Block chain traceability-based catering supply chain risk early warning method and system

Abstract

The invention discloses a blockchain traceability catering supply chain risk early warning method and system, and the method comprises the steps: deploying a multi-source sensor in a cold chain transportation link, collecting the temperature, humidity and vibration intensity data of food materials in real time, and carrying out the real-time binding with the GPS track of a transportation vehicle, and forming a space-time related transportation environment data set; setting a dynamic threshold based on the type of the transported food material, starting redundant sensor cross validation when sensor data in the transportation environment data set continuously exceeds the dynamic threshold for N times, and if the deviation between the validation data and the transportation environment data set exceeds a preset tolerance, marking the transportation environment data set as an abnormal data segment and triggering block chain evidence storage; and compressing the abnormal data segment and the GPS track fragment associated with the abnormal data segment into a data block, writing the data block into a lightweight alliance chain after the data block is subjected to double signature by a transportation party node and a receiving party node, and generating a traceability code containing a timestamp. According to the method, the supply chain risk disposal efficiency is improved through fusion of space-time gridding multi-source data and risk multi-dimensional assessment.

Description

Technical Field

[0001] The present application relates to the field of catering supply chain technology, and in particular to a catering supply chain risk warning method and system based on blockchain traceability. Background Art

[0002] Supply chain management refers to the coordinated management of the entire life cycle, from raw material procurement to product delivery to consumers, through the coordinated efforts of suppliers, manufacturers, logistics providers, and end-user sales networks. Its core goal is to optimize the efficiency of multi-link connections, ensure product quality and safety, and achieve a comprehensive balance between cost, timeliness, and resource utilization. In the food sector, supply chain management is particularly critical, encompassing key nodes such as raw material traceability, temperature-controlled warehousing, cold chain transportation, inventory turnover, and distribution networks. It focuses on addressing quality risks posed by food perishability, losses caused by fluctuations in the transportation environment, and information asymmetry in cross-party collaboration.

[0003] The existing Chinese patent, with the announcement number CN117993724B, relates to the field of supply chain management technology, specifically a blockchain-based catering supply chain safety management system and method. The system includes: a compliance verification module, a dynamic inventory management module, a food safety monitoring module, a supply chain configuration optimization module, a risk prediction and management module, a decision support system module, and a safety compliance assessment module. This invention optimizes the transparency and security of supply chain transactions through a decision tree algorithm and a logistic regression model, combined with blockchain smart contracts. It also optimizes supply chain response speed using an autoregressive model and a long-short-term memory network. It also ensures food quality and safety using an isolation forest algorithm and an autoencoder algorithm. It also applies genetic algorithms and support vector machines to optimize supply chain structure and strategy. Social network analysis and gated recurrent units provide data support for risk prediction. Multi-criteria decision analysis and data envelopment analysis are used to optimize management decisions, improving the efficiency and stability of the catering supply chain.

[0004] However, in the process of implementing relevant technical solutions, it was found that there were at least the following technical problems:

[0005] 1. A real-time dynamic binding mechanism for cold chain transport environmental parameters (temperature, humidity, and vibration) to GPS tracks was not established. Abnormal data determination relied on static thresholds and lacked cross-validation using redundant sensors, resulting in insufficient data credibility. 2. Risk assessment was limited in its single dimension, failing to integrate spatiotemporal characteristics with multi-indicator correlation analysis. Insufficient comparison between historical data and the real-time transport environment made it difficult to accurately locate risk nodes and trigger tiered responses. These deficiencies could easily lead to delayed warnings or misjudgments, impacting the efficiency of supply chain risk management. Summary of the Invention

[0006] In order to solve the above problems, an embodiment of the present invention provides a blockchain traceability catering supply chain risk warning method, which includes:

[0007] Deploy multi-source sensors in cold chain transportation to collect food temperature, humidity, and vibration intensity data in real time. These data are then linked to the GPS trajectory of transportation vehicles to form a spatiotemporally correlated transportation environment dataset.

[0008] A dynamic threshold is set based on the type of food being transported. When the sensor data in the transportation environment dataset exceeds the dynamic threshold N times in a row, where N is a preset value, redundant sensor cross-validation is initiated. If the deviation between the validation data and the transportation environment dataset exceeds the preset tolerance, the transportation environment dataset is marked as an abnormal data segment and blockchain evidence is triggered.

[0009] The abnormal data segment and its associated GPS trajectory segment are compressed into a data block, written into the lightweight alliance chain after being double-signed by the transporter and the recipient node, and a traceability code including a timestamp is generated;

[0010] Extract the spatiotemporal characteristics of the data block, compare it with historical normal transportation data on the same route, calculate the temperature fluctuation rate, the proportion of abnormal duration, and the cumulative value of vibration intensity, and generate a three-dimensional risk heat map;

[0011] If at least two dimensions in the three-dimensional risk heat map exceed the corresponding baseline values, the smart contract call will be activated to push a graded warning signal to the recipient, including the start and end time of the abnormal data segment, the associated responsible parties, and recommended re-inspection measures.

[0012] Furthermore, when the spatiotemporally correlated transportation environment dataset is generated, a sliding time window algorithm is used to perform time-series alignment between the sensor data and the GPS trajectory.

[0013] Furthermore, the dynamic threshold setting logic includes:

[0014] The temperature threshold is dynamically adjusted based on the remaining shelf life of the food, and is calculated using the following formula:

[0015] ;

[0016] Where, The upper limit of the standard storage temperature for food. is the total shelf life in days, The remaining days, It is a preset attenuation factor based on the food type; when the remaining shelf life days are reduced to a preset ratio, the nonlinear threshold tightening mechanism is triggered.

[0017] Furthermore, the lightweight alliance chain adopts a shard storage structure, where the abnormal data segment is stored in the shard where the receiver is located, and the GPS track segment is stored in the shard where the logistics party is located. Data association query is realized through cross-shard hash index; the storage method of the lightweight alliance chain includes:

[0018] First, in the data compression stage, the abnormal data segment and its corresponding GPS trajectory segment are packaged separately to form a data unit with an independent hash identifier.

[0019] Secondly, during the cross-shard index construction phase, a bidirectional hash pointer is generated for each pair of associated data. When an abnormal data segment is written to the receiving shard, the hash value of the abnormal data segment is automatically embedded in the metadata field of the corresponding GPS track segment. At the same time, the hash value of the GPS block is backfilled into the header of the abnormal data segment.

[0020] Finally, in the data query and verification phase, alliance chain members use the inter-shard consensus protocol to achieve cross-region data retrieval.

[0021] Furthermore, the method for generating the three-dimensional risk heat map includes: dividing the transportation route into geographical grid units, counting the cumulative value of abnormal indicators in each unit, and superimposing the grid units to generate a three-dimensional risk heat map.

[0022] Furthermore, the method for generating the graded warning signal includes: when the temperature fluctuation rate and the cumulative value of vibration intensity exceed the standard at the same time, it is marked as a first-level warning and it is recommended to suspend reception; when only a single dimension exceeds the standard, it is marked as a second-level warning and it is recommended to conduct partial sampling; after the warning signal is generated, the warning record is synchronously updated through the alliance chain shard storage architecture, and the warning level and disposal suggestion summary are written to the receiving party shard, and the corresponding environmental data proof fragment is stored in the logistics party shard.

[0023] Furthermore, the redundant sensor cross-validation method includes:

[0024] If the initial sensor data exceeds the threshold, the data of at least two backup sensors separated by a preset distance in the same compartment are retrieved;

[0025] Calculate the standard deviation between the sensor data. If the standard deviation exceeds the judgment threshold associated with the preset tolerance, it is judged as equipment failure data and discarded.

[0026] Otherwise, the mean value of each sensor data is taken as the valid abnormal data segment.

[0027] Furthermore, the method for obtaining historical normal transportation data of the same route includes:

[0028] Initiate data requests to other logistics companies through the federated learning protocol, and each company locally calculates the mean and variance of temperature and vibration indicators of historical transportation data;

[0029] Add differential privacy noise to the mean and variance and upload them to the aggregation node;

[0030] The aggregation node uses a dynamic weight allocation mechanism to generate a global historical baseline value, and the weight is determined based on the amount of data and data credibility of each enterprise.

[0031] The blockchain traceability catering supply chain risk warning method also includes:

[0032] A real-time environmental parameter feedback system is introduced during the generation of a three-dimensional risk heat map. The introduction method of the real-time environmental parameter feedback system includes:

[0033] Through the communication module on the logistics vehicle, the vehicle can access the meteorological platform in real time to obtain the rainfall intensity, wind speed and UV index of the current transportation route;

[0034] Call the real-time traffic data interface of the traffic management department to extract the vibration interference factor;

[0035] Environmental parameters are mapped into compensation coefficients. Environmental parameters include rainfall intensity, wind speed, UV index and vibration interference factor of the current transportation route; and the comparison threshold of the historical baseline value is dynamically adjusted.

[0036] The blockchain traceability catering supply chain risk early warning system includes:

[0037] The data set collection module deploys multi-source sensors in the cold chain transportation process to collect food temperature, humidity, and vibration intensity data in real time, and binds it to the GPS trajectory of the transportation vehicle in real time to form a transportation environment data set with temporal and spatial correlation;

[0038] The anomaly determination module sets a dynamic threshold based on the type of food being transported. When the sensor data in the transportation environment dataset exceeds the dynamic threshold N times in a row, where N is a preset value, redundant sensor cross-validation is initiated. If the deviation between the validation data and the transportation environment dataset exceeds the preset tolerance, the transportation environment dataset is marked as an abnormal data segment and blockchain evidence is triggered.

[0039] The data compression module compresses the abnormal data segment and its associated GPS trajectory segment into a data block, writes it into the lightweight alliance chain after being double-signed by the transporter and the recipient node, and generates a traceability code including a timestamp;

[0040] The risk visualization module extracts the spatiotemporal characteristics of data blocks, compares them with historical normal transportation data on the same route, calculates the temperature fluctuation rate, the proportion of abnormal duration, and the cumulative value of vibration intensity, and generates a three-dimensional risk heat map;

[0041] In the risk processing module, if at least two dimensions in the three-dimensional risk heat map exceed the corresponding baseline value, the smart contract call is activated to push a graded warning signal containing the start and end time of the abnormal data segment, the related responsible parties and the recommended re-inspection measures to the recipient.

[0042] The technical effects and advantages of the blockchain traceability catering supply chain risk early warning method and system provided by the present invention are as follows:

[0043] This invention overcomes the technical bottlenecks of low credibility of environmental monitoring data and delayed risk warning in cold chain transportation by constructing a spatiotemporal grid multi-source data fusion model and dynamic risk multi-dimensional assessment, achieving accurate verification of abnormal data and hierarchical response of risk nodes, effectively improving the efficiency of supply chain risk management. The invention proposes a spatiotemporal correlation data model of multi-source sensors and GPS trajectories, combined with dynamic threshold adjustment and redundant verification mechanism, to achieve reliable evidence storage and rapid positioning of abnormal data, reducing the misjudgment rate; constructs a three-dimensional risk heat map and a cross-shard blockchain storage architecture, integrating multi-dimensional spatiotemporal feature analysis of temperature fluctuations, vibration intensity, and abnormal duration, triggering hierarchical warnings driven by smart contracts, and improving the accuracy of risk identification and the efficiency of coordinated response. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 This is a flow chart of the blockchain traceability catering supply chain risk warning method in Example 1;

[0045] Figure 2 This is a flow chart of the blockchain traceability catering supply chain risk warning method in Example 2;

[0046] Figure 3 This is a connection diagram of the blockchain traceability catering supply chain risk warning system in Example 3. DETAILED DESCRIPTION

[0047] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0048] Example 1:

[0049] See also Figure 1 As shown, an embodiment of the present invention provides a blockchain traceability catering supply chain risk early warning method, the method comprising:

[0050] Deploy multi-source sensors in cold chain transportation to collect food temperature, humidity, and vibration intensity data in real time. These data are then linked to the GPS trajectory of transportation vehicles to form a spatiotemporally correlated transportation environment dataset.

[0051] A dynamic threshold is set based on the type of food being transported. When the sensor data in the transportation environment dataset exceeds the dynamic threshold N times in a row, where N is a preset value, redundant sensor cross-validation is initiated. If the deviation between the validation data and the transportation environment dataset exceeds the preset tolerance, the transportation environment dataset is marked as an abnormal data segment and blockchain evidence is triggered.

[0052] The abnormal data segment and its associated GPS trajectory segment are compressed into a data block, written into the lightweight alliance chain after being double-signed by the transporter and the recipient node, and a traceability code including a timestamp is generated;

[0053] Extract the spatiotemporal characteristics of the data block, compare it with historical normal transportation data on the same route, calculate the temperature fluctuation rate, the proportion of abnormal duration, and the cumulative value of vibration intensity, and generate a three-dimensional risk heat map;

[0054] If at least two dimensions in the three-dimensional risk heat map exceed the corresponding baseline values, the smart contract call will be activated to push a graded warning signal to the recipient, including the start and end time of the abnormal data segment, the associated responsible parties, and recommended re-inspection measures.

[0055] When generating a spatiotemporally correlated transportation environment dataset, a sliding time window algorithm is used to align the sensor data with the GPS trajectory to eliminate the offset error caused by device clock asynchrony; the sensor data includes food temperature, humidity, and vibration intensity data.

[0056] During cold chain transportation, clock synchronization issues across different devices can cause temporal misalignment between sensor data and GPS tracks. For example, the data recorded by a temperature sensor at 14:30:05 should match the vehicle's GPS track coordinates between 14:30:00 and 14:30:10. However, clock deviations of several to tens of seconds between devices can lead to data misalignment. Therefore, a sliding time window algorithm is used to achieve dynamic temporal alignment of multi-source data when generating a spatiotemporally correlated transportation environment dataset.

[0057] During the specific implementation, an independent data processing channel is first built for each transport vehicle. During the data collection stage, sensors such as temperature and humidity upload monitoring values ​​at a fixed frequency. At the same time, the on-board GPS device uploads longitude and latitude coordinates at set intervals. The two types of data streams are marked with the timestamps of their respective device clocks during the transmission process.

[0058] The timing alignment process includes:

[0059] Based on the time axis of GPS data, a sliding time window including front and back buffers is set. The time span of the sliding time window is preset to ensure that it can cover the possible clock offset range of the sensor data.

[0060] The sliding time window slides on the time axis in incremental steps. Each time it moves, the sensor data in the window is automatically extracted and matched with the GPS trajectory points for features. For example, when the window slides to the interval of 14:30:00-14:30:30, all sensor readings in that period are associated and mapped with GPS coordinates.

[0061] The optimal offset between the sensor time series and the GPS trajectory is calculated through a dynamic programming algorithm to automatically correct the clock differences between devices. This overlapping sliding mechanism can not only eliminate fixed offset errors, but also adapt to occasional clock drift that may occur during transportation.

[0062] The time-aligned dataset has precise spatiotemporal correlation characteristics. Each sensor reading not only contains a calibrated timestamp but is also tied to the vehicle's specific geographic location at that moment. For example, when a batch of beef experiences abnormal temperature fluctuations on a certain section of road during transportation, the system can accurately trace the longitude, latitude, and altitude information of the vehicle during that period, providing a reliable basis for subsequent analysis of whether the anomaly is caused by terrain changes or road bumps. This data processing method effectively avoids the problem of decreased data credibility caused by device clock asynchrony in traditional methods, and establishes a high-quality data foundation for subsequent risk warnings.

[0063] The dynamic threshold setting logic includes: dynamically adjusting the temperature threshold according to the remaining shelf life of the food, specifically calculated using the following relationship:

[0064] ;

[0065] Where, The upper limit of the standard storage temperature for food. is the total shelf life in days, The remaining days, It is a preset attenuation factor based on the food type; when the remaining shelf life days are reduced to a preset ratio, the nonlinear threshold tightening mechanism is triggered.

[0066] The lightweight alliance chain adopts a shard storage structure, in which abnormal data segments are stored in the shard where the receiver is located, and GPS track segments are stored in the shard where the logistics party is located. Data association query is realized through cross-shard hash index.

[0067] In the cold chain transportation abnormal data attestation scenario, the lightweight consortium chain adopts a shard storage architecture to achieve efficient data management and associated traceability. When the transporter and the recipient complete the dual-signature attestation, shard routing is automatically executed according to the data attributes. That is, abnormal temperature, humidity and other sensor data are distributed to the shard node group of the receiving enterprise, while the corresponding GPS track segment is stored in the shard node group of the logistics carrier.

[0068] Each shard uses a streamlined blockchain structure, maintaining only data blocks related to the business in that shard. The storage methods of the lightweight consortium chain include:

[0069] First, in the data compression stage, the abnormal data segment (such as the continuous temperature exceeding the standard record in the third hour of a certain transportation) and its corresponding GPS trajectory fragment (the longitude and latitude coordinate sequence every five minutes in the same time period) are packaged separately to form a data unit with an independent hash identifier. At this time, the data unit includes the abnormal data packet and the GPS packet. The abnormal data packet embeds the recipient's enterprise code, and the GPS packet carries the digital identity of the logistics party as the basis for fragmented routing.

[0070] Secondly, during the cross-shard index construction phase, a bidirectional hash pointer is generated for each pair of associated data. When an abnormal data segment is written to the receiving shard, the hash value of the abnormal data segment is automatically embedded in the metadata field of the corresponding GPS track segment. At the same time, the hash value of the GPS block is backfilled into the header of the abnormal data segment. This cross-anchoring mechanism can form a chain index structure. For example, in an investigation into a fresh food spoilage incident, auditors can jump directly to the logistics party's shard through the "track hash" field of the abnormal data segment to retrieve the vehicle's travel route details for that day.

[0071] Finally, in the data query and verification stage, the alliance chain members use the inter-shard consensus protocol to realize cross-regional data retrieval; when the recipient needs to trace the full chain of evidence of a warning event, after entering the traceability code, the abnormal data segment is first extracted from the local shard, and then a collaborative query request is initiated to the logistics party's shard based on its index field; for example, the abnormal vibration intensity record of a batch of frozen beef can be quickly linked to the bumpy trajectory of the vehicle passing through the mountain road during that period through the inter-shard hash index, thereby assisting in determining the responsibility for cargo damage; this design not only ensures the logistics party's independent control of the transportation trajectory data, but also maintains data relevance through cryptographic indexes, which can significantly reduce the redundancy of the entire chain data while improving traceability efficiency.

[0072] The method for generating a three-dimensional risk heat map includes: dividing the transportation route into geographical grid units, counting the cumulative value of abnormal indicators in each unit, and superimposing the grid units to generate a three-dimensional risk heat map.

[0073] In the cold chain transportation risk visualization link, the generation of a three-dimensional risk heat map relies on the comparison results of the spatiotemporally correlated transportation environment data set and historical data. Specifically, the complete transportation route is mapped to the geographic coordinate system, and a dynamic grid division algorithm is used to decompose the transportation trajectory coverage area into continuous geographic grid units; the geographic range of each unit is automatically adjusted according to the transportation speed, larger grids are used on highway sections to improve processing efficiency, and the grid granularity is reduced on complex urban sections to capture local risks.

[0074] The implementation process includes the first phase, the second phase and the final phase.

[0075] In the first stage, data spatial mapping is performed, and the vehicle GPS trajectory points are bound in chronological order with the sensor anomaly indicators of the corresponding time period (temperature fluctuation rate, anomaly duration ratio and cumulative vibration intensity value), forming a set of abnormal event points with geographic coordinates; for example, in a certain seafood transportation, the system accurately mapped three temperature-exceeding events detected between 14:25 and 14:35 to the coordinate points of the coastal highway bends that the vehicle passed during this period.

[0076] The second stage performs grid risk aggregation and conducts multi-dimensional statistics on abnormal event points in each geographic grid unit; the temperature fluctuation rate takes the fluctuation range value of all abnormal points in the geographic grid unit, the abnormal duration ratio is the ratio of the total abnormal duration in the calculation unit to the transportation time, and the cumulative vibration intensity value is the weighted sum of all vibration abnormality readings in the unit, so that grid cells close to mountainous areas may show high vibration cumulative values, while grids passing through areas with large day and night temperature differences are prone to temperature fluctuation rate peaks.

[0077] In the final stage, the three-dimensional risk indicators of each grid unit are converted into a visualization layer through a three-dimensional spatial rendering engine; dual encoding of gradient coloring and three-dimensional height mapping is adopted, that is, the plane dimension uses a color scale from light green to dark red to represent the risk level, and the vertical dimension uses columnar height to reflect the combined strength of different indicators; for example, when a grid is simultaneously colored red and has towering columns, it indicates that there are multi-dimensional cross-risks in the area, which may correspond to the synergistic effect of intensified vibration of cold chain equipment and temperature fluctuations caused by frequent braking on mountain roads.

[0078] The above implementation process can directly assist managers in identifying high-risk sections, optimizing transportation routes or strengthening cold chain equipment protection measures in specific sections.

[0079] The method for generating graded warning signals includes: when the temperature fluctuation rate and the cumulative value of vibration intensity exceed the standard at the same time, it is marked as a first-level warning and it is recommended to suspend reception; when only a single dimension exceeds the standard, it is marked as a second-level warning and it is recommended to conduct partial sampling.

[0080] In the cold chain transportation risk warning link, the generation of graded warning signals is based on the multi-dimensional cross-analysis of the spatiotemporal correlation transportation environment data set. In specific implementation, the pre-processed sensor data stream is dynamically associated with the GPS trajectory, and the two core risk indicators of temperature fluctuation rate and cumulative vibration intensity during transportation are calculated in real time; the temperature fluctuation rate is calculated by the ratio of the temperature standard deviation to the average value in the statistical window, reflecting the temperature control stability of the cold chain equipment; the cumulative vibration intensity uses the time domain integration algorithm to perform weighted processing on the three-axis acceleration sensor data to characterize the cumulative impact of mechanical vibration on the goods during transportation.

[0081] When the transportation monitoring system detects an abnormal signal, it triggers a hierarchical warning determination process, which includes:

[0082] Compare the current temperature fluctuation rate and the cumulative value of vibration intensity to see if they exceed the preset fluctuation threshold. The preset fluctuation threshold is manually set according to the risk data quantile of historical transportation tasks. If both the temperature fluctuation rate and the cumulative value of vibration intensity exceed the standard at the same time (exceed the preset fluctuation threshold), it will be automatically marked as a level one warning and a warning signal will be emitted. At this time, the warning signal will be associated with the real-time position of the transport vehicle, the duration of the exceeding standard and the environmental parameter curve, and similar historical cases will be matched through the risk knowledge base to generate a disposal plan similar to "it is recommended to suspend reception and start emergency refrigeration". For example, during a vaccine transport, when the vehicle entered a mountainous section with continuous bends, the frequent speed changes caused the vibration of the cold chain equipment to intensify and the refrigeration efficiency to decrease. After identifying that the two indicators exceeded the standard simultaneously, a level one warning was immediately pushed to the recipient.

[0083] When only a single dimension indicator exceeds the preset fluctuation threshold, the system triggers the secondary warning mechanism. At this time, the warning signal is supplemented with the specific dimension and confidence assessment of the exceeded indicator. For example, in the transportation of a batch of refrigerated fruits, although the vibration intensity remains within the normal range, the temperature fluctuation rate in the refrigerated compartment continues to be high due to the frequent opening and closing of the door. The disposal suggestion of "30% sampling inspection of the vulnerable area in the front of the cargo box" will be generated; this suggestion is combined with the cargo category feature library data to give priority to targeted inspection of the location of temperature-sensitive goods.

[0084] After the warning signal is generated, the warning record is updated synchronously through the alliance chain shard storage architecture, and the warning level and disposal suggestion summary are written in the receiving party's shard. At the same time, the corresponding environmental data proof fragment is stored in the logistics party's shard. The above-mentioned two-way evidence storage mechanism can ensure that during subsequent quality traceability, the complete warning decision-making basis chain can be quickly retrieved through cross-shard indexes. For example, in a cargo damage dispute, the recipient can verify the actual vibration spectrum and temperature control log of the cold chain equipment when the first-level warning is triggered, and accurately divide the responsibility for cargo damage.

[0085] Redundant sensor cross-validation methods include:

[0086] If the initial sensor data exceeds the dynamic threshold, the data of at least two backup sensors separated by a preset distance in the same compartment are retrieved;

[0087] Calculate the standard deviation between the sensor data. If the standard deviation exceeds the judgment threshold associated with the preset tolerance, it is judged as equipment failure data and discarded.

[0088] Otherwise, the mean value of each sensor data is taken as the valid abnormal data segment.

[0089] In the cold chain transportation data verification link, the redundant sensor cross-validation method realizes abnormal data identification through a spatially distributed sensor array. In the specific implementation, the main and backup sensor networking mode is adopted, and three groups of sensors of the same type are deployed at each monitoring point. Among them, the main sensor is responsible for normal monitoring tasks, and the two groups of backup sensors are installed at the positions with preset distances from the main sensor in the diagonal direction of the carriage, forming a triangular verification structure; the setting of the preset distance follows the thermodynamic distribution characteristics of the carriage, ensuring that each sensor can not only capture local environmental changes, but also cover adjacent areas to form data cross-validation capabilities.

[0090] When the primary sensor detects that the temperature or vibration index exceeds the dynamic threshold, the system automatically triggers the redundancy verification mechanism, which includes:

[0091] Retrieve the monitoring data stream of the backup sensor group at the corresponding moment and perform spatiotemporal alignment with the main sensor data. The alignment process includes time axis alignment and space axis alignment:

[0092] The time axis eliminates millisecond-level sampling delays between devices through timestamp matching; the spatial axis compensates for the gradient differences in environmental parameters caused by vehicle movement based on the coordinates of the sensor installation position; for example, during a vaccine transportation, the main sensor detected a sudden temperature rise of 2.8°C at 10:15:23, and immediately located the temperature readings of the three spatial nodes in the car at that moment for joint analysis.

[0093] The main sensor data is spatiotemporally aligned and multi-source data consistency check is performed. The consistency check includes calculating the standard deviation matrix of the corresponding indicators of the three groups of sensors. The standard deviation matrix includes the discrete degree assessment of three dimensions: instantaneous fluctuation, continuous abnormality duration and spatial gradient change. If any dimension in the standard deviation matrix exceeds the preset tolerance (the tolerance is set according to the characteristics of the transport category, such as fresh goods are allowed to have a fluctuation tolerance 30% higher than that of precision instruments), the equipment fault diagnosis process is triggered. The equipment fault diagnosis process distinguishes between sensor failure and environmental anomalies by comparing historical data patterns, equipment operation logs and environmental parameter correlations for the same period. For example, when the cold chain vehicle travels to an area with strong electromagnetic interference, the temperature readings of the three sensors show abnormal fluctuations of +5°C, -3°C and +12°C respectively. The system identifies the data discrete anomaly through standard deviation matrix analysis, and combines the GPS positioning information to determine that it is instantaneous fault data caused by external interference and discards it.

[0094] When the standard deviation matrix is ​​within the preset tolerance range, the weighted mean calculation algorithm is activated; the weighted mean calculation algorithm dynamically assigns weights based on the sensor's historical calibration records. For example, a sensor with a calibration deviation of less than 1% in the past three months is given a weight of 60%, and a newly replaced device is reduced to a weight of 30%.

[0095] For example, if the calibration status of three sensors in a refrigerated compartment is Class A (0.5%), Class B (1.2%), and Class A (0.7%), the composite values ​​of the abnormal data segments are fused with a weighting ratio of 45%-25%-30%, ultimately generating a reliable, spatially corrected anomaly record. This process also generates a data credibility assessment report, including metadata such as the maximum inter-sensor deviation, spatial compensation coefficient, and weighting basis.

[0096] Abnormal data segments that pass the redundant verification mechanism are doubly documented within the blockchain's sharded storage architecture. This dual documentation involves storing the original sensor data packet and verification process logs on the carrier's shard, and writing the weighted average results and credibility report on the consignee's shard. For example, in a cold chain dispute, the consignee retrieved the abnormal segment verification records stored on the blockchain and was able to trace back to the original vibration waveforms and spatial compensation algorithm parameters of the three sensors, effectively verifying the carrier's claim that "road bumps caused equipment false alarms." This redundant verification mechanism ensures data integrity and traceability while mitigating the commercial risks associated with raw data leaks.

[0097] Methods for obtaining historical normal transportation data on the same route include:

[0098] Initiate data requests to other logistics companies through the federated learning protocol, and each company locally calculates the mean and variance of temperature and vibration indicators of historical transportation data;

[0099] Add differential privacy noise to the mean and variance and upload them to the aggregation node;

[0100] The aggregation node uses a dynamic weight allocation mechanism to generate a global historical baseline value, and the weight is determined based on the amount of data and data credibility of each enterprise.

[0101] In the construction of cold chain transportation benchmark data, the acquisition of historical normal transportation data on the same route adopts a distributed collaborative learning framework, and sends data query requests to cooperative logistics companies (participants) participating in federated learning through encrypted communication channels. The data query requests include the spatial topological characteristics of the target transportation route (such as the altitude change nodes and typical climate sections passed through) and the cargo category code. The participants perform feature matching in the local database and screen out historical transportation records with the same refrigeration level and similar vehicle models as the current transportation task.

[0102] Each participant independently performs data preprocessing, which includes spatiotemporal slicing of sensor data for the selected transport task, dividing the continuous monitoring data stream into data segments according to preset durations, and eliminating non-transportation status data from the loading and unloading stages. For each data segment, the sliding window mean of the temperature fluctuation rate and the cumulative energy value of the vibration intensity are calculated to form a feature vector set containing spatiotemporal coordinates. During this process, abnormal data segments generated during equipment failures are automatically filtered out, and only valid data confirmed by cross-validation of redundant sensors is retained.

[0103] After data preprocessing is completed, all participants activate the privacy protection mechanism, which includes applying random perturbation noise to core indicators such as temperature mean and vibration variance; the random perturbation noise generation algorithm adopts a differential privacy model to balance data availability and privacy protection requirements by adaptively adjusting the noise intensity; for example, when the sample size of a transportation task provided by a company is small, the noise injection intensity is automatically increased to prevent the company's operating details from being inferred through a small amount of data. The characteristic data after applying random perturbation noise is transmitted to the central aggregation node through a homomorphic encryption channel to ensure that the transmission process cannot be decrypted by a third party.

[0104] After the central aggregation node receives the characteristic data of each participant, it starts the dynamic weight allocation calculation engine. The dynamic weight allocation calculation engine determines to follow the dual evaluation standards. The data volume dimension allocates the basic weight in proportion to the number of valid samples provided by each enterprise; the data quality dimension is dynamically adjusted through the credibility assessment model, and the credibility assessment model analyzes indicators such as the integrity of historical data submission, sensor calibration records and the proportion of abnormal data; for example, although a company provides a large amount of data, its annual sensor calibration loss rate reaches the preset threshold, its data weight will be downgraded; after the weight calculation result is confirmed by a multi-party security agreement, a global historical baseline value that integrates the characteristics of each enterprise is generated.

[0105] Global historical baseline values ​​serve as a benchmark for transport monitoring. When a new transport mission is initiated, real-time sensor data collected is dynamically compared with the baseline values ​​for the corresponding route. For example, in a cross-border vaccine transport mission, the temperature maintenance levels of the Nordic route in the same season in the historical baseline were used as a reference to automatically identify abnormal sensor readings caused by polar magnetic field interference during the current transport. During this process, the federated learning mechanism continuously updates the data contribution records of participants and automatically calculates data usage fees through smart contracts, incentivizing companies to maintain high-quality data sources.

[0106] During the subsequent data usage phase, consignees can request to view the traceability information generated by the baseline value. The system returns a desensitized, aggregated feature report, including metadata such as the number of participating companies, the level of data perturbation intensity, and a summary of weight distribution. This not only meets quality traceability requirements but also protects the commercial privacy of all participants. For example, if a fresh produce supplier complains about excessive transport temperatures, the carrier can demonstrate that the current transport environment is within a reasonable range by displaying the temperature fluctuation characteristics of similar goods in the baseline value for that route.

[0107] Example 2:

[0108] like Figure 2 As shown, this embodiment further improves the design based on Example 1. The difference is that in actual operation, it is found that the static historical baseline value cannot adapt to dynamic environmental changes, resulting in an increased false alarm rate. Based on this, the blockchain traceability catering supply chain risk warning method also includes:

[0109] A real-time environmental parameter feedback system is introduced during the generation of a three-dimensional risk heat map. The introduction method of the real-time environmental parameter feedback system includes:

[0110] Through the communication module on the logistics vehicle, the vehicle can access the meteorological platform in real time to obtain the rainfall intensity, wind speed and UV index of the current transportation route;

[0111] Call the real-time traffic data interface of the traffic management department to extract the vibration interference factor;

[0112] Environmental parameters are mapped into compensation coefficients. Environmental parameters include rainfall intensity, wind speed, UV index and vibration interference factor of the current transportation route; and the comparison threshold of the historical baseline value is dynamically adjusted.

[0113] At the same time, the real-time environmental parameter feedback system can form a closed loop with the generation of the three-dimensional risk heat map of Example 1. The method for generating the three-dimensional risk heat map after combining the real-time environmental parameter feedback system includes:

[0114] The historical normal transportation data on the same route (including temperature and vibration statistical characteristics) obtained from the federated learning platform is used as the initial baseline;

[0115] Superimpose real-time environmental compensation coefficients to generate a dynamic three-dimensional comparison space;

[0116] The temperature fluctuation rate of the abnormal data segment is converted into an equivalent fluctuation value according to the rainfall and then compared with the baseline;

[0117] The cumulative value of vibration intensity is combined with road maintenance grade data to implement segmented weighted calculation.

[0118] Exemplary:

[0119] During an inter-provincial fruit and vegetable transportation mission, the real-time environmental parameter feedback system detected a sudden downpour on the current road section. The meteorological compensation module automatically increased the temperature fluctuation baseline by 15%. At the same time, combined with the real-time information on flooded sections, the warning threshold of the cumulative vibration intensity was increased by 20%. When the on-board sensor detected a vibration spike caused by road bumps, the comparison model after dynamic compensation could accurately distinguish between normal road vibration and abnormal equipment vibration.

[0120] Without changing the original blockchain evidence storage architecture, the real-time environmental parameter feedback system improves the accuracy of cold chain transportation anomaly identification through dynamic coupling of environmental parameters and historical baselines, while reducing the amount of invalid evidence data caused by sudden weather changes, forming an intelligent early warning system that adapts to environmental changes.

[0121] Example 3:

[0122] like Figure 3 As shown, based on the same inventive concept as the blockchain traceable catering supply chain risk warning method in the aforementioned embodiment, this application provides a blockchain traceable catering supply chain risk warning system. The system and method embodiments in the embodiments of this application are based on the same inventive concept. The system includes:

[0123] The data set collection module deploys multi-source sensors in the cold chain transportation process to collect food temperature, humidity, and vibration intensity data in real time, and binds it to the GPS trajectory of the transportation vehicle in real time to form a transportation environment data set with temporal and spatial correlation;

[0124] The anomaly determination module sets a dynamic threshold based on the type of food being transported. When the sensor data in the transportation environment dataset exceeds the dynamic threshold N times in a row, where N is a preset value, redundant sensor cross-validation is initiated. If the deviation between the validation data and the transportation environment dataset exceeds the preset tolerance, the transportation environment dataset is marked as an abnormal data segment and blockchain evidence is triggered.

[0125] The data compression module compresses the abnormal data segment and its associated GPS trajectory segment into a data block, writes it into the lightweight alliance chain after being double-signed by the transporter and the recipient node, and generates a traceability code including a timestamp;

[0126] The risk visualization module extracts the spatiotemporal characteristics of data blocks, compares them with historical normal transportation data on the same route, calculates the temperature fluctuation rate, the proportion of abnormal duration, and the cumulative value of vibration intensity, and generates a three-dimensional risk heat map;

[0127] In the risk processing module, if at least two dimensions in the three-dimensional risk heat map exceed the corresponding baseline value, the smart contract call is activated to push a graded warning signal containing the start and end time of the abnormal data segment, the related responsible parties and the recommended re-inspection measures to the recipient.

[0128] Obviously, those skilled in the art may make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if such changes and modifications fall within the scope of the claims and their equivalents, the present invention is intended to include such changes and modifications.

[0129] The above is only a preferred specific implementation method of the embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes based on the technical solution and concept of the present application within the technical scope disclosed in the present application, and they should be covered by the scope of protection of the present application.

Claims

1. A blockchain-based catering supply chain risk warning method, characterized by: Methods include: Deploy multi-source sensors in cold chain transportation to collect food temperature, humidity, and vibration intensity data in real time. These data are then linked to the GPS trajectory of transportation vehicles to form a spatiotemporally correlated transportation environment dataset. A dynamic threshold is set based on the type of food being transported. When the sensor data in the transportation environment dataset exceeds the dynamic threshold N times in a row, where N is a preset value, redundant sensor cross-validation is initiated. If the deviation between the validation data and the transportation environment dataset exceeds the preset tolerance, the transportation environment dataset is marked as an abnormal data segment and blockchain evidence is triggered. The abnormal data segment and its associated GPS trajectory segment are compressed into a data block, written into the lightweight alliance chain after being double-signed by the transporter and the recipient node, and a traceability code including a timestamp is generated; Extract the spatiotemporal characteristics of the data block, compare it with historical normal transportation data on the same route, calculate the temperature fluctuation rate, the proportion of abnormal duration, and the cumulative value of vibration intensity, and generate a three-dimensional risk heat map; If at least two dimensions in the three-dimensional risk heat map exceed the corresponding baseline values, the smart contract call will be activated to push a graded warning signal to the recipient, including the start and end time of the abnormal data segment, the associated responsible parties, and recommended re-inspection measures.

2. The blockchain traceability catering supply chain risk early warning method according to claim 1 is characterized in that: When the spatiotemporally correlated transportation environment dataset is generated, a sliding time window algorithm is used to perform time series alignment between the sensor data and the GPS trajectory.

3. The blockchain traceability catering supply chain risk early warning method according to claim 1 is characterized in that: The setting logic of the dynamic threshold includes: The temperature threshold is dynamically adjusted based on the remaining shelf life of the food, and is calculated using the following formula: ； Where, The upper limit of the standard storage temperature for food. is the total shelf life in days, The remaining days, It is a preset attenuation factor based on the food type; when the remaining shelf life days are reduced to a preset ratio, the nonlinear threshold tightening mechanism is triggered.

4. The blockchain traceability catering supply chain risk early warning method according to claim 1 is characterized in that: The lightweight alliance chain adopts a shard storage structure, with abnormal data segments stored in the receiver's shard and GPS track segments stored in the logistics party's shard. Data association query is achieved through cross-shard hash indexing. The storage method of the lightweight alliance chain includes: First, in the data compression stage, the abnormal data segment and its corresponding GPS trajectory segment are packaged separately to form a data unit with an independent hash identifier. Secondly, during the cross-shard index construction phase, a bidirectional hash pointer is generated for each pair of associated data. When an abnormal data segment is written to the receiving shard, the hash value of the abnormal data segment is automatically embedded in the metadata field of the corresponding GPS track segment. At the same time, the hash value of the GPS block is backfilled into the header of the abnormal data segment. Finally, in the data query and verification phase, alliance chain members use the inter-shard consensus protocol to achieve cross-region data retrieval.

5. The blockchain traceability catering supply chain risk early warning method according to claim 1 is characterized in that: The method for generating the three-dimensional risk heat map includes: dividing the transportation route into geographical grid units, counting the cumulative value of abnormal indicators in each unit, and superimposing the grid units to generate the three-dimensional risk heat map.

6. The blockchain traceability catering supply chain risk early warning method according to claim 1 is characterized in that: The method for generating the graded warning signal includes: when the temperature fluctuation rate and the cumulative value of vibration intensity exceed the standard at the same time, it is marked as a first-level warning and it is recommended to suspend reception; when only a single dimension exceeds the standard, it is marked as a second-level warning and it is recommended to conduct partial sampling; after the warning signal is generated, the warning record is synchronously updated through the alliance chain shard storage architecture, and the warning level and disposal suggestion summary are written in the receiving party shard, and the corresponding environmental data proof fragment is stored in the logistics party shard.

7. The blockchain traceability catering supply chain risk early warning method according to claim 1 is characterized in that: The redundant sensor cross-validation method comprises: If the initial sensor data exceeds the threshold, the data of at least two backup sensors separated by a preset distance in the same compartment are retrieved; Calculate the standard deviation between the sensor data. If the standard deviation exceeds the judgment threshold associated with the preset tolerance, it is judged as equipment failure data and discarded. Otherwise, the mean value of each sensor data is taken as the valid abnormal data segment.

8. The blockchain traceability catering supply chain risk early warning method according to claim 1 is characterized in that: Methods for obtaining historical normal transportation data on the same route include: Initiate data requests to other logistics companies through the federated learning protocol, and each company locally calculates the mean and variance of temperature and vibration indicators of historical transportation data; Add differential privacy noise to the mean and variance and upload them to the aggregation node; The aggregation node uses a dynamic weight allocation mechanism to generate a global historical baseline value, and the weight is determined based on the amount of data and data credibility of each enterprise.

9. The blockchain traceability catering supply chain risk early warning method according to claim 1 is characterized in that: Also includes: A real-time environmental parameter feedback system is introduced during the generation of a three-dimensional risk heat map. The introduction method of the real-time environmental parameter feedback system includes: Through the communication module on the logistics vehicle, the vehicle can access the meteorological platform in real time to obtain the rainfall intensity, wind speed and UV index of the current transportation route; Call the real-time traffic data interface of the traffic management department to extract the vibration interference factor; Environmental parameters are mapped into compensation coefficients. Environmental parameters include rainfall intensity, wind speed, UV index and vibration interference factor of the current transportation route; and the comparison threshold of the historical baseline value is dynamically adjusted.

10. The blockchain traceability catering supply chain risk early warning system is characterized by: The system includes: The data set collection module deploys multi-source sensors in the cold chain transportation process to collect food temperature, humidity, and vibration intensity data in real time, and binds it to the GPS trajectory of the transportation vehicle in real time to form a transportation environment data set with temporal and spatial correlation; The anomaly determination module sets a dynamic threshold based on the type of food being transported. When the sensor data in the transportation environment dataset exceeds the dynamic threshold N times in a row, where N is a preset value, redundant sensor cross-validation is initiated. If the deviation between the validation data and the transportation environment dataset exceeds the preset tolerance, the transportation environment dataset is marked as an abnormal data segment and blockchain evidence is triggered. The data compression module compresses the abnormal data segment and its associated GPS trajectory segment into a data block, writes it into the lightweight alliance chain after being double-signed by the transporter and the recipient node, and generates a traceability code including a timestamp; The risk visualization module extracts the spatiotemporal characteristics of data blocks, compares them with historical normal transportation data on the same route, calculates the temperature fluctuation rate, the proportion of abnormal duration, and the cumulative value of vibration intensity, and generates a three-dimensional risk heat map; In the risk processing module, if at least two dimensions in the three-dimensional risk heat map exceed the corresponding baseline value, the smart contract call is activated to push a graded warning signal containing the start and end time of the abnormal data segment, the related responsible parties and the recommended re-inspection measures to the recipient.

Images