Dairy product intelligent management method based on internet of things technology

Abstract

The present application relates to the field of Internet of Things cold chain management and dairy product warehouse monitoring, in particular to a dairy product intelligent management method based on Internet of Things technology; a server establishes dairy product batch records and binds Internet of Things nodes, storage locations, target temperature zones and refrigeration execution mechanisms; nodes collect package surface temperatures, air temperature and humidity, door states and refrigeration outputs and upload them; the server calculates data reliability, adopts backup temperature correction to obtain risk calculation temperature when the data reliability is lower than a threshold value, calculates temperature exposure based on over-temperature degree and over-temperature duration, determines batch comprehensive deterioration risk in combination with external heating events, and generates batch state identification, storage location handling instructions and refrigeration execution mechanism target output to issue control.

Description

Technical Field

[0001] The present invention relates to the field of Internet of Things cold chain management and dairy product storage monitoring, in particular to an intelligent dairy product management method based on Internet of Things technology. Background Art

[0002] Existing dairy product cold chain management usually relies on scattered temperature recording, manual inspection and post-event traceability; in the related art, in order to grasp the status of dairy products during storage, transportation and handover, it is usually necessary to collect temperature, humidity and equipment operation information, and perform alarming and control based on the collected information; or, it is judged whether situations such as over-temperature and door opening occur through single-point sensing data; However, when the number of nodes exceeds a preset node number threshold, the communication packet loss rate exceeds a preset packet loss rate threshold, and local sensing anomalies occur frequently, the former approach is prone to the problems of discontinuous data, unclear batch association and delayed disposal, while the latter approach can hardly balance the accuracy of risk judgment and the real-time performance of control strategies; therefore, the intelligent dairy product management methods in the related art can hardly meet the requirements of batch-level risk assessment, anomaly correction and linked control at the same time. Summary of the Invention

[0003] The object of the present invention is to provide an intelligent dairy product management method based on Internet of Things technology, so as to solve the following technical problem: existing dairy product cold chain management technologies have deficiencies in data continuity, batch-level risk assessment and real-time performance of control strategies, and there is an urgent need for an intelligent dairy product management method that can associate batch spoilage risk with anomaly correction and dynamically optimize linked control. The present invention specifically provides the following technical solution: The present invention provides an intelligent dairy product management method based on Internet of Things technology, which is interactively executed by a server and Internet of Things nodes, comprising: the server establishes dairy product batch records, and binds each batch with corresponding Internet of Things node identifiers, location identifiers, target temperature zone identifiers and refrigeration actuator identifiers; the Internet of Things nodes communicate with packaging surface temperature sensors, air temperature and humidity sensors, door sensors and refrigeration actuators; the Internet of Things nodes collect packaging surface temperature, air temperature and humidity, door status and actual output of refrigeration actuators, and upload the collected data to the server; the server calculates the data credibility of the collected data, and corrects the packaging surface temperature by using a backup temperature when the data credibility is lower than a preset threshold to obtain a risk calculation temperature; the server calculates the temperature exposure based on the over-temperature degree that the risk calculation temperature is higher than the upper limit of the target temperature zone and the over-temperature duration, identifies external heating events, and determines the comprehensive spoilage risk of the batch according to the temperature exposure and the external heating events; the server generates batch status identifiers, location handling instructions and target output of refrigeration actuators according to the comprehensive spoilage risk of the batch, and issues them to the Internet of Things nodes for control.

[0004] Preferably, the collected data also includes light intensity, triaxial acceleration, position, compressor target output and actual output, fan speed, and evaporator temperature; among which, the sampling time collected by the IoT node is used for temperature exposure accumulation, and the reception time is used for communication delay determination; triaxial acceleration is used to identify handling events, and position is used to verify the binding of the storage location; for batches that meet the preset redundancy monitoring conditions, at least two IoT nodes are bound, the temperature data of the redundant IoT nodes are used to verify the data of the same batch, and a backup temperature is provided when the data of the target IoT node is abnormal.

[0005] As a preferred method, the data credibility of the collected data is calculated, and when the data credibility is lower than the preset threshold, the packaging surface temperature is corrected using the backup temperature to obtain the risk calculation temperature. This includes: normalizing the range out of bounds, the rate of temperature change out of bounds, the temperature deviation of adjacent storage locations, the inconsistency between the door status and the temperature and humidity changes, and the power decay into abnormal indicators, and calculating the data credibility based on each abnormal indicator and its weight. Set a first confidence threshold and a second confidence threshold that is greater than the first confidence threshold; record the packaging surface temperature collected by the IoT node to be evaluated as the target temperature; when the data confidence is less than the first confidence threshold, obtain the backup temperature sequentially from the nodes in the same batch with confidence reaching the threshold, the adjacent cargo location nodes, and the temperature response model, and use the backup temperature as the risk calculation temperature; When the data reliability is greater than or equal to the first reliability threshold and less than the second reliability threshold, the backup temperature is sequentially obtained from nodes in the same batch with reliability reaching the first reliability threshold, adjacent storage location nodes, and the temperature response model. The product of the data reliability and the target temperature is then added to 1, and the sum of the product of the difference in data reliability and the backup temperature is subtracted from the product of these two values ​​to determine the risk calculation temperature. The corresponding calculation relationship is as follows: ;in, Calculate temperature for risk assessment. For data credibility, For the target temperature, The backup temperature is set as the target temperature when the data credibility is greater than or equal to the second credibility threshold.

[0006] As a preferred option, the temperature response model obtains the estimated packaging surface temperature based on the air temperature, the actual output of the refrigeration actuator, and the thermal response parameters. The thermal response parameters are updated based on the temperature difference before and after control, the rate of temperature change, and the recovery time. When the data loss time exceeds the loss threshold, temperature conflicts occur in the same batch of nodes, or during an external heating event, the higher of the highest value of the reliable candidate temperature and the estimated packaging surface temperature plus a preset safety margin is used as the backup temperature. When a backup temperature cannot be obtained, the missing interval is marked, and the data uncertainty is calculated based on the missing duration, the time-weighted average of the reliability of nodes in the same batch, and the number of node temperature conflicts. The data uncertainty is used to correct the overall batch deterioration risk. When the temperature difference between two nodes in the same batch exceeds the conflict threshold for a preset duration, a temperature conflict is recorded.

[0007] Preferably, identifying external heating events includes: within the event determination window, when the door is opened and at least one of the following—the rate of change of air temperature, the amount of change of humidity, the air temperature difference between the first monitoring area near the door and the second monitoring area far from the door, and the amount of change of light intensity—reaches the corresponding trigger threshold, a candidate external heating event is generated; when door status data is missing, a candidate external heating event is generated when at least two of the following—the rate of change of air temperature, the amount of change of humidity, the air temperature difference, and the amount of change of light intensity—reach the corresponding trigger threshold. During the observation period, an external heating event is confirmed when the increase in packaging surface temperature reaches the temperature rise confirmation threshold, the event temperature exposure corresponding to the degree and duration of overheating during the candidate external heating event reaches the exposure confirmation threshold, or the duration reaches the duration confirmation threshold. After the door is closed and a preset detection time has elapsed, if the risk calculation temperature is higher than the risk calculation temperature when the door is closed or is still higher than the upper limit of the target temperature zone, the temperature recovery state is determined to be entered. When the risk calculation temperature is continuously within the target temperature zone, its rate of change does not exceed the stability rate threshold, and it remains stable for a preset time, the temperature recovery state is determined to be exited.

[0008] As a preferred embodiment, the server encapsulates the batch status identifier, the location handling instruction, and the target output of the refrigeration actuator into a data packet and sends it to the IoT node; the data packet includes a status control table, a threshold table, and control instructions applicable to the same batch, location, temperature zone, and refrigeration actuator. When the IoT node passes the field integrity verification of the sent data packet, the data packet version is higher than the current execution version, the data packet is within the valid time window, and the batch identifier, cargo location identifier, target temperature zone identifier, and refrigeration actuator identifier are consistent with the local record, the node verifies whether the target output of the refrigeration actuator is within the allowable output range, whether the minimum start-stop interval is met, whether the single-cycle output change does not exceed the maximum output change, and whether the minimum predicted temperature is not lower than the sum of the preset minimum allowable temperature and safety margin. The system uses a preset temperature response model to predict the packaging surface temperature sequence within a preset time domain and takes the minimum value as the lowest predicted temperature. When the verification passes, the status control table, threshold table, and control instructions are replaced in one go. When the verification fails, the system executes a pre-stored safe version. If there is no safe version, the output is limited to a preset safe output range and a local alarm is generated.

[0009] Preferably, during the period when communication between the IoT node and the server is interrupted, the data corresponding to the external heating event is divided into a baseline segment with a preset duration before the candidate event occurs, a main segment from the occurrence of the candidate event to the closing of the door, and a recovery segment from entering the temperature recovery state to exiting the temperature recovery state, and is used as an event data packet cache; the event data packet includes an event identifier, batch identifier, node identifier, sampling data sequence, confidence sequence, state switching record, control command record and execution receipt, and the server confirms its validity after receiving all three segments of data completely; After communication is restored, the IoT node sequentially sends the current status, unresolved alarms, the most recent execution receipt of the current version, high-priority event data packets, and other historical data and historical receipts. Data of the same priority are sorted according to sampling time, node identifier, and local incrementing sequence number. The sequence period identifier formed after resetting or wrapping the local incrementing sequence number is used as the cycle epoch. The server records the maximum consecutive sequence number, the maximum sampling time corresponding to the completely received data, and the range of sequence numbers to be retransmitted according to the node identifier and cycle epoch. The server determines that the data is complete if there are no missing sequence numbers within the range of sequence numbers to be retransmitted.

[0010] As a preferred method, the comprehensive batch deterioration risk is determined based on temperature exposure and external heating events, including: using the period from warehousing to outgoing as the cumulative batch management cycle, using a preset rolling window as the control management cycle, calculating the temperature difference exceeding the upper limit of the corresponding target temperature zone based on the risk calculation at each sampling time, and calculating the temperature exposure based on the temperature difference exceeding the upper limit of the corresponding target temperature zone and the time interval between adjacent sampling; and normalizing the maximum temperature rise and the event temperature exposure of the external heating event and weighting them according to preset weights to obtain the event intensity. The credibility of an event is determined based on the average credibility of nodes during the event period and the completeness of data on door status, rate of change of air temperature, amount of change of humidity, regional temperature difference, amount of change of light, and temperature rise of packaging surface. The basic risk is determined based on the number of valid events that all reach the corresponding effective thresholds for temperature exposure, event intensity, and event credibility, the maximum event intensity, and the nominal remaining shelf life. The basic risk is then adjusted upwards using data uncertainty to obtain the comprehensive batch spoilage risk.

[0011] As a preferred option, a first risk threshold, a second risk threshold, and a third risk threshold are set in ascending order. When the overall risk of batch deterioration is less than the first risk threshold, it is determined to be low risk; when it is greater than or equal to the first risk threshold and less than the second risk threshold, it is determined to be medium risk; when it is greater than or equal to the second risk threshold and less than the third risk threshold, it is determined to be relatively high risk; and when it is greater than or equal to the third risk threshold, it is determined to be high risk. Based on the risk level, the sampling period, sending period, buffer priority, alarm flag, and batch status identifier are determined respectively. For medium-risk batches or batches whose risk increment reaches the increment threshold, a sampling period one level shorter than the current level is selected from the sampling period table classified by period duration, and the sending priority is increased. For higher-risk batches, a sampling sequence of preset duration is retained before and after the comprehensive deterioration risk first reaches the second risk threshold, and a priority outbound identifier or a candidate instruction for cargo location handling is generated. For high-risk batches, a high-risk alarm is generated, and the batch status is set to a prohibited automatic release state.

[0012] Preferably, the server generates warehouse handling instructions and refrigeration actuator target outputs, including: using the control management cycle as the warehouse evaluation cycle, determining the warehouse heating level based on the warehouse air temperature exposure and the number of external heating events, and determining the delivery urgency level based on the remaining time from the current time to the delivery time window deadline; and determining the warehouse handling priority, outbound priority, and cold air allocation priority based on the risk level, warehouse heating level, nominal remaining shelf life level, and delivery urgency level. When multiple temperature zones share a compressor, if the compressor capacity is sufficient and there are no high-priority alarms, the target output of the refrigeration actuator is determined based on the weighted sum of the refrigeration demands of each temperature zone. If the capacity is insufficient or there are high-priority alarms, the target output of the refrigeration actuator is determined based on the demand of the highest-priority temperature zone, and the refrigeration capacity is reduced sequentially from the lower-priority temperature zones until the total compressor capacity constraint is met. When the predicted minimum temperature is lower than the sum of the preset minimum allowable temperature and the safety margin, or when the target output of the refrigeration actuator exceeds the allowable output range, alarm information requiring manual processing, a locked object identifier, and a constraint not met identifier are generated.

[0013] As can be seen from the above solutions, the advantages of the present invention are: 1. This method calculates the data reliability of the collected data, and when the reliability is lower than the threshold, it uses a backup temperature to correct the packaging surface temperature to obtain the risk calculation temperature. The above steps overcome the problem of data discontinuity caused by frequent local sensor anomalies, realize anomaly correction, and improve the accuracy of risk assessment data. 2. This method determines the comprehensive deterioration risk of a batch based on temperature exposure and external heating events, and generates batch status identifiers, cargo handling instructions, and target outputs of refrigeration actuators accordingly. This makes up for the shortcomings of the existing technology in handling lag, takes into account the accuracy of risk assessment and the real-time nature of control strategies, and meets the needs of batch-level linkage control. Attached Figure Description

[0014] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0015] Figure 1 This is a flowchart illustrating the intelligent management method for dairy products based on Internet of Things (IoT) technology proposed in this application. Detailed Implementation

[0016] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings.

[0017] Reference Figure 1 One embodiment of this application provides a smart management method for dairy products based on Internet of Things (IoT) technology, executed by interaction between a server and IoT nodes, including: The server establishes batch records for dairy products, binding each batch with its corresponding IoT node and its identifier, storage location and its identifier, target temperature zone and its identifier, and refrigeration actuator and its identifier. The IoT node communicates with the packaging surface temperature sensor, air temperature and humidity sensor, door sensor, and refrigeration actuator; The IoT nodes collect data on the surface temperature of the packaging, air temperature and humidity, door status, and the actual output of the refrigeration actuator, and upload the collected data to the server. The server calculates the data credibility of the collected data and obtains a backup temperature when the data credibility is lower than a preset threshold. The backup temperature is then used to correct the packaging surface temperature to obtain the risk calculation temperature. The server calculates the temperature exposure based on the degree and duration of temperature exceeding the upper limit of the target temperature zone, identifies external heating events, and determines the overall batch spoilage risk based on the temperature exposure and external heating events. The server generates batch status identifiers, storage location handling instructions, and target outputs for refrigeration actuators based on the overall batch spoilage risk, and sends them to IoT nodes for control.

[0018] The collected data also includes: light intensity, triaxial acceleration, position, compressor target output and actual output, fan speed and evaporator temperature; Among them, the server accumulates temperature exposure based on the sampling time collected by IoT nodes, and determines communication delay based on the receiving time; The server identifies handling events based on three-axis acceleration and verifies the binding of storage locations based on position. For a batch that meets the preset redundancy monitoring conditions, bind at least two IoT nodes, use the temperature data of the redundant IoT nodes to verify the data of the same batch, and use the temperature data of the redundant IoT nodes as backup temperature when the data of the IoT node is abnormal.

[0019] The data reliability of the collected data is calculated, and when the data reliability is lower than a preset threshold, the packaging surface temperature is corrected using a backup temperature to obtain the risk calculation temperature, including: The following abnormal indicators were normalized: exceeding the range limit, exceeding the temperature change rate limit, temperature deviation between adjacent storage locations, inconsistency between door status and temperature and humidity changes, and power decay. The data reliability was calculated based on each abnormal indicator and its weight. Set a first confidence threshold and a second confidence threshold that is greater than the first confidence threshold; The packaging surface temperature collected by the IoT node to be evaluated is recorded as the target temperature; When the data credibility is less than the first credibility threshold, the backup temperature is obtained sequentially from the same batch node with credibility reaching the first credibility threshold, the adjacent storage location node, and the temperature response model, and the backup temperature is used as the risk calculation temperature. When the data reliability is greater than or equal to the first reliability threshold and less than the second reliability threshold, the backup temperature is sequentially obtained from nodes in the same batch with reliability reaching the first reliability threshold, adjacent storage location nodes, and the temperature response model. The product of the data reliability and the target temperature is then added to 1, and the sum of the product of the difference in data reliability and the backup temperature is subtracted from the product of these two values ​​to determine the risk calculation temperature. The corresponding calculation relationship is as follows: ;in, Calculate temperature for risk assessment. For data credibility, For the target temperature, This is for backup temperature; When the data credibility is greater than or equal to the second credibility threshold, the target temperature is determined as the risk calculation temperature.

[0020] During the temporary storage, refrigerated transportation, and handover of dairy products in cold storage, the server continuously maintains batch records and establishes a correspondence between batches and nodes, storage locations, target temperature zones, and refrigeration actuators. When receiving data from heterogeneous nodes, the server uses an asynchronous queue based on node identifier hashing for distribution and adds a lifecycle timestamp based on the arrival time to implement flow-limiting concurrency control. Each batch of dairy products always corresponds to a specific collection node and control object within the system to complete temperature exposure accumulation, risk assessment, and disposal issuance according to the batch dimension. IoT nodes collect data on packaging surface temperature, air temperature and humidity, door status, and the actual output of the refrigeration actuator, and upload the sampling results to the server. After receiving the data, the server first calculates the data reliability and then determines whether to use the backup temperature to correct the packaging surface temperature. The backup temperature is obtained in the order of nodes in the same batch, adjacent storage location nodes, and temperature response model. When the backup temperature is activated, the server switches the logical status label of the data stream from the original direct sampling state to the cascaded compensation state. The maximum allowable pressure drop range, upper limit of natural environmental fluctuations, anti-diffusion extreme value, and basic fixed abnormal value are preset and stored on the server based on the equipment nominal parameters and historical natural distribution characteristics of the corresponding cold chain situation. When the data credibility is lower than the preset threshold, the original data is retained and the correction process is initiated; abnormal indicators are quantified according to the following: range exceeding the limit, temperature change rate exceeding the limit, temperature deviation between adjacent storage locations, inconsistency between door status and temperature and humidity changes, and power decay. The quantification logic for the inconsistency between the door status and temperature and humidity changes includes: when the door sensor shows that it is closed, but the temperature or humidity sensor shows that the rate of change of temperature or humidity within a preset time window exceeds the upper limit of natural environmental fluctuations, the excess amount is divided by the benchmark upper limit as the initial value of inconsistency; when the door sensor shows that it is open and the rate of change of temperature and humidity is lower than the preset lower limit of change rate, a preset basic fixed abnormal value is set; other matching states are set to 0. Among them, the initial value of the range exceeding the limit is quantified as the absolute deviation value of the actual sampling temperature of the node exceeding the nominal temperature measurement range of the sensor hardware; the initial value of the temperature change rate exceeding the limit is quantified as the difference between the absolute value of the rate of change of the node's temperature in two consecutive samplings and the preset upper limit of the natural fluctuation of the environment. For issues such as range exceeding limits, temperature change rate exceeding limits, temperature deviation between adjacent storage locations, inconsistent initial values, and power decay, range normalization is used to map them to the [0,1] interval. The normalization calculation process for temperature deviation between adjacent storage locations is as follows: the absolute temperature difference between nodes is subtracted from the preset dead zone tolerance, and the remaining value is divided by the maximum value of the natural distribution limit temperature difference of the corresponding temperature zone. Power decay is quantified as follows: when the node battery voltage is lower than the lower limit of the rated operating voltage, the voltage drop value is divided by the preset maximum allowable voltage drop range to obtain the normalized decay index, where 0 represents normal and 1 represents complete failure, forming a unified confidence result; the basic confidence level is set to 1. The final confidence level equals the basic confidence level minus the sum of the products of each normalized anomaly index and its preset weight, with the calculation result limited to the range [0,1]. The preset weights of each anomaly index are allocated according to their physical impact on measurement accuracy, with a weight of 0.4 for range exceeding limits and a weight of 0.2 for temperature change rate exceeding limits. The remaining weights are allocated based on equipment characteristics. In this embodiment, the normalized anomaly indices corresponding to range exceeding limits, temperature change rate exceeding limits, temperature deviation between adjacent storage locations, inconsistency between door status and temperature / humidity changes, and power attenuation are, in order: , , , , Then the data credibility according to Calculation; among which, the weights corresponding to the temperature deviation of adjacent storage locations, the inconsistency between the door status and temperature and humidity changes, and the power attenuation are 0.15, 0.15, and 0.10 respectively, the sum of the three weights is 0.40, and the sum of the five weights is 1; when the equipment is replaced, the above three weights can be adjusted according to the equipment calibration error, but the sum of the three weights should be kept at 0.40.

[0021] The first and second confidence thresholds are calibrated based on the statistical characteristics of the confidence intervals of historical normal cold chain nodes under similar circumstances, and are set to 0.6 and 0.85 respectively. The two are weighted and fused to obtain the risk calculation temperature. If the value is above the second confidence threshold, the target temperature is directly used for subsequent calculations. In the calculation of temperature exposure, the system uses the degree and duration of the risk-calculated temperature exceeding the upper limit of the target temperature zone as the basis, and accumulates the exposure according to the sampling time. That is, within each discrete sampling step, the extent by which the risk-calculated temperature exceeds the upper limit of the target temperature zone is multiplied by the sampling interval duration to obtain the step exposure increment, and then accumulated through a time-series integrator. When the door is opened and changes occur in air temperature, humidity, light intensity, and regional temperature difference, a judgment is made in conjunction with external heating events. If the packaging surface temperature continues to rise after the door is closed, or if the air temperature has dropped but the packaging surface temperature remains high, the batch will not be directly deemed safe simply because the air has recovered. Instead, the overall batch spoilage risk will be determined by combining the risk calculation temperature and external heating events. Based on this, the server generates batch status identifiers, cargo handling instructions, and target outputs for the refrigeration actuator, and sends them to the IoT nodes for execution. For batches requiring redundant monitoring, the system will bind at least two IoT nodes and use the temperature data of the redundant nodes to verify the data of the same batch; if the target node is abnormal, the backup temperature will be provided directly by the redundant node or the adjacent storage location node, thereby ensuring that the batch can continue to calculate the risk and generate control commands even if a single node is abnormal; the sampling time is used to accumulate the temperature exposure, and the receiving time is used to determine the communication delay and retransmission order, and does not participate in the risk accumulation. In one embodiment, the temperature response model estimates the packaging surface temperature based on the air temperature, the actual output of the refrigeration actuator, and the thermal response parameters, which are updated based on the temperature difference before and after control, the rate of temperature change, and the recovery time. When the duration of missing data exceeds the missing threshold, historical reliable temperature data is obtained as reliable candidate temperatures, and the higher of the highest reliable candidate temperature and the estimated packaging surface temperature plus the preset safety margin set based on the target temperature zone is used as the backup temperature. When a backup temperature cannot be obtained, the missing interval is marked, and the data uncertainty is calculated based on the missing duration, the time-weighted average of the reliability of nodes in the same batch, and the number of node temperature conflicts. The data uncertainty is then used to correct the overall batch deterioration risk. A temperature conflict is recorded when the temperature difference between two nodes in the same batch continuously exceeds the conflict threshold and reaches a preset duration.

[0022] Identify external heating events, including: Within the event determination window, when the door is opened and at least one of the following—the rate of change of air temperature, the amount of change of humidity, the air temperature difference between the first monitoring area near the door and the second monitoring area far from the door, and the amount of change of light—reaches the corresponding trigger threshold, a candidate external heating event is generated. When the gate status data is missing, a candidate external heating event is generated when at least two of the following factors—the rate of change of air temperature, the amount of change of humidity, the air temperature difference, and the amount of change of light intensity—reach the corresponding trigger threshold. During the observation period, an external heating event is confirmed to have occurred when the increase in packaging surface temperature reaches the temperature rise confirmation threshold, the event temperature exposure corresponding to the degree of overheating and duration during the candidate external heating event reaches the exposure confirmation threshold, or the duration reaches the duration confirmation threshold. After the door is closed and a preset detection period has elapsed, if the risk calculation temperature is greater than or equal to the risk calculation temperature when the door is closed, or is still greater than or equal to the upper limit of the target temperature zone, then the door is determined to enter the temperature recovery state. When the risk calculation temperature remains within the target temperature range, its rate of change does not exceed the stable rate threshold, and it remains stable for a preset duration, the temperature recovery state is exited.

[0023] The server encapsulates the batch status identifier, the location handling instruction, and the target output of the refrigeration actuator into a data packet and sends it to the IoT node. The data package sent out includes a status control table, threshold table, and control instructions applicable to the same batch, storage location, temperature zone, and refrigeration actuator; The packaging surface temperature sequence within a preset time domain is predicted using a preset temperature response model, and the minimum value is taken as the lowest predicted temperature. When the IoT node passes the field integrity verification of the sent data packet, the data packet version is higher than the current execution version, the data packet is within the valid time window, and the batch identifier, cargo location identifier, target temperature zone identifier, and refrigeration actuator identifier are consistent with the local record, the node verifies whether the target output of the refrigeration actuator is within the allowable output range, whether the minimum start-stop interval is met, whether the single-cycle output change does not exceed the maximum output change, and whether the minimum predicted temperature is not lower than the sum of the preset minimum allowable temperature and safety margin. If the verification passes, the status control table, threshold table, and control instructions are replaced in one go; if the verification fails, a pre-stored secure version is executed. When no secure version is available, the output will be limited to the preset secure output range and a local alarm will be generated.

[0024] The IoT nodes will continuously collect temperature and equipment status data and organize them into executable control data, and determine the correspondence between packaging surface temperature, air temperature, and the output of the refrigeration actuator. The temperature response model uses a first-order thermal capacity-thermal resistance network for dynamic simulation. The difference between the air temperature and the packaging surface temperature is converted into an estimated rate of change using thermal resistance-thermal capacity parameters. This rate of change is then combined with the cooling correction term corresponding to the actual output of the refrigeration actuator to estimate the packaging surface temperature. In practice, the system performs recursive estimation according to a discrete time step k. The estimation formula can be expressed as: ; in, To estimate the surface temperature of the packaging, For air temperature, The actual output of the refrigeration actuator. The environmental heat transfer coefficient is the converted value of the thermal resistance-heat capacity parameter. This is the cooling correction factor; before and after control, the system extracts the residual between the actual temperature change on the packaging surface and the model-estimated temperature change based on the measured temperature difference, temperature change rate, and recovery time. This residual is then compared in real-time using least-squares fitting for dynamic correction. and This enables continuous updating of thermal response parameters; The system maintains a historical data sliding window of a preset length N, where N is 10 control cycles. ; The dependent variables form a column vector. air temperature difference and negative cooling output Constructed as corresponding independent variables observation matrix Through the least squares analytical expression ; The regression coefficient that minimizes the sum of squared residuals is directly calculated and set for updating, thereby achieving dynamic adaptive parameter updating without model retraining; where, This serves as the time step index within the historical data sliding window. This refers to the actual surface temperature of the packaging. This represents the actual change in surface temperature of the packaging; superscript Represents the transpose operation of a matrix or vector; superscript This represents the inverse operation of a matrix; Before regression solving, the system uses singular value decomposition to obtain the ratio of the maximum to the minimum singular value of the observation matrix as a condition number for verification. When the condition number exceeds the preset anti-divergence threshold, the matrix is ​​determined to be ill-conditioned, the parameter iteration is forcibly terminated and the anti-divergence branch is triggered, the current parameter update is stopped and the thermal response parameters of the previous cycle are used to ensure the mathematical robustness of the parameter iteration process. The environmental heat transfer coefficient and the cooling correction term coefficient are all converted according to the sampling period and the time scale, so that the temperature difference term and the cooling output term correspond to the temperature increment within the same time step. When the data loss duration exceeds the preset loss duration threshold, or when temperature conflicts occur between nodes in the same batch, or when an external heating event is in progress, the backup temperature will no longer simply take the most recent value, but will take the higher value between the highest value of the credible candidate temperature and the estimated packaging surface temperature plus a safety margin. The loss threshold is set to be more than twice the single communication reconnection timeout, for example, 3 minutes. The conflict threshold for temperature conflicts is set to twice the sensor's factory nominal absolute error. The safety margin is set based on the empirical value of the cooling fluctuation tolerance of the target temperature zone, for example, 0.5℃ for the cold storage zone. If the backup temperature cannot be obtained, mark the missing interval and calculate the data uncertainty based on the missing duration, the time-weighted average of the reliability of nodes in the same batch, and the number of temperature conflicts. The system calculates the overall batch spoilage risk based on the proportion of missing duration, the reverse quantification result of average confidence, and the normalized result of the number of conflicts; among which, the weight of the proportion of missing duration is 0.5, the weight of the decrease in average confidence is 0.3, and the weight of the number of conflicts is 0.2. External heating events are identified by a combination of door status, air temperature change rate, humidity change, regional temperature difference, and light intensity change. When the door is open, a candidate event is generated if any one of these triggering conditions is met. If door status data is missing, at least two conditions must meet the trigger threshold before a candidate event is generated. The trigger thresholds and the confirmation thresholds for temperature rise and duration are established based on the measured boundary data of the standard thermal response of a specific dairy product in the corresponding target temperature zone. The trigger thresholds are determined based on the measured boundary data of the standard thermal response in the target temperature zone, including an air temperature change rate higher than 0.5℃ / min and a light intensity change exceeding 50 lux. Once a candidate event enters the observation period, it is confirmed based on the temperature rise of the packaging surface, the temperature exposure of the event, and the duration. After the door is closed, if the packaging surface temperature is still higher than the risk calculation temperature when the door is closed, or if it is still higher than the upper limit of the target temperature zone, it enters the temperature recovery state. When the packaging surface temperature is continuously within the target temperature zone, and the rate of change and duration meet the stability conditions, it exits the temperature recovery state. After the server encapsulates the batch status identifier, the location handling instruction, and the target output of the refrigeration actuator into a data packet, the node first performs integrity verification, version verification, time window verification, and local binding consistency verification. Then, it verifies whether the target output of the refrigeration actuator meets the allowable range, minimum start-stop interval, single-cycle change, and minimum predicted temperature requirements. The minimum predicted temperature is obtained through the temperature response model within a preset time domain, and the minimum value in the sequence is taken. If the verification passes, the state control table, threshold table, and control instructions are atomically replaced as a whole; if the verification fails, the node's internal state machine switches to the safety protection mode and rolls back to the pre-stored safety version; if no safety version is pre-stored, a hardware watchdog interrupt is used to limit the cooling execution output to a preset safe duty cycle range and generate a local alarm. In one embodiment, during the period when communication between the IoT node and the server is interrupted, the data corresponding to the external heating event is divided into a baseline segment with a preset duration before the candidate event occurs, a main segment from the occurrence of the candidate event to the closing of the door, and a recovery segment from entering the temperature recovery state to exiting the temperature recovery state, and is used as an event data packet cache. The event data packet includes an event identifier, batch identifier, node identifier, sampled data sequence, confidence sequence, state transition record, control command record, and execution receipt. The server confirms the validity of the data after receiving all three segments completely. After communication is restored, the IoT node sequentially sends the current status, unresolved alarms, the most recently executed receipt of the current version, high-priority event data packets, and other historical data and historical receipts; Data of the same priority are sorted according to sampling time, node identifier, and local incrementing sequence number; The sequence period identifier formed after the local incrementing sequence number is reset or wrapped around is used as the cycle epoch. The server records the maximum consecutive sequence number, the maximum sampling time corresponding to the completely received data, and the range of sequence numbers to be retransmitted according to the node identifier and cycle epoch. The server determines that the data is complete when there are no missing sequence numbers within the range of sequence numbers to be retransmitted.

[0025] The overall batch spoilage risk is determined based on temperature exposure and external heat events, including: The cumulative management cycle for batches is from warehousing to outbound, and the control management cycle is a preset rolling window. The temperature exposure is calculated based on the temperature difference between the sampling time and the upper limit of the corresponding target temperature zone and the time interval between adjacent sampling. The maximum temperature rise and event temperature exposure of the external heating event are normalized and weighted according to preset weights to obtain the event intensity; The credibility of an event is determined based on the average credibility of nodes during the event period and the completeness of data on door status, rate of change of air temperature, amount of change of humidity, regional temperature difference, amount of change of light, and temperature rise of packaging surface. The basic risk is determined based on the number of valid events whose temperature exposure, event intensity, and event credibility all reach the corresponding valid thresholds, the maximum event intensity, and the nominal remaining shelf life, using a preset basic risk mapping function. The basic risk mapping function is as follows: ; in, Basic risk, The number of valid events. To preset the event quantity threshold, For maximum event intensity, The nominal remaining shelf life, Total shelf life , , These are the basic risk weight coefficients assigned by prior engineering considerations, where... , , All are weight coefficients greater than 0 and less than 1, and satisfy the following conditions: In this embodiment, , , During project implementation, the above weighting coefficients can be recalibrated based on the prediction errors of historical batch samples, but the sum of the three should be kept at 1. The basic risk should be adjusted upward based on the uncertainty of the acquired data to obtain the comprehensive batch deterioration risk.

[0026] Set a first risk threshold, a second risk threshold, and a third risk threshold that increase sequentially; When the overall risk of deterioration of a batch is less than the first risk threshold, it is judged as low risk; when it is greater than or equal to the first risk threshold and less than the second risk threshold, it is judged as medium risk; when it is greater than or equal to the second risk threshold and less than the third risk threshold, it is judged as relatively high risk; when it is greater than or equal to the third risk threshold, it is judged as high risk. The sampling period, sending period, buffer priority, alarm flag, and batch status identifier are determined according to the risk level. For medium-risk batches or batches whose risk increment reaches the increment threshold, select a sampling period one level shorter than the current level from the sampling period table classified by period duration, and increase the transmission priority. For higher-risk batches, a sampling sequence with a preset time period is retained before and after the comprehensive deterioration risk first reaches the second risk threshold, and a priority outbound identifier or a candidate instruction for cargo location handling is generated; High-risk batches generate high-risk alarms and set the batch status to "disable automatic release".

[0027] The server generates warehouse handling instructions and refrigeration actuator target outputs, including: The control and management cycle is used as the evaluation cycle for the cargo location. The degree of heat exposure of the cargo location is determined based on the air temperature exposure of the cargo location and the number of external heat events. The degree of delivery urgency is determined based on the remaining time from the current time to the end of the delivery time window. The priority of cargo handling, outbound priority, and cold air allocation priority are determined based on the risk level, the degree of heat exposure of the cargo location, the nominal remaining shelf life, and the urgency of delivery. When multiple temperature zones share a compressor, provided the compressor capacity is sufficient and there are no high-priority alarms, the server obtains the pre-set nominal space heat capacity parameter, which characterizes the spatial heat capacity of each temperature zone, and calculates the cooling demand for each temperature zone. The cooling demand is equal to the product of the positive temperature difference between the upper limit of the corresponding target temperature zone and the lowest predicted temperature obtained using the pre-set temperature response model, the nominal space heat capacity parameter of the target temperature zone, and the reciprocal of the control cycle. The calculation formula is as follows: ; in, For cooling needs, This corresponds to the positive temperature difference between the upper limit of the target temperature zone and the lowest predicted temperature. This refers to the nominal space heat capacity parameter for this temperature range. To control the cycle, the target output of the refrigeration actuator is determined based on the weighted sum of the refrigeration demands in each temperature zone. When there is insufficient capacity or high-priority alarms, the target output of the refrigeration actuator is determined according to the demand of the highest priority temperature zone, and the refrigeration capacity is reduced sequentially from the low-priority temperature zones until the total capacity constraint of the compressor is met. When the predicted minimum temperature is lower than the sum of the preset minimum allowable temperature and the safety margin, or when the target output of the refrigeration actuator exceeds the allowable output range, an alarm message requiring manual processing, a locked object identifier, and a constraint not met identifier are generated.

[0028] The system segments data related to external heating events into baseline segments, main segments, and recovery segments before storing them in an edge ring buffer. When the buffer reaches a preset limit, the node triggers a spatial permutation strategy, discarding sample points of the baseline segment at preset thinning intervals, while retaining the main and recovery segments. The server confirms the validity of the data after receiving all three segments completely. After communication is restored, nodes prioritize sending current status, unresolved alarms, most recently executed receipts, and high-priority event data packets, followed by other historical data and historical receipts. Data of the same priority are sorted by sampling time, node identifier, and local incrementing sequence number. After the sequence number wraps around, data segments from different rounds are distinguished by the cyclic epoch. The server records the maximum consecutive sequence number, the maximum sampling time corresponding to the data that has been completely received, and the range of sequence numbers to be retransmitted according to the node identifier and cyclic epoch, and determines whether the retransmission is complete based on this. The server is configured with a calculation lock based on sequence number continuity: before the sequence number range to be retransmitted is cleared, the latest status data that arrives first is used to perform real-time feedforward estimation based on the temperature response model. At the same time, the recalculation of the comprehensive deterioration risk of the corresponding batch of the node is paused, and the current alarm flag that is reported first is accepted until the communication data is completed and the comprehensive deterioration risk calculation is unlocked. The calculation of batch comprehensive deterioration risk not only depends on temperature exposure, but also on the intensity and reliability of external heating events and the completeness of data. The system uses the period from warehousing to outbound as the batch cumulative management cycle and a preset rolling window as the control management cycle. It calculates the temperature difference when the temperature exceeds the upper limit of the target temperature zone and the cumulative temperature exposure between adjacent sampling time intervals according to risk. The event intensity is obtained by mapping the ratio of the maximum temperature rise of the event to the maximum permissible temperature rise of the target temperature zone, and the ratio of the temperature exposure of the event to the maximum permissible exposure of a single event, to the interval [0,1], and then weighting and fusing them according to their respective sensitivity coefficients. The sensitivity coefficients are calibrated according to the sensitivity of the target dairy product to transient thermal shock and steady-state heating, with the temperature rise sensitivity set to 0.6 and the exposure sensitivity set to 0.4. Event credibility is calculated by multiplying the average credibility of nodes during the event period with the data completeness as the ratio of the number of valid sampling points to the total number of theoretical sampling points. Only when the temperature exposure, event intensity, and event credibility all reach the valid threshold are the corresponding events counted in the number of valid events, and the basic risk is further formed by combining the maximum event intensity and the nominal remaining shelf life. In this process, data uncertainty is only used for upward adjustment and does not replace the basic risk alone. The adjustment logic can be as follows: multiply the basic risk by the product of the preset adjustment coefficient and the data uncertainty plus one; when the basic risk value is 0.5, the data uncertainty is 0.8 and the preset adjustment coefficient is 0.25, the basic risk is adjusted according to the formula 0.5×(1+0.25×0.8) to obtain the batch comprehensive deterioration risk of 0.6. The risk thresholds are progressively increased in three levels. The first, second, and third risk thresholds are equivalent to 30%, 60%, and 85% of the shelf-life loss rate of dairy products, respectively. When the overall spoilage risk of a batch falls into different ranges, the system adjusts the sampling cycle, sending cycle, buffer priority, alarm flag, and batch status identifier accordingly. For medium-risk batches, when the risk increment reaches the threshold, the sampling cycle will be automatically shortened by one level and the sending priority will be increased. For higher-risk batches, the sampling sequence before and after the risk first reaches the second risk threshold is retained, and a priority outbound identifier or a candidate instruction for warehouse location handling is generated; for high-risk batches, a high-risk alarm is generated, and the batch status is set to a prohibited automatic release state. When the server generates cargo handling instructions and refrigeration actuator target outputs, it determines the priority of each item through a preset multi-dimensional comprehensive evaluation mechanism. The system adopts a hierarchical mapping rule to transform the status of each dimension into a control basis. For the cargo location heat level, it presets the equivalent temperature exposure constant for a single external heat event. The system adds the actual calculated cargo location air temperature exposure to the product of the number of external heat events and the constant to obtain the comprehensive heat load. Based on the preset tiered threshold range into which the comprehensive heat load falls, it is quantified into a discrete score value of 1 to 5. Similarly, the remaining shelf life and delivery deadline are mapped to the corresponding score tiers. The system quantifies the risk level, the heat level of the storage location, the nominal remaining shelf life, and the urgency of delivery, and assigns them to their respective preset weights (e.g., the risk level is assigned the highest weight, and the remaining shelf life is assigned the second highest weight). The system then performs a weighted summation to obtain a comprehensive priority score. Based on the scores, from highest to lowest, the corresponding storage location handling priority, outbound priority, and cooling capacity allocation priority are determined. When multiple temperature zones share a compressor, if the compressor capacity is sufficient and there are no high-priority alarms, the target output of the refrigeration actuator is determined by a weighted summation of the cooling needs of each temperature zone. If the capacity is insufficient or there are high-priority alarms, the highest priority temperature zone will be prioritized, and then the cooling capacity will be gradually reduced from the lower priority temperature zones until the total capacity constraint is met. The specific reduction logic is as follows: the system calculates the difference between the total capacity and the total demand as the cooling capacity to be reduced. For temperature zones that are not the highest priority, the reciprocal of their comprehensive priority score is normalized within the set of temperature zones participating in the reduction. That is, the reciprocal of the score of a certain temperature zone is divided by the sum of the reciprocals of all scores in the set, thereby obtaining the dimensionless penalty allocation ratio. The cooling capacity to be reduced is multiplied by this ratio and allocated to each temperature zone as its target reduction value, thereby ensuring that the system allocation process strictly follows the law of conservation of energy. The nominal reduction amount of each low-priority temperature zone is calculated through the penalty allocation ratio. If the allocated nominal reduction amount is greater than the current demand of the temperature zone, the allocated cooling output is directly set to zero, and the reduction margin exceeding the current demand is overflowed and added to the reduction amount of the next lowest priority temperature zone. The allocation continues in the remaining temperature zones that can be reduced until the total cooling capacity to be reduced is allocated. If the predicted minimum temperature is already lower than the sum of the preset minimum allowable temperature and the safety margin, or if the target output of the refrigeration actuator exceeds the allowable range, the system will not continue to generate executable control quantities, but will instead output alarm information requiring manual processing, locked object identifiers, and unmet constraint identifiers.

[0029] The above are merely preferred embodiments of the present invention. It should be noted that those skilled in the art can make several improvements and modifications without departing from the principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for intelligent management of dairy products based on Internet of Things (IoT) technology, characterized in that: Executed by the interaction between the server and IoT nodes, including: The server establishes batch records for dairy products, binding each batch with its corresponding IoT node and its identifier, storage location and its identifier, target temperature zone and its identifier, and refrigeration actuator and its identifier. The IoT node communicates with the packaging surface temperature sensor, air temperature and humidity sensor, door sensor, and refrigeration actuator. The IoT node collects the packaging surface temperature, air temperature and humidity, door status, and actual output of the refrigeration actuator, and uploads the collected data to the server. The server calculates the data credibility of the collected data, and when the data credibility is lower than a preset threshold, it obtains a backup temperature, uses the backup temperature to correct the packaging surface temperature, and obtains the risk calculation temperature. The server calculates the temperature exposure based on the degree and duration of the temperature exceeding the upper limit of the target temperature zone, identifies external heating events, and determines the overall batch spoilage risk based on the temperature exposure and the external heating events. The server generates a batch status identifier, a cargo handling instruction, and a target output for the refrigeration actuator based on the overall spoilage risk of the batch, and sends them to the IoT node for control.

2. The intelligent management method for dairy products based on Internet of Things technology according to claim 1, characterized in that, The collected data also includes: light intensity, triaxial acceleration, position, compressor target output and actual output, fan speed and evaporator temperature; The server accumulates the temperature exposure based on the sampling time collected by the IoT node, and determines the communication delay based on the receiving time. The server identifies handling events based on the triaxial acceleration and verifies the location binding based on the position. For a batch that meets the preset redundancy monitoring conditions, bind at least two IoT nodes, use the temperature data of the redundant IoT nodes to verify the data of the same batch, and use the temperature data of the redundant IoT nodes as the backup temperature when the data of the IoT node is abnormal.

3. The intelligent management method for dairy products based on Internet of Things technology according to claim 1, characterized in that, Calculate the data reliability of the collected data, and when the data reliability is lower than a preset threshold, correct the packaging surface temperature using a backup temperature to obtain the risk calculation temperature, including: The following abnormal indicators are normalized: exceeding the range limit, exceeding the temperature change rate limit, temperature deviation between adjacent storage locations, inconsistency between door status and temperature and humidity changes, and power decay. The data reliability is calculated based on each of the abnormal indicators and their weights. Set a first confidence threshold and a second confidence threshold that is greater than the first confidence threshold; The packaging surface temperature collected by the IoT node to be evaluated is recorded as the target temperature; When the data credibility is less than the first credibility threshold, the backup temperature is obtained sequentially from the same batch node with credibility reaching the first credibility threshold, the adjacent storage location node, and the temperature response model, and the backup temperature is used as the risk calculation temperature. When the data reliability is greater than or equal to the first reliability threshold and less than the second reliability threshold, the backup temperature is sequentially obtained from nodes in the same batch with reliability reaching the first reliability threshold, adjacent storage location nodes, and the temperature response model. The product of the data reliability and the target temperature is then added to 1, and the sum of the product of the difference in data reliability and the backup temperature is subtracted from the product of these two values ​​to determine the risk calculation temperature. The corresponding calculation relationship is as follows: ;in, Calculate temperature for risk assessment. For data credibility, For the target temperature, This is for backup temperature; When the data credibility is greater than or equal to the second credibility threshold, the target temperature is determined as the risk calculation temperature.

4. The intelligent management method for dairy products based on Internet of Things technology according to claim 3, characterized in that, The temperature response model estimates the packaging surface temperature based on the air temperature, the actual output of the refrigeration actuator, and the thermal response parameters. The thermal response parameters are updated based on the temperature difference before and after control, the rate of temperature change, and the recovery time. When the duration of missing data exceeds the missing threshold, historical reliable temperature data is obtained as a reliable candidate temperature, and the higher of the highest value of the reliable candidate temperature and the estimated packaging surface temperature plus a preset safety margin set based on the target temperature zone is used as the backup temperature. When the backup temperature cannot be obtained, the missing interval is marked, and the data uncertainty is calculated based on the missing duration, the time-weighted average of the reliability of nodes in the same batch, and the number of node temperature conflicts. The data uncertainty is then used to correct the overall deterioration risk of the batch. A temperature conflict is recorded when the temperature difference between two nodes in the same batch continuously exceeds the conflict threshold and reaches a preset duration.

5. The intelligent management method for dairy products based on Internet of Things technology according to claim 1, characterized in that, Identify external heating events, including: Within the event determination window, when the door is opened and at least one of the following—the rate of change of air temperature, the amount of change of humidity, the air temperature difference between the first monitoring area near the door and the second monitoring area far from the door, and the amount of change of light—reaches the corresponding trigger threshold, a candidate external heating event is generated. When the door status data is missing, a candidate external heating event is generated when at least two of the air temperature change rate, humidity change, air temperature difference and light change reach the corresponding trigger threshold. During the observation period, when the increase in packaging surface temperature reaches the temperature rise confirmation threshold, the event temperature exposure corresponding to the degree of overheating and duration during the candidate external heating event reaches the exposure confirmation threshold, or the duration reaches the duration confirmation threshold, the external heating event is confirmed to have occurred. After the door is closed and a preset detection period has elapsed, if the calculated risk temperature is greater than or equal to the calculated risk temperature when the door is closed, or is still greater than or equal to the upper limit of the target temperature zone, then it is determined that the temperature recovery state has been entered. When the risk calculation temperature is continuously within the target temperature range, its rate of change does not exceed the stable rate threshold, and it remains stable for a preset duration, it is determined to exit the temperature recovery state.

6. The intelligent management method for dairy products based on Internet of Things technology according to claim 1, characterized in that, The server encapsulates the batch status identifier, the cargo handling instruction, and the target output of the refrigeration actuator into a data packet and sends it to the IoT node. The data packet sent out includes a status control table, threshold table, and control instructions applicable to the same batch, storage location, temperature zone, and refrigeration actuator. The packaging surface temperature sequence within a preset time domain is predicted using a preset temperature response model, and the minimum value is taken as the lowest predicted temperature. When the IoT node passes the field integrity verification of the sent data packet, the data packet version is higher than the current execution version, the data packet is within the valid time window, and the batch identifier, cargo location identifier, target temperature zone identifier, and refrigeration actuator identifier are consistent with the local record, the node verifies whether the target output of the refrigeration actuator is within the allowable output range, whether the minimum start-stop interval is met, whether the single-cycle output change does not exceed the maximum output change, and whether the minimum predicted temperature is not lower than the sum of the preset minimum allowable temperature and safety margin. If the verification passes, the status control table, the threshold table, and the control instructions are replaced in one go; if the verification fails, a pre-stored secure version is executed. When no secure version is available, the output will be limited to the preset secure output range and a local alarm will be generated.

7. The intelligent management method for dairy products based on Internet of Things technology according to claim 5, characterized in that, During the period when the communication between the IoT node and the server is interrupted, the data corresponding to the external heating event is divided into a baseline segment with a preset duration before the candidate event occurs, a main segment from the occurrence of the candidate event to the closing of the door, and a recovery segment from entering the temperature recovery state to exiting the temperature recovery state, and is used as an event data packet cache; The event data packet includes an event identifier, a batch identifier, a node identifier, a sampled data sequence, a confidence sequence, a state switching record, a control command record, and an execution receipt. The server confirms the validity of the data after receiving all three data segments completely. After communication is restored, the IoT node sequentially sends the current status, unresolved alarms, the most recently executed receipt of the current version, the event data packet with high priority, and the remaining historical data and historical receipts; Data of the same priority are sorted according to sampling time, node identifier, and local incrementing sequence number; The sequence period identifier formed after the local incrementing sequence number is reset or wrapped around is used as the cycle epoch. The server records the maximum consecutive sequence number, the maximum sampling time corresponding to the completely received data, and the range of sequence numbers to be retransmitted according to the node identifier and cycle epoch. The server determines that the data is complete when there are no missing sequence numbers in the range of sequence numbers to be retransmitted.

8. The intelligent management method for dairy products based on Internet of Things technology according to claim 4, characterized in that, The overall batch spoilage risk is determined based on the temperature exposure and the external heating event, including: The batch cumulative management cycle is taken as from warehousing to outbound, and the control management cycle is taken as a preset rolling window. The temperature exposure is calculated based on the temperature difference between the risk calculation at each sampling time and the temperature difference between adjacent sampling time intervals. The maximum temperature rise and the event temperature exposure of the external heating event are normalized and weighted according to preset weights to obtain the event intensity; The credibility of an event is determined based on the average credibility of nodes during the event period and the completeness of data on door status, rate of change of air temperature, amount of change of humidity, regional temperature difference, amount of change of light, and temperature rise of packaging surface. The basic risk is determined based on the number of valid events where the temperature exposure, event intensity, and event credibility all reach the corresponding effective threshold, the maximum event intensity, and the nominal remaining shelf life, using a preset basic risk mapping function. The basic risk mapping function is as follows: in, Basic risk, The number of valid events. To preset the event quantity threshold, For maximum event intensity, The nominal remaining shelf life, Total shelf life , , The basic risk weight coefficients, which are assigned by prior engineering, are adjusted upwards based on the uncertainty of the acquired data to obtain the comprehensive deterioration risk of the batch.

9. The intelligent management method for dairy products based on Internet of Things technology according to claim 8, characterized in that, Set a first risk threshold, a second risk threshold, and a third risk threshold that increase sequentially; When the overall risk of deterioration of the batch is less than the first risk threshold, it is determined to be low risk; when it is greater than or equal to the first risk threshold and less than the second risk threshold, it is determined to be medium risk; when it is greater than or equal to the second risk threshold and less than the third risk threshold, it is determined to be relatively high risk; when it is greater than or equal to the third risk threshold, it is determined to be high risk. The sampling period, sending period, buffer priority, alarm flag, and batch status identifier are determined according to the risk level. For medium-risk batches or batches whose risk increment reaches the increment threshold, select a sampling period one level shorter than the current level from the sampling period table classified by period duration, and increase the transmission priority. For higher-risk batches, a sampling sequence of a preset time period is retained before and after the comprehensive deterioration risk first reaches the second risk threshold, and a priority outbound identifier or a cargo location handling candidate instruction is generated; High-risk batches generate high-risk alarms and set the batch status to "disable automatic release".

10. The intelligent management method for dairy products based on Internet of Things technology according to claim 9, characterized in that, The server generates warehouse handling instructions and refrigeration actuator target outputs, including: The control and management cycle is used as the evaluation cycle for the cargo location. The degree of heat exposure of the cargo location is determined based on the air temperature exposure of the cargo location and the number of external heat events. The degree of delivery urgency is determined based on the remaining time from the current time to the end of the delivery time window. The priority of cargo handling, the priority of outbound delivery, and the priority of cold air allocation are determined based on the risk level, the heat level of the cargo location, the nominal remaining shelf life level, and the urgency level of delivery. When multiple temperature zones share a compressor, and the compressor capacity is sufficient and there are no high-priority alarms, the server obtains a pre-set nominal space heat capacity parameter characterizing the spatial heat capacity of each temperature zone, and calculates the cooling demand of each temperature zone. The cooling demand is equal to the product of the positive temperature difference between the upper limit of the target temperature zone and the lowest predicted temperature obtained using a preset temperature response model, the nominal space heat capacity parameter of the target temperature zone, and the reciprocal of the control cycle. The calculation formula is as follows: in, For cooling needs, This corresponds to the positive temperature difference between the upper limit of the target temperature zone and the lowest predicted temperature. The nominal space heat capacity parameter for this temperature range, To control the cycle, the target output of the refrigeration actuator is determined based on the weighted sum of the refrigeration demands of each temperature zone. When there is insufficient capacity or a high-priority alarm, the target output of the refrigeration actuator is determined according to the demand of the highest priority temperature zone, and the refrigeration capacity is reduced sequentially from the low-priority temperature zone until the total capacity constraint of the compressor is met. When the lowest predicted temperature is lower than the sum of the preset minimum allowable temperature and the safety margin, or when the target output of the refrigeration actuator exceeds the allowable output range, an alarm message requiring manual processing, a locked object identifier, and a constraint not met identifier are generated.

Images