e.g., adding 10% to the stocking due to live broadcast scheduling).

[0086] 603. Calculate the forecasted sales volume, the preset safety factor, the preset in-transit inventory, and the preset replenishment capacity to obtain the optimal inventory level;

[0087] In this embodiment, the calculation formula for the optimal inventory is as follows:

[0088] Optimal inventory level = (forecasted sales volume × safety factor) - in-transit inventory + replenishment capacity. The optimal inventory level reflects the overall demand fluctuations, in-transit inventory, and replenishment capacity, thus achieving a differentiated inventory strategy.

[0089] In this embodiment, by using the hierarchical analysis method and the demand attenuation coefficient, the three core indicators of clothing sentiment tendency, cross-platform dissemination speed, and coverage are extracted and weighted from the optimized event association network to objectively reflect user evaluation, market penetration speed and influence; the demand forecasting model generates sales forecasts based on weighted indicator scores, and formulates stocking decisions based on the gap in the stocking cycle and event-driven factors (such as live broadcast scheduling) to ensure timely supply of goods; the optimal inventory level is calculated through the formula of "forecasted sales × safety factor - in-transit inventory + replenishment capacity" to achieve differentiated inventory strategies for hot-selling and long-tail products; this solution effectively enhances the supply chain's responsiveness to market dynamics and resource allocation efficiency.

[0090] See also Figure 7 A seventh embodiment of a fabric evaluation and decision-making method in the embodiments of the present invention includes:

[0091] 701. Generate fabric demand data and clothing style data based on optimal inventory and stocking quantity decisions;

[0092] In this embodiment, fabric demand data includes calculations of fabric type (e.g., cotton, denim), usage (e.g., 1.2 meters per piece, totaling 540 meters), and performance parameters (e.g., weight, elasticity) based on inventory quantity (e.g., 450 pieces of a certain style) and unit consumption standards. Fabric priority (high-elasticity fabrics for popular styles, general-purpose fabrics for long-tail styles) is determined based on inventory structure (hot items / long-tail styles). Apparel style data includes user preferences based on sales forecasts (e.g., oversize styles for popular live-stream styles), outputting style structure parameters (length, sleeve shape, collar shape), and design elements (e.g., whether to add pockets, print placement), etc.

[0093] 702. Extract features from the fabric demand data and clothing style data to obtain fabric texture features and clothing regional features;

[0094] 703. Perform style transfer on the clothing regional features based on the preset style transfer model and fabric texture features to obtain the target predicted style;

[0095] In this embodiment, the style transfer model is a model trained by a neural network (such as CycleGAN or Neural StyleTransfer) by pre-learning style features such as texture patterns and color distribution from a large number of style images. The style transfer model (such as the CycleGAN neural network) maps fabric texture features to regional features of the garment to generate a target predicted style. The algorithm pre-learns the rules for how fabric textures and regional morphologies fit together (such as how coarse knit textures fit loose garments) and accurately transfers the extracted fabric textures to the corresponding garment regions, ensuring that the texture direction (such as stripes running the length of the garment) and density (such as denser texture at the neckline) conform to the design logic.

[0096] 704. Render the target predicted style according to a preset texture alignment error to obtain an initial rendering;

[0097] In this embodiment, rendering parameters are corrected based on texture alignment errors (e.g., the distortion of the fabric texture at the corners of the cuffs) to ensure that the texture fits naturally on the three-dimensional style and reduce visual distortion;

[0098] 705. Pre-inference is performed on the initial renderings according to a preset virtual model posture library to obtain clothing fit;

[0099] In this embodiment, a virtual model posture library (such as standing and sitting postures) is called to simulate the wearing state of clothing, calculate fit indicators (such as the amount of underarm wrinkles and waist tightness), and identify design defects (such as abnormal wrinkles caused by overly tight cuffs);

[0100] 706. Adjust the initial rendering according to the fit of the clothing to obtain the final rendering;

[0101] In this embodiment, the style parameters are optimized based on the fit results (e.g., loosening the cuffs by 1 cm), and the corrected final effect image is output to intuitively present the actual matching effect of the fabric and style;

[0102] 707. Generate fabric evaluation decisions based on the final renderings;

[0103] In this embodiment, fabric evaluation decisions range from inventory decisions to fabric demand;

[0104] In this embodiment, fabric demand and clothing style data are generated based on the optimal inventory and reserve quantity, and fabric texture and clothing regional features are obtained through feature extraction; the texture and style are matched with the help of the style transfer model, and the initial rendering is rendered in combination with the texture alignment error. The virtual model posture library then pre-infers the fit and adjusts the parameters to output the final rendering; this solution optimizes the matching degree between fabric and style, improves market acceptance; avoids design defects in advance, reduces development costs; reduces fabric waste rate, improves inventory turnover rate, and enhances resource utilization efficiency and market competitiveness.

[0105] The above describes a fabric evaluation and decision-making method in an embodiment of the present invention. The following describes a fabric evaluation and decision-making device in an embodiment of the present invention. Figure 8 In one embodiment of the present invention, a fabric evaluation and decision-making device includes:

[0106] The first data acquisition module 1 is used to acquire market performance data and clothing demand data;

[0107] Attenuation analysis module 2, used to perform attenuation analysis on market performance data and clothing demand data to obtain an attenuation factor;

[0108] The second data acquisition module 3 is used to obtain current sales data and current time;

[0109] The demand decay calculation module 4 is used to calculate the preset reference time, decay factor, current sales data and current time to obtain the demand decay coefficient;

[0110] The third data acquisition module 5 is used to obtain social media data and key events of the e-commerce platform;

[0111] Prediction module 6, used to predict social media data and key events of e-commerce platforms based on the demand attenuation coefficient, a preset demand forecasting model, and a preset association rule mining algorithm to obtain stocking quantity decisions and optimal inventory levels;

[0112] Fabric evaluation decision module 7, used to generate fabric evaluation decision based on stocking quantity decision and optimal inventory quantity;

[0113] In this embodiment, relying on the Zhixiaobu APP, enterprise ERP, POS and e-commerce backend system, comprehensive market performance data and clothing demand data are obtained to achieve a panoramic insight into external dynamics and internal characteristics; through attenuation analysis, the attenuation factor reflecting the natural attenuation rate of sales is obtained, and the demand attenuation coefficient is calculated by combining the benchmark time, current sales data and time, and the real demand intensity at different stages is dynamically quantified; based on this coefficient, the time series forecasting model and association rule mining algorithm are used to integrate social media data and key e-commerce events for prediction, generate stocking quantity decisions and optimal inventory quantities, and finally output fabric evaluation decisions; this solution realizes full-link data connection, enhances the timeliness and accuracy of demand perception, effectively balances the supply and demand relationship, reduces inventory backlogs and out-of-stock risks, provides data-driven reliable support for supply chain optimization of apparel companies, and improves market response efficiency and competitiveness.

[0114] Figure 9 This is a schematic diagram of the structure of a fabric evaluation and decision-making device provided in an embodiment of the present invention. This fabric evaluation and decision-making device 900 may vary significantly depending on its configuration or performance. It may include one or more processors (central processing units, CPUs) 910 (e.g., one or more processors), memory 920, and one or more storage media 930 (e.g., one or more mass storage devices) storing applications 933 or data 932. The memory 920 and storage medium 930 may be either transient or persistent storage. The program stored in the storage medium 930 may include one or more modules (not shown), each of which may include a series of instructions and operations within the fabric evaluation and decision-making device 900. Furthermore, the processor 910 may be configured to communicate with the storage medium 930, executing the series of instructions and operations stored in the storage medium 930 on the fabric evaluation and decision-making device 900 to implement the steps of the fabric evaluation and decision-making method provided in the aforementioned method embodiments.

[0115] A fabric evaluation and decision-making device 900 may also include one or more power supplies 940, one or more wired or wireless network interfaces 950, one or more input and output interfaces 960, and / or one or more operating systems 931, such as Windows Server, Mac OS X, Unix, Linux, FreeBSD, etc. It will be understood by those skilled in the art that Figure 9 The structure of a fabric evaluation and decision-making device shown does not constitute a limitation on the fabric evaluation and decision-making device, and may include more or fewer components than shown in the figure, or combine certain components, or arrange the components differently.

[0116] The present invention also provides a computer-readable storage medium, which may be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium. The computer-readable storage medium stores instructions that, when executed on a computer, cause the computer to execute the steps of a fabric evaluation decision method.

[0117] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described systems, devices, and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0118] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the portion that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes various media that can store program code, such as a USB flash drive, a mobile hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0119] Finally, it should be noted that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art will be able to modify the technical solutions described in the aforementioned embodiments or substitute equivalents for some of the technical features. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention shall be included within the scope of protection of the present invention.

Claims

1. A fabric evaluation decision method, characterized in that: include: Obtain market performance data and apparel demand data; Conduct attenuation analysis on market performance data and apparel demand data to obtain attenuation factors; Get current sales data and current time; Calculate the preset base time, decay factor, current sales data and current time to obtain the demand decay coefficient; Obtain social media data and key events from e-commerce platforms; Based on the demand decay coefficient, the preset demand forecasting model and the preset association rule mining algorithm, social media data and key events on the e-commerce platform are predicted to obtain stocking quantity decisions and optimal inventory levels; Generate fabric valuation decisions based on stocking quantity decisions and optimal inventory levels.

2. A fabric evaluation decision method according to claim 1, characterized in that: The method of predicting social media data and key events on e-commerce platforms based on the demand attenuation coefficient, a preset demand forecasting model, and a preset association rule mining algorithm to obtain stocking quantity decisions and optimal inventory levels includes: Perform heat analysis on social media data based on the preset pruning exact linear time algorithm to obtain hot events; An optimized event association network is constructed based on association rule mining algorithms, key events, and hot events on the e-commerce platform; The optimized event association network is predicted based on the demand forecasting model, hierarchical analysis method and demand decay coefficient to obtain the stocking quantity decision and optimal inventory level.

3. A fabric evaluation decision method according to claim 2, characterized in that: The method of performing heat analysis on social media data according to a preset pruning precise linear time algorithm to obtain heat events includes: Performing a popularity test on social media data based on a pruning exact linear time algorithm and a preset standard deviation difference condition and a preset standard deviation range condition to obtain a popularity test result; When the heat test result shows that the volatility of the sequence after the retrospective change point is greater than the volatility of the sequence before the retrospective change point and meets the standard deviation range conditions, a heat surge event is generated; Generate a heat event based on a sudden heat event; When the heat test result shows that the volatility of the sequence after the retrospective change point is less than the volatility of the sequence before the retrospective change point and meets the standard deviation range conditions, a heat drop event is generated; Generates a heat event based on a heat drop event.

4. A fabric evaluation decision method according to claim 3, characterized in that: The heat test of the social media data is performed according to the pruning exact linear time algorithm and the preset standard deviation difference condition and the preset standard deviation range condition to obtain the heat test result, including: Iteratively analyze social media data based on a preset time series starting point and a preset time window to obtain the optimal backtracking point; Backtracking analysis of social media data is performed based on the pruning exact linear time algorithm and the optimal backtracking point to obtain a backtracking change point sequence; The heat test is performed on the retrospective change point sequence according to the standard deviation difference condition and the standard deviation range condition to obtain the heat test result.

5. A fabric evaluation decision method according to claim 2, characterized in that: The optimized event association network constructed based on the association rule mining algorithm, key events and hot events of the e-commerce platform includes: Classify key events of the e-commerce platform to obtain multiple classified events; Perform association analysis on multiple classified events and hot events based on association rule mining algorithms to obtain association analysis results; The event association network is constructed based on the association analysis results and the preset association rules; Perform false verification on the event correlation network according to the preset independence test conditions to obtain false verification results; The event correlation network is filtered according to the false verification results to obtain an optimized event correlation network.

6. A fabric evaluation decision method according to claim 2, characterized in that: The method of performing prediction on the optimized event association network based on the demand forecasting model, the hierarchical analysis method, and the demand attenuation coefficient to obtain the stocking quantity decision and the optimal inventory quantity includes: A hierarchical analysis of the optimized event association network was conducted based on the hierarchical analysis method and the demand decay coefficient to obtain the clothing sentiment tendency score weight, cross-platform communication speed score weight, and coverage score weight; Based on the demand forecasting model and the preset stocking cycle, the clothing sentiment tendency score weight, cross-platform communication speed score weight, and coverage score weight are predicted to obtain predicted sales and stocking quantity decisions; The forecast sales volume, the preset safety factor, the preset in-transit inventory and the preset replenishment capacity are calculated to obtain the optimal inventory level.

7. A fabric evaluation and decision-making method according to claim 1, characterized in that: The method of generating a fabric evaluation decision based on the stocking quantity decision and the optimal inventory quantity includes: Generate fabric demand data and clothing style data based on optimal inventory and stocking quantity decisions; Extract features from fabric demand data and clothing style data to obtain fabric texture features and clothing area features; Perform style transfer on clothing regional features based on the preset style transfer model and fabric texture features to obtain the target predicted style; Render the target prediction style according to the preset texture alignment error to obtain the initial rendering; Pre-inference is performed on the initial renderings based on the preset virtual model pose library to obtain the clothing fit; Adjust the initial rendering according to the fit of the clothing to obtain the final rendering; Generate fabric evaluation decisions based on the final rendering.

8. A fabric evaluation and decision-making device, characterized in that: include: A first data acquisition module is used to acquire market performance data and clothing demand data; Attenuation analysis module, used to perform attenuation analysis on market performance data and clothing demand data to obtain attenuation factors; The second data acquisition module is used to obtain current sales data and current time; The demand decay calculation module is used to calculate the preset reference time, decay factor, current sales data and current time to obtain the demand decay coefficient; The third data acquisition module is used to obtain social media data and key events of e-commerce platforms; The forecasting module is used to predict social media data and key events on e-commerce platforms based on the demand decay coefficient, a preset demand forecasting model, and a preset association rule mining algorithm to determine stocking quantity decisions and optimal inventory levels. The fabric evaluation decision module is used to generate fabric evaluation decisions based on stocking quantity decisions and optimal inventory quantities.

9. A fabric evaluation and decision-making device, characterized in that: The fabric evaluation and decision-making device comprises: a memory and at least one processor, wherein the memory stores instructions; At least one of the processors calls the instructions in the memory to enable the fabric evaluation and decision-making device to perform each step of the fabric evaluation and decision-making method according to any one of claims 1 to 7.

10. A computer-readable storage medium having instructions stored thereon, characterized in that: When the instructions are executed by the processor, the steps of the fabric evaluation and decision-making method according to any one of claims 1 to 7 are implemented.

Images