A multi-level audit and dynamic early warning method for fresh food prices

Abstract

The present application relates to the technical field of data processing, and discloses a multi-level auditing and dynamic early warning method for fresh food prices. The method collects storage and logistics time sequence environmental parameters, calculates real-time quality decay coefficients in combination with fresh food category benchmark spoilage rates; non-linearly fuses procurement cost benchmarks, real-time quality decay coefficients and supply and demand fluctuation factors to generate a dynamic price auditing threshold interval that automatically narrows over time; receives declared prices and compares them with the dynamic threshold interval at the corresponding time; if the threshold is exceeded, an early warning is triggered, regular auditing is frozen and a high-level auditing node is forcibly introduced. The present application enables the auditing benchmark and the actual value state of fresh food to evolve synchronously, overcomes the failure of static thresholds in quality decay, avoids value loss caused by auditing lag, and realizes early intervention of risks.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and discloses a method for multi-level review and dynamic early warning of fresh produce prices. Background Technology

[0002] The current price review process in the fresh produce supply chain mainly relies on manual or semi-automated numerical comparison systems. In practice, when purchasing or sales personnel submit a declared price for fresh produce, the review system extracts this price and compares it with a fixed price range or static gross profit margin pre-stored in a database. This fixed price range is typically calculated based on historical average procurement costs, standard transportation costs, and a preset profit margin, and is written to a configuration file during system initialization or periodic maintenance. Based on the comparison results, the review system decides whether to proceed the declared price to the next review node. The entire process follows a preset, fixed-level, serial structure, requiring traversal of all preset nodes regardless of price fluctuations.

[0003] In practice, fresh produce is perishable and has a short shelf life, and its quality and value continuously change with temperature, humidity, and time during storage and logistics. However, existing auditing systems only compare declared prices with statically configured fixed thresholds, making the auditing benchmark independent of the real-time physical state of the fresh produce and changes in the external environment. When fresh produce deteriorates in quality due to abnormal temperature and humidity during logistics, the declared price still needs to be compared with a fixed price range that does not consider spoilage. Even if the declared price deviates from the current actual value of the fresh produce, the auditing system still determines it to be reasonable and proceeds with the normal flow of goods because the fixed threshold is not adjusted accordingly, resulting in a disconnect between the auditing benchmark and the dynamic value of the fresh produce.

[0004] The core technical problem with existing technologies is that static, fixed review thresholds cannot adaptively adjust to the dynamic decay of fresh produce's physical condition. Under the existing comparison mechanism, the declared price is compared with a static value detached from the physical environment. When the quality of fresh produce undergoes non-linear decay over time and due to environmental fluctuations, the review benchmark cannot reflect the downward trend in the actual value of the fresh produce. Low-value, deteriorated fresh produce still passes review according to the conventional high-value benchmark, resulting in a mismatch between the review benchmark and the actual value of the fresh produce, leading to review delays and inaccurate early warnings. Summary of the Invention

[0005] The purpose of this invention is to provide a method for multi-level review and dynamic early warning of fresh produce prices, which can effectively solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the technical solution adopted by the present invention is as follows: A method for multi-level review and dynamic early warning of fresh produce prices includes: collecting time-series environmental parameters of fresh produce in the warehousing and logistics process, combining them with the benchmark spoilage rate of fresh produce categories, and calculating the real-time quality decay coefficient. Obtain the procurement cost benchmark, and nonlinearly fuse the procurement cost benchmark, the real-time quality decay coefficient, and the supply and demand fluctuation factor to generate a dynamic price review threshold range that continuously evolves over time. The dynamic price review threshold range automatically narrows over time. Receive the declared price and compare the declared price with the dynamic price review threshold range at the corresponding time. If the declared price exceeds the limit, an early warning will be triggered and the regular review process will be frozen, forcibly introducing a higher-level review node. If the declared price does not exceed the limit, the review will be carried out along the regular review process path.

[0007] Preferably, the collection of time-series environmental parameters of fresh produce in the warehousing and logistics process, combined with the benchmark spoilage rate of fresh produce category, to calculate the real-time quality decay coefficient includes: extracting time-series temperature data, time-series humidity data, and time-series vibration frequency data in the warehousing and logistics process as the time-series environmental parameters. Based on the fresh produce category attributes, a first attenuation weight corresponding to the temperature time series data, a second attenuation weight corresponding to the humidity time series data, and a third attenuation weight corresponding to the vibration frequency time series data are determined. The temperature time series data, the humidity time series data, and the vibration frequency time series data are weighted and summed with the corresponding first attenuation weight, second attenuation weight, and third attenuation weight, respectively, to obtain the comprehensive environmental disturbance value; The comprehensive environmental disturbance value and the benchmark spoilage rate of the fresh produce category are integrated over time to output the real-time quality decay coefficient.

[0008] Preferably, before obtaining the procurement cost benchmark and nonlinearly fusing the procurement cost benchmark, the real-time quality decay coefficient, and the supply and demand fluctuation factor to generate a dynamic price review threshold range that continuously evolves over time, the method further includes: obtaining the historical transaction order volume and historical canceled order volume within the previous time window of the current region, and calculating the ratio of the historical canceled order volume to the historical transaction order volume as the regional supply and demand tightness. Retrieve the real-time transfer inbound and real-time consumption of adjacent related regions within the preceding time window, and calculate the ratio of the difference between the real-time consumption and the real-time transfer inbound as the cross-regional supply and demand flow. A nonlinear mapping is performed between the regional supply and demand tightness and the cross-regional supply and demand mobility, and the mapping result is determined as the supply and demand fluctuation factor.

[0009] Preferably, the procurement cost benchmark, the real-time quality decay coefficient, and the supply and demand fluctuation factor are nonlinearly fused to generate a dynamic price review threshold range that evolves continuously over time, including: using the procurement cost benchmark as the benchmark price origin, using the real-time quality decay coefficient as the price reduction index, and performing exponential decay calculation on the benchmark price origin to obtain the quality decay benchmark price. The supply and demand fluctuation factor is used as the price fluctuation correction coefficient. The price fluctuation correction coefficient is multiplied by the quality deterioration benchmark price to obtain the dynamic price center value. Based on the dynamic price center value, an upward and downward fluctuation ratio is set to generate the upper and lower thresholds of the dynamic price review threshold range. The upward and downward fluctuation ratio is inversely proportional to the time elapsed since the current time of entry into the warehouse, so that the dynamic price review threshold range automatically narrows over time.

[0010] Preferably, the step of receiving the declared price and comparing the declared price with the dynamic price review threshold range at the corresponding time includes: receiving the declared price to be reviewed and the fresh produce batch location identifier associated with the declared price; Based on the fresh produce batch location identifier, the local environmental offset in the corresponding storage microenvironment is extracted; Substitute the local environmental offset into the dynamic price review threshold range, and perform offset compensation on the upper and lower thresholds of the dynamic price review threshold range to generate a local dynamic price review threshold range corresponding to the fresh produce batch location identifier. The declared price is compared with the local dynamic price review threshold range to determine the deviation of the declared price from the local dynamic price review threshold range.

[0011] Preferably, if the declared price exceeds the limit, an early warning is triggered and the regular review process is frozen, and a higher-level review node is forcibly introduced, including: when the declared price is lower than the lower limit of the dynamic price review threshold range, calculating the downward offset difference between the declared price and the lower limit threshold. When the declared price is higher than the upper limit of the dynamic price review threshold range, calculate the upper offset difference between the declared price and the upper limit threshold; Input the lower offset difference or the upper offset difference into the review level mapping function, and output the corresponding higher-order review level identifier; Based on the higher-level review identifier, the regular review flow path is frozen, a redirected review path is generated pointing to the review terminal corresponding to the higher-level review identifier, and the warning is sent to the review terminal.

[0012] Preferably, the comprehensive environmental disturbance value and the benchmark spoilage rate of fresh produce are integrated over time to output the real-time quality decay coefficient, including: obtaining the rate of change of the temperature time-series data, the humidity time-series data and the vibration frequency time-series data in adjacent time slices; When the rate of change exceeds a preset rate of change threshold, extract the mutation environment data segment that exceeds the preset rate of change threshold; The mutation environment data segment is assigned a mutation acceleration weight, and the product of the mutation environment data segment and the mutation acceleration weight is added to the integral result of the integral operation. The integral result after superimposing the mutation acceleration weight is indexed, and the indexed result is determined as the real-time quality degradation coefficient to characterize the accelerating effect of environmental mutation on quality degradation.

[0013] Preferably, the step of retrieving the real-time transfer inbound volume and real-time consumption volume of adjacent related regions within the preceding time window, and calculating the ratio of the difference between the real-time consumption volume and the real-time transfer inbound volume as the cross-regional supply and demand flow, includes: obtaining logistics network status data between the adjacent related regions, wherein the logistics network status data includes the network congestion index and the delay time of transportation vehicle scheduling. The road network timeliness loss coefficient is calculated based on the road network congestion index and the vehicle dispatch delay time. Multiply the initial difference ratio between the real-time consumption and the real-time allocation and storage by the road network timeliness loss coefficient to attenuate and correct the initial difference ratio; The percentage of the difference after attenuation correction is determined as the cross-regional supply and demand mobility, in order to reflect the reduction in actual cross-regional supply capacity due to obstruction of the logistics network.

[0014] Preferably, based on the dynamic price center value, an upper and lower fluctuation ratio is set to generate the upper and lower thresholds of the dynamic price review threshold range, including: obtaining the historical application rejection records of the fresh food category within the historical review period; Extract the rejection price distribution density from the historical rejection records, and determine the historical rejection price concentration range based on the rejection price distribution density; Calculate the deviation distance between the dynamic price center value and the historical rejection price concentration interval; If the deviation distance is less than a preset deviation threshold, the value of the vertical fluctuation ratio is reduced; if the deviation distance is greater than or equal to the preset deviation threshold, the value of the vertical fluctuation ratio is maintained. Based on the adjusted upward and downward fluctuation ratios and the dynamic price center value, the upper and lower thresholds of the dynamic price review threshold range are generated.

[0015] Preferably, the local environment offset is substituted into the dynamic price review threshold range, and the upper and lower thresholds of the dynamic price review threshold range are offset compensated to generate a local dynamic price review threshold range corresponding to the fresh produce batch location identifier. This includes: performing spatial clustering of the fresh produce batch location identifier in the warehouse space coordinate system to divide it into multiple local environment clusters. Calculate the average of the local environment offsets corresponding to the location identifiers of all fresh produce batches within the same local environment cluster, and use this as the cluster-level environment offset. Substitute the cluster-level environment offset into the dynamic price review threshold range, replace the local environment offset for offset compensation, and generate a cluster-level dynamic price review threshold range corresponding to the local environment cluster. Fresh produce batches belonging to the same local environmental cluster will be compared using the cluster-level dynamic price review threshold range.

[0016] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. By collecting time-series environmental parameters from warehousing and logistics processes and combining them with the benchmark spoilage rate for fresh produce to calculate a real-time quality decay coefficient, the system nonlinearly integrates the procurement cost benchmark, the real-time quality decay coefficient, and supply and demand fluctuation factors to generate a dynamic price review threshold range that continuously evolves over time. This allows the review benchmark to adjust synchronously with the evolution of the physical state of the fresh produce. The dynamic threshold range automatically narrows over time, overcoming the failure of fixed threshold review when faced with the nonlinear decay of fresh produce quality over time. This ensures that the price review benchmark remains synchronized with the actual value of the fresh produce, preventing further loss of fresh produce value due to delayed review.

[0017] 2. Upon receiving a declared price, the system compares it to the corresponding dynamic price review threshold range. If the price exceeds the threshold, an alert is triggered, the regular review process is frozen, and a higher-level review node is forcibly introduced, enabling early intervention against review delay risks. Based on the location identifier of fresh produce batches, local environmental offsets are extracted to compensate for the threshold range, generating a local dynamic price review threshold range. This transforms the quantified physical state value into control instructions for the review process. When the declared price deviates from the local threshold range, the offset difference is calculated and mapped to a higher-level review level, generating a redirected review path, improving the accuracy and reliability of intercepting abnormal pricing.

[0018] 3. By introducing regional supply-demand tightness and cross-regional supply-demand mobility, a supply-demand fluctuation factor is calculated. In the cross-regional calculation, a road network timeliness loss coefficient is introduced to attenuate and correct the initial difference ratio, reflecting the reduction in actual supply capacity due to logistics network obstruction. When calculating the real-time quality degradation coefficient, a mutation acceleration weight is assigned to data segments of abrupt environmental changes exceeding a preset change rate threshold, characterizing the accelerated impact of environmental mutations on quality degradation. Spatial clustering is used to generate cluster-level dynamic price review threshold ranges, reducing the independent calculation overhead of individual batches and lowering the data processing latency and computational resource consumption of the review system. Attached Figure Description

[0019] Figure 1 This is the overall flowchart for calculating the real-time quality degradation coefficient of fresh produce according to the present invention; Figure 2 This is a flowchart illustrating the generation process of the fresh produce supply and demand fluctuation factor in this invention. Figure 3 A flowchart for generating the dynamic price review threshold range of this invention; Figure 4 This is a flowchart of the local / cluster-level dynamic price threshold offset compensation process of the present invention; Figure 5 A flowchart illustrating the calculation of the real-time quality degradation coefficient for environmental abrupt correction in this invention. Figure 6 This is a flowchart illustrating the multi-level review and dynamic early warning intervention process for fresh produce declared prices, as described in this invention. Detailed Implementation

[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] Please refer to Figure 1 This embodiment provides a multi-level price review and dynamic early warning system for fresh produce, deployed on a fresh produce supply chain management platform. It integrates with real-time data interfaces with warehousing environment monitoring, logistics tracking, procurement management, sales declaration, and review management systems. The system acquires structured data from each subsystem through standardized API interfaces and employs a distributed data processing architecture to process massive amounts of time-series environmental and business data in parallel, ensuring real-time performance and accuracy. The core logic of the system is to transform the physical quality and market supply and demand of fresh produce into quantifiable price review benchmarks. These benchmarks are dynamically adjusted according to the value of fresh produce and the market environment, and tiered review and early warning intervention are implemented based on the degree of price deviation.

[0022] In one embodiment, the system collects real-time environmental parameters of each fresh produce batch throughout the entire warehousing and logistics chain. The warehousing environment monitoring system deploys temperature sensors, humidity sensors, and triaxial accelerometers at different locations on each shelf level and in the refrigerated truck compartment within the warehouse space. The temperature sensors are digital, with a sampling interval of 1 minute, a measurement accuracy of ±0.1℃, and a measurement range of -20℃ to 60℃, meeting the low-temperature environmental requirements of fresh produce warehousing and logistics. The humidity sensors are capacitive, with a sampling interval of 1 minute, a measurement accuracy of ±2%RH, and a measurement range of 0%RH to 100%RH, exhibiting good long-term stability and anti-interference capabilities. The triaxial accelerometers have a sampling frequency of 100Hz and a measurement range of ±16g, capable of capturing various vibration signals generated during transportation. The system performs a Fast Fourier Transform on the raw acceleration data collected by the triaxial accelerometers to extract vibration signals within the frequency range of 10Hz to 100Hz, as vibrations in this frequency range cause the most significant physical damage to fresh produce. The system counts the number of vibrations within a given frequency range per minute as the vibration frequency, generating vibration frequency time-series data. Each time-series data point contains a unique timestamp, sensor identifier, and corresponding measurement value. The system associates and stores this data with the fresh produce batch identifier, forming a full-chain environmental parameter profile for each fresh produce batch.

[0023] The system retrieves the corresponding baseline spoilage rate and environmental parameter degradation weights for each fresh produce batch from the database. The baseline spoilage rate, calculated as the percentage of quality degradation per unit time under standard warehousing and logistics conditions through extensive accelerated aging experiments and actual supply chain tracking experiments, is expressed as 1 / hour. Different fresh produce categories exhibit significant differences in natural perishability, resulting in a wide range of baseline spoilage rate values. The environmental parameter degradation weights reflect the varying sensitivities of different fresh produce categories to three environmental factors: temperature, humidity, and vibration; the sum of these three degradation weights is 1. The system pre-stores a table of baseline spoilage rates and degradation weights for all fresh produce categories, which system administrators can periodically update based on actual operational data.

[0024] The system normalizes the collected temperature, humidity, and vibration frequency time-series data to eliminate the influence of different environmental parameter dimensions. The normalization process uses a min-max normalization method, mapping the values ​​of each parameter to a range of 0 to 1. For temperature time-series data, the normalization formula is:

[0025] in, The actual temperature measurement value at time t. This is the minimum suitable storage temperature for this type of fresh produce. This is the maximum suitable storage temperature for this type of fresh produce. When the actual temperature is lower than... or higher hour, The values ​​are 0 or 1, respectively. Similarly, the system performs the same normalization process on the humidity time series data and the vibration frequency time series data to obtain normalized humidity time series data. and normalized vibration frequency time series data .

[0026] The system weights and sums the normalized temperature time-series data, humidity time-series data, and vibration frequency time-series data with their corresponding first, second, and third attenuation weights, respectively, to obtain the comprehensive environmental disturbance value. The formula for calculating the comprehensive environmental disturbance value is as follows:

[0027] in, The first attenuation weight is temperature. The second attenuation weight (humidity). This is the third attenuation weight (vibration). (Comprehensive environmental disturbance value) The value ranges from 0 to 1. The larger the value, the more severe the adverse impact of the current environmental conditions on the quality of fresh produce.

[0028] The system integrates the comprehensive environmental disturbance value with the benchmark spoilage rate for fresh produce over time, outputting a real-time quality degradation coefficient. The integration calculation uses a trapezoidal numerical integration method, starting from the time the fresh produce batch enters the warehouse. up to the current moment The time interval is divided into multiple time steps of equal length, with each time step having the same length as the sampling interval of the environmental parameters, i.e., 1 minute. The system calculates the average comprehensive environmental disturbance value within each time step, then multiplies it by the time step length to obtain the integral contribution value for that time step. The system sums the integral contribution values ​​of all time steps to obtain the total integral result. The formula for calculating the real-time quality degradation coefficient is:

[0029] in, As the benchmark spoilage rate for fresh produce, This is the integral result of the comprehensive environmental disturbance value from the time of entry into the warehouse to the current time. Real-time quality degradation coefficient. The value range is from 0 to 1, when This indicates that the quality of the fresh produce has not deteriorated at all. This indicates that the fresh produce has completely lost its quality and has no sales value. Table 1 shows the comparison of the benchmark spoilage rate and decay weight for different fresh produce categories.

[0030] Table 1. Comparison of Benchmark Perishability and Decay Weight for Different Fresh Produce Categories

[0031] Table 1 lists the baseline spoilage rates of common fresh produce categories and the corresponding attenuation weights for environmental parameters. The system automatically retrieves the corresponding parameters for calculation based on the category identifier of the fresh produce batch. The differences in baseline spoilage rates among different categories reflect their varying natural perishability; for example, berries have the highest baseline spoilage rate, while root vegetables have the lowest. The differences in attenuation weights reflect their varying sensitivities to different environmental factors; for example, poultry and seafood are most sensitive to temperature, therefore the first attenuation weight is the largest; berries are most sensitive to vibration, therefore the third attenuation weight is the largest.

[0032] The system retrieves the procurement cost benchmark for the corresponding fresh produce batch from the procurement management system. The procurement cost benchmark is the unit purchase price of that batch of fresh produce plus the unit standard transportation cost and the unit standard warehousing cost. The unit purchase price is the total purchase amount of the batch of fresh produce divided by its total weight; the unit standard transportation cost is the total transportation cost from the production site to the warehousing center divided by its total weight; and the unit standard warehousing cost is the daily warehousing cost under standard warehousing conditions divided by its total weight. The procurement cost benchmark is the basis for calculating the dynamic price review threshold range and reflects the initial cost value of fresh produce.

[0033] refer to Figure 3 The system uses the procurement cost benchmark as the origin of the benchmark price and the real-time quality degradation coefficient as the price reduction index. It then performs exponential degradation calculations on the origin of the benchmark price to obtain the quality degradation benchmark price. The quality degradation benchmark price reflects the basic value of fresh produce at its current quality state, and the calculation formula is as follows:

[0034] in, As a benchmark for procurement costs, Let be the real-time quality degradation coefficient at time t. As time progresses, the real-time quality degradation coefficient... Gradually increasing, quality deterioration benchmark price The gradual decrease accurately reflects the decline in value of fresh produce due to quality degradation.

[0035] The system uses supply and demand fluctuation factors as price fluctuation correction coefficients, and multiplies these coefficients by the quality degradation benchmark price to obtain the dynamic price center value. The dynamic price center value reflects the reasonable price center for fresh produce under the current quality and market supply and demand conditions. The calculation formula is as follows:

[0036] in, Let be the supply and demand volatility factor at time t. The value of the supply and demand volatility factor ranges from 0.8 to 1.5. When the market supply exceeds demand, prices should fluctuate downwards; when This indicates that market supply is insufficient to meet demand, and prices should rise.

[0037] The system sets upward and downward fluctuation ratios based on the dynamic price center value, generating upper and lower thresholds for the dynamic price review threshold range. The upward and downward fluctuation ratios are inversely proportional to the time elapsed since the entry into the warehouse, ensuring that the dynamic price review threshold range automatically narrows over time. The formula for calculating the upward and downward fluctuation ratios is:

[0038] in, This is the initial upward and downward fluctuation ratio. The narrowing coefficient, The time of entry into the warehouse. This is the current moment. Initial upward and downward fluctuation ratio. Based on the price fluctuation characteristics of fresh produce categories, the settings are set for fresh produce categories with large price fluctuations (such as aquatic products and poultry). The value is 0.2; for fresh produce categories with relatively small price fluctuations (such as root vegetables). The value is 0.1. Narrowing coefficient. Based on the shelf life setting of fresh produce categories, for fresh produce categories with shorter shelf lives (such as leafy vegetables and berries). The value is 0.05 / hour; for fresh produce with a long shelf life (such as root vegetables). The value is 0.01 / hour.

[0039] The formulas for calculating the upper and lower thresholds of the dynamic price review threshold range are as follows:

[0040] in, Let t be the upper limit of the dynamic price approval threshold range. This represents the lower limit of the dynamic price approval threshold range at time t. As time progresses... The increase, the fluctuation ratio Gradually decrease, therefore and The difference between them gradually decreases, and the dynamic price review threshold range automatically narrows. When a batch of fresh produce is close to its expiration date... When the dynamic price threshold approaches 0, the dynamic price review threshold range approaches the dynamic price center value. At this point, the review of declared prices is more stringent, which can effectively prevent fresh products nearing their expiration date from being sold at unreasonably high prices.

[0041] refer to Figure 6When sales or purchasing personnel submit price declarations through the sales declaration system, the system receives the declared prices pending review, along with the associated fresh produce batch identifier. The system first verifies the validity of the declared prices, including checking if the price is a positive number, if it matches the unit of measurement for the fresh produce batch, and if it contains any illegal characters. If the declared price is invalid, the system directly rejects it and sends an invalid declaration notification to the submitter, explaining the reason for rejection. If the declared price is valid, the system... Retrieve the corresponding dynamic price review threshold range The declared price Compare with the interval.

[0042] like or The system determines that the declared price exceeds the limit. The system generates a warning message, which includes the fresh produce batch identifier, fresh produce category name, warehousing time, current real-time quality decay coefficient, dynamic price review threshold range, declared price, type of exceeding the limit (below the lower limit or above the upper limit), and time of exceeding the limit. Simultaneously, the system freezes the regular review workflow, no longer sending review tasks to regular review nodes. The system forcibly introduces a higher-level review node, redirecting review tasks to the review terminal corresponding to the higher-level review node, and sending the warning message and relevant review materials to that review terminal, including the full-chain environmental parameter file of the fresh produce batch, procurement cost details, historical sales data, etc. If the system determines that the declared price does not exceed the limit, it will send the review task to the next review node along the preset regular review process and continue to execute the regular review process.

[0043] This embodiment collects time-series environmental parameters of fresh produce during warehousing and logistics, calculates a real-time quality decay coefficient based on the benchmark spoilage rate of fresh produce, and nonlinearly fuses the procurement cost benchmark, the real-time quality decay coefficient, and supply and demand fluctuation factors to generate a dynamic price review threshold range that continuously evolves and automatically narrows over time. By comparing the declared price with the corresponding dynamic price review threshold range, the review benchmark and the actual value of the fresh produce are adjusted synchronously. When the declared price exceeds the limit, an early warning is triggered, and a higher-level review node is forcibly introduced, preventing abnormal pricing from passing the review due to a disconnect between the review benchmark and the actual value of the fresh produce.

[0044] In one embodiment, reference Figure 2Before generating the dynamic price approval threshold range, the system calculates the supply and demand fluctuation factor. The system first obtains the historical transaction volume and historical canceled order volume within the previous time window for the current region. The previous time window uses a sliding window mechanism, with the window length set according to the sales cycle of fresh produce, typically ranging from 24 to 72 hours. The system updates the historical data within the previous time window every hour to ensure data real-time performance. The system then extracts all completed and confirmed orders from the sales management system for the current region within the previous time window, and calculates the total number of these orders as the historical transaction volume. The system also extracts all submitted but ultimately cancelled orders from the current region within the preceding time window, including orders cancelled by users and orders cancelled by the system due to insufficient inventory or other reasons. The total number of these orders is then counted as the historical cancelled order volume. .

[0045] The system calculates the ratio of historical canceled orders to historical completed orders as the regional supply-demand balance. The calculation formula is as follows:

[0046] Regional supply and demand tightness reflects the current strength of consumer demand for this fresh produce category within the region. A higher level indicates a large number of order cancellations, weak consumer demand, and an oversupply in the market; when A lower value indicates fewer order cancellations, stronger consumer demand, and a supply shortage in the market. For example, when... When this occurs, it means that for every 100 completed orders, there are 20 cancelled orders, indicating strong consumer demand; when This indicates that for every 100 completed orders, there are 50 cancelled orders, suggesting relatively weak consumer demand.

[0047] The system retrieves the real-time inbound and outbound quantities of adjacent regions within a specified time window. Adjacent regions are those geographically adjacent to the current region and connected by regular fresh produce logistics channels. The system defines adjacent regions as those within 50 kilometers of each other and with a logistics channel travel time of less than 2 hours. The list of adjacent regions is configured by the system administrator during system initialization and can be dynamically adjusted based on changes in the logistics network. The system extracts the total amount of fresh produce transferred from external suppliers or other regions to each adjacent region within the specified time window from the supply chain management system as the real-time inbound quantity. Simultaneously, the sum of the total fresh produce sales and total losses in each adjacent related region within the preceding time window is extracted as the real-time consumption. .

[0048] The system calculates the ratio of the difference between real-time consumption and real-time allocation and warehousing as the initial cross-regional supply and demand flow rate. The calculation formula is as follows:

[0049] Initial cross-regional supply and demand mobility reflects the supply and demand balance in adjacent related regions. When When positive, it indicates that the consumption in adjacent related areas exceeds the amount transferred into storage, resulting in a supply gap that may require replenishment from the current area; when... When the value is negative, it indicates that the amount of fresh produce transferred into the warehouse from adjacent related regions is greater than the amount consumed, indicating a supply surplus. This may lead to the transfer of fresh produce to the current region, increasing the supply in the current region.

[0050] The system acquires logistics network status data between adjacent and related areas. This data includes a network congestion index and vehicle dispatch delay time. The system obtains real-time logistics network status data from a third-party traffic management platform via a standardized API interface, with an update frequency of 5 minutes. The network congestion index is calculated by the third-party traffic management platform based on vehicle speeds, ranging from 0 to 1. 0 indicates smooth traffic with vehicle speeds reaching the design speed; 1 indicates complete congestion with vehicle speeds below 5 km / h. Vehicle dispatch delay time is provided by the logistics management system. The system calculates the average time interval between the issuance of all vehicle dispatch instructions and the actual departure of the vehicle within the past hour as the current vehicle dispatch delay time, expressed in hours.

[0051] The system calculates the road network timeliness loss coefficient based on the road network congestion index and the delay time of transportation vehicle scheduling. The calculation formula is as follows:

[0052] in, The road network congestion index. For the delay time of transportation vehicle scheduling, The congestion impact coefficient, This represents the scheduling delay impact coefficient. and The values ​​of all are in the range of 0 to 1, and ,generally The value is 0.6. The value is 0.4. Road network timeliness loss coefficient. The value range is from 0 to 1. The smaller the value, the more severe the obstruction of the logistics network and the greater the time loss in cross-regional transfers. When the calculated value... When it is less than 0, the system will A value of 0 indicates that cross-regional allocation is completely impossible.

[0053] The system multiplies the initial cross-regional supply and demand mobility by the road network timeliness loss coefficient, and applies a decay correction to the initial difference ratio to obtain the corrected cross-regional supply and demand mobility. The calculation formula is as follows: By introducing a road network timeliness loss coefficient, the reduction in cross-regional supply capacity caused by obstructed logistics networks can be accurately reflected. When logistics network congestion is severe or transportation vehicle scheduling delays are long, even if there are supply gaps in adjacent regions, fresh produce cannot be promptly transferred from the current region to supplement them. Therefore, it is necessary to adjust the initial cross-regional supply and demand mobility by attenuation to avoid overestimating the impact of cross-regional supply capacity on current regional prices.

[0054] The system performs a nonlinear mapping between regional supply and demand tightness and cross-regional supply and demand mobility, and determines the mapping result as the supply and demand volatility factor. The calculation formula is as follows:

[0055] in, This represents the regional supply and demand impact coefficient. This represents the cross-regional supply and demand impact coefficient. and The values ​​of all are in the range of 0 to 1, and ,generally The value is 0.6. The value is 0.3. When the regional supply and demand tightness... At lower levels, Larger supply and demand volatility factors Increased supply and demand mobility leads to upward price fluctuations; when cross-regional supply and demand mobility increases... When it is high, the supply and demand volatility factor An increase in the supply-demand volatility factor leads to upward price fluctuations. Conversely, when regional supply and demand are highly concentrated or inter-regional supply and demand mobility is low, the supply-demand volatility factor... As the quantity decreases, the price fluctuates downwards. Table 2 shows the time-delay coefficients corresponding to different logistics network conditions.

[0056] Table 2. Time Delay Coefficients for Different Logistics Network Statuses

[0057] Table 2 lists the calculation results of the network timeliness loss coefficient under different logistics network conditions. When the network congestion index is high or the transportation vehicle dispatch delay is long, the network timeliness loss coefficient decreases significantly. For example, when the network congestion index is 0.5 and the transportation vehicle dispatch delay is 1.5 hours, the network timeliness loss coefficient is only 0.10, indicating that the actual supply capacity of cross-regional allocation is only 10% of the theoretical supply capacity. When the calculated... When it is less than 0, the system will A value of 0 indicates that cross-regional allocation is completely impossible. In this case, the cross-regional supply and demand mobility is 0, and the supply and demand fluctuation factor is determined solely by the regional supply and demand tightness.

[0058] This embodiment calculates regional supply-demand tightness and cross-regional supply-demand mobility, and introduces a road network timeliness loss coefficient to attenuate and correct cross-regional supply-demand mobility. This accurately reflects the current supply-demand situation in the region and adjacent related regions, as well as the impact of the logistics network on cross-regional supply capacity. By performing a nonlinear mapping between regional supply-demand tightness and cross-regional supply-demand mobility to obtain a supply-demand volatility factor, the dynamic price center value can be adjusted according to changes in market supply and demand, further improving the accuracy of the dynamic price review threshold range.

[0059] In one embodiment, reference Figure 4 The system receives the declared prices to be reviewed, along with the associated fresh produce batch location identifiers. The fresh produce batch location identifier contains the specific location information of that batch of fresh produce within the storage space, including warehouse number, shelf number, layer number, and column number. Based on the fresh produce batch location identifier, the system extracts the local environmental offset within the corresponding storage microenvironment. The local environmental offset refers to the degree of difference between the actual environmental conditions at that storage location and the average environmental conditions of the storage space.

[0060] The system acquires environmental parameter measurements from all sensors within the warehouse environment monitoring system at the current moment, calculating the average temperature, average humidity, and average vibration frequency of the warehouse space. The system then extracts the actual temperature, actual humidity, and actual vibration frequency from the three nearest sensors to the location of the specific fresh produce batch, calculating the average of these three sensor measurements as the actual environmental parameter value for that storage location. The system calculates the difference between the actual environmental parameter value and the average environmental parameter value of the warehouse space to obtain the local temperature offset. Local humidity offset and local vibration frequency offset .

[0061] The system calculates the local environment offset by weighted summing of the offsets of these three local environment parameters with their corresponding attenuation weights. The calculation formula is as follows:

[0062] in, , , Using the same decay weight as before, , , These are the normalized ranges for temperature, humidity, and vibration frequency, respectively. Local environmental offset. The value range is from -1 to 1, when When positive, it indicates that the environmental conditions at the storage location are worse than the average environmental conditions in the warehouse space, and the quality of fresh produce deteriorates faster; when... When the value is negative, it indicates that the environmental conditions at the storage location are better than the average environmental conditions in the warehouse space, and the fresh produce deteriorates more slowly.

[0063] The system substitutes the local environmental offset into the dynamic price review threshold range, compensates for the offset between the upper and lower thresholds of the dynamic price review threshold range, and generates a local dynamic price review threshold range corresponding to the location identifier of the fresh produce batch. The calculation formula is as follows:

[0064] When local environment offset When positive, the local dynamic price review threshold range shifts downwards overall, reflecting the actual situation that fresh produce in that storage location deteriorates faster and has lower value; when When the value is negative, the local dynamic price review threshold range shifts upward as a whole, reflecting the actual situation that the fresh produce in that storage location deteriorates more slowly and has a higher value.

[0065] The system compares the declared price with the local dynamic price review threshold range to determine the deviation amount of the declared price from the local dynamic price review threshold range. If the declared price is lower than the local lower threshold, the deviation amount is the local lower threshold minus the declared price; if the declared price is higher than the local upper threshold, the deviation amount is the declared price minus the local upper threshold; if the declared price is within the range, the deviation amount is 0.

[0066] In a preferred embodiment, to reduce the independent computational overhead of individual batches and lower the data processing latency and computational resource consumption of the auditing system, the system performs spatial clustering of fresh produce batch location identifiers within the warehouse space coordinate system, dividing them into multiple local environment clusters. The warehouse space coordinate system takes the southwest corner of the warehouse space as its origin, with the X-axis extending eastward along the length of the warehouse space, the Y-axis extending northward along the width of the warehouse space, and the Z-axis extending upward along the height of the warehouse space. Each shelf location corresponds to a rectangular area in the coordinate system, and each shelf's layer number and column number correspond to the coordinate values ​​of the Z-axis and X-axis, respectively. The system converts the warehouse number, shelf number, layer number, and column number in the fresh produce batch location identifiers into three-dimensional coordinate points. .

[0067] The system employs the K-means clustering algorithm to cluster the 3D coordinates of all fresh produce batches. The initial cluster centers are randomly selected, and clustering terminates when the change in cluster centers is less than a preset threshold (0.1 meters) or when the maximum number of iterations (100) is reached. The number of clusters, K, is set based on the size of the storage space and the deployment density of environmental sensors, typically ranging from 5 to 20. After clustering, each local environmental cluster contains multiple geographically adjacent fresh produce batch location identifiers, and these location identifiers correspond to relatively similar storage microenvironment conditions.

[0068] The system calculates the average local environment offset corresponding to the location identifiers of all fresh produce batches within the same local environment cluster, and uses this as the cluster-level environment offset. The calculation formula is as follows:

[0069] in, This refers to the number of fresh produce batches within this local environmental cluster. Let be the local environmental offset of the i-th fresh produce batch.

[0070] The system substitutes the cluster-level environment offset into the dynamic price review threshold range, replaces the local environment offset for offset compensation, and generates the cluster-level dynamic price review threshold range corresponding to the local environment cluster. The calculation formula is as follows:

[0071] The system applies the cluster-level dynamic price review threshold range to compare the declared prices of all fresh produce batches belonging to the same local environment cluster. The system re-clusters spatially every hour and updates the cluster-level environmental offset and cluster-level dynamic price review threshold range for each local environment cluster. When a new fresh produce batch arrives at the warehouse, the system assigns it to the corresponding local environment cluster based on its location identifier and applies the cluster-level dynamic price review threshold range for review. Table 3 shows the environmental parameters and cluster-level environmental offsets of different local environment clusters in a fresh produce storage center.

[0072] Table 3. Environmental parameters and cluster-level environmental offsets for different local environmental clusters in a fresh food storage center.

[0073] Table 3 lists the environmental parameters and corresponding cluster-level environmental offsets for five local environmental clusters in a fresh food storage center. The environmental parameters of different local environmental clusters differ significantly, resulting in different cluster-level environmental offsets. For example, cluster 3 is located near the air outlet of the refrigeration unit, where temperature, humidity, and vibration frequency are all higher than the average of the storage space. Therefore, its cluster-level environmental offset is positive and relatively large, and the fresh food batches within this cluster are subject to a more downward-shifted cluster-level dynamic price review threshold range. Cluster 2 is located inside the storage space, far from the entrance and the refrigeration unit, where temperature, humidity, and vibration frequency are all lower than the average of the storage space. Therefore, its cluster-level environmental offset is negative, and the fresh food batches within this cluster are subject to an upward-shifted cluster-level dynamic price review threshold range.

[0074] This embodiment extracts the local environmental offset corresponding to the location of fresh produce batches and compensates for this offset in the dynamic price review threshold range, generating a local dynamic price review threshold range. This range reflects the impact of different storage microenvironments on the degradation of fresh produce quality, further improving the accuracy of the price review benchmark. By generating cluster-level dynamic price review threshold ranges through spatial clustering, multiple fresh produce batches within the same local environment cluster are reviewed using the same threshold range. This reduces the independent computational overhead of individual batches, lowers the data processing latency and computational resource consumption of the review system, and simultaneously ensures the accuracy of the review benchmark.

[0075] In one embodiment, reference Figure 5 When calculating the real-time quality degradation coefficient, the system acquires the rate of change of temperature, humidity, and vibration frequency time-series data in adjacent time slices. The time slice length is set to 15 minutes, and the system calculates the rate of change of each environmental parameter every 15 minutes. The system calculates the average temperature, average humidity, and average vibration frequency within the current time slice, and then compares them with the average parameter values ​​of the previous time slice to calculate the rate of change. For example, the formula for calculating the temperature rate of change is:

[0076] in, The average temperature of the time slice at time t is This represents the length of the time slice (0.25 hours). Similarly, the system calculates the rate of change in humidity and the rate of change in vibration frequency.

[0077] When the rate of change of any environmental parameter exceeds a preset rate of change threshold, the system extracts the abrupt environmental data segment exceeding the preset rate of change threshold. The preset rate of change threshold is set according to the environmental tolerance characteristics of fresh produce. For example, the temperature rate of change threshold for leafy vegetables is typically 2℃ / hour, the humidity rate of change threshold is typically 10% / hour, and the vibration frequency rate of change threshold is typically 20 times / minute / hour. The abrupt environmental data segment refers to the environmental data within a continuous time interval from the moment the rate of change exceeds the threshold to the moment the rate of change recovers below the threshold.

[0078] The system assigns a mutation acceleration weight to data segments in mutated environments. The mutation acceleration weight is greater than 1, and its specific value is determined based on the degree to which the rate of change exceeds a threshold. The formula for calculating the mutation acceleration weight is:

[0079] in, This is the mutation acceleration factor, which is usually set to 1; This represents the actual rate of change of environmental parameters; A preset threshold for the rate of change is set. The greater the rate of change exceeds the threshold, the larger the value of the mutation acceleration weight. For example, when the temperature change rate is 1.5 times the threshold, the mutation acceleration weight is 1.5; when the temperature change rate is 2 times the threshold, the mutation acceleration weight is 2.

[0080] The system adds the product of the mutated environment data segment and the mutated acceleration weight to the integral result of the integral operation. Then, it exponentializes the resulting integral and uses this exponentialized result as the real-time quality degradation coefficient to characterize the accelerating effect of environmental mutations on quality degradation. The formula for calculating the real-time quality degradation coefficient considering environmental mutations is as follows:

[0081] in, The number of data segments in the mutated environment. The mutation acceleration weight is the value corresponding to the i-th mutation environment data segment. Let i be the start time of the i-th mutation environment data segment. This represents the end time of the i-th mutation environment data segment. By introducing mutation acceleration weights, the accelerated impact of environmental mutations (such as a sudden increase in temperature due to refrigeration equipment failure, or a sudden increase in vibration due to a violent collision during transportation) on the deterioration of fresh produce quality can be accurately reflected, thus avoiding the problem that rapid quality deterioration caused by environmental mutations is not reflected in the audit criteria in a timely manner.

[0082] In a preferred embodiment, when setting the upward and downward fluctuation ratio based on the dynamic price center value, the system retrieves historical rejection records for the fresh produce category within the historical review period. The length of the historical review period is typically 1 to 3 months, and the system updates the historical rejection record database every week. The system extracts all rejected price records for the fresh produce category within the historical review period from the review system database. Each record includes information such as the declared price, rejection time, rejection reason, and reviewer identifier.

[0083] The system extracts the price distribution density of rejected applications from historical rejection records. The system divides the price range for this fresh produce category into multiple equally wide price sub-ranges, typically 0.1 yuan / kg. The system counts the number of rejected prices within each price sub-range and calculates the price distribution density for each sub-range, which is the proportion of rejected prices in that sub-range to the total number of rejected prices. The system accumulates the distribution densities in descending order until the accumulated value reaches 0.75. The price sub-range covered at this point is the historical concentration range of rejected prices. This historical concentration range of rejected prices reflects the price range where abnormal pricing most frequently occurred historically.

[0084] The system calculates the deviation distance between the dynamic price center value and the historical rejection price concentration range. The deviation distance is the absolute value of the difference between the dynamic price center value and the midpoint of the historical rejection price concentration range. If the deviation distance is less than a preset deviation threshold, it indicates that the current dynamic price center value is relatively close to the historical rejection price concentration range, and the risk of abnormal pricing is high. In this case, the system reduces the upward and downward fluctuation ratio, typically by 20%. If the deviation distance is greater than or equal to the preset deviation threshold, it indicates that the current dynamic price center value differs significantly from the historical rejection price concentration range, and the risk of abnormal pricing is low. In this case, the system maintains the current upward and downward fluctuation ratio. The preset deviation threshold is typically 25% of the width of the historical rejection price concentration range.

[0085] Based on the adjusted upward and downward fluctuation ratios and the dynamic price center value, the system generates upper and lower thresholds for the dynamic price review threshold range. By incorporating historical rejection records to adjust the upward and downward fluctuation ratios, the system can dynamically adjust the strictness of the review according to the distribution of historical abnormal pricing. When the risk of abnormal pricing is high, the review threshold range is narrowed to improve the accuracy of abnormal pricing interception; when the risk of abnormal pricing is low, a wider review threshold range is maintained to ensure the efficiency of normal pricing review.

[0086] When the declared price is lower than the lower limit of the dynamic price review threshold range, the system calculates the downward offset difference between the declared price and the lower limit threshold. When the declared price exceeds the upper limit of the dynamic price review threshold range, the system calculates the upward offset difference between the declared price and the upper limit threshold. .

[0087] The system inputs the lower or upper offset difference into the review level mapping function and outputs the corresponding higher-level review level identifier. The review level mapping function is a piecewise function that maps the offset difference to different higher-level review levels. The parameters of the review level mapping function are configured by the system administrator based on the value of the fresh produce category and the company's management requirements. For high-value fresh produce categories, the threshold is set lower, and a smaller offset difference will trigger a higher-level review; for low-value fresh produce categories, the threshold is set higher, and a larger offset difference will trigger a higher-level review.

[0088] The system freezes the regular review workflow based on the higher-level review hierarchy identifier, preventing review tasks from being sent to regular review nodes. The system generates a redirected review path pointing to the review terminal corresponding to this higher-level review hierarchy identifier and sends warning information and relevant review materials to that terminal, including the real-time quality degradation coefficient of the fresh produce batch, dynamic price review threshold range, declared price, offset difference, end-to-end environmental parameter file, procurement cost details, and historical sales data. Reviewers at the higher-level review node manually review the declared price based on this information. If the review is successful, the system forwards the review task to the next node; if the review fails, the system rejects the declared price and notifies the applicant of the reason for rejection. The mapping of higher-level review hierarchy corresponding to different offset differences is shown in Table 4.

[0089] Table 4. Mapping table of higher-level review levels corresponding to different offset differences.

[0090] Table 4 lists the higher-level review levels corresponding to different offset difference ranges. The larger the offset difference, the more serious the deviation of the declared price from the dynamic price review threshold range, requiring review by higher-level reviewers. Reviewers at different levels have different review permissions. Higher-level reviewers can not only approve or reject declared prices, but also adjust the dynamic price review threshold range or modify system parameters according to the actual situation to deal with special circumstances.

[0091] This embodiment introduces a sudden acceleration weight when calculating the real-time quality decay coefficient, which accurately reflects the accelerating impact of environmental mutations on the quality decay of fresh produce, thus improving the accuracy of the real-time quality decay coefficient calculation. By adjusting the upward and downward fluctuation ratios using historical rejection records, the strictness of the review can be dynamically adjusted according to the distribution of historical abnormal pricing, further improving the accuracy of abnormal pricing interception. By calculating the offset difference and mapping it to the corresponding higher-level review level, higher-level review nodes are introduced according to the degree of abnormality, making the review process more reasonable and ensuring that abnormal pricing is handled in a timely and effective manner.

Claims

1. A method for multi-level review and dynamic early warning of fresh produce prices, characterized in that, include: Collect time-series environmental parameters of fresh produce during warehousing and logistics, and calculate the real-time quality degradation coefficient by combining them with the benchmark spoilage rate of fresh produce categories. Obtain the procurement cost benchmark, and nonlinearly fuse the procurement cost benchmark, the real-time quality decay coefficient, and the supply and demand fluctuation factor to generate a dynamic price review threshold range that continuously evolves over time. The dynamic price review threshold range automatically narrows over time. Receive the declared price and compare the declared price with the dynamic price review threshold range at the corresponding time. If the declared price exceeds the limit, an early warning will be triggered and the regular review process will be frozen, forcibly introducing a higher-level review node. If the declared price does not exceed the limit, the review will be carried out along the regular review process path.

2. The method for multi-level review and dynamic early warning of fresh produce prices according to claim 1, characterized in that, The collection of time-series environmental parameters of fresh produce in the warehousing and logistics process, combined with the benchmark spoilage rate of fresh produce categories, to calculate the real-time quality decay coefficient includes: extracting time-series temperature data, time-series humidity data, and time-series vibration frequency data in the warehousing and logistics process as the time-series environmental parameters. Based on the fresh produce category attributes, a first attenuation weight corresponding to the temperature time series data, a second attenuation weight corresponding to the humidity time series data, and a third attenuation weight corresponding to the vibration frequency time series data are determined. The temperature time series data, the humidity time series data, and the vibration frequency time series data are weighted and summed with the corresponding first attenuation weight, second attenuation weight, and third attenuation weight, respectively, to obtain the comprehensive environmental disturbance value; The comprehensive environmental disturbance value and the benchmark spoilage rate of the fresh produce category are integrated over time to output the real-time quality decay coefficient.

3. The method for multi-level review and dynamic early warning of fresh produce prices according to claim 1, characterized in that, Before obtaining the procurement cost benchmark, and nonlinearly fusing the procurement cost benchmark, the real-time quality decay coefficient, and the supply and demand fluctuation factor to generate a dynamic price review threshold range that continuously evolves over time, the method further includes: obtaining the historical transaction order volume and historical canceled order volume within the previous time window of the current region, and calculating the ratio of the historical canceled order volume to the historical transaction order volume as the regional supply and demand tightness. Retrieve the real-time transfer inbound and real-time consumption of adjacent related regions within the preceding time window, and calculate the ratio of the difference between the real-time consumption and the real-time transfer inbound as the cross-regional supply and demand flow. A nonlinear mapping is performed between the regional supply and demand tightness and the cross-regional supply and demand mobility, and the mapping result is determined as the supply and demand fluctuation factor.

4. The method for multi-level review and dynamic early warning of fresh produce prices according to claim 1, characterized in that, The procurement cost benchmark, the real-time quality decay coefficient, and the supply and demand fluctuation factor are nonlinearly fused to generate a dynamic price review threshold range that evolves continuously over time. This includes: using the procurement cost benchmark as the origin of the benchmark price, using the real-time quality decay coefficient as the price reduction index, and performing an exponential decay calculation on the origin of the benchmark price to obtain the quality decay benchmark price. The supply and demand fluctuation factor is used as the price fluctuation correction coefficient. The price fluctuation correction coefficient is multiplied by the quality deterioration benchmark price to obtain the dynamic price center value. Based on the dynamic price center value, an upward and downward fluctuation ratio is set to generate the upper and lower thresholds of the dynamic price review threshold range. The upward and downward fluctuation ratio is inversely proportional to the time elapsed since the current time of entry into the warehouse, so that the dynamic price review threshold range automatically narrows over time.

5. The method for multi-level review and dynamic early warning of fresh produce prices according to claim 1, characterized in that, The process of receiving the declared price and comparing it with the dynamic price review threshold range at the corresponding time includes: receiving the declared price to be reviewed and the fresh produce batch location identifier associated with the declared price; Based on the fresh produce batch location identifier, the local environmental offset in the corresponding storage microenvironment is extracted; Substitute the local environmental offset into the dynamic price review threshold range, and perform offset compensation on the upper and lower thresholds of the dynamic price review threshold range to generate a local dynamic price review threshold range corresponding to the fresh produce batch location identifier. The declared price is compared with the local dynamic price review threshold range to determine the deviation of the declared price from the local dynamic price review threshold range.

6. The method for multi-level review and dynamic early warning of fresh produce prices according to claim 1, characterized in that, If the declared price exceeds the limit, an early warning is triggered and the regular review process is frozen, and a higher-level review node is forcibly introduced, including: when the declared price is lower than the lower limit of the dynamic price review threshold range, calculating the downward offset difference between the declared price and the lower limit threshold. When the declared price is higher than the upper limit of the dynamic price review threshold range, calculate the upper offset difference between the declared price and the upper limit threshold; Input the lower offset difference or the upper offset difference into the review level mapping function, and output the corresponding higher-order review level identifier; Based on the higher-level review identifier, the regular review flow path is frozen, a redirected review path is generated pointing to the review terminal corresponding to the higher-level review identifier, and the warning is sent to the review terminal.

7. The method for multi-level review and dynamic early warning of fresh produce prices according to claim 2, characterized in that, The real-time quality decay coefficient is output by integrating the comprehensive environmental disturbance value and the benchmark spoilage rate of the fresh produce category over time. This includes: acquiring the rate of change of the temperature time-series data, the humidity time-series data and the vibration frequency time-series data in adjacent time slices. When the rate of change exceeds a preset rate of change threshold, extract the mutation environment data segment that exceeds the preset rate of change threshold; The mutation environment data segment is assigned a mutation acceleration weight, and the product of the mutation environment data segment and the mutation acceleration weight is added to the integral result of the integral operation. The integral result after superimposing the mutation acceleration weight is indexed, and the indexed result is determined as the real-time quality degradation coefficient to characterize the accelerating effect of environmental mutation on quality degradation.

8. The method for multi-level review and dynamic early warning of fresh produce prices according to claim 3, characterized in that, The step of retrieving the real-time transfer inbound volume and real-time consumption volume of adjacent related regions within the preceding time window, and calculating the ratio of the difference between the real-time consumption volume and the real-time transfer inbound volume as the cross-regional supply and demand flow, includes: obtaining logistics network status data between the adjacent related regions, wherein the logistics network status data includes the network congestion index and the transportation vehicle scheduling delay time. The road network timeliness loss coefficient is calculated based on the road network congestion index and the vehicle dispatch delay time. Multiply the initial difference ratio between the real-time consumption and the real-time allocation and storage by the road network timeliness loss coefficient to attenuate and correct the initial difference ratio; The percentage of the difference after attenuation correction is determined as the cross-regional supply and demand mobility, in order to reflect the reduction in actual cross-regional supply capacity due to obstruction of the logistics network.

9. The method for multi-level review and dynamic early warning of fresh produce prices according to claim 4, characterized in that, Based on the dynamic price center value, the upper and lower fluctuation ratios are set to generate the upper and lower thresholds of the dynamic price review threshold range, including: obtaining the historical application rejection records of the fresh food category within the historical review period; Extract the rejection price distribution density from the historical rejection records, and determine the historical rejection price concentration range based on the rejection price distribution density; Calculate the deviation distance between the dynamic price center value and the historical rejection price concentration interval; If the deviation distance is less than a preset deviation threshold, the value of the vertical fluctuation ratio is reduced; if the deviation distance is greater than or equal to the preset deviation threshold, the value of the vertical fluctuation ratio is maintained. Based on the adjusted upward and downward fluctuation ratios and the dynamic price center value, the upper and lower thresholds of the dynamic price review threshold range are generated.

10. The method for multi-level review and dynamic early warning of fresh produce prices according to claim 5, characterized in that, Substituting the local environment offset into the dynamic price review threshold range, offset compensation is performed on the upper and lower thresholds of the dynamic price review threshold range to generate a local dynamic price review threshold range corresponding to the fresh produce batch location identifier, including: spatial clustering of the fresh produce batch location identifier in the warehouse space coordinate system to divide multiple local environment clusters; Calculate the average of the local environment offsets corresponding to the location identifiers of all fresh produce batches within the same local environment cluster, and use this as the cluster-level environment offset. Substitute the cluster-level environment offset into the dynamic price review threshold range, replace the local environment offset for offset compensation, and generate a cluster-level dynamic price review threshold range corresponding to the local environment cluster. Fresh produce batches belonging to the same local environmental cluster will be compared using the cluster-level dynamic price review threshold range.

Images