Intelligent price comparison method for fresh food purchase based on big data

Abstract

The application discloses a kind of based on big data's fresh procurement intelligent comparison price method, belong to fresh supply chain digitization technical field.The method is by gathering multi-source procurement data and carrying out cleaning integration to build original comparison price dataset, according to original comparison price dataset to offer, loss, cold chain and time window data are quantitatively treated to obtain multidimensional comparison price parameter, multidimensional comparison price parameter is jointly converted and forms dynamic saleable equivalent cost set with reasonableness verification, again by successive comparison, logic correction and stability marking obtain accurate comparison price result, finally, actual arrival and loss feedback data are compared to dynamic calibration comparison price parameter, obtain updated comparison price parameter set.The application is favorable to solve the problem that traditional comparison price benchmark is not unified, risk and price are disconnected, improve the accuracy and scientificity of fresh procurement comparison price, and support dynamic optimization and whole-process cost controllable of procurement decision.

Description

Technical Field

[0001] This invention relates to the field of digital technology for fresh food supply chains, and in particular to a smart price comparison method for fresh food procurement based on big data. Background Technology

[0002] With the continuous advancement of modernization in commerce and trade, centralized procurement and distribution, and intelligent supply chains, instant retail and daily fresh food delivery businesses are experiencing rapid growth. The demand for daily replenishment in terminal retail scenarios such as chain fresh food pre-positioned warehouses and community stores is becoming increasingly prominent. Fresh food procurement has entered a new stage characterized by high frequency, small batches, and concurrent quotations from multiple suppliers. In such scenarios, fresh produce is characterized by high freshness sensitivity, diverse specifications and grades, and strict constraints on delivery time windows. The core need for procurement price comparison has shifted from simply comparing nominal prices to a comprehensive consideration of the equivalent cost corresponding to the actual salable result after delivery. Traditional static price comparison models are no longer suitable for the refined procurement decision-making needs in this scenario.

[0003] Currently, most existing fresh produce procurement price comparison technologies employ a core process of: price collection, price ranking, and manual confirmation. This involves collecting price information, historical transaction data, and basic performance information from multiple suppliers to rank suppliers for the same procurement need using price ranking or simple weighted scoring. This is then combined with human experience, fixed threshold rules, or supplier priorities to determine the final procurement target. Some optimization schemes introduce historical on-time delivery rates, out-of-stock rates, or quality inspection pass rates as auxiliary correction factors. However, the overall approach still relies on nominal prices as the core basis for price comparison, failing to systematically integrate and quantify multi-dimensional influencing factors, and lacking a dynamic utilization and feedback correction mechanism for data throughout the entire procurement process. Existing technologies still have significant shortcomings, making it difficult to meet the refined and dynamic price comparison needs of fresh produce procurement: First, the price comparison benchmarks are not uniform, and the specifications, grades, and packaging dimensions of different supplier quotations are not standardized and converted, which easily leads to situations where the quoted price is low but the effective saleable quantity is small, resulting in higher actual saleable costs; Second, freshness risk and fulfillment risk are not included in the price comparison decision for the same batch, and key data such as cold chain temperature trajectory and the matching degree of the receiving window are not linked with price data, making it impossible to identify suppliers with low prices but high fulfillment risks in advance, which can easily lead to problems such as delivery mismatch, high losses, and temporary additional procurement; Third, there is a lack of a joint conversion mechanism for multi-dimensional parameters, which cannot unify scattered indicators such as price, sellability, and freshness preservation into equivalent costs that can be directly ranked, resulting in insufficient scientific validity of price comparison results; Fourth, there is no sound feedback and correction mechanism, which cannot dynamically update the price comparison parameters based on actual delivery and loss feedback data, leading to distortion of price comparison results after long-term use, and making it difficult to adapt to the characteristics of large price fluctuations and unstable quality of fresh produce. Therefore, this solution proposes a smart price comparison method for fresh food procurement based on big data. Summary of the Invention

[0004] The purpose of this invention is to provide a smart price comparison method for fresh produce procurement based on big data, so as to solve the above-mentioned technical problems.

[0005] To solve the above-mentioned technical problems, the present invention adopts the following technical solution: a smart price comparison method for fresh produce procurement based on big data, comprising the following steps: By aggregating and integrating procurement data from multiple sources, the original price comparison dataset is obtained. Based on the original price comparison dataset, multi-dimensional parameter processing is performed to obtain multi-dimensional price comparison parameters; By jointly converting the multi-dimensional price comparison parameters, a dynamic set of salable equivalent costs is obtained; The dynamic set of salable equivalent costs is sorted to obtain the price comparison results; The price comparison results are corrected by combining actual delivery and loss feedback data to obtain an updated price comparison parameter set.

[0006] As a preferred embodiment of the intelligent price comparison method for fresh produce procurement based on big data described in this invention, multi-dimensional parameter processing is performed on the original price comparison dataset to obtain multi-dimensional price comparison parameters, including: Based on the original price comparison dataset, the supplier quotations are converted according to specifications and grades to obtain a standardized price comparison unit price; Based on the historical acceptance and loss data in the original price comparison dataset, a sales rate model is performed to obtain basic sales rate parameters. Based on the cold chain temperature trajectory in the original price comparison dataset and the store receiving time window data, freshness preservation and time window matching correction are performed to obtain freshness preservation parameters and time window matching parameters. The standardized price comparison unit price, basic sellability parameter, freshness retention parameter, and time window matching parameter are integrated to form the multi-dimensional price comparison parameter.

[0007] As a preferred embodiment of the intelligent price comparison method for fresh produce procurement based on big data described in this invention, the standardized price comparison unit price, basic sellability parameters, freshness retention parameters, and time window matching parameters are integrated to form the multi-dimensional price comparison parameters, including: The standardized price comparison unit price, basic sellability parameter, freshness retention parameter and time window matching parameter are matched and associated to obtain a set of associated parameters; The set of related parameters is standardized in terms of dimensions and its validity is verified to form multi-dimensional price comparison parameters.

[0008] As a preferred embodiment of the intelligent price comparison method for fresh produce procurement based on big data described in this invention, the set of related parameters is standardized in terms of dimensions and validated for effectiveness, forming multi-dimensional price comparison parameters, including: The parameters in the aforementioned set of related parameters are uniformly transformed to obtain a set of dimensionally normalized parameters. Based on preset parameter thresholds, the set of dimension-normalized parameters is judged item by item to obtain the set of parameters to be screened. Abnormal and missing data are removed from the set of parameters to be screened to obtain valid and compliant parameter data; The valid and compliant parameter data are categorized and integrated to obtain multi-dimensional price comparison parameters.

[0009] As a preferred embodiment of the intelligent price comparison method for fresh produce procurement based on big data described in this invention, the multi-dimensional price comparison parameters are jointly converted to obtain a dynamic set of salable equivalent costs, including: Substitute the multi-dimensional price comparison parameters into the preset conversion formula to obtain the initial equivalent cost data; The initial equivalent cost data is grouped and aggregated to obtain grouped equivalent cost data; The numerical rationality of the grouped equivalent cost data is verified to obtain calibrated equivalent cost data; The calibrated equivalent cost data are then uniformly organized to obtain a dynamic set of marketable equivalent costs.

[0010] As a preferred embodiment of the intelligent price comparison method for fresh food procurement based on big data described in this invention, the initial equivalent cost data is grouped and aggregated to obtain grouped equivalent cost data, including: The correlation dimension information of the initial equivalent cost data is categorized and aggregated to obtain preliminary grouped data; The initial grouped data is subjected to duplicate item identification and merging processing to obtain deduplicated grouped data; Based on the deduplicated grouped data, dimensional alignment and format unification processing are performed to obtain grouped equivalent cost data.

[0011] As a preferred embodiment of the intelligent price comparison method for fresh produce procurement based on big data described in this invention, the dynamic set of salable equivalent costs is sorted to obtain the price comparison results, including: The data of the dynamic salable equivalent cost set is compared one by one to obtain preliminary sorting data; The preliminary sorted data is marked with quality and stability to obtain the marked sorted data; Based on the labeled and sorted data, high-cost items are removed to obtain filtered and sorted data; The filtered and sorted data are then normalized in terms of serial number and format to generate a price comparison result.

[0012] As a preferred embodiment of the intelligent price comparison method for fresh produce procurement based on big data described in this invention, the data of the dynamic set of salable equivalent costs is compared sequentially to obtain preliminary sorting data, including: The equivalent cost values ​​in the dynamic set of salable equivalent costs are compared pairwise to obtain the order of magnitude between the data. Based on the aforementioned size order relationship, the data is arranged according to a preset sorting rule to obtain the initial sorted data; The initial sorted data is logically corrected to obtain preliminary sorted data.

[0013] As a preferred embodiment of the intelligent price comparison method for fresh food procurement based on big data described in this invention, the initial sorting data is logically corrected to obtain preliminary sorting data, including: The initial sorted data is subjected to correlation information extraction to obtain a sorting verification dataset; Perform a sequential logical check on the sorting verification dataset to identify the sorting order compliance judgment result; Based on the sorting order compliance determination result, the sorting verification dataset is retrieved to obtain non-compliant data events; The original equivalent cost calculation records of the non-compliant data events were traced to obtain the order deviation data; The order deviation data is adjusted and its arrangement position is temporarily stored to obtain preliminary corrected sorting data; A full-domain logical review is performed on the preliminary corrected sorting data to obtain the preliminary sorting data.

[0014] As a preferred embodiment of the intelligent price comparison method for fresh produce procurement based on big data described in this invention, the price comparison results are corrected by incorporating actual delivery and loss feedback data to obtain an updated set of price comparison parameters, including: The price comparison results and actual delivery and loss feedback data are collected synchronously to construct a correction and verification dataset; A deviation comparison analysis was performed on the corrected verification dataset to obtain the actual difference information; The actual difference information is converted by combining preset weights to generate corresponding correction coefficients; Based on the corresponding correction coefficients, the parameters in the price comparison results are adjusted item by item to obtain preliminary corrected price comparison parameters; The rationality of the initially revised price comparison parameters is verified to obtain the updated price comparison parameter set.

[0015] The beneficial effects of this invention are: 1. This invention constructs a complete original price comparison dataset by aggregating multi-source procurement data such as supplier quotations, product specification data, and historical acceptance records. It also uniformly converts the packaging specifications, net weight information, grade labels, and tolerance ranges of different supplier quotations into standardized price comparison unit prices under standard product categories. This completely solves the problems of inconsistent price comparison benchmarks, inability to directly compare quotations of different specifications and grades, and the potential distortion of actual salable costs in existing technologies. It not only unifies the core criteria for price comparison in fresh produce procurement but also significantly improves the accuracy and fairness of price comparison calculations. It avoids the procurement pitfall of low quotations but low effective salable quantities, providing a solid foundation of data support for subsequent accurate price comparisons.

[0016] 2. By modeling the sellability of historical acceptance and loss data, basic sellability parameters are obtained. Analysis of cold chain temperature trajectory and store receiving window data yields freshness preservation parameters and window matching parameters. These parameters are then combined with standardized comparative pricing units into a preset formula for joint calculation, generating a dynamic sellable equivalent cost set that comprehensively reflects price, quality, freshness, and timeliness. Key risk factors such as quality loss, cold chain freshness, and fulfillment timeliness are simultaneously incorporated into the same batch price comparison decision-making process. This completely solves the problems of price and risk disconnect in existing technologies, focusing only on nominal low prices while ignoring actual losses and fulfillment risks, resulting in low-price bids but high actual losses and fulfillment mismatches. It effectively reduces trimming losses and spoilage rates after procurement, reduces chain problems such as stockouts and temporary additional procurement caused by mismatched delivery windows, and lowers the overall cost and fulfillment risk of fresh produce procurement.

[0017] 3. By comparing, logically correcting, and verifying the dynamic set of available equivalent costs one by one, the system first sorts the suppliers by cost value, and then combines historical out-of-stock rates, delay rates, cancellation rates, and other fulfillment data to mark and correct the supply stability. Finally, it outputs the main supplier ranking, backup supplier ranking, and order splitting price comparison results. This solves the problems of relying solely on nominal prices for price comparison, making decisions in a single and crude manner, and failing to take into account both cost and supply stability. It significantly improves the scientificity and rationality of price comparison decisions for fresh produce procurement, ensuring the economy of procurement costs and reducing the risk of stockouts caused by a single supplier. It is suitable for the high-frequency, small-batch procurement needs of chain fresh produce pre-positioned warehouses and community stores for daily rolling replenishment.

[0018] 4. By comparing, logically correcting, and verifying the dynamic set of available equivalent costs one by one, the system first sorts the suppliers by cost value, and then combines the suppliers' historical out-of-stock rate, delay rate, cancellation rate, and other fulfillment data to mark and correct the supply stability. Finally, it outputs the main supplier ranking, backup supplier ranking, and order splitting price comparison results, which greatly improves the scientificity and rationality of fresh food procurement price comparison decisions. It not only ensures the economy of procurement costs, but also reduces the risk of stockouts caused by a single supplier. It is suitable for the high-frequency, small-batch procurement needs of chain fresh food front warehouses and community stores for daily rolling replenishment. Attached Figure Description

[0019] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments will be briefly described below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation on the scope of this application.

[0020] Figure 1 This is a schematic diagram of a smart price comparison method for fresh produce procurement based on big data proposed in this invention. Figure 2 This is a bar chart comparing the core technical indicators of the present invention with those of traditional price comparison methods. Detailed Implementation

[0021] To make the technical means, creative features, and achieved objectives and effects of this invention easier to understand, the invention is further described below with reference to specific embodiments. However, the following embodiments are merely preferred embodiments of this invention and not all of them. Other embodiments obtained by those skilled in the art based on the embodiments described herein without creative effort are all within the protection scope of this invention. Unless otherwise specified, the experimental methods in the following embodiments are conventional methods, and the materials and reagents used in the following embodiments are commercially available unless otherwise specified.

[0022] like Figure 1 As shown, this embodiment provides a smart price comparison method for fresh food procurement based on big data. The following describes the specific implementation of the present invention in detail, taking the "morning replenishment batch" of a chain fresh food front warehouse as an example, with the procurement categories being two kinds of fresh products: Grade A cucumbers and loose-leaf lettuce.

[0023] All preset parameters are set based on actual application data in the fresh food supply chain industry, experimental verification of 300+ batches of fresh food procurement, and combined with the procurement scenario requirements of chain fresh food pre-positioning warehouses. The specific setting principles are as follows: The formula for dynamic salable equivalent cost is derived based on the characteristics of fresh produce: freshness determines the salable quantity and time window determines the fulfillment effectiveness. Correction coefficient weights: These are determined using the analytic hierarchy process, taking into account the influence of each parameter on the price comparison results. The weight of dynamic salable equivalent cost is 0.4, while the weights of basic salability rate, freshness retention, and time window matching are 0.2 each. High cost threshold: Based on the procurement data statistics of the past 30 days, after eliminating suppliers that exceed this threshold, the overall procurement cost can be reduced by more than 15%. Minimum sample threshold: Ensure that the sample size for availability rate modeling meets the statistical significance requirements and avoid parameter distortion caused by insufficient sample size; Freshness Sensitivity Coefficient: This coefficient is used to quantify the sensitivity of fresh produce to cold chain temperatures. The larger the coefficient, the greater the impact of temperature fluctuations on the product and the faster its freshness deteriorates. Based on the temperature sensitivity characteristics of different fresh produce categories, leafy vegetables (loose-leaf lettuce) are more temperature sensitive than fruit vegetables (Grade A cucumbers). Therefore, the coefficient for loose-leaf lettuce (1.2) is higher than that for Grade A cucumbers (0.8).

[0024] In this embodiment, the method includes the following steps: By aggregating and integrating procurement data from multiple sources, the original price comparison dataset is obtained. The document specifies that this procurement batch is a morning replenishment batch for the chain fresh food pre-positioning warehouse, with the time window set from 06:00 to 08:00 on the same day. The procurement requirements are 100kg of Grade A cucumbers and 80kg of loose-leaf lettuce. The collected multi-source data includes: The collected data includes: supplier quotation data, historical acceptance and loss data, cold chain temperature trajectory data, store receiving time window data, and other auxiliary data; Supplier quotation data: Quotations were collected from 5 qualified suppliers: Suppliers A, B, C, D, and E. The quotations for Grade A cucumbers were as follows: Supplier A: 3.8 yuan / jin (packaging specifications: 20 jin / box, net weight: 19.5 jin / box, Grade A, tolerance ±0.5 jin); Supplier B: 3.7 yuan / jin (packaging specifications: 15 jin / box, net weight: 14.8 jin / box, Grade A, tolerance ±0.2 jin); Supplier C: 3.9 yuan / jin (bulk, Grade A, no packaging tolerance); Supplier D: 3.6 yuan / jin (packaging specifications: 10 jin / box, net weight: 9.7 jin / box, Grade A-, tolerance ±0.3 jin); Supplier E: 3.85 yuan / jin (packaging specifications: 25 jin / box, net weight: 24.6 jin / box, Grade A, tolerance ±0.4 jin). The prices for loose-leaf lettuce are as follows: Supplier A: 2.5 yuan / jin (packaging specifications: 10 jin / box, net weight: 9.6 jin / box, grade A, tolerance ±0.4 jin); Supplier B: 2.4 yuan / jin (packaging specifications: 8 jin / box, net weight: 7.8 jin / box, grade A-, tolerance ±0.2 jin); Supplier C: 2.6 yuan / jin (bulk, grade A, no packaging tolerance); Supplier D: 2.3 yuan / jin (packaging specifications: 12 jin / box, net weight: 11.7 jin / box, grade B, tolerance ±0.3 jin); Supplier E: 2.55 yuan / jin (packaging specifications: 15 jin / box, net weight: 14.8 jin / box, grade A, tolerance ±0.2 jin).

[0025] Historical Acceptance and Loss Data: Data was collected from these five suppliers over the past 30 days regarding arrival and acceptance records, trimming loss records, spoilage records, and return records for Grade A cucumbers and loose-leaf lettuce. For example: Supplier A received 15 batches of Grade A cucumbers in the past 30 days, totaling 3000 kg. 2850 kg passed inspection, with 120 kg of trimming loss, 30 kg of spoilage, and no returns. Supplier A also received 12 batches of loose-leaf lettuce, totaling 1800 kg. 1620 kg passed inspection, with 144 kg of trimming loss, 36 kg of spoilage, and 0 kg returned.

[0026] Cold chain temperature trajectory data: Collect cold chain transportation plans corresponding to each supplier's quotation, including estimated transportation time, cold chain equipment model, and real-time temperature trajectory records. For example: Supplier A's cold chain transportation time is 1.5 hours, and the temperature is maintained at 2-4℃ with no over-temperature records; Supplier D's cold chain transportation time is 2 hours, during which the temperature rises to 5.5℃ for 10 minutes, with a cumulative over-temperature time of 10 minutes.

[0027] Store receiving time window data: The receiving time window for this morning's replenishment batch at the front warehouse is 08:00-09:00, with a shelf preparation time of 30 minutes. The tolerable delivery deviation range is ±15 minutes, meaning that deliveries within 07:45-09:15 are considered acceptable, while deliveries exceeding this range are considered to have reduced time window matching.

[0028] Other supporting data includes: each supplier's qualification documents, on-time delivery rate, out-of-stock rate, delay rate, cancellation rate and other fulfillment data in the past 30 days. For example: Supplier A has an on-time delivery rate of 98% in the past 30 days, an out-of-stock rate of 1%, a delay rate of 2%, and no order cancellation records; Supplier D has an on-time delivery rate of 85% in the past 30 days, an out-of-stock rate of 5%, a delay rate of 10%, and an order cancellation rate of 2%.

[0029] The collected multi-source data was cleaned, deduplicated, and correlated to remove invalid data. Related data from the same supplier and the same product category were linked and bound to construct the original price comparison dataset. For example, supplier A's Grade A cucumber price, historical acceptance loss data, cold chain temperature data, and delivery window commitment data were linked and bound to form a complete record of supplier and product category data. The final original price comparison dataset contains 10 valid records from 5 suppliers and 2 product categories. Each record includes core fields such as supplier identifier, product category, price information, specifications, historical loss, cold chain information, and performance information.

[0030] Based on the original price comparison dataset, multi-dimensional parameter processing is performed to obtain multi-dimensional price comparison parameters; Specifically, the following processing steps are included: Based on the original price comparison dataset, the supplier quotations are converted according to specifications and grades to obtain a standardized price comparison unit price. This is used to eliminate the differences in packaging specifications, net weight, and grade between different supplier quotations, and to uniformly convert all quotations to the standard product category unit preset by the enterprise, so that the quotations can be directly compared. Specifically, based on the enterprise's master procurement data and acceptance standards, the standard product category units for this procurement are set as "per kg of marketable Grade A cucumbers" and "per kg of marketable Grade A loose-leaf lettuce." Among them, the standard grade requirements for Grade A cucumbers are: no rot, no deformities, length ≥20cm, and trimming loss rate ≤5%; the standard grade requirements for Grade A loose-leaf lettuce are: no yellowing, no rot, intact leaves, and trimming loss rate ≤8%.

[0031] Formula used: Specifications and grades are converted, including: The standardized price comparison unit price (unit: yuan / kg) for the corresponding product category of the i-th supplier. The original price quoted by the i-th supplier (unit: yuan / jin). The net weight conversion quantity (unit: jin / kg) corresponding to the i-th supplier's quotation is determined jointly by the packaged net weight, historical sampling net weight deviation, and the company's receiving measurement rules, using the following formula: = Net weight marked on packaging * (1 - historical sampling net weight deviation rate); where each kg corresponds to 2 jin, so the final net weight of the packaging needs to be converted to the net weight per kg; This is a specification grade conversion factor, representing the equivalent ratio between the currently quoted product and the standard category unit. It is determined by combining the grade conversion table in the company's acceptance standards with samples from manual inspections over the past 30 days. Grade A products... =1, Grade A goods =0.95, Grade B product =0.90.

[0032] Specific conversion example in this embodiment: Supplier A: Original Quotation =3.8 yuan / jin, the packaging indicates a net weight of 19.5 jin / box, and the historical net weight deviation rate is 0.01 (1%). =19.5*(1-0.01)=19.305 jin / box, each box has a gross weight of 20 jin, which translates to a net weight per kg: 19.305 jin / 20 jin*10 jin / kg=9.6525 jin / kg (1kg=2 jin, here we convert to jin / kg for easier calculation); Grade A. =1.0; then the standardized price comparison unit price =3.8 / (9.6525*1.0)≈0.394 yuan / jin, which is equivalent to 0.788 yuan / kg (this is just an example calculation; the actual calculation can be done by unifying the units).

[0033] Supplier D: Original Quotation =3.6 yuan / jin, the packaging indicates a net weight of 9.7 jin / box, and the historical net weight deviation rate is 0.02 (2%). =9.7*(1-0.02)=9.506 jin / box, which translates to net weight per kg: 9.506 jin / 10 jin*10 jin / kg=9.506 jin / kg; grade is A-. =0.95; then the standardized price comparison unit price =3.6 / (9.506*0.95)≈0.398 yuan / jin, which is equivalent to 0.796 yuan / kg.

[0034] Following the above method, complete the standardized price comparison unit price conversion for all suppliers and the two product categories, and finally obtain the standardized price comparison unit price set. For example: Standardized price comparison unit price (yuan / kg) for Grade A cucumbers: A=0.788, B=0.772, C=0.780, D=0.796, E=0.790; Standardized price comparison unit price (yuan / kg) for loose-leaf lettuce: A=0.510, B=0.502, C=0.520, D=0.498, E=0.521.

[0035] Based on the historical acceptance and loss data in the original price comparison dataset, a saleability model is performed to obtain basic saleability parameters. This step is used to quantify the supplier's historical quality performance into saleability parameters, reflecting the proportion of goods that can be directly put into the sales process after arrival, transforming "how much can be bought" into "how much can ultimately be sold", and avoiding the distortion of real costs caused by ignoring historical losses.

[0036] Specifically, using statistical modeling, based on historical acceptance records, trimming loss records, spoilage records, and return records in the original price comparison dataset, the basic availability rate for each supplier's corresponding product category is calculated. The formula is: If the number of batches delivered to the corresponding product category by the supplier in the past 30 days is less than the minimum sample threshold, a mixed calculation method of "data weight of the most recent 7 days 0.6 + average data weight of the category 0.4" will be adopted.

[0037] A specific calculation example using loose-leaf lettuce: Supplier A: In the past 30 days, 12 batches of loose-leaf lettuce have arrived, with 10 batches meeting the minimum sample threshold, totaling 1800kg. 1620kg passed inspection. Therefore, the basic sellable rate is... =1620 / 1800=0.9 (90%).

[0038] Supplier D: 8 batches of loose-leaf lettuce arrived in the past 30 days, less than the minimum sample threshold of 10 batches. The total arrival volume in the last 7 days was 500kg, with 425kg passing inspection. The sellability rate for the last 7 days is 425 / 500 = 0.85. The average sellability rate for loose-leaf lettuce is 0.88, therefore the basic sellability rate is... =0.85*0.6+0.88*0.4=0.862 (86.2%).

[0039] The final set of basic sellability parameters is obtained, for example: basic sellability of Grade A cucumbers: A=0.95, B=0.93, C=0.94, D=0.88, E=0.92; basic sellability of loose-leaf lettuce: A=0.90, B=0.87, C=0.89, D=0.862, E=0.88.

[0040] Based on the cold chain temperature trajectory and store receiving window data in the original price comparison dataset, freshness preservation and window matching correction are performed to obtain freshness preservation parameters and window matching parameters. This step is used to quantify the freshness decay risk in the cold chain transportation process and the matching degree of the receiving window into parameters and incorporate them into the price comparison system to solve the problem of price and fulfillment risk being disconnected in the existing technology.

[0041] Freshness retention parameters Combining the freshness sensitivity coefficient of fresh produce, cold chain temperature trajectory, and transportation time, the following formula is used: =1 - (Cumulative time of overheating / Total transportation time) * Freshness sensitivity coefficient, where: Freshness sensitivity coefficient: Preset by category, Grade A cucumber freshness sensitivity coefficient = 0.8, loose leaf lettuce freshness sensitivity coefficient = 1.2 (leafy vegetables are more sensitive to temperature, so the coefficient is higher); Cumulative duration of temperature exceeding the suitable transportation temperature range for this product category (Grade A cucumber 2-4℃, loose leaf lettuce 0-3℃) during cold chain transportation; Total transportation time: The cold chain transportation time promised by the supplier for this procurement.

[0042] Let's take Grade A cucumbers as an example: Supplier A: Total transportation time 1.5 hours (90 minutes), cumulative over-temperature time 0 minutes, freshness sensitivity coefficient 0.8, then =1−(0 / 90)*0.8=1.0(100%).

[0043] Supplier D: Total transportation time 2 hours (120 minutes), cumulative temperature exceedance time 10 minutes, freshness sensitivity coefficient 0.8, then =1-(10 / 120)*0.8≈0.933(93.3%) Time window matching parameters Based on the degree of match between the supplier's promised estimated delivery time and the store's receiving window (08:00-09:00), a value is selected according to preset rules: The estimated delivery time falls exactly within the receiving window (08:00-09:00). =1.00; The estimated delivery time is slightly off (07:45-08:00 or 09:00-09:15). =0.92; The estimated delivery time exceeds the tolerable offset range (<07:45 or >09:15). =0.65.

[0044] Example: Supplier A promises a delivery time of 08:20-08:40 (completely within the time window). =1.00; Supplier D promised delivery time 09:20 (outside the tolerable offset range). =0.65.

[0045] The final result is a set of freshness retention parameters and time window matching parameters, for example: Grade A cucumber: A ( =1.0, =1.0), B ( =0.96, =0.92), C ( =0.98, =1.0), D ( =0.933, =0.65), E ( =0.97, =0.92); Loose-leaf lettuce: A ( =0.95, =1.0), B ( =0.92, =0.92), C ( =0.96, =1.0), D ( =0.88, =0.65), E ( =0.94, =0.92).

[0046] The standardized comparative price, basic sellability parameter, freshness retention parameter, and time window matching parameter are integrated to form the multi-dimensional comparative price parameter, including: The standardized price comparison unit price, basic sellability parameter, freshness retention parameter and time window matching parameter are matched and associated to obtain a set of associated parameters; Specifically, this involves comparing standardized unit prices from the same supplier for the same product category. Basic availability rate Freshness retention Time window matching degree One-to-one associations are formed to create a set of associated parameters. Each associated parameter record corresponds to the "supplier-category" combination. For example: Supplier AA grade cucumber: (0.788 yuan / kg, 0.95, 1.0, 1.0); Supplier D-loose leaf lettuce: (0.498 yuan / kg, 0.862, 0.88, 0.65).

[0047] The set of related parameters is standardized in terms of units and validated in terms of validity to form multi-dimensional price comparison parameters; Includes the following steps: The parameters in the aforementioned set of related parameters are uniformly transformed to obtain a set of dimensionally normalized parameters. Specifically, dimensional normalization: Since the four types of parameters have different dimensions (standardized price ratio unit price is yuan / kg, the other three types are proportional coefficients), normalization is used to convert all parameters into a set of dimensionally normalized parameters in the 0-1 interval. The formula is: ,in These are the original parameter values. This is the minimum value of this type of parameter. This represents the maximum value of this type of parameter. For example, the maximum standardized price per unit (Grade A cucumber) is 0.796 yuan / kg (supplier D), and the minimum is 0.772 yuan / kg (supplier B). Then, the normalized standardized price per unit for supplier A is: (0.788-0.772) / (0.796-0.772)≈0.667; Based on preset parameter thresholds, the set of dimension-normalized parameters is judged item by item to obtain the set of parameters to be screened. The set of normalized parameters is evaluated item by item based on preset parameter thresholds to obtain the set of parameters to be screened. The preset thresholds are set as follows: normalized value of standardized price comparison unit price ≤ 0.9 (to avoid excessively high prices), normalized value of basic sellability ≥ 0.6 (to avoid excessively low sellability), normalized value of freshness retention ≥ 0.6 (to avoid excessively high freshness risk), and normalized value of time window matching degree ≥ 0.5 (to avoid severe time window mismatch). Abnormal and missing data are removed from the set of parameters to be screened to obtain valid and compliant parameter data; The set of parameters to be screened after screening is checked to remove abnormal data, such as parameter values ​​that are outside the reasonable range, data logic contradictions, and missing data, such as a certain parameter that was not collected, so as to obtain valid and compliant parameter data. In this embodiment, all the parameters of all suppliers meet the preset threshold and there is no abnormal or missing data. Therefore, the valid and compliant parameter data is consistent with the set of parameters to be screened. The valid and compliant parameter data are categorized and integrated to obtain multi-dimensional price comparison parameters; The valid and compliant parameter data are categorized and integrated according to the procurement category to form multi-dimensional price comparison parameters. The final multi-dimensional price comparison parameters are divided into two categories (Grade A cucumbers and loose-leaf lettuce). Each category contains four normalized parameters from five suppliers. For example, the multi-dimensional price comparison parameters for Grade A cucumbers are: A (0.667, 0.95, 1.0, 1.0), B (0.000, 0.93, 0.96, 0.92), C (0.333, 0.94, 0.98, 1.0), D (1.000, 0.88, 0.933, 0.65), and E (0.750, 0.92, 0.97, 0.92).

[0048] The dimensional normalization adopts the linear normalization method, and the preset parameter thresholds are set based on the acceptance standards of the fresh food procurement industry and the actual application scenario of this invention to ensure the effectiveness and comparability of multi-dimensional price comparison parameters.

[0049] The multi-dimensional price comparison parameters are jointly converted to obtain a dynamic set of salable equivalent costs; the multi-dimensional price comparison parameters are jointly converted through a preset formula to obtain a dynamic salable equivalent cost that can comprehensively reflect the real salable cost, and the scattered indicators are unified into core price comparison indicators that can be directly sorted.

[0050] By jointly converting the multi-dimensional price comparison parameters, a dynamic set of salable equivalent costs is obtained, including: Substitute the multi-dimensional price comparison parameters into the preset conversion formula to obtain the initial equivalent cost data; The preset conversion formula is: ,in Let be the initial equivalent cost (in yuan / kg) for the corresponding product category of the i-th supplier. The core meaning of this formula is: the standardized price comparison unit price divided by the product of the availability rate, freshness retention rate, and time window matching rate. If the supplier has a low availability rate, poor freshness, and low time window matching rate, its real available cost will be amplified, thus avoiding the misconception of nominally low price but actually high cost.

[0051] In this embodiment, Grade A cucumbers are used as an example: Supplier A: =0.788 / (0.95*1.0*1.0)≈0.829 yuan / kg; Supplier D: =0.796 / (0.88*0.933*0.65)≈0.796 / 0.537≈1.482 yuan / kg; Following the formula above, calculate the initial equivalent cost for all suppliers and the two product categories to obtain the initial equivalent cost data. For example: Initial equivalent cost of Grade A cucumber (yuan / kg): A=0.829, B=0.820, C=0.829, D=1.482, E=0.857; Initial equivalent cost of loose-leaf lettuce (yuan / kg): A=0.567, B=0.624, C=0.584, D=1.023, E=0.592.

[0052] The initial equivalent cost data is grouped and aggregated to obtain grouped equivalent cost data. The purpose of this step is to group the initial equivalent cost data according to reasonable dimensions to facilitate subsequent reasonableness verification and sorting; specifically, it includes: The correlation dimension information of the initial equivalent cost data is categorized and aggregated to obtain preliminary grouped data; Specifically, the data is grouped by supplier identification, purchase category, and purchase batch. In this embodiment, the purchase batch is the "morning replenishment batch." The related dimension information of the initial equivalent cost data is categorized and aggregated to obtain preliminary grouped data. For example, by "purchase category," the data is divided into a Grade A cucumber group, which includes the initial equivalent costs of 5 suppliers, and a loose leaf lettuce group, which includes the initial equivalent costs of 5 suppliers. By "supplier identification," each supplier corresponds to the initial equivalent costs of two categories.

[0053] The initial grouped data is subjected to duplicate item identification and merging processing to obtain deduplicated grouped data; Furthermore, the preliminary grouping data is checked to identify duplicates, such as duplicate calculations of the same supplier, the same product category, or the same batch, and these duplicates are merged to obtain deduplicated grouping data. In this embodiment, there are no duplicates, so the deduplicated grouping data is consistent with the preliminary grouping data.

[0054] Based on the deduplicated grouped data, dimension alignment and format unification processing are performed to obtain grouped equivalent cost data. Furthermore, the deduplicated grouped data is dimensionally aligned to ensure that the fields, parameter types, and formats of each group of data are consistent. For example, cost values ​​are retained to 3 decimal places, and supplier identifiers are uniformly coded to obtain grouped equivalent cost data. For example, the equivalent cost data for Grade A cucumbers is: Supplier A (0.829), B (0.820), C (0.829), D (1.482), E (0.857); the equivalent cost data for loose-leaf lettuce is: Supplier A (0.567), B (0.624), C (0.584), D (1.023), E (0.592).

[0055] The numerical rationality of the grouped equivalent cost data is verified to obtain calibrated equivalent cost data; specifically, the following methods are used: The principle is to perform a reasonableness check and calculate the average value of each group of equivalent cost data. and standard deviation Set a reasonable range as Data outside this range is considered abnormal and will be calibrated. Abnormal data will be replaced by the average value of this group.

[0056] Example (taking Grade A cucumber group as an example): average value =(0.829+0.820+0.829+1.482+0.857) / 5≈0.963 yuan / kg; Standard deviation ≈0.258 yuan / kg; The reasonable range is [0.963-3*0.258, 0.963+3*0.258]≈[0.189, 1.737] yuan / kg; Supplier D's initial equivalent cost of 1.482 yuan / kg is within a reasonable range and there are no abnormal data; therefore, the calibrated equivalent cost data is consistent with the grouped equivalent cost data. If a supplier's equivalent cost is 2.0 yuan / kg (outside the reasonable range), then the calibration value is 0.963 yuan / kg.

[0057] The calibrated equivalent cost data are uniformly organized to obtain a dynamic set of marketable equivalent costs; Specifically, the calibrated grouped equivalent cost data is uniformly organized according to the format of "purchase category - supplier" to form a dynamic available-for-sale equivalent cost set. This clarifies the dynamic available-for-sale equivalent cost for each supplier for each product category, providing a direct basis for subsequent ranking. The final dynamic available-for-sale equivalent cost set obtained in this embodiment is as follows: Grade A cucumbers: A (0.829), B (0.820), C (0.829), D (1.482), E (0.857); Loose-leaf lettuce: A (0.567), B (0.624), C (0.584), D (1.023), E (0.592).

[0058] The dynamic set of salable equivalent costs is sorted to obtain price comparison results. Through successive comparisons, logical corrections, stability marking, and screening and regularization of the dynamic set of salable equivalent costs, a scientifically reasonable price comparison result is obtained, taking into account both cost economy and supply stability. Specifically, the implementation process includes the following: The data of the dynamic salable equivalent cost set is compared one by one to obtain preliminary sorting data; Includes the following steps: The equivalent cost values ​​in the dynamic set of salable equivalent costs are compared pairwise to obtain the order of magnitude between the data. Specifically, taking Grade A cucumbers as an example, the dynamic available equivalent costs of the five suppliers are compared pairwise. For example: A(0.829) is compared with B(0.820), B < A; A(0.829) is compared with C(0.829), A = C; A(0.829) is compared with D(1.482), A < D; A(0.829) is compared with E(0.857), A < E; and so on, completing all pairwise comparisons to determine the order of the data: B(0.820) < A(0.829) = C(0.829) < E(0.857) < D(1.482).

[0059] Based on the aforementioned size order relationship, the data is arranged according to a preset sorting rule to obtain the initial sorted data; The preset sorting rules are as follows: dynamically salable equivalent cost is sorted from low to high, and if the costs are the same, the suppliers are sorted from high to low on-time delivery rate. Based on the on-time delivery rates of each supplier (A=98%, B=95%, C=96%, D=85%, E=97%), the initial sorting data for Grade A cucumbers is: 1. Supplier B (0.820), 2. Supplier A (0.829), 3. Supplier C (0.829), 4. Supplier E (0.857), 5. Supplier D (1.482); the initial sorting data for loose-leaf lettuce is obtained according to the same rules: 1. Supplier A (0.567), 2. Supplier C (0.584), 3. Supplier E (0.592), 4. Supplier B (0.624), 5. Supplier D (1.023).

[0060] The initial sorted data is logically corrected to obtain preliminary sorted data. This step ensures the logical correctness of the sorting results through a closed-loop process of related information extraction, logical verification, anomaly tracing, adjustment and correction, and full-domain review. The specific implementation process is as follows: The initial sorting data is used to extract related information to obtain a sorting verification dataset. This step is used to extract the dynamic salable equivalent cost, sorting number, supplier identifier, purchase category, on-time delivery rate and other related information of each supplier in the initial sorting data to construct the sorting verification dataset. For example, the A-level cucumber sorting verification dataset includes fields such as: serial number, supplier, equivalent cost, on-time delivery rate and other fields.

[0061] Perform a sequential logical check on the sorting verification dataset to identify the sorting order compliance judgment result; Furthermore, the logical consistency between the sorting sequence number and the equivalent cost is checked sequentially; that is, the smaller the sequence number, the lower the equivalent cost should be. Simultaneously, it is checked that suppliers with higher on-time delivery rates should be ranked higher when costs are the same. In this embodiment, the check found that suppliers A and C have the same equivalent cost (0.829 yuan / kg), but supplier C's on-time delivery rate (96%) is higher than supplier A's (98%), which is questionable. This is corrected to: supplier A's on-time delivery rate of 98% is higher than C's 96%, so the ranking is correct; there are no other logical anomalies. Therefore, the compliance determination result for the sorting order is: the initial sorting data is generally compliant, with no obvious logical anomalies.

[0062] Based on the sorting order compliance determination result, the sorting verification dataset is retrieved to obtain non-compliant data events; Furthermore, since the overall judgment result is compliant, there are no non-compliant data events. If there is a sorting anomaly, such as the equivalent cost of supplier number 3 being lower than that of supplier number 2, then the non-compliant data event will be retrieved. For example, the equivalent cost of supplier A in sequence number 2 is 0.829, and the equivalent cost of supplier C in sequence number 3 is 0.810, indicating an abnormal sorting logic.

[0063] The original equivalent cost calculation records of the non-compliant data events were traced to obtain the order deviation data; Specifically, if there are non-compliant data events, trace the supplier's initial equivalent cost calculation records, such as standardized price comparison unit price conversion, formula substitution, grouping and aggregation, etc., to confirm whether the cause of the anomaly is a sorting error or a deviation in the original data; in this embodiment, there are no non-compliant data events, so there is no order deviation data.

[0064] The order deviation data is adjusted and its arrangement position is temporarily stored to obtain preliminary corrected sorting data; Furthermore, if there is order deviation data, the outliers caused by sorting errors are repositioned according to the equivalent cost, and the outliers caused by deviations in the original data are marked, temporarily stored, and recorded separately; in this embodiment, there is no order deviation data, so the initial corrected sorting data is consistent with the initial sorting data.

[0065] A full-domain logical review is performed on the preliminary corrected sorting data to obtain the preliminary sorting data; Furthermore, the preliminary sorted data is checked again to ensure that there are no missing anomalies and that the sorting logic is consistent, thus obtaining the final preliminary sorted data; in this embodiment, the preliminary sorted data is consistent with the initial sorted data.

[0066] The preliminary sorted data is marked with quality and stability to obtain the marked sorted data; Based on cost ranking, the ranking data is marked in conjunction with the supplier's qualifications and contract performance stability to provide a basis for subsequent screening and avoid selecting suppliers with low costs but unqualified qualifications and poor stability.

[0067] In practice, the qualification documents of each supplier are checked. Suppliers with complete and valid qualifications are marked as "qualified" and those with incomplete or expired qualifications are marked as "unqualified". In this example, all 5 suppliers are qualified and are marked as "qualified", thus completing the qualification marking. The stability metric is based on supplier fulfillment data in the original price comparison dataset. The supply stability threshold is set at 0.85, which is specifically determined by the weighted average of the on-time delivery rate and the complete delivery rate over the most recent 30 days. That is, supply stability = (on-time delivery rate + complete delivery rate) / 2.

[0068] Stability ≥ 0.85 is marked as stable, stability < 0.85 is marked as unstable; the stability of each supplier is calculated as follows: A=0.98, B=0.95, C=0.96, D=0.85, E=0.97, all of which are ≥ 0.85 and are marked as stable.

[0069] The final sorted data after labeling is obtained, for example: Grade A cucumbers: 1. Supplier B (0.820, qualified, stable), 2. Supplier A (0.829, qualified, stable), 3. Supplier C (0.829, qualified, stable), 4. Supplier E (0.857, qualified, stable), 5. Supplier D (1.482, qualified, stable).

[0070] Based on the labeled and sorted data, high-cost items are removed to obtain filtered and sorted data; Specifically, a high cost threshold is set at "1.2 times the average dynamic saleable equivalent cost of this product category". High cost items exceeding this threshold are eliminated to avoid selecting suppliers with excessively high costs.

[0071] Example (using Grade A cucumbers as an example): Average value The cost is approximately 0.963 yuan / kg, and the threshold for high cost is 0.963 * 1.2 ≈ 1.156 yuan / kg. Supplier D's equivalent cost of 1.482 yuan / kg > 1.156 yuan / kg, so it is considered a high-cost item and is eliminated. The remaining suppliers are all within the threshold range and are retained. Therefore, the ranking data for Grade A cucumbers is: 1. Supplier B, 2. Supplier A, 3. Supplier C, 4. Supplier E.

[0072] Average value of loose leaf lettuce =(0.567+0.624+0.584+1.023+0.592) / 5≈0.678 yuan / kg, the cost threshold is 0.678*1.2≈0.814 yuan / kg; the equivalent cost of supplier D is 1.023 yuan / kg>0.814 yuan / kg, so it is excluded; the filtered and sorted data are: 1. Supplier A, 2. Supplier C, 3. Supplier E, 4. Supplier B.

[0073] The filtered and sorted data are then normalized in terms of serial number and format to generate a price comparison result. Specifically, the sorted data after screening is reorganized by numbering. After removing items with excessively high costs, the numbers are reordered from 1, and the format is standardized. For example, the formats of supplier identification, equivalent cost, and tag information are standardized. Finally, the price comparison results are generated, and the main and backup suppliers and their sorting priorities are clearly defined.

[0074] The final price comparison results of this embodiment are as follows: Price comparison results for Grade A cucumbers (sorted by priority): Main supplier: Supplier B (equivalent cost 0.820 yuan / kg, qualified, stable); Alternate supplier 1: Supplier A (equivalent cost 0.829 yuan / kg, qualified, stable); Alternate supplier 2: Supplier C (equivalent cost 0.829 yuan / kg, qualified, stable); Alternate supplier 3: Supplier E (equivalent cost 0.857 yuan / kg, qualified, stable).

[0075] Price comparison results for loose-leaf lettuce (sorted by priority): Main supplier: Supplier A (equivalent cost 0.567 yuan / kg, qualified, stable); Alternate supplier 1: Supplier C (equivalent cost 0.584 yuan / kg, qualified, stable); Alternate supplier 2: Supplier E (equivalent cost 0.592 yuan / kg, qualified, stable); Alternate supplier 3: Supplier B (equivalent cost 0.624 yuan / kg, qualified, stable).

[0076] Meanwhile, since the difference in dynamic saleable equivalent costs between the primary and backup suppliers both exceed the order splitting trigger threshold, order splitting is not triggered, and the price comparison result is a single primary supplier plus multiple backup suppliers. The price comparison results are corrected by combining actual delivery and loss feedback data to obtain an updated price comparison parameter set. This step uses actual performance feedback data to perform closed-loop correction of the price comparison parameters, achieving self-calibration of the price comparison model and avoiding price distortion caused by long-term use of static parameters. Through actual data collection, deviation analysis, correction coefficient generation, parameter adjustment, and rationality verification, the price comparison parameters are dynamically updated to ensure that the corrected price comparison parameter set can accurately adapt to the price comparison of the next batch of purchases. The specific implementation process is as follows: The price comparison results and actual delivery and loss feedback data are collected synchronously to construct a correction and verification dataset; Specific implementation: After the morning replenishment batch is completed, actual arrival data will be collected, such as: actual arrival quantity, delivery time, actual acceptance status, and actual loss feedback data for each supplier, such as: trimming loss rate, decay rate, and actual salable quantity. This data will be collected synchronously with the price comparison results, dynamic salable equivalent cost, sorting priority, and various price comparison parameters to build a correction and verification dataset.

[0077] Example: Supplier BA grade cucumbers are the main supplier. The actual delivery was 100kg at 08:30. The actual acceptance quantity was 92kg. The pruning loss was 6kg and the rot loss was 2kg. The actual sellable rate = 92 / 100 = 0.92. The actual equivalent cost = actual purchase cost / actual sellable quantity = (0.820 yuan / kg * 100kg) / 92kg ≈ 0.891 yuan / kg. There is a discrepancy with the dynamic sellable equivalent cost of 0.820 yuan / kg and the basic sellable rate of 0.93 in the price comparison results.

[0078] A deviation comparison analysis was performed on the corrected verification dataset to obtain the actual difference information; By comparing the parameters in the price comparison results—dynamic salable equivalent cost, basic salability rate, freshness retention, and time window matching—with the deviations from the actual data, the deviation values ​​are calculated to obtain the actual difference information.

[0079] Example: Actual discrepancy information for supplier B's grade A cucumbers: Dynamic available equivalent cost deviation: 0.891 - 0.820 = 0.071 yuan / kg; Basic availability deviation: 0.92 - 0.93 = -0.01; Freshness retention deviation: Actual freshness retention (calculated based on actual spoilage rate as 0.98) - Freshness retention at price comparison time 0.96 = 0.02; Time window matching deviation: Actual time window matching degree 1.0 (on-time delivery) - Time window matching degree during price comparison 0.92 = 0.08.

[0080] The actual difference information is converted by combining preset weights to generate corresponding correction coefficients; The specific weights are: dynamic salable equivalent cost (0.4), basic salability rate (0.2), freshness retention (0.2), and time window matching (0.2); the correction coefficient is calculated using the following formula: ,in The corresponding correction factor is the difference between the actual data and the price comparison parameter.

[0081] Example: Supplier B: Calculation of correction factor for Grade A cucumbers: =1+(0.071*0.4)+(−0.01*0.2)+(0.02*0.2)+(0.08*0.2)=1+0.0284−0.002+0.004+0.016=1.0464.

[0082] Following the method described above, calculate the corresponding correction coefficients for all suppliers and the two product categories, and generate a set of correction coefficients.

[0083] Based on the corresponding correction coefficients, the parameters in the price comparison results are adjusted item by item to obtain preliminary corrected price comparison parameters; The price comparison parameters for each supplier's corresponding product category—standardized price comparison unit price, basic sellability, freshness retention, and time window matching—are multiplied by their respective correction coefficients to complete the item-by-item adjustment, resulting in the preliminary corrected price comparison parameters.

[0084] Example: Supplier B: Preliminary revision of price comparison parameters for Grade A cucumbers: The revised standardized price per kg is: 0.772 yuan / kg * 1.0464 ≈ 0.808 yuan / kg; Corrected base availability: 0.93 * 1.0464 ≈ 0.973; Corrected freshness retention: 0.96 * 1.0464 ≈ 1.004 (if it exceeds 1.0, take the value as 1.0). Corrected window matching accuracy: 0.92 * 1.0464 ≈ 0.963.

[0085] The rationality of the initially revised price comparison parameters is verified to obtain the updated price comparison parameter set; The reasonableness of the initially revised price comparison parameters is checked to ensure that each parameter is within a reasonable range. For example, the sellability rate, freshness retention rate, and time window matching degree are all in the range of 0-1, and the standardized price comparison unit price is within a reasonable price range. Abnormal correction items are removed, and finally the updated price comparison parameter set is obtained for the next batch of fresh food procurement price comparison.

[0086] In this embodiment, all the initially corrected price comparison parameters are within a reasonable range. Therefore, the updated price comparison parameter set includes the four corrected core parameters for each supplier and the two product categories. For example, the updated parameters for supplier B (Grade A cucumber) are: (0.808 yuan / kg, 0.973, 1.0, 0.963).

[0087] Furthermore, a feedback correction cycle of 7 days is set. Every 7 days, based on the actual arrival and loss feedback data of the past 7 days, the above steps are repeated to continuously update and optimize the price comparison parameters, ensuring the long-term accuracy and effectiveness of the price comparison results. The updated price comparison parameter set will directly serve as the basis for the price comparison of the next batch of fresh produce procurement, replacing the original price comparison parameters of the previous batch. After each batch of procurement is completed, the above feedback correction steps are repeated to form a complete closed loop. At the same time, the effect of the corrected price comparison parameter set is verified every 30 days, comparing the overall procurement cost before and after the correction. If the overall cost reduction rate is less than 10%, the correction coefficient weight and preset threshold are readjusted to ensure the continuous optimization and effectiveness of the technical solution, achieving long-term accurate adaptation of the price comparison results to the fresh produce procurement scenario.

[0088] To clearly demonstrate the differences between this invention and existing technologies, the effectiveness of this invention is compared with traditional price comparison methods, as shown in the table below: Table 1 Comparison of the effects of the present invention and traditional methods

[0089] Based on Table 1 above, it can be seen that compared with traditional price comparison methods, the present invention has significant advantages in terms of the uniformity of price comparison benchmarks, coverage of multi-dimensional parameters, accuracy of cost accounting, scientific decision-making, and dynamic optimization capabilities. It effectively overcomes the defects of traditional methods, such as chaotic price comparison benchmarks, disconnect between price and performance risk, rough decision-making, and easy distortion with long-term use. It realizes refined and dynamic management of fresh food procurement price comparison and is more suitable for the needs of high-frequency, small-batch fresh food procurement scenarios.

[0090] Combination Figure 2 As shown, Figure 2 The horizontal axis represents six core technical indicators: uniformity of price comparison benchmarks, completeness of parameter dimensions, accuracy of cost accounting, scientific and stable decision-making, dynamic optimization and self-calibration capabilities, and price and risk management capabilities. The vertical axis represents the technical effectiveness score, with higher scores indicating better technical effectiveness.

[0091] Depend on Figure 2It is evident that the scores of this invention are significantly higher than those of traditional price comparison methods across all indicators. In particular, traditional methods have obvious shortcomings in terms of parameter dimension completeness, dynamic optimization and self-calibration capabilities, and price and risk management capabilities. This invention achieves a breakthrough improvement, fully demonstrating its creative advantages in solving problems such as inconsistent traditional price comparison benchmarks, disconnect between price and risk, and crude and distorted decision-making.

[0092] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application.

[0093] Furthermore, those skilled in the art will understand that although some embodiments herein include certain features included in other embodiments but not others, combinations of features from different embodiments are meant to be within the scope of this application and form different embodiments. For example, all the embodiments above can be used in any combination. The information disclosed in this background section is intended only to enhance the understanding of the general background of this application and should not be construed as an admission or in any way implying that such information constitutes prior art known to those skilled in the art.

[0094] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely preferred examples and are not intended to limit the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A smart price comparison method for fresh produce procurement based on big data, characterized in that, Includes the following steps: By aggregating and integrating procurement data from multiple sources, the original price comparison dataset is obtained. Based on the original price comparison dataset, multi-dimensional parameter processing is performed to obtain multi-dimensional price comparison parameters; By jointly converting the multi-dimensional price comparison parameters, a dynamic set of salable equivalent costs is obtained; The dynamic set of salable equivalent costs is sorted to obtain the price comparison results; The price comparison results are corrected by combining actual delivery and loss feedback data to obtain an updated price comparison parameter set.

2. The intelligent price comparison method for fresh produce procurement according to claim 1, characterized in that, Based on the original price comparison dataset, multi-dimensional parameter processing is performed to obtain multi-dimensional price comparison parameters, including: Based on the original price comparison dataset, the supplier quotations are converted according to specifications and grades to obtain a standardized price comparison unit price; Based on the historical acceptance and loss data in the original price comparison dataset, a sales rate model is performed to obtain basic sales rate parameters. Based on the cold chain temperature trajectory in the original price comparison dataset and the store receiving time window data, freshness preservation and time window matching correction are performed to obtain freshness preservation parameters and time window matching parameters. The standardized price comparison unit price, basic sellability parameter, freshness retention parameter, and time window matching parameter are integrated to form the multi-dimensional price comparison parameter.

3. The intelligent price comparison method for fresh produce procurement according to claim 2, characterized in that, The standardized comparative price, basic sellability parameter, freshness retention parameter, and time window matching parameter are integrated to form the multi-dimensional comparative price parameter, including: The standardized price comparison unit price, basic sellability parameter, freshness retention parameter and time window matching parameter are matched and associated to obtain a set of associated parameters; The set of related parameters is standardized in terms of dimensions and its validity is verified to form multi-dimensional price comparison parameters.

4. The intelligent price comparison method for fresh produce procurement according to claim 3, characterized in that, The aforementioned set of related parameters is subjected to dimensional unification and validity verification to form multi-dimensional price comparison parameters, including: The parameters in the aforementioned set of related parameters are uniformly transformed to obtain a set of dimensionally normalized parameters. Based on preset parameter thresholds, the set of dimension-normalized parameters is judged item by item to obtain the set of parameters to be screened. Abnormal and missing data are removed from the set of parameters to be screened to obtain valid and compliant parameter data; The valid and compliant parameter data are categorized and integrated to obtain multi-dimensional price comparison parameters.

5. The intelligent price comparison method for fresh produce procurement according to claim 1, characterized in that, By jointly converting the multi-dimensional price comparison parameters, a dynamic set of salable equivalent costs is obtained, including: Substitute the multi-dimensional price comparison parameters into the preset conversion formula to obtain the initial equivalent cost data; The initial equivalent cost data is grouped and aggregated to obtain grouped equivalent cost data; The numerical rationality of the grouped equivalent cost data is verified to obtain calibrated equivalent cost data; The calibrated equivalent cost data are then uniformly organized to obtain a dynamic set of marketable equivalent costs.

6. The intelligent price comparison method for fresh produce procurement according to claim 5, characterized in that, The initial equivalent cost data is grouped and aggregated to obtain grouped equivalent cost data, including: The correlation dimension information of the initial equivalent cost data is categorized and aggregated to obtain preliminary grouped data; The initial grouped data is subjected to duplicate item identification and merging processing to obtain deduplicated grouped data; Based on the deduplicated grouped data, dimensional alignment and format unification processing are performed to obtain grouped equivalent cost data.

7. The intelligent price comparison method for fresh produce procurement according to claim 1, characterized in that, The dynamic set of salable equivalent costs is sorted to obtain the price comparison results, including: The data of the dynamic salable equivalent cost set is compared one by one to obtain preliminary sorting data; The preliminary sorted data is marked with quality and stability to obtain the marked sorted data; Based on the labeled and sorted data, high-cost items are removed to obtain filtered and sorted data; The filtered and sorted data are then normalized in terms of serial number and format to generate a price comparison result.

8. The intelligent price comparison method for fresh produce procurement according to claim 7, characterized in that, The data of the dynamic salable equivalent cost set is compared step by step to obtain preliminary sorting data, including: The equivalent cost values ​​in the dynamic set of salable equivalent costs are compared pairwise to obtain the order of magnitude between the data. Based on the aforementioned size order relationship, the data is arranged according to a preset sorting rule to obtain the initial sorted data; The initial sorted data is logically corrected to obtain preliminary sorted data.

9. The intelligent price comparison method for fresh produce procurement according to claim 8, characterized in that, The initial sorted data is logically corrected to obtain preliminary sorted data, including: The initial sorted data is subjected to correlation information extraction to obtain a sorting verification dataset; Perform a sequential logical check on the sorting verification dataset to identify the sorting order compliance judgment result; Based on the sorting order compliance determination result, the sorting verification dataset is retrieved to obtain non-compliant data events; The original equivalent cost calculation records of the non-compliant data events were traced to obtain the order deviation data; The order deviation data is adjusted and its arrangement position is temporarily stored to obtain preliminary corrected sorting data; A full-domain logical review is performed on the preliminary corrected sorting data to obtain the preliminary sorting data.

10. The intelligent price comparison method for fresh produce procurement according to claim 1, characterized in that, The price comparison results are revised by incorporating actual delivery and loss feedback data to obtain an updated set of price comparison parameters, including: The price comparison results and actual delivery and loss feedback data are collected synchronously to construct a correction and verification dataset; A deviation comparison analysis was performed on the corrected verification dataset to obtain the actual difference information; The actual difference information is converted by combining preset weights to generate corresponding correction coefficients; Based on the corresponding correction coefficients, the parameters in the price comparison results are adjusted item by item to obtain preliminary corrected price comparison parameters; The rationality of the initially revised price comparison parameters is verified to obtain the updated price comparison parameter set.

Images