A cloud-edge collaborative management method and system for food supply chain transparency

Abstract

The application relates to the technical field of food supply chain management, and discloses a cloud-edge collaborative management method and system for food supply chain transparency, wherein the method comprises the following steps: acquiring time sequence data of a temperature and humidity sensor, running state data of refrigeration equipment and door opening and closing event log data of a cold chain transportation edge node; performing sliding window extreme value detection to generate an extreme value event sequence; associating the equipment state data to generate an abnormal event tuple with causal association annotation; coding to generate a collaborative synchronization pre-request data packet carrying a downstream associated node identifier and a predicted arrival time window; after network recovery, uploading to the cloud end in priority; the cloud end issues an advance collaborative collection instruction to a downstream transit warehouse edge node; performing cross-link consistency checking to generate a cross-link deviation index; generating a cross-link breakage causal link reconstruction report and writing into a supply chain transparency state view.

Description

Technical Field

[0001] This invention relates to the field of food supply chain management technology, specifically to a cloud-edge collaborative management method and system for making the food supply chain transparent. Background Technology

[0002] Cold chain logistics involves multiple stages, including transportation, transshipment, and warehousing, as food flows from production to retail. Edge nodes are deployed at each stage, and data collected is aggregated to the cloud through a cloud-edge collaborative architecture to achieve end-to-end status monitoring and quality traceability.

[0003] In the existing solution, each edge node independently collects data at fixed intervals and uploads it to the cloud when the network is available. The cloud relies on the uploaded data from each link to maintain a transparent view of the supply chain.

[0004] The above solution has the following shortcomings: When a cold chain transport vehicle passes through a signal blind spot, if a chain break occurs, although the edge node can detect the anomaly locally, it cannot report it to the cloud in real time due to network unavailability. The cloud cannot detect the anomaly and therefore will not issue collaborative collection instructions to the downstream transit warehouse edge nodes. After the network is restored, although the cloud receives the chain break anomaly record from the transport end, the transit warehouse only has regular low-frequency sparse data at the same time, resulting in a serious misalignment of data in terms of time granularity between the two links. The cloud cannot verify whether the impact of the chain break heat exposure extends to the downstream links, creating a traceability blind spot in the cross-link causal chain in the supply chain transparency status view. Summary of the Invention

[0005] This invention provides a cloud-edge collaborative management method and system for making the food supply chain transparent, solving the technical problems in related technologies such as the inability to achieve cross-link data linkage verification of abnormal temperature and humidity events during cold chain transportation with downstream transit links, and the difficulty in tracing the cause and effect of chain breakage.

[0006] This invention discloses a cloud-edge collaborative management method for making the food supply chain transparent, comprising: Acquire the time-series data of temperature and humidity sensors and the operating status data of refrigeration equipment within the current acquisition window of the cold chain transportation edge node; Perform sliding window extreme value detection on the time-series data of the temperature and humidity sensor, mark extreme value points that deviate from the safety threshold as extreme value events, and generate an extreme value event sequence; For each extreme event in the extreme event sequence, associate it with the device state change records within the same time window to generate an abnormal event tuple with causal association annotation; The abnormal event tuples are encoded into event snapshot data blocks, and the downstream related node identifiers and expected arrival time windows are extracted from the locally stored supply chain topology association information to generate collaborative synchronization pre-request data packets. When network connectivity is restored, the event snapshot data block and the collaborative synchronization pre-request data packet are uploaded to the cloud with a forced priority. The cloud parses the collaborative synchronization pre-request data packet and sends an advance collaborative collection instruction to the downstream transit warehouse edge node before the start boundary of the expected arrival time window; The downstream transit warehouse edge nodes collect data in a high-frequency acquisition mode within a specified time window and upload it to the cloud. Based on temperature transmission attenuation constraint rules and weight conservation tolerance rules, the cloud performs cross-link consistency verification on the event snapshot data blocks at the transportation end and the high-frequency collaborative collection data at the transit warehouse end, generating a cross-link deviation index. Based on the cross-link deviation index and the causal association annotation, a cross-link breakage causal link reconstruction report is generated and written into the supply chain transparency status view.

[0007] Furthermore, the sliding window extreme value detection includes: extracting local maximum and local minimum values ​​and their corresponding timestamps within each sliding window; calculating the magnitude and duration of each extreme value deviating from the safety threshold; marking extreme value points whose deviation exceeds a preset magnitude threshold and whose duration exceeds a preset duration threshold as extreme value events; wherein, the step length of the sliding window is less than the window width, so that there is an overlapping area between adjacent windows.

[0008] Furthermore, the generation method of the causal association label includes: calculating the time offset between the timestamp of the extreme event occurrence and the timestamp of the most recent device state change; when the time offset is positive and the absolute value is less than a preset causal time window, it is labeled as device-triggered; when no device event can be matched within the preset causal time window, it is labeled as external environment type; the abnormal event tuple includes the extreme value amplitude, duration, associated device event type and the time offset.

[0009] Furthermore, it also includes performing event aggregation processing on the extreme event sequence: when the time interval between adjacent extreme events is less than a preset aggregation interval, the adjacent extreme events are merged into an aggregated extreme event; the extreme value amplitude of the aggregated extreme event is the maximum amplitude value among the merged events, and the duration is the total time span from the start of the first merged event to the end of the last merged event.

[0010] Furthermore, the estimated arrival time window in the collaborative synchronization pre-request data packet is calculated based on the current time and the remaining estimated transportation time from the current location to the termination node, and includes the start and end boundaries of the estimated arrival time; the supply chain topology association information is static configuration data that is issued by the cloud and cached locally on the edge node before the transportation task is started.

[0011] Furthermore, the advance collaborative acquisition instruction includes the parameter types that need to be acquired at high frequency and the acquisition duration; the cloud determines the collaborative acquisition level based on the extreme value amplitude and duration in the event snapshot data block: when the extreme value amplitude exceeds the first level threshold and the duration exceeds the first duration threshold, the collaborative acquisition level is high level, and the acquisition frequency is a preset high multiple of the base frequency; when the extreme value amplitude or duration is lower than the first level threshold, the collaborative acquisition level is standard level, and the acquisition frequency is a preset standard multiple of the base frequency.

[0012] Furthermore, the generation method of the cross-stage deviation index includes: calculating the expected arrival temperature based on the temperature transmission attenuation constraint rule. ,in Set the temperature of the carriage after the cooling system has been restored. This is the peak temperature of chain breakage. The temperature decay coefficient is The time interval from the moment the chain is restored to the moment the goods arrive at the transit warehouse; calculate the cross-chain deviation index: ,in This refers to the actual receiving temperature at the transit warehouse. This represents the maximum allowable threshold for temperature deviation. The weight at the point of origin of the transport. The weight received by the transit warehouse. To preset the weight tolerance value, and The weighting coefficients are satisfied. .

[0013] Furthermore, the cross-link causal link reconstruction report includes the start time and location of the link break event, the event type of the associated equipment, the temperature transmission path across the handover point, and the residual risk level; the residual risk level is determined based on the value range of the cross-link deviation index: when the cross-link deviation index is below the first risk threshold, it is low risk; when it is between the first and second risk thresholds, it is medium risk; and when it is above the second risk threshold, it is high risk.

[0014] Furthermore, the event snapshot data block and the cooperative synchronization pre-request data packet occupy an independent reserved storage area in the local transmission queue. The reserved storage area has a fixed capacity and is not overwritten by regular data. When the reserved storage area is close to full, the edge node performs downsampling compression on the newly generated regular acquisition data and writes it into the ordinary storage area.

[0015] This invention discloses a cloud-edge collaborative management system for making the food supply chain transparent, comprising: The edge acquisition module is used to acquire time-series data of temperature and humidity sensors and operating status data of refrigeration equipment within the current acquisition window of the cold chain transportation edge node; The extreme value detection module is used to perform sliding window extreme value detection on the time-series data of the temperature and humidity sensor to generate an extreme value event sequence; The causal association module is used to associate each extreme event in the extreme event sequence with device state change records within the same time window, and generate anomaly event tuples with causal association annotations. The encoding and pre-request generation module is used to encode the abnormal event tuple into an event snapshot data block and generate a collaborative synchronization pre-request data packet carrying the downstream associated node identifier and the expected arrival time window; The priority transmission module is used to upload the event snapshot data block and the collaborative synchronization pre-request data packet to the cloud with a forced priority when the network connection is restored; The collaborative instruction issuing module is used to parse the collaborative synchronization pre-request data packet in the cloud and issue an advance collaborative collection instruction to the downstream transit warehouse edge node before the start boundary of the expected arrival time window; A high-frequency collaborative acquisition module is used for the downstream transit warehouse edge nodes to acquire data in a high-frequency acquisition mode within a specified time window and upload it to the cloud. The cross-link verification module is used to perform consistency verification on the event snapshot data block at the transportation end and the high-frequency collaborative collection data at the transit warehouse end based on the temperature transmission attenuation constraint rule and the weight conservation tolerance rule, and generate a cross-link deviation index. The causal link reconstruction module is used to generate a cross-link broken causal link reconstruction report based on the cross-link deviation index and the causal association label, and write it into the supply chain transparency status view.

[0016] The beneficial effects of this invention are as follows: This invention generates a collaborative synchronization pre-request data packet carrying the identifier of downstream associated nodes and the expected arrival time window during the anomaly detection stage at the edge end. This allows edge nodes to prepare the associated information required for cross-link collaboration in advance when the network is unavailable. After the network is restored, the cloud can issue an advance collaborative collection instruction to the downstream transit warehouse edge nodes before the goods actually arrive. This enables the downstream transit warehouse edge nodes to have high-frequency collaborative collection data with time granularity matching the chain break event at the transportation end within the handover time window. This aligns the data at the transportation end and the transit warehouse end in terms of time resolution, providing a data foundation for cross-link consistency verification.

[0017] Sliding window extreme value detection ensures that instantaneous anomalies caused by chain breakage are not lost due to the smoothing effect of regular data compression logic. Causal association annotation preserves the causal information between extreme events and changes in the operating status of refrigeration equipment. The cross-link chain breakage causal link reconstruction report, generated based on the cross-link deviation index, associates chain breakage events at the transportation end with the receiving status at the transit warehouse end, eliminating cross-link traceability blind spots caused by signal blind spots and low-frequency data collection, and restoring full-link quality traceability capabilities in the transparent supply chain status view. Attached Figure Description

[0018] Figure 1 This is a flowchart of the cloud-edge collaborative management method for making the food supply chain transparent, provided in an embodiment of the present invention. Figure 2 This is a schematic diagram comparing the time-series temperature changes and safety thresholds of the VEH-031 carriage provided in an embodiment of the present invention; Figure 3 This is a schematic diagram illustrating the linkage between compressor speed and temperature in a refrigeration device provided in an embodiment of the present invention; Figure 4 This is a schematic diagram of the distribution of temperature deviation from the threshold amplitude in sliding window extreme value detection provided by an embodiment of the present invention; Figure 5 This is a schematic diagram of the high-frequency collaborative temperature acquisition timing of the transit warehouse WH-07 provided in an embodiment of the present invention; Figure 6 This is a schematic diagram comparing the weight of goods departing from the transportation terminal with the weight received at the transit warehouse, provided in an embodiment of the present invention. Figure 7 This is a schematic diagram of the event timeline provided in the embodiments of the present invention: from compressor shutdown to completion of cross-stage verification; Figure 8 This is a schematic diagram illustrating the correspondence between the collaborative acquisition frequency level and the acquisition interval provided in an embodiment of the present invention; Figure 9 This is a schematic diagram comparing key temperature values ​​in a cross-stage temperature transfer path provided in an embodiment of the present invention. Detailed Implementation

[0019] The subject matter described herein will now be discussed with reference to exemplary embodiments. It should be understood that these embodiments are discussed only to enable those skilled in the art to better understand and implement the subject matter described herein, and changes may be made to the function and arrangement of the elements discussed without departing from the scope of this specification. Various processes or components may be omitted, substituted, or added as needed in the examples. Furthermore, some features described in the examples may be combined in other examples.

[0020] At least one embodiment of the present invention discloses a cloud-edge collaborative management method for food supply chain transparency, see [link to relevant documentation]. Figure 1This includes the following steps: Step 1: Obtain multi-source data from edge nodes in cold chain transportation.

[0021] At the edge node of cold chain transportation, time-series data from temperature and humidity sensors, refrigeration equipment operating status data, and door opening / closing event log data are acquired within the current acquisition window. The temperature and humidity sensor time-series data consists of sequences of temperature and humidity values ​​recorded at fixed sampling intervals. The refrigeration equipment operating status data includes compressor start / stop status, refrigeration power level, and status transition timestamps. The door opening / closing event log data records the timestamps of each door opening and closing.

[0022] Step 2: Perform sliding window extreme value detection on the time series data of the temperature and humidity sensor to generate an extreme value event sequence.

[0023] Perform sliding window extreme value detection on time-series data from temperature and humidity sensors. Extract local maximum and local minimum values ​​and their corresponding timestamps within each sliding window. Calculate the magnitude and duration of each extreme value's deviation from a preset safety threshold. Mark extreme value points whose deviation exceeds a preset magnitude threshold and whose duration exceeds a preset duration threshold as extreme value events, and arrange them in chronological order to generate an extreme value event sequence.

[0024] It should be noted that the sliding window in the above-mentioned sliding window extreme value detection is a time window with a preset window width and a preset step length. The window width covers the typical duration range of a single chain break event, and the step length is smaller than the window width to ensure that there is an overlapping area between adjacent windows, avoiding the extreme value event being missed because it happens to be located at the window boundary. The preset safety threshold is the upper temperature limit and lower temperature limit specified in food safety regulations. The deviation magnitude is the absolute value of the difference between the extreme value and the safety threshold. The duration is the time span during which the deviation in the same direction continuously exceeds the safety threshold.

[0025] Step 3: Associate extreme events with the operating status data of the refrigeration equipment to generate anomaly event tuples with causal association labels.

[0026] For each extreme event in the extreme event sequence, retrieve records of refrigeration equipment operating status changes and door opening / closing events within the same time window. Calculate the time difference between the extreme event occurrence timestamp and the most recent equipment status change timestamp. Generate anomaly event tuples with causal association annotations. Each anomaly event tuple contains the extreme value magnitude, duration, associated equipment event type, and time offset.

[0027] It should be noted that the associated device event type refers to the category of device state changes that are most temporally relevant to the extreme event, specifically including three types: refrigeration equipment shutdown, refrigeration power reduction, and door opening. The time offset is the signed time difference between the start timestamp of the extreme event and the occurrence timestamp of the associated device event. A positive time offset indicates that the device event occurred before the extreme event; a negative time offset indicates that the extreme event occurred before the device event. Causal association labeling is determined based on the sign and absolute value of the time offset: when the time offset is positive and its absolute value is less than the preset causal time window, it is labeled "Device-triggered"; when no device event can be matched within the causal time window, it is labeled "External Environment".

[0028] In this embodiment, to avoid accidental triggering of subsequent collaborative processes due to sporadic noise from a single extreme event, event aggregation processing is also performed on the extreme event sequence. When the time interval between adjacent extreme events is less than a preset aggregation interval, the adjacent extreme events are merged into an aggregated extreme event. The extreme value amplitude of the aggregated extreme event is taken as the maximum amplitude value among the merged events, and the duration is taken as the total time span from the start of the first merged event to the end of the last merged event. In subsequent steps, the same causal association labeling and collaborative pre-request generation are performed on each aggregated extreme event as on a single extreme event.

[0029] Step 4: Encode the exception event tuple and generate the cooperative synchronization pre-request data packet.

[0030] The abnormal event tuples are encoded into event snapshot data blocks, and a mandatory transmission priority is marked. Simultaneously, the downstream associated node identifiers and estimated arrival time windows of the current transportation segment are extracted from the locally stored supply chain topology association information. A collaborative synchronization pre-request data packet is generated, carrying the downstream associated node identifiers and estimated arrival time windows.

[0031] It should be noted that the supply chain topology association information is static configuration data distributed from the cloud and cached locally on the edge nodes before the transportation task begins. This information includes the starting node identifier, ending node identifier, intermediate node identifiers, estimated transportation time between each node, and the type attributes of each node for the current transportation segment. The downstream associated node identifier is the transit warehouse node identifier corresponding to the ending node of the current transportation segment. The estimated arrival time window is calculated based on the current time and the remaining estimated transportation time from the current location to the ending node, and includes the start and end boundaries of the estimated arrival time.

[0032] Event snapshot data blocks are encoded in a structured binary format, including an event identifier field, an exception event tuple sequence field, a collection window time range field, and an edge node identifier field. The forced transmission priority is the highest priority level in the transmission queue, ensuring that event snapshot data blocks are transmitted before regular periodic data collection once the network recovers.

[0033] Step 5: Queue locally and prioritize uploading event snapshot data blocks and collaborative synchronization pre-request packets when the network recovers.

[0034] Event snapshot data blocks and collaborative synchronization pre-request packets are queued in the local transmission queue. Edge nodes continuously probe network connectivity. When network connectivity is restored, event snapshot data blocks and collaborative synchronization pre-request packets are uploaded to the cloud with forced priority.

[0035] In this embodiment, to prevent local storage overflow during prolonged network unavailability, edge nodes implement hierarchical storage management for their local transmission queues. Data blocks with mandatory transmission priority occupy a dedicated reserved storage area with a fixed capacity that is not overwritten by regular data. When the reserved storage area is nearing full capacity, the edge node downsamples and compresses newly generated regular data before writing it into the ordinary storage area, ensuring the integrity of the mandatory priority data.

[0036] Step 6: The cloud parses the collaborative synchronization pre-request data packet and sends the advance collaborative collection instruction to the downstream nodes.

[0037] After receiving the pre-request data packet for collaborative synchronization, the cloud analyzes the downstream related node identifiers and estimated arrival time windows in the packet based on the supply chain topology graph. Before the start boundary of the estimated arrival time window, an advance collaborative data collection instruction is issued to the edge nodes of the downstream transit warehouse. This instruction includes the types of parameters requiring high-frequency data collection and the duration of the collection.

[0038] It should be noted that the parameter types in the advance collaborative data collection command include temperature and humidity data, cargo receiving and weighing data, and cold storage environment data. The data collection duration is the time span between the start and end boundaries of the expected arrival time window, plus a preset buffer period. Before issuing the advance collaborative data collection command, the cloud verifies the current online status of downstream related nodes. If a downstream node is currently unreachable, the advance collaborative data collection command is marked as pending issuance and repeatedly attempted to be issued at a preset retry interval until the downstream node responds or the end boundary of the expected arrival time window is exceeded.

[0039] In this embodiment, to match the parameter type and acquisition frequency of high-frequency acquisition with the severity of the link failure event, the cloud determines the collaborative acquisition level based on the extreme value amplitude and duration in the event snapshot data block. When the extreme value amplitude exceeds the first level threshold of the safety threshold deviation and the duration exceeds the first duration threshold, the collaborative acquisition level is high, and the corresponding acquisition frequency is a preset high multiple of the base frequency; when the extreme value amplitude or duration is lower than the first level threshold, the collaborative acquisition level is standard, and the corresponding acquisition frequency is a preset standard multiple of the base frequency.

[0040] Step 7: The downstream transit warehouse edge nodes perform high-frequency collaborative data collection and upload data.

[0041] After receiving the advance coordinated data collection instruction, the edge node of the downstream transit warehouse switches to high-frequency data collection mode within a specified time window. It collects temperature and humidity data, cargo receiving and weighing data, and cold storage environment data at the collection frequency specified in the advance coordinated data collection instruction. The high-frequency collected data, along with associated trigger identifiers, is uploaded to the cloud. These associated trigger identifiers mark that the high-frequency collected data was triggered by the coordinated data collection instruction, distinguishing it from regular periodic data collection.

[0042] Step 8: Perform cross-process consistency verification in the cloud and generate a cross-process deviation index.

[0043] After receiving event snapshot data blocks from the transportation end and high-frequency collaborative collection data from the transit warehouse end within the handover time window, the cloud performs cross-link consistency verification based on the temperature transmission attenuation constraint rules and weight conservation tolerance rules of the corresponding edges on the supply chain topology diagram.

[0044] The peak temperature of the chain break at the transportation end is compared with the cargo temperature at the time of receipt in the transit warehouse. Based on the temperature transmission attenuation constraint rule, the expected arrival temperature after attenuation from the peak temperature of the chain break over the remaining transportation time is calculated. The difference between the expected arrival temperature and the actual receiving temperature in the transit warehouse is taken as the temperature deviation component. Based on the weight conservation tolerance rule, the difference between the cargo weight at the departure point of the transportation end and the weighing data received in the transit warehouse is compared with a preset tolerance range. The difference exceeding the tolerance range is taken as the weight deviation component. The temperature deviation component and the weight deviation component are combined to generate a cross-stage deviation index.

[0045] It should be noted that the temperature transfer attenuation constraint rule is: the peak temperature after the chain break decreases exponentially over the remaining transport time, approaching the set temperature of the carriage after cooling recovery. The formula for calculating the expected arrival temperature is: ; in, The expected temperature (unit: °C). Set the temperature (unit: °C) for the carriage after cooling recovery. Peak chain breakage temperature (unit: °C). Represents an exponential function. Temperature decay coefficient (unit: ), This is the time interval (in seconds) from the moment the chain is restored to the moment the goods arrive at the transit warehouse. Temperature decay coefficient. The parameters are determined based on the vehicle's thermal insulation performance parameters and the cargo's heat capacity properties, and are stored in the edge attributes of the supply chain topology graph.

[0046] The weight conservation tolerance rule is as follows: the difference between the weight departing from the transport terminal and the weight received at the transit warehouse shall not exceed the preset tolerance value. The preset tolerance value takes into account the normal loss range of goods during transportation due to factors such as moisture evaporation.

[0047] The formula for calculating the cross-stage deviation index is: ; in, This is a cross-stage deviation index. The expected temperature (unit: °C). The actual receiving temperature of the transit warehouse (unit: °C). This represents the maximum allowable threshold for temperature deviation (unit: °C). The weight at the point of origin (unit: kg). Weight received by the transit warehouse (unit: kg). Preset weight tolerance value (unit: kg). and These are the weighting coefficients for the temperature deviation component and the weight deviation component, respectively, and they satisfy... Each component is divided by a reference quantity of the same dimension. and This process involves making each component dimensionless before performing a weighted summation, thereby eliminating the influence of different physical quantities' dimensions on the calculation results.

[0048] Step 9: Generate a cross-linked cause-and-effect chain reconstruction report and write it into the supply chain transparency status view.

[0049] Based on the cross-stage deviation index and causal association annotations in the event snapshot data block, a cross-stage chain break causal link reconstruction report is generated. The cross-stage chain break causal link reconstruction report annotates the start time and location of the chain break event, the event type of the associated equipment, the temperature transfer path across the handover point, and the residual risk level of the downstream stage.

[0050] The residual risk level is determined based on the range of values ​​for the cross-process deviation index. When the cross-process deviation index is below the first risk threshold, the residual risk level is low; when the cross-process deviation index is between the first and second risk thresholds, the residual risk level is medium; and when the cross-process deviation index is above the second risk threshold, the residual risk level is high.

[0051] Write the cross-link breakage causal link reconstruction report into the corresponding link interval in the supply chain transparency status view to complete the reconstruction of the causal link from the breakage event at the transportation end to the receiving status at the transit warehouse end.

[0052] On the other hand, this invention also proposes a cloud-edge collaborative management system for food supply chain transparency, including: The edge acquisition module is used to acquire time-series data of temperature and humidity sensors, operating status data of refrigeration equipment, and door opening and closing event log data within the current acquisition window of the cold chain transportation edge node.

[0053] The extreme value detection module is used to perform sliding window extreme value detection on the time series data of temperature and humidity sensors and generate extreme value event sequences.

[0054] The causal association module is used to associate each extreme event in the extreme event sequence with records of changes in the operating status of refrigeration equipment and records of door opening and closing events within the same time window, generating anomaly event tuples with causal association annotations.

[0055] The encoding and pre-request generation module is used to encode abnormal event tuples into event snapshot data blocks and generate collaborative synchronization pre-request data packets carrying downstream associated node identifiers and expected arrival time windows.

[0056] The priority transmission module is used to upload event snapshot data blocks and collaborative synchronization pre-request packets to the cloud with a forced priority when the network connection is restored.

[0057] The collaborative instruction issuance module is used to parse the downstream associated node identifier and expected arrival time window in the collaborative synchronization pre-request data packet in the cloud, and issue advance collaborative collection instructions to the edge nodes of the downstream transit warehouse.

[0058] The high-frequency collaborative acquisition module is used by the edge nodes of the downstream transit warehouse to collect temperature and humidity data, cargo receiving and weighing data, and cold storage environment data in a high-frequency acquisition mode within a specified time window.

[0059] The cross-stage verification module is used to perform consistency verification on the event snapshot data blocks at the transportation end and the high-frequency collaborative collection data at the transit warehouse end based on the temperature transmission attenuation constraint rule and the weight conservation tolerance rule, and generate a cross-stage deviation index.

[0060] The causal link reconstruction module is used to generate a cross-link broken causal link reconstruction report based on the cross-link deviation index and causal association annotation, and write it into the supply chain transparency status view.

[0061] By way of example, the system proposed in this invention may also include various features and combinations thereof in the method embodiments, which will not be elaborated here.

[0062] This invention generates a pre-request data packet for collaborative synchronization, carrying the identifiers of downstream associated nodes and the expected arrival time window, during the anomaly detection phase at the edge. This allows edge nodes to prepare all the necessary information for cross-link collaboration in advance when the network is unavailable. After the network is restored, the cloud can parse the pre-request data packet for collaborative synchronization and issue an advance collaborative collection instruction to the downstream transit warehouse edge nodes before the goods actually arrive. This ensures that the downstream transit warehouse edge nodes have high-frequency collaborative collection data with time granularity matching the chain break event at the transportation end within the handover time window. As a result, when the cloud performs cross-link consistency verification, the data at the transportation end and the transit warehouse end are aligned in time resolution, enabling the calculation of the residual impact of chain break heat exposure after the handover based on temperature transmission attenuation constraints and weight conservation tolerance rules. Sliding window extreme value detection ensures that the instantaneous anomaly of the chain break is not smoothly lost by the conventional data compression logic, and causal association labeling preserves the causal information between the extreme value event and the changes in the operating status of the refrigeration equipment. The cross-link breakage causal link reconstruction report, generated based on the cross-link deviation index, associates the breakage event at the transportation end with the receiving status at the transit warehouse end, eliminating the cross-link traceability blind spots caused by signal blind spots and low-frequency collection in the supply chain transparency status view, and restoring the full-link quality traceability capability.

[0063] See Figures 2-9 The following application examples are proposed: A food cold chain logistics company is responsible for transporting fresh aquatic products from processing plants at the production site to transit warehouses in multiple cities, and then distributing them to retail outlets. In month B of a certain year, refrigerated transport vehicle VEH-031 performed a transport mission from the processing plant at the production site to transit warehouse WH-07 in the city. The cargo was frozen salmon, with a loading weight of 486 kg, and the set temperature of the vehicle compartment was [temperature missing]. The upper limit of the safety threshold is The transport route passes through a mountain tunnel complex, a section of which is a known signal blind spot. The vehicle will spend approximately 47 minutes traveling within this blind spot. The total estimated transport time for WH-07 from the production and processing plant is 4 hours and 20 minutes, with the estimated arrival time window between 13:40 and 14:10 on the same day.

[0064] The cold chain transportation edge node NODE-031 is deployed on the VEH-031 vehicle control unit, and the transit warehouse edge node NODE-WH07 is deployed on WH-07. The cloud maintains a supply chain topology graph covering this transportation segment, with edge attributes recording the temperature decay coefficient. And the tolerance value for cargo weight.

[0065] During the data acquisition window from 09:50 to 10:40 on the same day, NODE-031 continuously acquired three types of multi-source data. Temperature and humidity sensors collected data at 30-second intervals, recording the sequence of temperature and humidity values. The refrigeration equipment's operating status data recorded a compressor shutdown event at 10:07, with the power level dropping from level 3 to 0. No door opening / closing event logs were recorded during this window. At 10:05, the vehicle entered a tunnel signal blind spot, and NODE-031 switched to local storage mode.

[0066] Table 1 Sample of raw multi-source acquisition data from Step 1 The extreme value detection module performs sliding window extreme value detection on the temperature time series data in Table 1, with a window width of 15 minutes and a step length of 5 minutes. Within the time period from 10:15 to 10:30, the local maximum temperature peak occurs at 10:22:30, with a temperature of [temperature value missing]. Exceeding the upper limit of the safety threshold The deviation range is The threshold-exceeding state lasted from 10:15:00 to 10:28:00, a duration of 13 minutes. Both the deviation magnitude and duration exceeded the preset threshold, and this extreme point was marked as extreme event EV-001.

[0067] Table 2. Results of Sliding Window Extreme Value Detection The causal correlation module retrieves records of refrigeration equipment status changes within the same time window for EV-001. The compressor shutdown event occurred at 10:07:00, EV-001 started at 10:15:00, and the time offset is... (Positive values, device events precede extreme value events). Smaller than the preset causal time window ( The causal relationship is labeled as "device-triggered". There are no door opening / closing event records within the same time window. Since EV-001 is the only extreme event within this window, there is no need to perform event aggregation; the abnormal event tuple AET-001 is generated directly.

[0068] Table 3 Results of Abnormal Event Tuple Generation The encoding and pre-request generation module encodes AET-001 into an event snapshot data block (snapshot number SNAP-031-B0923) and marks it with a mandatory transmission priority. NODE-031 extracts information from locally cached supply chain topology association information: the downstream associated node is identified as NODE-WH07, and based on the current time of 10:23 and the remaining estimated transportation time of 3 hours and 17 minutes, the estimated arrival time window is calculated to be from 13:40 to 14:10. A collaborative synchronization pre-request data packet PRE-REQ-001 is generated, carrying the NODE-WH07 identifier and the estimated arrival time window.

[0069] Table 4. Contents of the Cooperative Synchronization Pre-Request Data Packet SNAP-031-B0923 and PRE-REQ-001 were written to the reserved storage area of ​​NODE-031's local transmission queue. NODE-031 continued to monitor the network status. The vehicle exited the tunnel signal dead zone at 10:52, and network connectivity was restored. NODE-031 immediately uploaded SNAP-031-B0923 and PRE-REQ-001 to the cloud with forced priority, and the upload was completed at 10:53:18. During this network outage, the forced priority data in the reserved storage area was fully retained, while the regular collected data was compressed according to the downsampling strategy and written to the ordinary storage area.

[0070] The cloud received PRE-REQ-001 at 10:53:18 and, based on the supply chain topology diagram, identified the downstream node NODE-WH07 and its estimated arrival time window of 13:40 to 14:10. The current time is 2 hours and 47 minutes remaining until the estimated arrival start boundary, meeting the conditions for early delivery. The cloud also reads the extreme value amplitude from SNAP-031-B0923. With a duration of 13 minutes, the extreme value exceeds the first-level threshold, and the duration exceeds the first-level duration threshold. Therefore, the collaborative data collection level is determined to be high, and the collection frequency is set to four times the base frequency (once every 2 minutes), i.e., once every 30 seconds. The cloud verifies that NODE-WH07 is currently online and issues the advance collaborative data collection command CMD-WH07-001 to NODE-WH07 at 10:54:05.

[0071] Table 5. Contents of Advance Collaborative Data Acquisition Instructions After receiving CMD-WH07-001, NODE-WH07 switched to high-frequency acquisition mode at 13:30, collecting temperature and humidity, cargo receiving and weighing, and cold storage environment data every 30 seconds. The actual arrival time of the cargo was 13:58, and the unloading and weighing completion time was 14:06. All high-frequency acquisition data, along with the associated trigger identifier "PRE-REQ-001," was uploaded to the cloud and categorized and tagged in the cloud storage layer along with the regular periodic data from NODE-WH07.

[0072] Table 6. Sample of high-frequency collaborative data collected from transit warehouses After receiving high-frequency collaboratively acquired data from SNAP-031-B0923 and NODE-WH07, the cloud-based cross-link consistency verification module performs cross-link consistency verification. (Cycle break peak temperature) The set temperature of the carriage The time interval from the time the chain was restored (10:28) to the time the goods arrived at the transit warehouse (13:58). Temperature decay coefficient Substituting the values ​​into the temperature transfer attenuation constraint formula, the expected arrival temperature is calculated. Actual receiving temperature (First stable reading at 14:00:30 after cargo arrival). Departure weight from the transport end. The transit warehouse receives weight Preset weight tolerance value Weight difference The tolerance range is not exceeded; the weight deviation component is 0. Temperature deviation references the maximum threshold. Weighting coefficient , Therefore, the cross-stage deviation index is calculated. .

[0073] Table 7 Key Parameters for Cross-Process Consistency Verification Based on the cross-stage deviation index calculated in step 8 The causal relationship of AET-001 is marked as "device-triggered". The cloud-based causal link reconstruction module generates a cross-link broken causal link reconstruction report (report number RPT-031-B0923) and writes it into the link intervals corresponding to VEH-031 to WH-07 in the supply chain transparency status view.

[0074] Table 8 Core Fields of Cross-Segment Causal Link Reconstruction Report Throughout the process, the data started with the temperature and humidity time-series data collected locally by NODE-031 in the signal dead zone. After extreme value detection, EV-001 was identified, and AET-001 was generated through causal association annotation. Finally, it was encoded and encapsulated to form SNAP-031-B0923 and PRE-REQ-001. After network recovery, these two sets of data were uploaded to the cloud with mandatory priority. Two hours and 47 minutes before the goods arrived, the cloud issued a high-frequency acquisition command CMD-WH07-001 to NODE-WH07. During the goods handover, the high-frequency collaborative acquisition data (Table 6) generated by NODE-WH07 was aligned with the event snapshot data (Tables 3 and 4) and verification parameters (Table 7) in terms of time granularity, providing a comparable two-end data basis for the cross-stage consistency verification in step 8. After the verification result is written into RPT-031-B0923, it enters the supply chain transparency status view, completing the complete causal link closure from compressor shutdown triggering chain break, temperature peak exceeding threshold, and goods handover to transit warehouse receiving status, eliminating cross-link traceability blind spots caused by both signal blind spots and low-frequency acquisition.

[0075] The embodiments of the present invention have been described above. However, the embodiments are not limited to the specific implementation methods described above. The specific implementation methods described above are merely illustrative and not restrictive. Those skilled in the art can make more equivalent embodiments under the guidance of the present embodiments, and all of them are within the protection scope of the present embodiments.

Claims

1. A cloud-edge collaborative management method for making the food supply chain transparent, characterized in that, include: Acquire the time-series data of temperature and humidity sensors and the operating status data of refrigeration equipment within the current acquisition window of the cold chain transportation edge node; Perform sliding window extreme value detection on the time-series data of the temperature and humidity sensor, mark extreme value points that deviate from the safety threshold as extreme value events, and generate an extreme value event sequence; For each extreme event in the extreme event sequence, associate it with the device state change records within the same time window to generate an abnormal event tuple with causal association annotation; The abnormal event tuples are encoded into event snapshot data blocks, and the downstream related node identifiers and expected arrival time windows are extracted from the locally stored supply chain topology association information to generate collaborative synchronization pre-request data packets. When network connectivity is restored, the event snapshot data block and the collaborative synchronization pre-request data packet are uploaded to the cloud with a forced priority. The cloud parses the collaborative synchronization pre-request data packet and sends an advance collaborative collection instruction to the downstream transit warehouse edge node before the start boundary of the expected arrival time window; The downstream transit warehouse edge nodes collect data in a high-frequency acquisition mode within a specified time window and upload it to the cloud. Based on temperature transmission attenuation constraint rules and weight conservation tolerance rules, the cloud performs cross-link consistency verification on the event snapshot data blocks at the transportation end and the high-frequency collaborative collection data at the transit warehouse end, generating a cross-link deviation index. Based on the cross-link deviation index and the causal association annotation, a cross-link breakage causal link reconstruction report is generated and written into the supply chain transparency status view.

2. The cloud-edge collaborative management method for food supply chain transparency according to claim 1, characterized in that, The sliding window extreme value detection includes: Extract local maximum and local minimum values ​​and their corresponding timestamps within each sliding window; calculate the magnitude and duration of each extreme value's deviation from the safety threshold; mark extreme points where the deviation magnitude exceeds a preset magnitude threshold and the duration exceeds a preset duration threshold as extreme value events; wherein, the step length of the sliding window is less than the window width, so that there is an overlapping area between adjacent windows.

3. The cloud-edge collaborative management method for food supply chain transparency according to claim 1, characterized in that, The methods for generating the causal association annotations include: Calculate the time offset between the timestamp of the extreme event occurrence and the timestamp of the most recent device state change; when the time offset is positive and its absolute value is less than a preset causal time window, it is marked as device-triggered; when no device event can be matched within the preset causal time window, it is marked as external environment type; the abnormal event tuple includes the extreme value amplitude, duration, associated device event type and the time offset.

4. The cloud-edge collaborative management method for food supply chain transparency according to claim 2, characterized in that, It also includes performing event aggregation processing on the extreme event sequence: When the time interval between adjacent extreme events is less than the preset aggregation interval, the adjacent extreme events are merged into an aggregated extreme event; the extreme value amplitude of the aggregated extreme event is the maximum amplitude value among the merged events, and the duration is the total time span from the start of the first merged event to the end of the last merged event.

5. The cloud-edge collaborative management method for food supply chain transparency according to claim 1, characterized in that, The estimated arrival time window in the collaborative synchronization pre-request data packet is calculated based on the current time and the remaining estimated transportation time from the current location to the termination node, and includes the start and end boundaries of the estimated arrival time; the supply chain topology association information is static configuration data that is issued by the cloud and cached locally on the edge node before the transportation task is started.

6. The cloud-edge collaborative management method for food supply chain transparency according to claim 1, characterized in that, The advance collaborative acquisition instruction includes the parameter types that need to be acquired at high frequency and the acquisition duration; the cloud determines the collaborative acquisition level based on the extreme value amplitude and duration in the event snapshot data block: when the extreme value amplitude exceeds the first level threshold and the duration exceeds the first duration threshold, the collaborative acquisition level is high level, and the acquisition frequency is a preset high multiple of the base frequency; when the extreme value amplitude or duration is lower than the first level threshold, the collaborative acquisition level is standard level, and the acquisition frequency is a preset standard multiple of the base frequency.

7. The cloud-edge collaborative management method for food supply chain transparency according to claim 1, characterized in that, The generation methods for the cross-stage deviation index include: The expected arrival temperature is calculated based on the aforementioned temperature transfer attenuation constraint rule: ,in Set the temperature of the carriage after the cooling system has been restored. This is the peak temperature of chain breakage. The temperature decay coefficient is The time interval from the moment the chain is restored to the moment the goods arrive at the transit warehouse; Calculate the cross-stage deviation index: ,in This refers to the actual receiving temperature at the transit warehouse. This represents the maximum allowable threshold for temperature deviation. The weight at the point of origin of the transport. The weight received by the transit warehouse. To preset the weight tolerance value, and The weighting coefficients are satisfied. .

8. The cloud-edge collaborative management method for food supply chain transparency according to claim 1, characterized in that, The cross-linkage causal link reconstruction report includes the start time and location of the link breakage event, the event type of the associated equipment, the temperature transmission path across the handover point, and the residual risk level. The residual risk level is determined based on the range of the cross-link deviation index: when the cross-link deviation index is below the first risk threshold, it is low risk; when it is between the first and second risk thresholds, it is medium risk; and when it is above the second risk threshold, it is high risk.

9. The cloud-edge collaborative management method for food supply chain transparency according to claim 1, characterized in that, The event snapshot data block and the collaborative synchronization pre-request data packet occupy an independent reserved storage area in the local transmission queue. The reserved storage area has a fixed capacity and is not overwritten by regular data. When the reserved storage area is close to full, the edge node performs downsampling compression on the newly generated regular acquisition data and writes it into the ordinary storage area.

10. A cloud-edge collaborative management system for food supply chain transparency, used to execute the cloud-edge collaborative management method for food supply chain transparency as described in any one of claims 1 to 9, characterized in that, include: The edge acquisition module is used to acquire time-series data of temperature and humidity sensors and operating status data of refrigeration equipment within the current acquisition window of the cold chain transportation edge node; The extreme value detection module is used to perform sliding window extreme value detection on the time-series data of the temperature and humidity sensor to generate an extreme value event sequence; The causal association module is used to associate each extreme event in the extreme event sequence with device state change records within the same time window, and generate anomaly event tuples with causal association annotations. The encoding and pre-request generation module is used to encode the abnormal event tuple into an event snapshot data block and generate a collaborative synchronization pre-request data packet carrying the downstream associated node identifier and the expected arrival time window; The priority transmission module is used to upload the event snapshot data block and the collaborative synchronization pre-request data packet to the cloud with a forced priority when the network connection is restored; The collaborative instruction issuing module is used to parse the collaborative synchronization pre-request data packet in the cloud and issue an advance collaborative collection instruction to the downstream transit warehouse edge node before the start boundary of the expected arrival time window; A high-frequency collaborative acquisition module is used for the downstream transit warehouse edge nodes to acquire data in a high-frequency acquisition mode within a specified time window and upload it to the cloud. The cross-link verification module is used to perform consistency verification on the event snapshot data block at the transportation end and the high-frequency collaborative collection data at the transit warehouse end based on the temperature transmission attenuation constraint rule and the weight conservation tolerance rule, and generate a cross-link deviation index. The causal link reconstruction module is used to generate a cross-link broken causal link reconstruction report based on the cross-link deviation index and the causal association label, and write it into the supply chain transparency status view.

Images