Whole-chain tracing method and system based on fresh products

Abstract

The invention provides a full-chain tracing method and system based on fresh products, and aims to solve the problems that real-time monitoring and dynamic management are difficult to realize, supply chain risk assessment is insufficient and data integration and safety are poor in the prior art. Static product basic attributes are combined with dynamic market demands, environmental changes and sales trends, and real-time tracking and dynamic management of fresh products in links of production, transportation, storage and sales are realized. According to the method, through real-time data acquisition and multi-dimensional risk matrix analysis, the problem of insufficient supply chain risk estimation and management is effectively solved, and inventory and transportation scheduling is optimized. And through a full-chain data association network and a distributed storage mechanism, the integration and security of the data are ensured, and the overall transparency and controllability of the supply chain are improved.

Description

Technical Field

[0001] The present invention relates to the field of supply chain management, and particularly to a whole-chain traceability method and system for fresh products Background Art

[0002] The whole-chain traceability method for fresh products is a management method that ensures the transparency of the whole process of products from the source to consumers by tracking the status data of fresh products in various links such as production, transportation, storage, and sales. Its purpose is to improve the visibility and controllability of the supply chain and ensure the quality and safety of products. However, the existing whole-chain traceability methods for fresh products mainly rely on the recording of static information, usually only focusing on the basic attributes and historical information of products at each node in the supply chain. This method is unable to cope with the rapid changes in market demand, the uncertainties during transportation, and the dynamic impact of environmental conditions on product quality, resulting in difficulties in real-time monitoring and dynamic management. In addition, the existing methods lack an effective supply chain risk assessment mechanism, leading to the inability to predict and prevent supply chain risks such as inventory fluctuations and transportation delays. At the same time, the data at each node in the supply chain is often scattered and difficult to be effectively integrated, forming information islands, which affects the comprehensiveness and accuracy of traceability. Moreover, the existing technologies also have deficiencies in data security and integrity, easily resulting in data tampering or loss, and unable to ensure the whole-chain traceability management of the supply chain Summary of the Invention

[0003] In view of the above actual situation, the present invention proposes a whole-chain traceability method and system for fresh products to solve the problems in the prior art that real-time monitoring and dynamic management cannot be achieved, the supply chain risk assessment is insufficient, and the data integration and security are poor

[0004] A whole-chain traceability method for fresh products, the method comprising the following steps

[0005] S1, obtaining the predicted feature vectors of the target fresh product and competing products, the predicted feature vectors including environmental conditions, market demand, seasonal factors, and user behavior characteristics

[0006] S2, constructing a prediction mapping function based on the predicted feature vectors, the prediction mapping function including a sales volume prediction model of the target product, a sales volume prediction model of competing products, and a residual adjustment model for generating a sales volume prediction result

[0007] S3. Obtain and construct a comprehensive traceability identification code, which includes a static traceability identification code and a dynamic traceability identification code. The static traceability identification code is used to mark the basic attribute information of the product, and the dynamic traceability identification code is generated according to the prediction mapping function. The comprehensive traceability identification code is used to mark each product unit in the supply chain, and through data mapping, the traceability identification code is associated with the information of each node in the supply chain to form a traceability mapping set.

[0008] S4. Based on the comprehensive traceability identification code, conduct real-time data collection on fresh products in the supply chain nodes. The real-time data collection includes production data, transportation data, inventory dynamic data, and product transfer data during the sales process, so as to obtain a real-time monitoring data set of the supply chain.

[0009] S5. Based on the real-time monitoring data set of the supply chain and the traceability mapping set, conduct a multi-dimensional risk matrix analysis of the supply chain. The analysis includes the impact of inventory fluctuations, transportation delays, and environmental factors, and generate a supply chain risk assessment model.

[0010] S6. Based on the supply chain risk assessment model and the real-time monitoring data set of the supply chain, conduct cross-verification and correlation analysis on the data of each node in the supply chain. By comprehensively processing the data of the production, transportation, storage, and sales links, construct a full-chain data association network.

[0011] S7. Based on the full-chain data association network, use distributed storage and node consensus mechanism to achieve full-chain traceability management. The full-chain traceability management includes full-chain data recording and tracking of the production, transportation, storage, and sales links, ensuring the security, integrity, and traceability of the data of each node in the supply chain, generating a full-chain traceability record set for the transparent management of the supply chain.

[0012] Further, the prediction mapping function in step S2 is: Y adjusted,t = ω1Y target,t + ω2Y competitor,t + R target,t , where ω1 and ω2 are weight coefficients, satisfying ω1 + ω2 = 1; R target,t is the residual adjustment value used to correct the prediction result, Y target is the predicted sales volume of the target commodity, and Y competitor is the predicted sales volume of competing commodities.

[0013] Further, the static traceability identification code in step S3 is used to mark the basic attribute information of the product, including the production date, place of origin, and product type data. The dynamic traceability identification code contains dynamic information such as future market demand, environmental changes, and sales trends. Its specific calculation formula is as follows: T dynamic = Y adjusted,t + A t, where Y adjusted,t is the sales volume prediction value, A t is the adjustment factor, T dynamic is the dynamic traceability identification code, and the adjustment factor includes the supply chain volatility index SCI t and the environmental condition index ECI t .

[0014] Furthermore, in the S4 step, the supply chain real-time monitoring data set D realtime is D realtime ={D prod , D trans , D stock , D sales}, where D prod is the production node data, D tran is the transportation node data, D stock is the storage node data, and D sales is the sales node data.

[0015] Furthermore, in the S5 step, the risk assessment model is constructed by combining the inventory volatility quantification model, the transportation delay quantification model, and the environmental factor quantification model. The risk assessment model R total (t) has the following specific calculation formula: R total (t)=α1·I(t)+α2·T(t)+α3·E(t), where α1, α2, and α3 are the weight coefficients of the inventory volatility quantification model, the transportation delay quantification model, and the environmental factor quantification model, representing the influence degree of different risks on the overall supply chain. I(t) is the inventory volatility quantification model, T(t) is the transportation delay quantification model, and E(t) is the environmental factor quantification model.

[0016] Furthermore, the S6 step includes the following sub-steps:

[0017] S61, perform cross-validation on the supply chain real-time monitoring data set. The cross-validation includes data consistency analysis and data time-series consistency analysis;

[0018] S62, based on the supply chain real-time monitoring data set after cross-validation, perform association analysis to construct the data relationship between each node. The association analysis includes correlation analysis and dependence analysis;

[0019] S63, based on the results of the association analysis, generate a full-chain data association network.

[0020] Furthermore, the S7 step includes the following sub-steps:

[0021] S71. Based on the full-chain data association network, perform distributed storage configuration on the data of each key node in the supply chain to ensure the secure storage and redundant backup of the data of each node.

[0022] S72. Based on the dataset with distributed storage configuration, use the node consensus mechanism to perform consistency verification on the key data to ensure that the data records on all nodes are consistent and cannot be tampered with.

[0023] S73. Based on the dataset after consistency verification, generate the full-chain traceability records of each node and integrate the traceability record set into the full-chain traceability management system to achieve transparent management of all links in the supply chain.

[0024] Furthermore, the full-chain data association network is G=(N, E, W), where N={N prod , N trans , N stock , N sales} represents each link node in the supply chain, E={e ij} represents the edges between nodes, that is, the association relationships between different supply chain links; W={w ij} represents the weights of the edges, and its specific calculation formula is as follows: w ij =α·Corr(D i , D j )+β·P(D j |D i ), where w ij is the association weight between nodes D i and D j , Corr(D i , D j ) is the data correlation coefficient between the two, P(D j ∣D i ) is the conditional probability, and α and β are weight adjustment coefficients.

[0025] Furthermore, the node consensus mechanism is the practical Byzantine fault tolerance consensus mechanism or the proof-of-stake consensus mechanism.

[0026] In addition, the present application also proposes a full-chain traceability system for fresh products, characterized in that the system includes the following units:

[0027] An acquisition unit, configured to acquire the predicted feature vectors of the target fresh product and competing products, where the predicted feature vectors include environmental conditions, market demand, seasonal factors, and user behavior characteristics.

[0028] A prediction mapping unit for constructing a prediction mapping function based on the prediction feature vector. The prediction mapping function includes a sales volume prediction model for the target commodity, a sales volume prediction model for competing commodities, and a residual adjustment model for generating a sales volume prediction result;

[0029] An integrated traceability identification code construction unit for obtaining and constructing an integrated traceability identification code. The integrated traceability identification code includes a static traceability identification code and a dynamic traceability identification code. The static traceability identification code is used to mark the basic attribute information of the product, and the dynamic traceability identification code is generated according to the prediction mapping function; the integrated traceability identification code is used to mark each product unit in the supply chain, and through data mapping, the traceability identification code is associated with the information of each node in the supply chain to form a traceability mapping set;

[0030] A real-time data collection unit for performing real-time data collection on fresh products in supply chain nodes based on the integrated traceability identification code. The real-time data collection includes production data, transportation data, inventory dynamic data, and product transfer data during the sales process, so as to obtain a supply chain real-time monitoring data set;

[0031] A risk assessment unit for performing multi-dimensional risk matrix analysis of the supply chain based on the supply chain real-time monitoring data set and the traceability mapping set. The analysis includes the impacts of inventory fluctuations, transportation delays, and environmental factors, and generates a supply chain risk assessment model;

[0032] A data correlation analysis unit for performing cross-verification and correlation analysis on the data of each node in the supply chain based on the supply chain risk assessment model and the supply chain real-time monitoring data set, and constructing a full-chain data correlation network through comprehensive processing of the data in the production, transportation, storage, and sales links;

[0033] A full-chain traceability management unit for implementing full-chain traceability management based on the full-chain data correlation network using distributed storage and node consensus mechanisms. The full-chain traceability management includes full-chain data recording and tracking of the production, transportation, storage, and sales links, ensuring the security, integrity, and traceability of the data of each node in the supply chain, and generating a full-chain traceability record set for the transparent management of the supply chain.

[0034] A fresh product full-chain traceability method and system proposed in this application realizes the whole-process dynamic monitoring and management of fresh products from production to sales. By combining static and dynamic information through the integrated traceability identification code, it improves the real-time performance and transparency of supply chain data, and can perform multi-dimensional analysis on various risk factors in the supply chain, accurately evaluate problems such as inventory fluctuations and transportation delays, thereby improving the predictability and response ability of the supply chain. In addition, by constructing a full-chain data correlation network, it ensures the security, integrity, and traceability of the data of each node, and guarantees the overall stability and efficiency of the supply chain. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 FIG. 1 is a schematic flow chart of a method for full-chain traceability of fresh products proposed by the present invention;

[0036] Figure 2 FIG. 2 is a schematic flow chart of constructing a full-chain data association network in a method for full-chain traceability of fresh products proposed by the present invention;

[0037] Figure 3 FIG. 3 is a schematic flow chart of generating a full-chain traceability record set based on a full-chain data association network in a method for full-chain traceability of fresh products proposed by the present invention;

[0038] Figure 4 FIG. 4 is a schematic structural diagram of a full-chain traceability system for fresh products provided by an embodiment of the present application; DETAILED DESCRIPTION OF THE EMBODIMENTS

[0039] The following will clearly and completely describe the simulation technical route in the embodiments of the present invention with reference to the accompanying drawings of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0040] To make the above objects, features, and advantages of the present invention more obvious and understandable, the present invention will be further described in detail below with reference to the drawings and specific embodiments.

[0041] In the supply chain management of fresh products, a series of practical problems are faced, such as strong perishability of products, strict requirements for transportation and storage conditions, and large fluctuations in market demand. For example, the storage time, temperature, and humidity environmental conditions of fresh products at different supply chain nodes will directly affect the quality of the products. At the same time, due to the seasonality of fresh products and the uncertainty of consumer demand, inventory management and sales forecasting become more complex. In addition, traditional supply chain management methods usually rely on static information traceability, which can only record historical data in production, transportation, and sales, and cannot reflect real-time changes. This static traceability method is prone to data disconnection in the supply chain and cannot respond in time to sudden situations such as changes in market demand, transportation delays, or inventory fluctuations. Therefore, the present application proposes a method for full-chain traceability of fresh products. Please refer to the attached Figure 1 As shown in the figure, the method includes the following steps:

[0042] S1. Obtain the predicted feature vectors of the target fresh product and competing products, where the predicted feature vectors include environmental conditions, market demand, seasonal factors, and user behavior characteristics;

[0043] Specifically, the process of obtaining the predicted feature vectors of the target fresh produce and competing products is to collect multi-dimensional data related to the fresh produce supply chain and integrate this data into the basic features for subsequent model analysis. The components of the predicted feature vectors include environmental conditions, market demand, seasonal factors, and user behavior characteristics. The acquisition and construction of these elements are the prerequisites for realizing supply chain traceability and management.

[0044] Environmental conditions refer to the external physical condition data related to the storage, transportation, and production of fresh produce. In one or more embodiments, this includes temperature, humidity, and light. The external physical condition data is collected through Internet of Things devices and recorded at different links in the supply chain. In this embodiment, the climate data recorded during the production process and the cold chain temperature control data during the transportation link will both become the input data for the environmental condition dimension. These data are used for subsequent sales volume prediction and product freshness quality analysis; Market demand data includes historical sales data, order data, and market research results. The system analyzes the sales records, changes in consumer preferences, and market capacity information of the target fresh produce and competing products to obtain feature data that can reflect market demand fluctuations. In one or more embodiments, the market demand data is obtained through online e-commerce platforms, offline retail channels, and market report channels, and after processing, it provides data support for subsequent prediction models; Seasonal factors involve the climate cycles or holiday influencing factors related to product supply and demand. The production and demand of fresh produce are greatly affected by seasonal fluctuations. The system determines seasonal characteristics through historical sales data, meteorological information, and consumption patterns related to specific festivals. By analyzing these factors, predictions can be made in advance for future supply and demand during specific periods; User behavior characteristics are obtained by analyzing the shopping behavior data of consumers on different platforms, including purchase records, product reviews, and search record data sources. This type of data forms the data input for the user behavior characteristic dimension through the extraction and analysis of information related to fresh produce consumption behavior.

[0045] In terms of data processing, the system standardizes the data in the above four dimensions to eliminate the differences between different data sources, and uses feature selection and dimensionality reduction features for feature extraction. In this embodiment, principal component analysis and correlation analysis are used to extract the feature data that affects sales volume prediction. The preprocessing stage and the feature extraction stage in this step are the most common processing methods, so no further elaboration will be provided here. After processing, various types of data are integrated into multi-dimensional predicted feature vectors containing environmental conditions, market demand, seasonal factors, and user behavior characteristics, providing a data basis for subsequent model construction and data analysis.

[0046] S2. Based on the predicted feature vector, construct a prediction mapping function, which includes a sales volume prediction model for the target product, a sales volume prediction model for competing products, and a residual adjustment model for generating a sales volume prediction result;

[0047] Specifically, the sales volume prediction model for the target product constructs a prediction model adaptable to the fluctuations in the fresh product market through regression analysis combined with time series prediction methods. The input feature vector is X target , whose dimension n includes environmental conditions, market demand, seasonal factors, and user behavior characteristics. The sales volume prediction model is constructed by combining a linear regression model and a time series model. Specifically, a linear regression model is used to represent the basic sales volume prediction: where β0 is the intercept, β i is the regression coefficient of the i-th variable in the feature vector, and ∈ target is the error term. At the same time, in order to more accurately reflect the time dynamics of the market, a long short-term memory network (LSTM) is further used to process the time series dependence of sales volume. This model is suitable for data with time dependence. Therefore, based on the input feature vector, data from past time steps is combined for prediction. The LSTM model formula is: Y target,t = f LSTM (X target,t , h t-1 , C t-1 ), where X target,t is the feature vector at the current time step, h t-1 is the hidden state at the previous time step, C t-1 is the memory state at the previous time step, and f LSTM represents the recursive calculation process of the LSTM model. The model is trained with historical data to combine the time series characteristics of the target product and market fluctuations to achieve dynamic prediction of future sales volume.

[0048] The sales volume prediction model for competing products is used to analyze the sales volume changes of similar competing products in the market and then evaluate their impact on the sales volume of the target product. The feature vector of competing products is X competitor , whose dimension m contains feature data related to competing products. Specifically, similar to the sales volume prediction of the target product, the sales volume prediction of competing products first performs basic prediction through a linear regression model: where α0 is the intercept, α i is the regression coefficient of competing product features, and ∈ competitor is the error term. In order to capture the sales volume fluctuations of competing products, a support vector regression (SVR) model is further used to establish a non-linear relationship. The SVR model optimizes the kernel function to map the input features to a higher-dimensional space for more accurate prediction. Its model formula is: Ycompetitor = f SVR (X competitor ), in the SVR, in this embodiment, the radial basis kernel function (RBF) is used as the kernel function, and the formula is: K(x i , x j ) = exp(-γ||x i - x j || 2 ), where x i and x j are different dimensions of the feature vector, and γ is the hyperparameter of the kernel function, which determines the influence range of the kernel. By optimizing the kernel function and hyperparameters, the SVR model can effectively handle the non-linear changes in the sales volume of competing products.

[0049] The residual adjustment model is to correct the errors in the sales volume prediction of the target product and competing products, and a residual adjustment model is constructed. This model is based on historical prediction errors and uses the autoregressive moving average model (ARIMA) to model and adjust the residuals. In this embodiment, where the residual R target is the difference between the actual sales volume of the target product and the comprehensive predicted sales volume: R target,t = Y actual,t - Y predicted,t . Next, the ARIMA model is used to model the residuals. The ARIMA model consists of three parts: autoregressive AR, differencing I, and moving average MA. Its formula is: where c is the constant term, φ i is the autoregressive coefficient, θ j is the moving average coefficient, and ∈ t-j is the historical error term of the residuals. By adjusting the model parameters p and q, the dynamic characteristics of the residuals can be effectively captured, and then the future sales volume prediction results can be corrected. The adjusted final sales volume prediction result Y adjusted,t is expressed as: Y adjusted,t = Y predicted,t + R target,t . Through the residual adjustment model, the deviation of the target product sales volume prediction is compensated, further improving the prediction accuracy.

[0050] Finally, when constructing the prediction mapping function f map , the sales volume prediction model of the target product, the sales volume prediction model of the competing product, and the residual adjustment model are combined to obtain the final sales volume prediction function. The prediction mapping function is expressed as: f map (X target , X competitor ) = Y adjusted,t , where X target and X competitor are the feature vectors of the target product and the competing product respectively, and Y adjusted,tIs the final sales volume prediction result after residual adjustment. Specifically, in this embodiment, a prediction mapping function f is constructed map Is based on the sales volume prediction Y of the product target,t And the sales volume prediction Y of the competing product competitor To construct a weighted comprehensive prediction model and combine the prediction results of both. The weighting method determines the relative weights between the target product and the competing product according to the intensity of market competition or through the fitting of historical data. The formula is as follows: Y predicted,t = ω1Y target,t + ω2Y competitor , where ω1 and ω2 are weight coefficients and satisfy ω1 + ω2 = 1. Use the residual adjustment model to correct the weighted comprehensive prediction result Y predicted,t To obtain the final sales volume prediction value Y adjusted,t : Y adjusted,t = Y predicted,t + R target,t , where R target,t Is a residual model constructed based on historical residual data to adjust the comprehensive prediction value, reflecting market fluctuations, seasonal changes, or other unconsidered factors not captured in the previous prediction model. In summary, the final prediction mapping function is: Y adjusted,t = ω1Y target,t + ω2Y competitor,t + R target,t , where ω1 and ω2 are weight coefficients and satisfy ω1 + ω2 = 1; R target,t Is the residual adjustment value used to correct the prediction result, Y target The predicted sales volume of the target product, Y competitor The predicted sales volume of the competing product.

[0051] S3. Obtain and construct a comprehensive traceability identification code, which includes a static traceability identification code and a dynamic traceability identification code. The static traceability identification code is used to mark the basic attribute information of the product, and the dynamic traceability identification code is generated according to the prediction mapping function and contains dynamic information such as future market demand, environmental changes, and sales trends; the comprehensive traceability identification code is used to mark each product unit in the supply chain, and through data mapping, the traceability identification code is associated with the information of each node in the supply chain to form a traceability mapping set;

[0052] Specifically, the static traceability identification code T static Is used to record the stable attribute information of the product in the supply chain, including the production date, origin, and product type data. These information do not change significantly over time and are mainly used to track the source and basic characteristics of the product in the supply chain; the dynamic traceability identification code T dynamic Is based on the prediction mapping function f mapThe obtained final sales volume prediction value Y adjusted,t This prediction value has fully considered various influencing factors, such as environmental conditions, market demand, seasonal changes, and user behavior. Therefore, the dynamic traceability identification code does not need to include the independent calculation of these factors additionally, but directly adopts the sales volume prediction result. However, in the application, in addition to the sales volume prediction value Y map generated based on the prediction mapping function f adjusted,t , a real-time adjustment factor A t is also introduced dynamic to reflect the actual changes and market fluctuations in the supply chain. That is to say, the dynamic traceability identification code not only considers the predicted sales volume, but also dynamically adjusts the sales volume prediction result by combining the adjustment factor. The dynamic traceability identification code provides a real-time prediction of future market demand and a dynamic reflection of the supply chain status, enabling the production, logistics, inventory, and sales links of the supply chain to be adjusted and optimized according to this information. The specific expression is as follows: T adjusted,t =Y t +A t , where the adjustment factor A t dynamically adjusts the sales volume prediction result. Specifically, it can be generated through the following mechanism. In this embodiment, the adjustment factor A t is obtained by weighted calculation of the following real-time data, including: supply chain volatility index SCI t , environmental condition index ECI t . The calculation formula of the adjustment factor is: A t =α·SCI t +β·ECI where: I t represents inventory data, and ΔI t is the inventory change rate; D t represents the transportation delay time, and ΔD t is the transportation time change rate; P t represents the production rate, and ΔP t is the change situation of the production rate; w1, w2, and w3 are weighting coefficients in different dimensions.

[0053] The environmental condition index is comprehensively calculated through external factors such as weather data, natural disaster alerts, and logistics delays: ECI t =m1·W t +m2·N t +m3·L t , where: W t represents the weather condition, based on the meteorological prediction for a certain future period in the weather forecast; N tIndicates natural disaster information, based on relevant disaster warning systems; L t Represents historical data on logistics delays; m1, m2, and m3 are weighting coefficients for different dimensions.

[0054] Final comprehensive traceability identification code T combined Composed of a static traceability identification code and a dynamic traceability identification code, used to comprehensively track the flow of products in the supply chain. Its structure is: T combined =(T static , T dynamic ), where T static Provides basic information about the product, and T dynamic Then reflects the future sales forecast of the product in real time. By combining these two parts, the comprehensive traceability identification code can track products at each node in the supply chain, forming traceability of the entire product life cycle.

[0055] Through the comprehensive traceability identification code, each product unit in the supply chain can be marked. To ensure the transparency of the entire supply chain, the system associates the traceability identification code with the information at each node in the supply chain to generate a traceability mapping set M traceability . The traceability mapping set defines the mapping relationship between the comprehensive traceability identification code and each node in the supply chain: M traceability ={(T combined , N i )|N i ∈Supply_chain_Nodes}, through which dynamic tracking and management of each link in the supply chain can be realized, ensuring the traceability of the entire chain from production to sales of products. It is worth mentioning that steps S1 - S3 introduce new elements of prediction, real - time adjustment, and competitive analysis on the basis of providing full - chain traceability, providing a more dynamic and intelligent framework for supply chain management. Compared with traditional supply chain traceability methods, this greatly improves the timeliness and accuracy of data, not only being able to cope with current supply chain changes but also being able to predict and solve future possible problems in advance.

[0056] S4, based on the comprehensive traceability identification code, real - time data collection is carried out on fresh products in the supply chain nodes. The real - time data collection includes production data, transportation data, inventory dynamic data, and product transfer data during the sales process, thereby obtaining a real - time monitoring data set of the supply chain;

[0057] In step S4, based on the comprehensive traceability identification code system, real - time data collection is carried out on fresh products in the supply chain. During this process, the comprehensive traceability identification code, as the unique identifier, is closely associated with the data of each node in the supply chain to achieve tracking of the status of each product. The objects of real - time data collection cover production data, transportation data, inventory dynamic data, and sales data of all links in the supply chain, ensuring real - time visibility of the entire supply chain.

[0058] In specific implementation, the comprehensive traceability identification code T combined includes a static identification code T static and a dynamic identification code T dynamic . The static part provides the basic attribute information of the product (such as production date, place of origin, category), and the dynamic part is based on the sales volume prediction data and adjustment factors generated by the prediction mapping function f map . Specifically, after the data of each link in the supply chain is collected in real time, these data will be integrated and associated through the dynamic traceability identification code. For example, the dynamic traceability identification code of a fresh product includes the sales volume prediction and adjustment factors of the product. When the product is transported, the transportation status collected by the system, such as transportation time and temperature, will be associated with the product through the dynamic traceability identification code. The role of the dynamic traceability identification code is to help the system combine the real-time data collected with the future sales volume and supply chain status of the product, so as to form complete product tracking information. Generally speaking, the comprehensive traceability identification code not only covers the static attributes of the product, but also reflects the dynamic factors affecting sales and circulation in the supply chain in real time.

[0059] In the process of real-time data collection, through the static identification code T static associate the basic attribute information of the product with the sensor devices or management systems in the supply chain nodes, so as to automatically identify the product in the production, transportation, inventory, and sales links. In one or more embodiments, in the production link, the production batch, quality inspection information, and raw material source of the product are automatically collected by the system's Internet of Things devices and associated with the comprehensive traceability identification code. The system matches the static information of the product through scanning or automatic identification devices to ensure that the basic attributes of the product are accurately recorded in the real-time monitoring dataset.

[0060] Then, based on the dynamic identification code T dynamic , dynamically obtain environmental data, transportation information, inventory status, and sales trends through each node of the supply chain. The dynamic identification code not only includes the sales volume prediction data obtained through the prediction mapping function f map , but also combines the real-time adjustment factor A t , reflecting the current changes in the supply chain. This means that dynamic adjustments can be made according to the predicted sales volume and real-time changes in the supply chain. For example, in the transportation link, the dynamic identification code is associated with the vehicle's GPS data and transportation temperature control data to generate a transportation status dataset D trans . This dataset will be automatically updated according to the adjustment factor A t , such as the real-time position of the vehicle and the temperature fluctuation information during transportation are immediately fed back to the monitoring system.

[0061] In inventory management, the real-time collected inventory dynamic data D stockIncluding the inbound and outbound time and inventory change data, combined with the comprehensive traceability identification code T combined So that each fresh product can be identified in the inventory link. The inventory data is matched with the comprehensive traceability identification code through the sensor device in the warehouse management system, so as to monitor the inventory change in real time, combined with the sales forecast value Y adjusted,t And adjustment factors to ensure the effective management of the inventory status in the supply chain.

[0062] In the sales link, the system collects the product circulation data D sales through the sales terminal (such as the POS machine). The real-time collected sales data includes the product shelving time, sales speed, and consumer feedback. After the dynamic identification code is associated with these data, it is updated to the system to realize the dynamic tracking of the sales process.

[0063] Finally, based on the product basic attributes recorded by the static identification code and the real-time status reflected by the dynamic identification code, the system constructs a real-time monitoring data set D of the supply chain realtime , and this data set is represented by the following formula: D realtime ={D prod ,D trans ,D stock ,D sales}, where the production node data D prod records the detailed information of product production, and each data set is dynamically associated with each link in the supply chain through the comprehensive traceability identification code T combined to ensure the consistency and real-time nature of the data.

[0064] S5. Based on the real-time monitoring data set of the supply chain and the traceability mapping set, conduct a multi-dimensional risk matrix analysis of the supply chain. The analysis includes the impact of inventory fluctuations, transportation delays, and environmental factors, and generates a supply chain risk assessment model;

[0065] Specifically, the supply chain risk matrix analysis is based on real-time data in multiple dimensions. Through the traceability mapping set, the comprehensive traceability identification code of each product is mapped to specific nodes in the supply chain (such as production nodes, transportation nodes, inventory nodes, and sales nodes), enabling the system to collect relevant real-time data from these nodes. For example, if the system needs to analyze the transportation delay of a certain batch of products at a specific transportation node, the traceability mapping set provides the corresponding relationship between this batch of products and the transportation node through the comprehensive traceability identification code, ensuring that real-time data is collected at the accurate node. The real-time data in multiple dimensions will be classified into various risk factors. In one or more embodiments, the risk factors include inventory fluctuations, transportation delays, and environmental factors, and thus are divided into different dimensions and integrated and analyzed in matrix form. Each risk factor dimension represents a different type of risk source, and the matrix analysis integrates these risk factors to quantify the risks in the supply chain.

[0066] Specifically, its risk assessment model is constructed by combining an inventory fluctuation quantification model, a transportation delay quantification model, and an environmental factor quantification model. Among them, the inventory fluctuation quantification model quantifies inventory fluctuations, which are represented by the inventory change rate. The inventory change rate is calculated by the following formula: where S(t) is the inventory quantity at time t, and ΔS(t) is the change in inventory, representing the change in the current inventory compared to the previous moment. The greater the inventory fluctuation, the higher the risk of supply chain interruption. By monitoring the real-time data of inventory changes, potential future supply shortages or surpluses can be predicted. The transportation delay quantification model has a great impact on the timeliness and stability of the supply chain. Transportation delay can be measured by the difference between the actual transportation duration and the expected duration: where D(t) is the actual transportation time, and D expected is the expected transportation time. By calculating the deviation of the transportation time, the reliability of the transportation link can be evaluated. If the delay exceeds a certain threshold, this link may become a potential risk point in the supply chain. The environmental factor quantification model quantifies the environmental impact. Environmental factors include external uncontrollable factors such as climate change and natural disasters, which may directly affect the production, transportation, or storage process. The impact of environmental factors can be represented by an environmental risk index: E(t) = ω1·W(t) + ω2·N(t) + ω3·L(t), where W(t) represents the change in weather conditions (such as temperature, rainfall, etc.), N(t) represents the occurrence probability of natural disasters, L(t) represents the historical data of logistics interruption, and ω1, ω2, ω3 are the weighting coefficients of each factor, reflecting the influence weights of different factors on the supply chain. By monitoring environmental data in real time, the impact of natural disasters and climate change factors on each node of the supply chain can be evaluated. Based on the risk factors in the above three dimensions, a multi-dimensional risk matrix R(t) is constructed to comprehensively evaluate the risks in the supply chain. This matrix is defined as: Each row in the matrix represents the risk status of the supply chain at a certain time point, and each column represents different risk dimensions. By analyzing the matrix, the system can track the change trend of supply chain risks over time and identify high-risk time periods and links. The supply chain risk assessment model is generated based on the risk matrix R(t). By performing weighted analysis on each dimension in the matrix, the overall risk assessment model R total (t) is calculated: R totalR(t) = α1·I(t) + α2·T(t) + α3·E(t), where α1, α2, and α3 are the weight coefficients of each risk dimension, representing the impact of different risks on the overall supply chain. As the real-time data in the supply chain is continuously updated, the risk assessment model will continuously adjust the weight coefficients and risk indicators to reflect the latest supply chain status. At the same time, the risk model can be iteratively optimized through the accumulation of historical data to predict potential future risks. It is worth mentioning that the dynamic traceability identification code collects and feeds back various types of data in the supply chain in real time to help identify and adjust the current status, such as the real-time situation of production, inventory, transportation, and sales. It adjusts by predicting future demand and supply chain conditions, but it is more of an immediate response mechanism to ensure that the supply chain can adjust to dynamic changes in a short period of time. The risk assessment model, on the other hand, makes long-term predictions and evaluations based on this real-time data to build a global and comprehensive evaluation framework. Its purpose is to detect potential long-term risks in advance and help supply chain managers make more strategic decisions. The risk assessment model is not just about dealing with immediate risks, but evaluating potential risk trends at a systematic level, conducting in-depth analysis, and thus formulating long-term coping strategies. That is to say, the risk assessment model is not only used to deal with immediate risks, but by analyzing long-term data, it discovers systematic risks and then helps enterprises make strategic decisions. The impacts of these decisions are usually medium- and long-term, such as optimizing the supply chain structure, formulating new procurement strategies, and adjusting transportation routes. The synergistic effect of the dynamic traceability identification code and the risk assessment model is to ensure that the supply chain can respond to short-term changes and make long-term plans.

[0067] S6. Based on the supply chain risk assessment model and the supply chain real-time monitoring data set, cross-validate and conduct correlation analysis on the data of each supply chain node, and construct a full-chain data correlation network by comprehensively processing the data of the production, transportation, storage, and sales links;

[0068] Specifically, the purpose of data cross-validation is to ensure the consistency and integrity of the data between each supply chain node. In the D realtime data set, the production node data D prod records the detailed information of product production, the transportation node data D tran records the dynamic data of the logistics link, the storage node data D stock describes the warehousing situation, and the sales node data D sales records the transfer information of the sales process. The cross-validation process includes data consistency analysis and data time-series consistency analysis. The data consistency analysis is to ensure the coordination of data between different links and conduct logical consistency analysis on the data of each node. For example, the batch number and production quantity in the production node D prod should be consistent with those in the transportation node Dtrans matches the shipment batch number and the quantity transported in it; similarly, the transportation data D trans 's arrival information should be consistent with the storage node D stock 's warehousing data. Specifically, it is manifested as: D prod (bach_id) = D trans (batch_id) = D stock (batch_id) = D sales (batch_id). Through such batch number consistency checks, it is ensured that the data flow of products in the production, transportation, storage, and sales processes is consistent and complete. If data inconsistency occurs in a certain link, the system will mark the anomaly and further conduct fault analysis. The data time series consistency analysis is that D trans 's transportation time should be coordinated with D prod 's production completion time, and the warehousing time in the storage data D stock must also match the transportation arrival time within a reasonable time window. Mathematically, it can be expressed as: T trans (arrival) - T prod (completion) < ΔT trans_prod , where ΔT trans_prod represents the reasonable time difference range in the transportation link. If the time difference exceeds this range, it may indicate delays or anomalies in the transportation link.

[0069] After cross - verification, the data relationships between each node are constructed through association analysis. The goal of association analysis is to reveal the interactions between production, transportation, storage, and sales in order to better understand the synergy effects of each link in the supply chain. Based on the real - time monitoring data set of the supply chain, analyze the data dependency relationships among the production, transportation, storage, and sales links. Model the mutual influences among these links through correlation analysis and dependency analysis. Correlation analysis uses the Pearson correlation coefficient to measure the degree of data correlation between each node: where D i and D j respectively represent the data of two supply chain nodes, and Corr(D i , D j ) reflects the linear correlation between the two data sets. By analyzing these correlations, it is possible to identify which fluctuations in supply chain nodes will significantly affect other nodes. For example, if D prod has a high correlation with D sales , it indicates that the production quantity has a direct impact on sales. To more precisely analyze the dependencies between nodes, a conditional probability model is introduced simultaneously, and a dependency relationship model of the production, transportation, storage, and sales links is established through a Bayesian network. Its mathematical expression is: Here, P(D j|D i ) represents the conditional probability of node D given the known data i Under the condition of data, node D j The conditional probability of the data. By analyzing these conditional probabilities, the causal dependence relationship between nodes can be clarified. Especially when an abnormality occurs in a certain link, it can be traced which link's data abnormality is the root cause.

[0070] Based on the association analysis, a full-chain data association network of the supply chain is finally generated. This network contains the data dependence relationships and risk propagation paths of each node in the supply chain, thus providing basic support for the optimization, prediction, and management of the supply chain. Specifically, in the supply chain association network, network nodes represent different links of the supply chain (production, transportation, storage, sales), and the attributes of the nodes include their respective real-time data, risk assessment results, and the time characteristics of the nodes. For example, node N prod represents the production node, and its attributes include: Nprod = {Dprod, Rprod, Tprod}, D prod is the real-time data of the production node; R prod is the risk assessment value of the production node (provided by R total (t)); T prod is the time characteristic of the production node. Other nodes such as N trans , N stock , N sales are constructed in a similar way, representing the transportation, storage, and sales links respectively. The edges between nodes represent the relevance and dependence between each link. The weight of the edge is calculated through relevance and dependence analysis, and the formula is expressed as: w ij =α·Corr(D i , D j ) + β·P(D j |D i ), where w ij is the association weight between node D i and D j , Corr(D i , D j ) is the data correlation coefficient between the two, P(D j |D i ) is the conditional probability, and α and β are weight adjustment coefficients. An edge with a high weight indicates a strong dependence between two links, while a low weight indicates a weak association between them. According to the association weight w ij , the existence or non-existence of edge e ij can be defined, and the formula is expressed as: That is, when the association weight w ij between two nodes is greater than or equal to the preset threshold, there is an edge e between node D i and D j ​ij = 1, otherwise there is no edge e ij = 0. The presence (1) or absence (0) of an edge is not only a binary relation in mathematics, but also represents the dependencies and interactions between supply chain links in reality. For supply chain management, this means being able to determine which nodes have data dependencies or business associations, and which nodes are relatively independent. This can help managers clearly identify which links need to be focused on, especially in abnormal situations (such as delays or inventory backlogs), and the root cause of problems can be traced through the nodes with edges.

[0071] Based on the above analysis, the full-chain data association network constructed in this implementation is represented by the graph structure G = (N, E, W), where N = {N prod , N trans , N stock , N sales} represents the various link nodes in the supply chain; E = {e ij} represents the edges between nodes, that is, the association relationships between different supply chain links; W = {w ij} represents the weights of the edges, calculated by the above formula. The full-chain data association network not only reveals the interactions between nodes by analyzing the relevance and dependence of each link, but also clarifies the influence intensity of each link and the risk propagation path. For example, if w ij is relatively large, it indicates a strong dependence between the production node and the transportation node, and fluctuations in the production link may significantly affect the normal operation of the transportation link.

[0072] To sum up, as shown in the attached Figure 2 , step S6 includes the following sub-steps:

[0073] S61, perform cross-validation on the real-time monitoring data set of the supply chain, and the cross-validation includes data consistency analysis and data temporal consistency analysis;

[0074] S62, based on the real-time monitoring data set of the supply chain after cross-validation, perform association analysis to construct the data relationships between nodes, and the association analysis includes correlation analysis and dependence analysis;

[0075] S63, based on the results of the association analysis, generate a full-chain data association network by constructing a mutual influence model of each node in the supply chain.

[0076] S7, based on the full-chain data association network, use distributed storage and node consensus mechanism to implement full-chain traceability management. The full-chain traceability management includes full-chain data recording and tracking of production, transportation, storage, and sales links, ensuring the security, integrity, and traceability of data of each node in the supply chain, generating a full-chain traceability record set for the transparent management of the supply chain.

[0077] Specifically, based on the constructed full-chain data association network G = (N, E, W), through distributed storage and consensus mechanism, the tracking and recording of the production, transportation, storage, and sales links in the supply chain are realized. Thus, ensuring the data security, integrity, and traceability of each supply chain node, and finally generating a full-chain traceability record set for enhancing the transparency and security of supply chain management.

[0078] The data of supply chain nodes is stored distributively to ensure redundant storage of data and synchronous management across nodes, so as to avoid single-point failures and enhance data security and accessibility. The specific implementation includes a distributed storage architecture and a redundant storage strategy. The distributed storage architecture uses a distributed file system, and in this embodiment, the IPFS system is used for distributed data storage. The data on each node N is split into small data blocks, and a unique identifier is generated through the hash algorithm H(x). These data blocks are stored on different distributed nodes. The specific formula is as follows: H(x) = hash(x), where x represents the data to be stored, and H(x) is the unique identifier of this data. H(x) is used as the unique identifier for data consistency verification and protection in specific links. If the data of a certain node changes, it needs to be verified through its independent unique identifier, rather than through the identifier of the entire chain. The distributed storage system ensures redundant storage of data on different nodes to achieve fault recovery and data security. The redundant storage strategy is to ensure data security. Each data block x is stored on at least k different storage nodes, where k is the fault tolerance coefficient of the system: redundancy = k, and the system sets k ≥ 3 to ensure the integrity and recoverability of data when any two storage nodes fail.

[0079] To ensure data consistency among nodes in a distributed storage environment, the system introduces a node consensus mechanism. The consensus mechanism ensures that all nodes in the supply chain reach an agreement on the update of key data, avoids data tampering, and guarantees the transparency and credibility of traceability management. In one or more embodiments, the PBFT (Practical Byzantine Fault Tolerance) or PoS (Proof of Stake) consensus mechanism is used to ensure consistency among nodes in data processing and verification. For each new piece of supply chain data (such as a new production record, transportation information, etc.), the consensus mechanism combines the confirmation information of each node to reach the consensus threshold f set in the network, that is: where n is the number of nodes participating in the consensus. When more than f nodes reach a consensus, the new supply chain data can be written into the full-chain data record. This mechanism prevents the risk of malicious nodes tampering with data and ensures the integrity and security of the full-chain data.

[0080] Through distributed storage and consensus mechanisms, traceability management is carried out on the data of each link in the supply chain. When each supply chain node operates, its data will be written into the full-chain traceability record set, which includes data on four main links: production, transportation, storage, and sales. The specific implementation is as follows:

[0081] Traceability of production nodes: Production node N prod 's data D prod is stored on multiple nodes through the consensus mechanism and the distributed storage system, generating a unique identifier H(D prod ). In this embodiment, the traceability record set contains detailed information on production time, production batch number, and raw material source.

[0082] The production part R in the traceability record set trace-prod is expressed as: R trace-prod ={H(D prod ), T prod , B prod , S prod}, where T prod is the production time, B prod is the batch number, and S prod is other information related to the production link, which is the product specification in this embodiment.

[0083] Traceability of transportation nodes: Transportation node N trans 's data D trans is recorded and stored to generate a transportation record, which includes transportation time, transportation route, and transportation vehicle information in this embodiment. The generated traceability record part is: R trace-trans ={H(D trans ), T trans , P trans , V trans}, where P trans is the transportation route and V trans is the transportation vehicle information.

[0084] Traceability of storage nodes: Storage node N stock 's data D stock describes the storage location, warehousing time, and inventory quantity of the product in this embodiment. The generated traceability record part is: R trace-stock ={H(D stock ), T stock , L stock , Q stock}, where L stock is the storage location and Q stock is the inventory quantity.

[0085] Traceability of sales nodes: Sales node N sales 's data D salesRecord the sales information of the product, which includes the sales time, sales location, and sales batch number in this embodiment. The traceability record is: R trace-sales ={H(D sales ),T sales ,B sales ,P sales}, where B sales is the sales batch number, and P sales is the sales point information.

[0086] Combine the traceability records of each node to generate a full-chain traceability record set R trace , which covers all key nodes in the supply chain to ensure that the data of each node is traceable throughout the process from production to sales. The final traceability record set includes: R trace ={R trace-prod ,R trace-trans ,R trace-stock ,R trace-sales}. Through the full-chain traceability record set, supply chain managers can easily query the complete supply chain path of any product, including all links of its production, transportation, storage, and sales, ensuring transparent management of the supply chain and enhancing risk control capabilities.

[0087] At the same time, in order to ensure the security and integrity of the traceability records, the full-chain traceability management uses a multi-level security mechanism, including encrypted storage, hash verification, and anti-tampering mechanism. The encrypted storage is to encrypt all traceability data D before storage, using an asymmetric encryption algorithm. In this embodiment, RSA or elliptic curve encryption (ECC) is used to ensure that only users with the corresponding decryption key can access sensitive data. The hash verification is to verify the hash value H(x) of each data block through a consensus mechanism to ensure that the data has not been tampered with. Each data operation generates a new hash value and compares it with the original hash value to verify the data integrity. The anti-tampering mechanism is to ensure that the traceability records cannot be tampered with through blockchain or distributed ledger technology, and all data cannot be maliciously modified after being written into the traceability record set.

[0088] In summary, as shown in the attached Figure 3 figure, the S7 step includes the following sub-steps:

[0089] S71, Based on the full-chain data association network, perform distributed storage configuration on the data of each key node in the supply chain to enable secure storage and redundant backup of each node's data;

[0090] S72, Based on the data set of the distributed storage configuration, use the node consensus mechanism to perform consistency verification on the key data to ensure that the data records on all nodes are consistent and cannot be tampered with;

[0091] S73. Based on the dataset after consistency verification, generate the full-chain traceability records for each node, and integrate the traceability record set into the full-chain traceability management system to enable transparent management of all links in the supply chain.

[0092] Based on the description of the above embodiments of the fresh product full-chain traceability method, an embodiment of the present application also discloses a fresh product full-chain traceability system. The fresh product full-chain traceability system may be a computer program (including program code) that runs the above-mentioned fresh product full-chain traceability method. Please refer to the attached Figure 4 As shown, the fresh product full-chain traceability system may run the following units:

[0093] An acquisition unit 110, configured to acquire the predicted feature vectors of the target fresh product and competing products, where the predicted feature vectors include environmental conditions, market demand, seasonal factors, and user behavior characteristics;

[0094] A prediction mapping unit 120, configured to construct a prediction mapping function based on the predicted feature vectors. The prediction mapping function includes a sales volume prediction model for the target product, a sales volume prediction model for competing products, and a residual adjustment model, and is used to generate a sales volume prediction result;

[0095] A comprehensive traceability identification code construction unit 130, configured to acquire and construct a comprehensive traceability identification code. The comprehensive traceability identification code includes a static traceability identification code and a dynamic traceability identification code. The static traceability identification code is used to mark the basic attribute information of the product, and the dynamic traceability identification code is generated according to the prediction mapping function; the comprehensive traceability identification code is used to mark each product unit in the supply chain, and through data mapping, the traceability identification code is associated with the information of each node in the supply chain to form a traceability mapping set;

[0096] A real-time data acquisition unit 140, configured to perform real-time data acquisition on the fresh products in the supply chain nodes based on the comprehensive traceability identification code. The real-time data acquisition includes production data, transportation data, inventory dynamic data, and product transfer data during the sales process, so as to obtain a real-time monitoring dataset of the supply chain;

[0097] A risk assessment unit 150, configured to perform multi-dimensional risk matrix analysis of the supply chain based on the real-time monitoring dataset of the supply chain and the traceability mapping set. The analysis includes the impacts of inventory fluctuations, transportation delays, and environmental factors, and generates a supply chain risk assessment model;

[0098] A data association analysis unit 160, configured to perform cross-verification and association analysis on the data of each node in the supply chain based on the supply chain risk assessment model and the real-time monitoring dataset of the supply chain, and construct a full-chain data association network by comprehensively processing the data of the production, transportation, storage, and sales links;

[0099] The full-chain traceability management unit 170 is used to implement full-chain traceability management based on the full-chain data association network, using distributed storage and node consensus mechanisms. The full-chain traceability management includes recording and tracking the full-chain data in the production, transportation, storage, and sales links, ensuring the security, integrity, and traceability of the data of each node in the supply chain, generating a full-chain traceability record set, and being used for the transparent management of the supply chain.

[0100] The above are only the preferred embodiments of the present invention. It should be understood that the present invention is not limited to the form disclosed herein, should not be regarded as excluding other embodiments, but can be used in various other combinations, modifications, and environments, and can be changed within the scope of the concept described herein through the above teachings or the technology or knowledge in related fields. And the changes and alterations made by those skilled in the art without departing from the spirit and scope of the present invention shall all fall within the protection scope of the appended claims of the present invention.

Claims

1. A full-chain traceability method for fresh products, characterized in that: The method comprises the following steps: S1, obtaining prediction feature vectors of target fresh products and competing products, wherein the prediction feature vectors include environmental conditions, market demand, seasonal factors and user behavior characteristics; S2, constructing a prediction mapping function based on the prediction feature vector, wherein the prediction mapping function includes a sales prediction model of the target product, a sales prediction model of the competing product, and a residual adjustment model, and is used to generate a sales prediction result; S3, obtaining and constructing a comprehensive traceability identification code, wherein the comprehensive traceability identification code includes a static traceability identification code and a dynamic traceability identification code, wherein the static traceability identification code is used to mark basic attribute information of the product, and the dynamic traceability identification code is generated according to the prediction mapping function; the comprehensive traceability identification code is used to mark each product unit in the supply chain, and the traceability identification code is associated with each node information in the supply chain through data mapping to form a traceability mapping set; S4, based on the comprehensive traceability identification code, real-time data collection is performed on the fresh products in the supply chain nodes, wherein the real-time data collection includes production data, transportation data, inventory dynamic data, and product flow data during the sales process, thereby obtaining a supply chain real-time monitoring data set; S5, based on the supply chain real-time monitoring data set and the traceability mapping set, performing a supply chain multi-dimensional risk matrix analysis, the analysis including the impact of inventory fluctuations, transportation delays and environmental factors, and generating a supply chain risk assessment model; S6, based on the supply chain risk assessment model and the supply chain real-time monitoring data set, cross-validates and analyzes the data of each node in the supply chain, and constructs a full-chain data association network by comprehensively processing the data of production, transportation, storage and sales links; S7, based on the full-chain data association network, uses distributed storage and node consensus mechanism to realize full-chain traceability management. The full-chain traceability management includes full-chain data recording and tracking of production, transportation, storage and sales links, ensuring the security, integrity and traceability of data at each node in the supply chain, and generating a full-chain traceability record set for transparent management of the supply chain.

2. A full-chain traceability method for fresh products according to claim 1, characterized in that: The prediction mapping function described in step S2 is: adjusted,t =ω1Y target,t +ω2Y competitor,t +R target,t , where ω1 and ω2 are weight coefficients, satisfying ω1+ω2=1; R target,t is the residual adjustment value, which is used to correct the prediction results, Y target The predicted sales volume of the target product, Y competitor Forecast sales of competing products.

3. A full-chain traceability method for fresh products according to claim 1, characterized in that: The static traceability identification code in step S3 is used to mark the basic attribute information of the product, including production date, origin, and product type data; the dynamic traceability identification code contains dynamic information on future market demand, environmental changes, and sales trends, and its specific calculation formula is as follows: dynamic =Y adjusted,t +A t , where Y adjusted,t is the sales forecast value, A t is the adjustment factor, T dynamic is a dynamic traceability identification code, and the adjustment factor includes the supply chain volatility index SCI t , Environmental Condition Index ECI t .

4. A full-chain traceability method for fresh products according to claim 3, characterized in that: The supply chain real-time monitoring dataset D in step S4 realtime D realtime ={D prod ,D trans ,D stock ,D sales }, where D prod is the production node data, D tran is the transport node data, D stock To store node data, D sales It is the sales node data.

5. A full-chain traceability method for fresh products according to any one of claims 1 to 4, characterized in that: The risk assessment model in step S5 is constructed by combining the inventory fluctuation quantification model, the transportation delay quantification model, and the environmental factor quantification model. total (t) The specific calculation formula is as follows: R total (t) = α1·I(t)+α2·T(t)+α3·E(t), where α1, α2, and α3 are the weight coefficients of the inventory fluctuation quantification model, the transportation delay quantification model, and the environmental factor quantification model, representing the impact of different risks on the overall supply chain. I(t) is the inventory fluctuation quantification model, T(t) is the transportation delay quantification model, and E(t) is the environmental factor quantification model.

6. A method for full-chain traceability of fresh products according to claim 5, characterized in that: Step S6 includes the following sub-steps: S61, cross-validating the supply chain real-time monitoring data set, wherein the cross-validation includes data consistency analysis and data time series consistency analysis; S62, based on the cross-validated supply chain real-time monitoring data set, performing association analysis to construct a data relationship between each node, wherein the association analysis includes phase relationship analysis and dependency analysis; S63, based on the result of association analysis, generate a full-chain data association network.

7. A method for full-chain traceability of fresh products according to claim 5, characterized in that: The S7 step includes the following sub-steps: S71, based on the full-chain data association network, performs distributed storage configuration for key node data in the supply chain, so that each node data can be securely stored and redundantly backed up; S72, based on the data set configured with distributed storage, uses the node consensus mechanism to verify the consistency of key data, so that the data records on all nodes are consistent and cannot be tampered with; S73, based on the consistency-verified data set, generates the full-chain traceability record of each node, and integrates the traceability record set into the full-chain traceability management system, enabling transparent management of all links in the supply chain.

8. A method for full-chain traceability of fresh products according to claim 6, characterized in that: The full-chain data association network is G = (N, E, W), where N = {N prod ,N trans ,N stock ,N sales } represents each link node in the supply chain, E={e ij } represents the edge between nodes, that is, the relationship between different supply chain links; W = {w ij } represents the weight of the edge, and its specific calculation formula is as follows: ij =α·Corr(D i ,D j )+β·P(D j |D i ), where w ij For node D i and D j The correlation weight between i ,D j ) is the data correlation coefficient between the two, P(D j ∣D i ) is the conditional probability, α and β are the weight adjustment coefficients.

9. A method for full-chain traceability of fresh products according to claim 7, characterized in that: The node consensus mechanism is a practical Byzantine fault-tolerant consensus mechanism or a proof-of-stake consensus mechanism.

10. A full-chain traceability system for fresh products, characterized in that: The system comprises the following units: An acquisition unit, used to acquire prediction feature vectors of target fresh products and competing commodities, wherein the prediction feature vectors include environmental conditions, market demand, seasonal factors, and user behavior characteristics; A prediction mapping unit, configured to construct a prediction mapping function based on the prediction feature vector, wherein the prediction mapping function includes a sales prediction model of a target product, a sales prediction model of a competing product, and a residual adjustment model, and is configured to generate a sales prediction result; a comprehensive traceability identification code construction unit, used to obtain and construct a comprehensive traceability identification code, wherein the comprehensive traceability identification code includes a static traceability identification code and a dynamic traceability identification code, wherein the static traceability identification code is used to mark basic attribute information of the product, and the dynamic traceability identification code is generated according to the prediction mapping function; the comprehensive traceability identification code is used to mark each product unit in the supply chain, and the traceability identification code is associated with each node information in the supply chain through data mapping to form a traceability mapping set; A real-time data collection unit, used to collect real-time data of fresh products in supply chain nodes based on the comprehensive traceability identification code, wherein the real-time data collection includes production data, transportation data, inventory dynamic data, and product flow data during the sales process, so as to obtain a real-time monitoring data set for the supply chain; A risk assessment unit, for performing a supply chain multi-dimensional risk matrix analysis based on a supply chain real-time monitoring data set and a traceability mapping set, wherein the analysis includes the impact of inventory fluctuations, transportation delays, and environmental factors, and generating a supply chain risk assessment model; The data association analysis unit is used to cross-validate and conduct association analysis on the data of each node in the supply chain based on the supply chain risk assessment model and the supply chain real-time monitoring data set, and to build a full-chain data association network by comprehensively processing the data of production, transportation, storage and sales links; The full-chain traceability management unit is used to implement full-chain traceability management based on the full-chain data association network, using distributed storage and node consensus mechanism. The full-chain traceability management includes full-chain data recording and tracking of production, transportation, storage and sales links, ensuring the security, integrity and traceability of data at each node in the supply chain, and generating a full-chain traceability record set for transparent management of the supply chain.

Images