Food safety system for controlling transmission by dynamic message queue and composite encryption network

Abstract

The invention relates to the technical field of food safety, and discloses a food safety system of dynamic message queue and composite encryption network control transmission, which comprises the following steps: acquiring multi-source monitoring data in a transport carriage, and carrying out risk weight configuration on the multi-source monitoring data to obtain a data risk weight; dividing transmission priorities of the multi-source monitoring data according to the data risk weights, and monitoring data stream features of priority queues; performing queue throughput rate adjustment on the multi-source monitoring data according to the monitoring result and the transmission priority to obtain an adjusted structured data set; performing composite encryption on the adjusted structured data set, and performing composite decryption on the encrypted data to obtain decrypted structured data; and performing abnormal probability analysis on the temperature data in the decrypted structured data to obtain a cold chain abnormal probability. According to the invention, double-flow detection is carried out through the image data and the cold chain abnormal probability, and the comprehensiveness and accuracy of food safety detection are improved.

Description

Technical Field

[0001] This invention relates to the field of food safety technology, and in particular to a food safety system that uses dynamic message queues and composite encrypted network control transmission. Background Technology

[0002] With the development of society and the economy, more and more people are beginning to pay attention to health, especially the safety of food they consume.

[0003] Currently, in the cold chain logistics field, food safety monitoring relies on the real-time perception and analysis of transportation environmental parameters and cargo status. However, in traditional methods, the monitoring of environmental parameters and cargo status are mostly independent systems, with data transmission and analysis links separated. For example, temperature anomaly warnings only indicate environmental risks and cannot directly link them to the specific location or degree of cargo spoilage. Visual inspection relies solely on appearance judgment and does not combine the temporal characteristics of temperature fluctuations, resulting in a break in the dynamic correlation between "environmental risk" and "cargo spoilage," making accurate warnings difficult. Meanwhile, environmental and image data are mostly transmitted without differentiation, but high-priority data may be delayed due to network congestion; the lack of single or complete encryption methods can easily lead to the leakage of sensitive data and pose information security risks.

[0004] Therefore, data fragmentation, delayed early warning, and insecure transmission in existing technologies have become urgent problems to be solved. Summary of the Invention

[0005] This invention provides a food safety system with dynamic message queues and composite encrypted network control transmission, the main purpose of which is to solve the problems of insufficient comprehensiveness and accuracy in food safety testing.

[0006] To achieve the above objectives, the present invention provides a food safety system for dynamic message queues and composite encrypted network control transmission. The system includes a configuration acquisition module, a monitoring module, an adjustment module, an encryption / decryption module, an analysis module, and a detection module, wherein: The configuration acquisition module is used to acquire multi-source monitoring data inside the transport vehicle, configure risk weights for the multi-source monitoring data, and obtain data risk weights. The partitioning monitoring module is used to partition the transmission priority of the multi-source monitoring data according to the data risk weight, and monitor the data flow characteristics of the priority queue; The adjustment module is used to adjust the queue throughput of the multi-source monitoring data according to the monitoring results and the transmission priority, so as to obtain an adjusted structured dataset. The encryption and decryption module is used to perform compound encryption on the adjusted structured dataset and compound decryption on the encrypted data to obtain decrypted structured data. The analysis module is used to perform anomaly probability analysis on the temperature data in the decrypted structured data to obtain the cold chain anomaly probability. The detection module is used to perform dual-stream detection on the image data in the decrypted structured data and the cold chain anomaly probability to obtain the deterioration area detection result.

[0007] Optionally, when the configuration acquisition module performs risk weight configuration on the multi-source monitoring data to obtain data risk weights, it includes: A multidimensional risk factor matrix is ​​constructed based on the multi-source monitoring data, and ineffective factors in the multidimensional risk factor matrix are filtered out to obtain an initial factor data set; The dispersion of each factor in the initial factor dataset is determined by using preset historical data, and the initial dynamic weight of each factor is calculated based on the dispersion. The regional risk characteristics in transportation are obtained, and the perishable weight of goods in the initial dynamic weight is adjusted according to the regional risk characteristics to obtain the adjustment factor weight. Data risk weights are generated based on the adjusted factor weights and the multi-source monitoring data.

[0008] Optionally, when the monitoring division module performs the task of prioritizing the transmission of the multi-source monitoring data according to the data risk weight, it includes: Based on the data risk weights, the multi-source monitoring data are initially sorted by data priority to obtain a basic transmission priority queue. Dynamic transmission conditions are configured for the secondary priority data in the basic transmission priority queue to obtain the dynamic transmission conditions for the secondary priority data. Based on the dynamic transmission conditions, the packet loss rate and latency gradient changes of the basic transmission priority queue are monitored in real time. The transmission priority of the multi-source monitoring data is obtained by dynamically adjusting the data priority of the basic transmission priority queue based on the packet loss rate and the delay gradient change.

[0009] Optionally, when the adjustment module performs queue throughput adjustment on the multi-source monitoring data based on the monitoring results and the transmission priority to obtain an adjusted structured dataset, it includes: Based on the monitoring results and the transmission priority, the multi-source monitoring data is allocated a basic throughput to obtain an initial basic throughput. Obtain the round-trip time of the data transmission link probe packet, and calculate the network quality index based on the round-trip time; Dynamic throughput adjustment conditions are constructed based on the initial base throughput, network quality index, and data risk weight; The multi-source monitoring data is subjected to queue throughput adjustment based on the aforementioned throughput dynamic adjustment conditions to obtain an adjusted structured dataset.

[0010] Optionally, when the encryption / decryption module performs compound decryption on the encrypted data to obtain decrypted structured data, it includes: The encrypted data is decrypted at the hardware layer to obtain the hardware-decrypted data. The hardware-layer decrypted data is then decrypted at the transport layer to obtain hardware-transmitted decrypted data. The hardware-transmitted decrypted data is decrypted at the application layer, and the decryption process is subjected to decryption circuit breaker authentication. Decrypted structured data is generated based on the application layer decryption results and authentication results.

[0011] Optionally, when the analysis module performs anomaly probability analysis on the temperature data in the decrypted structured data to obtain the cold chain anomaly probability, it includes: Fluctuation characteristics of temperature data in the decrypted structured data are analyzed to obtain temperature data fluctuation characteristics. Based on the temperature data fluctuation characteristics, abrupt signal filtering is performed on the temperature data to obtain the frequency domain abrupt characteristics of the temperature data. Extract vehicle vibration intensity data from the decrypted structured data; A comprehensive risk assessment dimension is constructed using the vehicle vibration intensity data and the frequency domain mutation characteristics. The probability of cold chain anomalies is calculated based on the comprehensive risk assessment dimensions.

[0012] Optionally, when the analysis module calculates the probability of cold chain anomalies based on the comprehensive risk assessment dimensions, it includes: Determine the core indicators for the comprehensive risk assessment dimensions; Feature values ​​are extracted from the core indicators to obtain indicator feature values; Establish a mapping rule between feature intervals and anomaly probabilities based on the pre-defined historical feature anomaly correspondence; Based on the mapping rule, the anomaly probability of the indicator feature value is obtained by performing anomaly probability matching on the indicator feature value. The cold chain anomaly probability is generated based on the anomaly probability of the aforementioned indicators and the risk weights of the core indicators.

[0013] Optionally, when the detection module performs dual-stream detection of the image data in the decrypted structured data and the cold chain anomaly probability to obtain the deteriorated region detection result, it includes: The cold chain anomaly probability is converted into a spatial heat map; The original features of the image data in the decrypted structured data are extracted to obtain the original image features; The original features of the image are weighted and enhanced based on the spatial heatmap to obtain enhanced features; Based on the enhanced features, the boundaries of the altered regions and the alteration confidence level are located, and the boundaries of the altered regions and the alteration confidence level are mapped to the probability distribution of the spatial heat map; The detection results for the deteriorated areas are determined based on the corresponding results.

[0014] Optionally, when the detection module converts the cold chain anomaly probability into a spatial heatmap, it includes: Establish a mapping relationship between the coordinates of the temperature sensors inside the transport vehicle and the image data in the decrypted structured data; The cold chain anomaly probability is loaded into the target pixel region of the carriage image according to the mapping relationship, resulting in a discrete set of anomaly probability pixels. Spatial interpolation is performed on the probability values ​​of adjacent pixels in the set of abnormal probability pixels to obtain a continuous probability distribution; Extract the texture features from the image data, and correct the continuous probability distribution based on the texture features to obtain a spatial heatmap.

[0015] Optionally, when the detection module performs the correction of the continuous probability distribution based on the texture features to obtain a spatial heatmap, it includes: Determine the physical region corresponding to the texture feature; The influence of the temperature distribution is analyzed based on the texture features and the physical region, and the regional correction rule of the continuous probability distribution is determined based on the influence rule. The continuous probability distribution is adjusted according to the regional correction rule to obtain a spatial heatmap.

[0016] This invention acquires multi-source monitoring data and configures risk weights to clarify the risk impact of different data, providing a basis for subsequent priority allocation and highlighting the importance of key risk data. High-priority data (such as over-temperature alarms) is transmitted first, shortening the response time of critical information and preventing risk spread due to transmission delays. Monitoring data stream characteristics can detect congestion, interruptions, and other anomalies in real time, providing a basis for timely adjustments to transmission strategies and ensuring the stability of the data link. By dynamically adjusting throughput, the transmission channel for high-priority data can be prioritized during data surges, avoiding overall congestion. Composite encryption combines multiple encryption algorithms, which, compared to single encryption... The encryption is more difficult to crack, resisting diverse information security threats; the standardized decryption process ensures that data is completely restored after transmission, guaranteeing security without affecting the normal use of data; through probability quantification, ambiguous temperature anomalies can be transformed into decision-making values, facilitating the setting of early warning thresholds (such as triggering an alarm when the probability exceeds 70%), achieving refined risk management, and reducing false alarms or missed alarms; dual-flow detection image data and cold chain anomaly probability processing provide a visual representation of the cargo's condition, and combined with anomaly probabilities, cross-verification of spoilage risks can be performed. Even if the temperature is temporarily normal, if the image shows signs of spoilage, the comprehensiveness and accuracy of detection are greatly improved. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments of the present invention will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 A schematic diagram of a food safety system for dynamic message queues and composite encrypted network control transmission provided in an embodiment of the present invention; Figure 2 This is a schematic diagram of multi-source monitoring data provided in an embodiment of the present invention; Figure 3 This is a schematic diagram of a risk factor matrix provided in an embodiment of the present invention; Figure 4 This is a schematic diagram of an initial factor data set provided in an embodiment of the present invention; Figure 5 This is a schematic diagram illustrating the numerical range → probability correspondence rule provided in an embodiment of the present invention.

[0019] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0020] To enable those skilled in the art to better understand the technical solutions of this disclosure, and to fully understand and implement the process of how this disclosure applies technical means to solve technical problems and achieve corresponding technical effects, the technical solutions in the embodiments of this disclosure will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this disclosure, not all embodiments. The embodiments of this disclosure and the various features within them can be combined with each other without conflict, and the resulting technical solutions are all within the protection scope of this disclosure. All other embodiments obtained by those skilled in the art based on the embodiments of this disclosure without creative effort should fall within the protection scope of this disclosure.

[0021] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this disclosure are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this disclosure described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0022] This application provides a food safety system using dynamic message queuing and composite encrypted network control transmission. This system can be executed by software or hardware installed on terminal or server devices. The server includes, but is not limited to, a single server, a server cluster, a cloud server, or a cloud server cluster. The server can be a standalone server or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDNs), and big data and artificial intelligence platforms.

[0023] Reference Figure 1 The diagram shown is a functional block diagram of a food safety system with dynamic message queue and composite encrypted network control transmission provided in an embodiment of the present invention.

[0024] The food safety system 100 with dynamic message queue and composite encrypted network control transmission described in this invention can be installed in an electronic device. Depending on the functions implemented, the food safety system 100 with dynamic message queue and composite encrypted network control transmission may include a configuration acquisition module 101, a monitoring module 102, an adjustment module 103, an encryption / decryption module 104, an analysis module 105, and a detection module 106. The module described in this invention can also be called a unit, which refers to a series of computer program segments that can be executed by the processor of an electronic device and can perform a fixed function, and which are stored in the memory of the electronic device.

[0025] In this embodiment, the functions of each module / unit are as follows: The acquisition configuration module 101 is used to acquire multi-source monitoring data inside the transport vehicle, configure risk weights for the multi-source monitoring data, and obtain data risk weights.

[0026] In this embodiment of the invention, the acquisition refers to the real-time or periodic collection of various data reflecting the environment inside the transport vehicle, the status of the goods, and the operation of the equipment, such as temperature values, humidity values, vibration frequency, presence of smoke, and changes in the position of the goods, through various monitoring devices installed inside the transport vehicle, such as temperature sensors, humidity sensors, vibration detectors, smoke detectors, and cameras. The risk weight configuration refers to assigning a corresponding importance weight to each type of data based on the degree of correlation between different types of monitoring data and transportation risks (e.g., abnormal temperature has a greater impact on the risk of cold chain goods, so the weight value is high; slight vibration has a smaller impact on the risk of ordinary goods, so the weight value is low), thereby quantifying the contribution of various types of data in risk assessment.

[0027] Specifically, various types of data (such as environmental, cargo, and equipment-related data) are first collected through multiple monitoring devices inside the carriage. Then, based on the degree of impact of these data on transportation safety (such as cargo damage, transportation accidents, etc.), different risk weights (importance scores) are assigned to each type of data. Finally, the risk weight values ​​corresponding to each type of data are obtained, providing a quantitative basis for subsequent comprehensive risk assessment.

[0028] In this embodiment of the invention, when the configuration acquisition module performs risk weight configuration on the multi-source monitoring data to obtain data risk weights, it includes: A multidimensional risk factor matrix is ​​constructed based on the multi-source monitoring data, and ineffective factors in the multidimensional risk factor matrix are filtered out to obtain an initial factor data set; The dispersion of each factor in the initial factor dataset is determined by using preset historical data, and the initial dynamic weight of each factor is calculated based on the dispersion. The regional risk characteristics in transportation are obtained, and the perishable weight of goods in the initial dynamic weight is adjusted according to the regional risk characteristics to obtain the adjustment factor weight. Data risk weights are generated based on the adjusted factor weights and the multi-source monitoring data.

[0029] Specifically, assuming the multi-source monitoring data inside the transport vehicle is as follows: Figure 2 As shown: Risk factor types: environmental factors, cargo condition factors, and equipment safety factors; The specific environmental factors are temperature and humidity. The real-time temperature monitoring value is 25℃, and the data validity is valid. The real-time humidity monitoring value is --- (equipment malfunction, no value), and the data validity is invalid. The specific environmental factors are vibration frequency and tilt angle. The real-time vibration frequency monitoring value is 0.8Hz, and the data validity is valid. The real-time vibration frequency monitoring value is 3°, and the data validity is valid. The specific equipment safety factors are the carriage door lock status and refrigeration system operating parameters. The real-time monitoring value of the carriage door lock status is 1 (1 = locked / 0 = unlocked), and the data validity is valid. The real-time monitoring value of the refrigeration system operating parameters is 220V (voltage), and the data validity is valid.

[0030] like Figure 3 As shown: A risk factor matrix is ​​constructed with factor type as the row and real-time monitoring value as the column. The factor types are: temperature, humidity, vibration frequency, tilt angle, carriage door lock status, and refrigeration system operating parameters. The real-time monitoring values ​​corresponding to each factor are: 25℃, invalid value, 0.8Hz, 3°, 1, and 220V, respectively.

[0031] like Figure 4 As shown: The invalid values ​​in the multi-dimensional risk factor matrix constructed above are deleted, such as the humidity value. At the same time, the unit symbol is removed, and only the numerical value is retained. The data is then classified according to the factor type to form a structured dataset, i.e., the initial factor dataset. The factors in the set are of the following types: environmental factors, cargo status factors, and equipment safety factors. The environmental factor is named temperature, and the standardized monitoring value for temperature is 25. The factors corresponding to the cargo status factors are vibration frequency and tilt angle. The standardized monitoring value of vibration frequency is 0.8, and the standardized monitoring value of tilt angle is 3. The factors corresponding to the equipment safety factors are the carriage door lock status and the refrigeration system operating parameters. The standardized monitoring value for the carriage door lock status is 1, and the standardized monitoring value for the refrigeration system operating parameters is 220.

[0032] Furthermore, the dispersion of each factor in the initial factor dataset is determined by using preset historical data. For example, historical data on the transportation of similar foods over the past three months are used to statistically analyze the fluctuation range of each factor: Temperature: safe range 4-10℃, historical data shows that 60% of the events caused by abnormal temperature led to spoilage of goods, with a standard deviation of 3.2 (large fluctuation); Vibration frequency: safe range ≤1Hz, historical data shows that 10% of the events caused by vibration led to damage of goods, with a standard deviation of 0.3 (small fluctuation); Refrigeration voltage: safe range 210-230V, historical data shows that 10% of the events caused by abnormal voltage led to refrigeration failure, with a standard deviation of 0.2 (small fluctuation); Thus, the dispersion is ranked as follows: Temperature (3.2) > Vibration frequency (0.3) > Refrigeration voltage (0.2).

[0033] The safety interval is set based on the approximate results obtained from summarizing historical data. The standard deviation is calculated using the classic standard deviation formula in statistics, which is a mature statistical analysis technique and will not be elaborated on here.

[0034] Furthermore, the initial dynamic weight is assigned based on the degree of dispersion using the entropy method; for example, if the dispersion of temperature is 3.2, the initial dynamic weight of temperature is assigned to 0.4 using the entropy method; if the dispersion of vibration frequency is 0.3, the initial dynamic weight of temperature is assigned to 0.15. Similarly, the entropy method is existing technology and will not be elaborated here.

[0035] Furthermore, the latitude and longitude coordinates of the transport vehicle are obtained in real time through positioning systems such as GPS and Beidou. Combined with electronic map data, the current administrative region (such as XX province XX city), terrain features (such as mountainous areas, plains, coastal areas), and climate zones (such as tropical, temperate, and frigid zones) are analyzed. When passing through tropical regions, the regional characteristics related to the "perishability of goods" are extracted. For example, the regional characteristics of tropical regions are high temperature and high humidity, with an average daily temperature of 32°C and humidity of 85%.

[0036] For perishable goods currently being transported, sensitive risk factors are preset. For example, the sensitive factors for chilled meat are temperature, humidity, and transport bumps, with an initial weight of 0.3 for temperature and 0.2 for humidity. Based on the correlation between regional risk characteristics and the sensitive factors of the goods, adjustment rules are preset. These adjustment rules are pre-configured and stored, and support dynamic updates. For example, when the regional risk characteristic is temperature > 30°C, the corresponding sensitive factor is temperature, with an adjustment rule of weight × 1.5, and the adjustment rule for humidity is weight × 1.2. When transporting chilled meat through a tropical coastal area, where the temperature is 32°C and the humidity is 85%, the adjusted weight for the temperature factor is 0.3 × 1.5 = 0.45, and the adjusted weight for the humidity factor is 0.2 × 1.2 = 0.24. Finally, the weight values ​​of all factors are integrated to form the final adjustment factor weights.

[0037] Furthermore, the adjusted factor weights are correlated with the real-time multi-source monitoring data for calculation. For example, temperature: 32℃ × 0.45 = 14.4, humidity: 85% × 0.24 = 20.4, to obtain the risk contribution value of a single factor. The risk contribution values ​​of all factors are integrated, and the formula is expressed as: 14.4 + 20.4 = 34.8. After normalization, the data risk weight is 34.

[0038] In this embodiment of the invention, a multi-dimensional risk factor matrix is ​​constructed to transform scattered multi-source monitoring data into a structured risk factor system, avoiding data fragmentation and omissions, and ensuring that all core indicators related to transportation risks are fully included in the assessment scope. The perishable weight of goods is adjusted in combination with regional risk characteristics, breaking through the "one-size-fits-all" weight model and making risk assessment adaptable to real-time transportation scenarios. By combining the adjusted weights with real-time monitoring data, the abstract "risk" is transformed into a quantifiable weight value, providing a clear quantitative basis for subsequent risk warning and decision-making.

[0039] The partitioning monitoring module 102 is used to partition the transmission priority of the multi-source monitoring data according to the data risk weight, and monitor the data flow characteristics of the priority queue.

[0040] In this embodiment of the invention, based on the generated "data risk weight," the multi-source monitoring data (such as temperature, vibration, smoke, etc.) collected in the transport compartment are prioritized for transmission. That is, data with higher risk weight (such as real-time temperature data of perishable goods) is assigned a higher transmission priority and must be sent first; data with lower risk weight (such as light data in the compartment) has a lower priority and is sent one at a time. At the same time, the system monitors the status characteristics of the data streams (i.e., priority queues) arranged according to this priority in real time, such as data transmission speed, whether there is congestion, and whether high-priority data is delayed, to ensure that high-risk related data can be transmitted in a timely manner, providing a guarantee for risk warning.

[0041] In this embodiment of the invention, when the monitoring division module performs the task of prioritizing the transmission of the multi-source monitoring data according to the data risk weight, it includes: Based on the data risk weights, the multi-source monitoring data are initially sorted by data priority to obtain a basic transmission priority queue. Dynamic transmission conditions are configured for the secondary priority data in the basic transmission priority queue to obtain the dynamic transmission conditions for the secondary priority data. Based on the dynamic transmission conditions, the packet loss rate and latency gradient changes of the basic transmission priority queue are monitored in real time. The transmission priority of the multi-source monitoring data is obtained by dynamically adjusting the data priority of the basic transmission priority queue based on the packet loss rate and the delay gradient change.

[0042] In detail, based on the generated data risk weights, the multi-source monitoring data (such as temperature, vibration, humidity, equipment status, etc.) are initially prioritized. Specifically, the data risk weight values ​​are sorted from high to low, and the corresponding data transmission priorities are arranged from high to low in turn. For example, the "real-time temperature data of cold chain goods" with the highest risk weight is listed as the highest priority and must be transmitted first; the "cargo tilt angle data" with the second highest risk weight is listed as the medium priority; and the "lighting data inside the carriage" with the lowest risk weight is listed as the second priority. Finally, a "basic transmission priority queue" is formed according to a fixed initial order, which clarifies the initial transmission order of each data.

[0043] Furthermore, for "secondary priority data" with lower priority in the basic queue, such as non-critical environmental parameters and routine equipment status data, dynamic triggering conditions are set for their transmission, rather than fixed real-time transmission. When setting threshold triggering conditions, transmission is triggered when the monitored value of secondary priority data exceeds a preset safety range. For example, when the light intensity inside the carriage suddenly increases abnormally, the originally low-priority light data is transmitted immediately. When setting dynamic adjustment conditions for time intervals, the transmission frequency is set according to the overall data transmission pressure, such as once every 5 minutes when the network is smooth and extended to 10 minutes when the network is congested. When setting triggering conditions associated with high-priority data, when high-priority data is abnormal, such as temperature exceeding the standard, the transmission of related secondary priority data (such as humidity) is triggered simultaneously to assist in risk assessment. These conditions are solidified into rules, forming "dynamic transmission conditions for secondary priority data".

[0044] Furthermore, based on the dynamic transmission conditions of secondary priority data, the system tracks the transmission status of the entire basic transmission priority queue in real time, focusing on monitoring two core indicators: packet loss rate and latency gradient change. Packet loss rate refers to the proportion of high, medium, and secondary priority data lost during transmission within a unit of time (e.g., whether high-priority temperature data is lost due to network problems). Latency gradient change refers to recording the time delay from acquisition to reception of different priority data and analyzing the trend of latency change (e.g., whether the latency of high-priority data gradually increases, and whether the latency of secondary priority data exceeds the threshold). During the monitoring process, the real-time data is compared with the preset normal range (e.g., the high-priority packet loss rate should be ≤1%, and the latency should be ≤1 second) to determine whether there is any abnormality in the current transmission.

[0045] Furthermore, based on the changes in packet loss rate and latency gradient detected in the previous step, the basic transmission priority queue is optimized in real time. If the packet loss rate of high-priority data is too high or the latency gradient continues to rise (e.g., temperature data latency exceeds 3 seconds), its priority is further increased, and even the transmission resources of secondary-priority data are temporarily occupied, such as suspending the transmission of some secondary-priority data to prioritize the high-priority data. If secondary-priority data still cannot be transmitted due to latency after dynamic transmission conditions are triggered (e.g., abnormal lighting), its priority is temporarily increased to ensure that no critical abnormal information is missed. If the overall transmission status returns to normal (packet loss rate and latency return to a safe range), the priority is reverted to the basic queue status. After dynamic adjustment, a "transmission priority of multi-source monitoring data" adapted to the current transmission environment is finally formed.

[0046] In this embodiment of the invention, the importance of data is clearly defined by risk weights, ensuring that high-risk, high-value data is transmitted first, avoiding the squeezing of resources by low-priority data for important information, and laying a reasonable foundation for subsequent transmission strategies. By tracking key indicators (packet loss rate, latency) in real time during the transmission process, changes in the network environment and data transmission status are promptly grasped, providing an objective basis for priority adjustment and avoiding the rigidity of transmission strategies caused by relying on static rules.

[0047] The adjustment module 103 is used to adjust the queue throughput of the multi-source monitoring data according to the monitoring results and the transmission priority, so as to obtain an adjusted structured dataset.

[0048] In this embodiment of the invention, the queue processing efficiency (throughput) of multi-source monitoring data is dynamically adjusted by combining the monitoring results of data transmission status (such as packet loss rate, latency, etc.) and the determined transmission priority, and finally a dataset with a regular structure and transmission efficiency adapted to the current state is formed.

[0049] Specifically, taking multi-source monitoring data of cold chain transportation as an example, its transmission priority is: temperature > humidity > vibration > light intensity; the transmission priority queue is: temperature (high), humidity (medium), vibration (secondary), light intensity (low); monitoring results: network congestion, high-priority temperature data throughput decreases (only 30 data points / second, normally ≥50 data points / second), and secondary-priority vibration data causes queue blockage due to frequent transmission; the processing speed of temperature data is increased from 30 data points / second to 60 data points / second (occupying more bandwidth), vibration data is reduced from 10 data points / second to 5 data points / second, and light intensity data is reduced from 8 data points / second to 2 data points / second (reducing the bandwidth occupation of non-critical data).

[0050] In this embodiment of the invention, when the adjustment module performs queue throughput adjustment on the multi-source monitoring data based on the monitoring results and the transmission priority to obtain an adjusted structured dataset, it includes: Based on the monitoring results and the transmission priority, the multi-source monitoring data is allocated a basic throughput to obtain an initial basic throughput. Obtain the round-trip time of the data transmission link probe packet, and calculate the network quality index based on the round-trip time; Dynamic throughput adjustment conditions are constructed based on the initial base throughput, network quality index, and data risk weight; The multi-source monitoring data is subjected to queue throughput adjustment based on the aforementioned throughput dynamic adjustment conditions to obtain an adjusted structured dataset.

[0051] In this embodiment of the invention, the initial throughput rate is allocated to data of different priorities by combining the real-time monitoring results of data transmission and the preset transmission priority.

[0052] For example, if the current total network bandwidth is detected to be 100 packets / second, and the transmission priority is "temperature (high) > humidity (medium) > vibration (secondary) > illumination (low)", the allocation rule is: high priority accounts for 40%, medium priority accounts for 30%, secondary priority accounts for 20%, and low priority accounts for 10%. The initial base throughput is: temperature: 100 × 40% = 40 packets / second; humidity: 100 × 30% = 30 packets / second; vibration: 100 × 20% = 20 packets / second; illumination: 100 × 10% = 10 packets / second.

[0053] The allocation rules are formulated through real-time algorithm optimization. For example, if the system detects that the humidity suddenly exceeds the safe range, it can temporarily increase the bandwidth ratio for humidity (e.g., from 30% to 40%) to ensure more timely data transmission, and then reduce it back down after the humidity returns to normal.

[0054] When obtaining the round-trip time of the data transmission link probe packet, if a probe packet is sent to the data transmission link (such as the network from the terminal of the transport vehicle to the cloud platform) and the measured round-trip time is 20ms, the preset maximum allowable RTT is 100ms, the network quality index = 100 - (20 / 100) × 100 = 80 (good network quality).

[0055] Furthermore, by integrating three core parameters (initial base throughput, network quality index, and data risk weight), dynamic adjustment rules for throughput are formulated. The logic of the rules is as follows: when the network quality index is high (good network), the throughput of each priority level can be appropriately increased (especially high-risk weight data); when the network quality index is low (poor network), the throughput of high-risk weight data is prioritized, while the throughput of low-risk data is reduced; the higher the data risk weight, the higher the "protection priority" during adjustment (i.e., resources are reduced last when the network is poor).

[0056] For example, given the initial base throughput: 40 data points / second for temperature, 30 data points / second for humidity, and a network quality index of 80 (good), and data risk weights: 0.6 for temperature (high) and 0.3 for humidity (medium); the adjustment conditions are as follows: if the network quality index is ≥70 (good), the throughput of high-risk data (temperature) is multiplied by 1.2, and that of medium-risk data (humidity) is multiplied by 1.1; if the network quality index is <50 (poor), the throughput of high-risk data is multiplied by 0.9 (slight reduction), and that of medium-risk data is multiplied by 0.7 (significant reduction).

[0057] Furthermore, based on the aforementioned dynamic adjustment conditions, and combined with the current network quality index and data risk weights, the throughput of each priority data is adjusted in real time. The adjusted data is stored in a structured format (e.g., timestamp + value + priority) to form an adjusted structured dataset. For example, if the current network quality index is 80 (meeting the "≥70" condition), the following adjustments are made according to the rules: Temperature (high risk): 40 × 1.2 = 48 data points / second; Humidity (medium risk): 30 × 1.1 = 33 data points / second; Vibration (secondary risk): 20 × 1.1 = 22 data points / second; Illumination (low risk): 10 × 1.0 = 10 data points / second, resulting in an adjusted structured dataset {Time: 10:01; Temperature: 32℃; Throughput: 48; Priority: High}.

[0058] In this embodiment of the invention, the basic throughput is allocated based on monitoring results and transmission priority, ensuring that high-priority and high-importance data receive initial resource allocation, avoiding delays in critical data due to insufficient resources, and guaranteeing the basic needs of core data transmission. By calculating the network quality index through round-trip latency, the current network status (such as congestion and stability) can be quantitatively assessed, providing an objective basis for subsequent throughput adjustments and avoiding blind resource allocation. Adjusting the queue throughput according to dynamic conditions allows for flexible optimization of resource allocation when the network fluctuates or data risks change (such as prioritizing high-risk data when the network deteriorates), ultimately forming a well-structured and highly efficient dataset, improving the overall reliability and efficiency of data transmission.

[0059] The encryption / decryption module 104 is used to perform compound encryption on the adjusted structured dataset and compound decryption on the encrypted data to obtain decrypted structured data.

[0060] In this embodiment of the invention, the structured dataset formed after throughput adjustment is first subjected to "composite encryption" processing using a combination of multiple encryption methods (such as simultaneously using a symmetric encryption algorithm to encrypt the data content and an asymmetric encryption algorithm to encrypt the symmetric key) to ensure the security of the data during transmission or storage. Then, when the data needs to be used, it is subjected to "composite decryption" processing using a combination of corresponding decryption methods (such as first using the asymmetric key to decrypt to obtain the symmetric key, and then using the symmetric key to decrypt the data content), and finally the decrypted structured data that can be used directly and has a complete structure is restored.

[0061] In this embodiment of the invention, when the encryption / decryption module performs compound decryption on the encrypted data to obtain decrypted structured data, it includes: The encrypted data is decrypted at the hardware layer to obtain the hardware-decrypted data. The hardware-layer decrypted data is then decrypted at the transport layer to obtain hardware-transmitted decrypted data. The hardware-transmitted decrypted data is decrypted at the application layer, and the decryption process is subjected to decryption circuit breaker authentication. Decrypted structured data is generated based on the application layer decryption results and authentication results.

[0062] In this embodiment of the invention, the encrypted data is first decrypted by a dedicated hardware device (such as a dongle or chip-level decryption module). The hardware decryption relies on the physical key built into the device (such as a 16-bit key embedded in the chip). Only matching hardware can unlock the data, preventing it from being illegally read at the hardware level. For example, the encrypted data is #2Lk9$%. The hardware key (built into the device) is 1234-ABCD-5678-EFGH (written into the hardware at the factory and cannot be modified). The decryption rule is defined as follows: after the hardware verifies the key match, the hardware encryption identifier (such as #) in the data header is automatically removed and some garbled characters are restored. The hardware decryption result is 2Lk9$.

[0063] Furthermore, the data decrypted at the hardware layer undergoes transport layer encryption (such as SSL / TLS encryption) during network transmission. Decryption requires a transport layer key (such as a session key). This transport layer key is negotiated and generated by the sender and receiver before transmission (dynamically generated via a "handshake protocol," and is different for each transmission). For example, the hardware-layer decrypted data is: 2Lk9$%, and the transport layer key (dynamically negotiated and generated) is: 7890-XYZA-1234-BCDE (dedicated to this transmission, expires upon expiration). The decryption rule is: use the transport layer key to remove transport layer encryption identifiers (such as '&') and restore the intermediate garbled text. This rule is predefined by the transmission protocol (ensuring consistency between the sender and receiver). The transport layer decryption result is: Lk9 temperature.

[0064] Furthermore, the data decrypted at the transport layer is finally encrypted at the application layer (such as custom encryption by the software system to protect specific business data), and needs to be decrypted using an application layer key (such as a key derived from the system login password). Simultaneously, a "circuit breaker authentication" is triggered during decryption—if the number of decryption failures exceeds a threshold (e.g., 3 times), the data is automatically locked (circuit breaker) to prevent brute-force attacks. For example, hardware transmission decrypted data: Lk9 temperature, application layer key (generated after user login): user_2023_temp (bound to user permissions, different users have different keys), decryption rule: remove the business identifier (e.g., "Lk9") using the application layer key to restore the complete business data; simultaneously, an authentication threshold of "maximum 3 decryption failures" is set (the rule is defined by the system security policy to prevent malicious attempts). Assuming this decryption is successful (1 attempt), the application layer decryption result is: temperature = 8℃, time = 10:00 (complete business data), and the circuit breaker authentication result is: authentication passed (circuit breaker not triggered).

[0065] Furthermore, combining the application layer decryption result (complete business data) and the circuit breaker authentication result (whether the security verification passed), the data is organized into a structured format (such as tables or JSON). If authentication fails (e.g., triggering a circuit breaker), "decryption failed" is returned; if authentication succeeds, directly usable structured data is output. For example, application layer decryption result: temperature = 8℃, time = 10:00; circuit breaker authentication result: authentication passed; generation rules: format the data according to business requirements (e.g., add field names and units), the rules are defined by the system data standard; decrypt structured data {data type: temperature; value: 32℃; time: 10:00; decryption status: successful}.

[0066] In this embodiment of the invention, the hardware layer key cannot be read or tampered with by software. Even if the upper-layer software is compromised, the decryption result of the hardware layer can still maintain its original security. The hardware chip is designed specifically for encryption and decryption, and its processing speed is much faster than that of software, reducing the occupation of resources such as CPU, making it suitable for high-frequency data processing scenarios.

[0067] The analysis module 105 is used to perform anomaly probability analysis on the temperature data in the decrypted structured data to obtain the cold chain anomaly probability.

[0068] This invention extracts all temperature-related data (such as the temperature of the vehicle compartment and the temperature of refrigerated equipment recorded at different time points in cold chain transportation) from a decrypted and well-structured dataset (such as a table or standardized data containing information such as time, temperature value, and decryption status). Through statistical analysis and algorithm models (such as calculating the frequency and fluctuation range of temperature values ​​deviating from the normal range), it assesses the possibility of abnormal conditions (such as exceeding the safe range, sudden rises or falls) in these temperature data. Finally, it obtains a quantified probability value (i.e., the probability of cold chain anomalies) to determine whether there are potential risks in the cold chain system.

[0069] In this embodiment of the invention, when the analysis module performs anomaly probability analysis on the temperature data in the decrypted structured data to obtain the cold chain anomaly probability, it includes: Fluctuation characteristics of temperature data in the decrypted structured data are analyzed to obtain temperature data fluctuation characteristics. Based on the temperature data fluctuation characteristics, abrupt signal filtering is performed on the temperature data to obtain the frequency domain abrupt characteristics of the temperature data. Extract vehicle vibration intensity data from the decrypted structured data; A comprehensive risk assessment dimension is constructed using the vehicle vibration intensity data and the frequency domain mutation characteristics. The probability of cold chain anomalies is calculated based on the comprehensive risk assessment dimensions.

[0070] In this embodiment of the invention, all temperature values ​​are extracted from the decrypted standardized data (e.g., in tabular form, containing columns for "time" and "temperature (°C)"), and the characteristics of the temperature such as "fluctuation amplitude", "whether it is within the normal range", and "fluctuation frequency" are analyzed.

[0071] In detail, assuming the decrypted data is a 10-hour temperature record: Time: 0:00 → 1:00 → 2:00 → 3:00 → 4:00 → 5:00 → 6:00 → 7:00 → 8:00 → 9:00; Temperature: 2 → 2.1 → 1.9 → 3 → 2.5 → 4 → 3.8 → 5 → 4.5 → 6; Analysis of fluctuation characteristics: Normal range: 2-8℃, all temperatures are within the normal range; Fluctuation amplitude: Highest temperature 6℃, lowest 1.9℃, overall fluctuation range 4.1℃; Fluctuation frequency: Hourly temperature change is between 0.1-1.1℃, with small fluctuations in the first 3 hours (≤0.2℃), and gradually increasing fluctuations in the last 7 hours (≥0.5℃); The obtained fluctuation characteristics: The temperature is within the normal range, but the fluctuation amplitude and frequency increase in the later period.

[0072] Specifically, based on the aforementioned "fluctuation amplitude" characteristics, "mutation rules" are set. The formulation of these rules relies on industry experience; for example, a temperature difference greater than 1°C between two consecutive temperature changes is considered a "mutation." Signals that meet the mutation rules are then selected, and the "frequency" of mutations is statistically analyzed.

[0073] In detail, the adjacent temperature differences were calculated using the above temperature data: 1:00 - 0:00: 0.1℃ (no sudden change); 2:00 - 1:00: -0.2℃ (no sudden change); 3:00 - 2:00: +1.1℃ (sudden change); 4:00 - 3:00: -0.5℃ (no sudden change); 5:00 - 4:00: +1.5℃ (sudden change); 6:00 - 5:00: -0.2℃ (no sudden change); 7:00 - 6:00: +1.2℃ (sudden change); 8:00 - 7:00: -0.5℃ (no sudden change); 9:00 - 8:00: +1.5℃ (sudden change); Four sudden changes were selected, and the frequency domain characteristics (time interval of the sudden changes) were statistically analyzed: the first sudden change occurred at 3:00, the second at 5:00 (2-hour interval), the third at 7:00 (2-hour interval), and the fourth at 9:00 (2-hour interval); the obtained frequency domain sudden change characteristics: a total of 4 temperature sudden changes occurred, with an interval of 2 hours, which belong to "high-frequency periodic sudden changes".

[0074] Furthermore, from the decrypted structured data (with a newly added "Vibration Intensity (g)" column, where g is the unit of vibration, 1g≈9.8m / s²), vibration data corresponding to the time of the temperature record is directly extracted; assuming that the vibration data corresponding to 10 hours of the temperature data after the above decryption are: 0.1→0.12→0.09→0.3→0.25→0.4→0.38→0.5→0.45→0.6.

[0075] Furthermore, by combining the aforementioned "temperature frequency domain mutation characteristics" and "vibration intensity data," core indicators (dimensions) reflecting risk are selected, and a "risk level standard" is set for each dimension. Three assessment dimensions are constructed: vibration intensity during mutation: the vibration value corresponding to each temperature mutation (because vibration may cause temperature control equipment malfunction, thus triggering temperature mutation); the vibration values ​​corresponding to the above four mutations at 3:00, 5:00, 7:00, and 9:00 are 0.3g, 0.4g, 0.5g, and 0.6g, respectively; risk standard: vibration > 0.3g is "high-risk vibration" (the risk standard threshold is set by the industry sector of transporting a certain type of food, assuming this threshold is 0.3); number of high-frequency mutations: 4 mutations (under normal circumstances, mutations should be ≤ 2 within 10 hours, exceeding this is considered high-frequency); maximum vibration intensity: i.e., the maximum vibration value of 0.6g (far exceeding the 0.3g upper limit).

[0076] Furthermore, scores are given for the above three dimensions (0-10 points, the higher the score, the higher the risk), and the total score corresponds to the probability of anomaly (industry experience: 0-10 points get 0% probability, 11-20 points get 30%, 21-30 points get 80%).

[0077] Among the indicators, the vibration intensity during the mutation was >0.3g in 3 out of 4 mutations (0.4, 0.5, and 0.6g), scoring 8 points; the number of high-frequency mutations was 4 (far exceeding 2), scoring 9 points; the maximum vibration intensity was 0.6g (far exceeding 0.3g), scoring 9 points; the total score was 8+9+9=26 points. The corresponding anomaly probability was in the 21-30 point range, with an anomaly probability of 80%.

[0078] In this embodiment of the invention, when the analysis module calculates the probability of cold chain anomalies based on the comprehensive risk assessment dimensions, it includes: Determine the core indicators for the comprehensive risk assessment dimensions; Feature values ​​are extracted from the core indicators to obtain indicator feature values; Establish a mapping rule between feature intervals and anomaly probabilities based on the pre-defined historical feature anomaly correspondence; Based on the mapping rule, the anomaly probability of the indicator feature value is obtained by performing anomaly probability matching on the indicator feature value. The cold chain anomaly probability is generated based on the anomaly probability of the aforementioned indicators and the risk weights of the core indicators.

[0079] In this embodiment of the invention, key indicators with the greatest impact on cold chain anomalies are selected from two major dimensions: "vehicle vibration intensity" and "temperature frequency domain abrupt change characteristics." Assuming that analysis of 1000 past cold chain transportation anomaly events (such as meat spoilage) reveals four indicators most closely associated with anomalies: temperature abrupt change amplitude (the maximum value of a single sudden temperature increase / decrease); temperature abrupt change frequency (the number of sudden temperature changes per hour); vibration intensity peak value (the maximum single vibration intensity of the vehicle during transportation); and vibration duration (the number of minutes a single strong vibration lasts). These four indicators account for over 80% of the frequency of occurrence in historical anomaly events and are therefore identified by experts as core indicators.

[0080] Furthermore, the specific values ​​(i.e., "characteristic values") of each core indicator are calculated from the actual transportation data. For example, assuming a truck transporting frozen meat collects 2 hours of real-time data: In the temperature data, the largest sudden temperature rise is from -18℃ to -12℃, an increase of 6℃ → Temperature Sudden Change Amplitude Characteristic Value = 6℃; Four sudden temperature changes occur within 2 hours, averaging 2 times per hour → Temperature Sudden Change Frequency Characteristic Value = 2 times / hour; When the vehicle passes through a bumpy road section, the maximum vibration intensity reaches 11g (g is the unit of gravitational acceleration) → Vibration Intensity Peak Characteristic Value = 11g; One strong vibration lasts for 4 minutes → Vibration Duration Characteristic Value = 4 minutes.

[0081] Furthermore, using historical data, the probability of anomalies (such as spoilage risk) corresponding to different indicator value ranges is statistically analyzed to form a correspondence rule of "value range → probability". Analyzing data from 500 normal transport trips and 500 abnormal transport trips, the following mapping rule is obtained (example only), such as... Figure 5 As shown: Key indicators: Temperature change amplitude, temperature change frequency, peak vibration intensity, and vibration duration; The numerical ranges corresponding to the temperature change amplitudes are: 0-3℃, 3-6℃, and >6℃; the probability of anomaly is 5% for 0-3℃ (almost no risk), 30% for 3-6℃ (medium risk), and 70% for >6℃ (high risk). The numerical ranges corresponding to the frequency of temperature abrupt changes are: 0-1 times / hour, 1-3 times / hour, and >3 times / hour; the probability of anomaly is 5% for 0-1 times / hour, 20% for 1-3 times / hour, and 50% for >3 times / hour. The peak vibration intensity corresponds to the following numerical ranges: 0-5g, 5-10g, and >10g; the anomaly probability for 0-5g is 5%, for 5-10g it is 25%, and for >10g it is 60%. The numerical ranges corresponding to the duration of vibration are: 0-2 minutes, 2-5 minutes, and >5 minutes; the probability of anomaly is 5% for 0-2 minutes, 20% for 2-5 minutes, and 40% for >5 minutes.

[0082] The above-mentioned intervals and probabilities are obtained by statistically analyzing "the number of times that cold chain abnormalities (such as meat spoilage) eventually occurred when this interval appeared in history ÷ the total number of times (for example, "vibration intensity peak > 10g" has occurred 100 times in history, of which 60 times it eventually spoiled, so the probability is 60%).

[0083] Furthermore, the "feature values" calculated in the above steps are mapped to the table ranges to find the probability of anomalies for each indicator; the feature value matching rules are used: Temperature change amplitude = 6℃ → belongs to the "3-6℃" range → corresponding probability of anomaly = 30%; Temperature change frequency = 2 times / hour → belongs to the "1-3 times / hour" range → corresponding probability of anomaly = 20%; Vibration intensity peak value = 11g → belongs to the ">10g" range → corresponding probability of anomaly = 60%; Vibration duration = 4 minutes → belongs to the "2-5 minutes" range → corresponding probability of anomaly = 20%.

[0084] Furthermore, different core indicators have varying degrees of impact on cold chain anomalies (i.e., different "weights"). Multiplying each weight by its respective anomaly probability and summing these results yields the final overall risk probability. Assuming that, based on expert evaluation and historical data impact analysis, the risk weights (i.e., the proportion of importance) of the four core indicators are: Temperature fluctuation range: 30% (the greatest impact, as sudden temperature changes have the most direct damage to frozen products); Peak vibration intensity: 25% (strong vibrations may cause refrigeration equipment malfunctions); Temperature fluctuation frequency: 25% (frequent fluctuations can easily lead to deterioration); Vibration duration: 20% (continuous vibration has the next highest impact).

[0085] Calculate the overall cold chain anomaly probability: (Probability of temperature change amplitude × weight) + (Probability of peak vibration intensity × weight) + (Probability of temperature change frequency × weight) + (Probability of vibration duration × weight) = (30% × 30%) + (60% × 25%) + (20% × 25%) + (20% × 20%) = 9% + 15% + 5% + 4% = 33% That is, the probability of abnormality in this cold chain transportation is 33% (with a moderate risk of spoilage).

[0086] In this embodiment of the invention, core indicators for comprehensive risk assessment are determined, focusing on key influencing factors, avoiding interference from irrelevant data, and improving assessment efficiency. Feature values ​​are extracted from the core indicators to obtain indicator feature values, transforming abstract data into specific quantifiable values ​​and providing "standardized materials" for subsequent analysis. A mapping rule between feature intervals and anomaly probabilities is established, giving risk assessment a "historical basis," avoiding subjective judgment, and making the probabilities more credible. The anomaly probabilities of the indicators are matched according to the mapping rule, linking real-time data with historical patterns to quickly locate the current risk level.

[0087] The detection module 106 is used to perform dual-stream detection on the image data in the decrypted structured data and the cold chain anomaly probability to obtain the deterioration area detection result.

[0088] In this embodiment of the invention, image data (such as real-time images of goods in cold chain transportation, like photos of the appearance of meat and medicine) from a decrypted and structurally organized dataset is combined with the previously calculated cold chain anomaly probability (i.e., the possibility of a problem in the cold chain system, such as 33%). This data is then analyzed using "dual-flow detection" (a technique that simultaneously processes image information and probability data). On one hand, the images are used to identify whether the goods show signs of spoilage (such as darkening of color or the appearance of mold). On the other hand, the cold chain anomaly probability is used to determine whether these signs are caused by a cold chain problem. Ultimately, the specific areas of the goods that have spoiled or may spoil (such as "20% of the lower left corner of the meat has spoiled") are accurately located, resulting in a clear detection result for spoiled areas.

[0089] In this embodiment of the invention, when the detection module performs dual-stream detection of the image data in the decrypted structured data and the cold chain anomaly probability to obtain the deterioration region detection result, it includes: The cold chain anomaly probability is converted into a spatial heat map; The original features of the image data in the decrypted structured data are extracted to obtain the original image features; The original features of the image are weighted and enhanced based on the spatial heatmap to obtain enhanced features; Based on the enhanced features, the boundaries of the altered regions and the alteration confidence level are located, and the boundaries of the altered regions and the alteration confidence level are mapped to the probability distribution of the spatial heat map; The detection results for the deteriorated areas are determined based on the corresponding results.

[0090] In this embodiment of the invention, the probability of anomalies at different locations in the cold chain is visually displayed on a "map" (similar to a temperature heat map in a weather forecast) using varying shades of color; the darker the color (e.g., red), the higher the probability of anomalies at that location, and the lighter the color (e.g., blue), the lower the probability.

[0091] For example, suppose a cold chain truck compartment is divided into 5 zones (numbered 1-5). The previously calculated abnormal probabilities for each zone are: Zone 1 (80%), Zone 2 (60%), Zone 3 (30%), Zone 4 (50%), and Zone 5 (70%). According to the rules (which come from the industry's general standard for risk visualization: red for probability >70%, yellow for 50%-70%, and blue for <50%), the converted heat map is: Zone 1 (red), Zone 5 (red), Zone 2 (yellow), Zone 4 (yellow), and Zone 3 (blue).

[0092] Furthermore, basic information reflecting the condition of the goods (such as color, texture, shape, etc.) is extracted from the decrypted photos of the goods (such as real-time images of frozen meat inside the carriage) and converted into specific numerical values.

[0093] For example, taking a photo of frozen beef inside a train carriage as an example, we can use image recognition tools (similar to the "color recognition" function in mobile phone photography) to extract features: Area 1 (top left corner beef): Color brightness value 20 (0-100, the lower the value, the darker), texture roughness value 8 (0-10, the higher the value, the rougher). Area 5 (top right corner beef): Color brightness value 25, texture roughness value 6; Region 3 (middle beef): Color brightness value 40, texture roughness value 3.

[0094] These values ​​are the "original image features," reflecting the details of the goods' appearance.

[0095] Furthermore, the image features of high-probability areas (darker colors) in the heatmap are made "more prominent" by using the anomaly probability of that area as a weight and multiplying it by the original feature value to strengthen the feature influence of high-risk areas.

[0096] For example, combining the heatmap probabilities (region 1: 80%, region 5: 70%) and the original features: The original features of region 1 (brightness 20, roughness 8), multiplied by a weight of 80%, result in a brightness of 20 × 0.8 = 16 and a roughness of 8 × 0.8 = 6.4 after enhancement. The original features of region 5 (brightness 25, roughness 6), multiplied by a weight of 70%, result in a brightness of 25 × 0.7 = 17.5 and a roughness of 6 × 0.7 = 4.2 after enhancement. (Region 3 has a low probability and a weight of 30%. After enhancement, the feature value will be even smaller, and the influence will be weakened.)

[0097] Among them, high-probability areas are more likely to deteriorate. By weighting these areas, their characteristics will receive more attention in subsequent analyses, thus avoiding interference from the normal characteristics of low-risk areas.

[0098] Furthermore, based on the enhanced features (such as low brightness and rough texture, which are typical appearances of deterioration), the specific location (boundary) of the deterioration of the goods and "how accurate this judgment is" (confidence level) are determined, and then the anomaly probability of the location in the heat map is checked to see if it matches. For example, the enhanced features show that region 1 has a brightness of 16 (significantly dark) and a roughness value of 6.4 (significantly rough), which matches the characteristics of historically spoiled beef (the rule is based on statistics from 500 photos of spoiled beef: when the brightness is <20 and the roughness value is >5, 90% of the beef is spoiled). Therefore, the metamorphic boundary of the location region 1 is "from the upper left corner (coordinates x=10cm, y=10cm) to (x=30cm, y=30cm)", with a confidence level of 85% (that is, there is an 85% certainty that metamorphosis occurs here).

[0099] The corresponding heatmap shows that the probability of anomalies in this area is 80%, both of which are high values ​​and match each other (the high-probability area does indeed show high-confidence deterioration characteristics).

[0100] Furthermore, by combining the boundaries and confidence levels of the altered areas with the probability of the heat map, we can ultimately determine "where the alteration occurs and how likely it is to occur".

[0101] For example, based on the above description, the boundary of region 1 is clear (range of 10-30cm), with a confidence level of 85% (high), and it matches the heat map with a high probability of 80%. Therefore, the detection result is determined to be: "The beef in the upper left corner of the carriage (10cm, 10cm) - (30cm, 30cm) area has spoiled, with a confidence level of 85%". If the confidence level of a certain area is low (e.g., 30%) or does not match the probability of the heat map (e.g., the probability of the heat map is 20% but the confidence level is 70%), it will not be judged as a deteriorated area to avoid misjudgment.

[0102] In this embodiment of the invention, when the detection module converts the cold chain anomaly probability into a spatial heat map, it includes: Establish a mapping relationship between the coordinates of the temperature sensors inside the transport vehicle and the image data in the decrypted structured data; The cold chain anomaly probability is loaded into the target pixel region of the carriage image according to the mapping relationship, resulting in a discrete set of anomaly probability pixels. Spatial interpolation is performed on the probability values ​​of adjacent pixels in the set of abnormal probability pixels to obtain a continuous probability distribution; Extract the texture features from the image data, and correct the continuous probability distribution based on the texture features to obtain a spatial heatmap.

[0103] In this embodiment of the invention, the actual location of each temperature sensor in the carriage (e.g., "1 meter from the top left corner") is determined and mapped to the pixel position in the carriage photo (image data) (e.g., "pixel in row 200 and column 150" in the photo), which is equivalent to "marking the position of the sensor on the photo".

[0104] For example, suppose there are 3 temperature sensors inside the carriage: Sensor A: Its actual location is the middle of the front wall of the carriage (2 meters from the front of the carriage and 1 meter from the left side wall). Sensor B: The actual location is on the left side of the rear wall of the carriage (5 meters from the front of the carriage and 0.5 meters from the left side wall). Sensor C: The actual location is the right side of the middle of the carriage (3.5 meters from the front of the carriage and 0.8 meters from the right side wall).

[0105] Use the scale conversion of the train carriage photos (rule: based on the shooting distance and the actual size of the carriage, 1 meter of actual distance corresponds to 100 pixels on the photo, for example, if the carriage is 6 meters long, the photo height is 600 pixels): Sensor A corresponds to image pixels (200, 100) (2 meters × 100 = 200 pixels, 1 meter × 100 = 100 pixels). Sensor B corresponds to image pixels (500, 50); Sensor C corresponds to image pixels (350, 520) (right side wall distance 0.8 meters, total width 6 meters → 600 pixels, 600 - 0.8 × 100 = 520).

[0106] A mapping relationship between "actual sensor location and photo pixel" has been established, which allows the sensor's corresponding location to be found on the photo later.

[0107] Furthermore, the probability of cold chain anomalies calculated by each sensor is mapped to the corresponding pixel location of that sensor in the photo, forming points with probability values ​​(similar to marking "80% risk here" on a map).

[0108] For example, based on the mapping relationship above, the anomaly probabilities of the three sensors are known: Sensor A (corresponding to pixels 200, 100): probability 70%; Sensor B (corresponding to pixels 500, 50): probability 90%; Sensor C (corresponding to 350,520 pixels): probability 40%.

[0109] According to the mapping relationship, the probability values ​​are labeled on these three pixels in the photo to obtain a discrete set of points: {(200,100):70%,(500,50):90%,(350,520):40%}.

[0110] Furthermore, when there are only a few scattered risk points in the photo, the scattered points can be connected into a continuous probability distribution by calculating the "intermediate value" between adjacent points (similar to "estimating the temperature of the intermediate area when the temperatures of two places are known").

[0111] For example, sensor A (70%) and sensor B (90%) are located close to each other in the image (300 pixels horizontally and 50 pixels vertically), and are considered adjacent points. Calculate the probability of the midpoint using "linear interpolation": The pixel (350, 75) between A and B (horizontal middle 200+150=350, vertical middle 100-25=75) is equidistant from A and B, therefore the probability is (70%+90%)÷2=80%. The pixel closer to B (450, 60) is closer to B (accounting for 1 / 5 of the total distance), and its probability is 90% × 4 / 5 + 70% × 1 / 5 = 86%.

[0112] After repeating this operation on all adjacent points, the photo will no longer show isolated points, but rather a continuous probability change from 90% (near B) to 70% (near A) and then to 40% (near C), for example, the overall probability of a certain area is between 60% and 80%.

[0113] Furthermore, observe the texture of the goods in the photos (e.g., whether frozen goods have frost or clumps, these texture changes may indicate the actual condition), adjust the previous continuous probability distribution with texture features (e.g., increase the probability of places with abnormal textures and decrease the probability of places with normal textures), and finally form a color-intuitive heat map. For example, extract texture features from photos of the cargo compartment (using image analysis tools, with the following rules: a texture feature value > 50 indicates "signs of melting and chaotic texture," suggesting possible spoilage; < 30 indicates "uniform frost and normal texture," referencing texture statistics from 1000 photos of normal / spoiled goods): The texture feature value near sensor B (original probability 90%) is 70 (>50, texture abnormality), therefore the probability is increased to 95% after correction; The texture feature value near sensor A (original probability 70%) is 40 (between the middle), and the probability remains unchanged at 70%. The texture feature value near sensor C (original probability 40%) is 20 (<30, texture is normal), and the probability is reduced to 30% after correction.

[0114] The corrected continuous probability distribution is displayed using colors: 95% in dark red, 70% in orange, and 30% in light blue, forming the final spatial heat map, which can intuitively show that "the darkest red is on the left side of the rear wall of the carriage (near B), indicating the highest risk."

[0115] Through the above steps, from the sensor location corresponding to the photo pixels to marking risk points, then to completing the risks in the middle area, and finally combining the correction of the cargo appearance texture, a heat map that can accurately reflect the risk distribution inside the carriage is finally obtained.

[0116] In this embodiment of the invention, when the detection module performs the correction of the continuous probability distribution based on the texture features to obtain a spatial heatmap, it includes: Determine the physical region corresponding to the texture feature; The influence of the temperature distribution is analyzed based on the texture features and the physical region, and the regional correction rule of the continuous probability distribution is determined based on the influence rule. The continuous probability distribution is adjusted according to the regional correction rule to obtain a spatial heatmap.

[0117] Specifically, by identifying the shapes of different textures in the image (such as line density, color depth variation, pattern regularity, etc.), they are matched to the actual physical space inside the carriage (such as regular grid lines corresponding to the ventilation grille area, continuous dark block textures corresponding to the cargo stacking area, and smooth, textureless areas corresponding to the inner wall of the carriage).

[0118] Furthermore, by combining texture features with the attributes of the corresponding physical regions, the characteristics of temperature distribution are summarized. For example, due to air circulation, the temperature in the ventilation grille area is prone to fluctuations with the outside environment, and the probability of anomalies needs to be increased; due to strong insulation, the temperature in the cargo stacking area changes slowly, and the probability of anomalies needs to be reduced. The patterns are then transformed into clear correction rules, such as "probability value of ventilation grille area × 1.2, probability value of cargo stacking area × 0.8".

[0119] Furthermore, according to the correction rules, the abnormal probability values ​​of areas such as ventilation grilles and cargo stacks in the continuous probability distribution are adjusted in a targeted manner, so that the color depth of the heat map (representing the level of abnormal probability) is more in line with the actual temperature propagation characteristics, and finally a spatial heat map that can accurately reflect the cold chain anomalies in the carriage is generated. Among them, the determination of specific values ​​in the correction rules (such as multiplying the cargo area by 0.7) is derived by combining the physical area attributes corresponding to the texture features, historical data patterns and temperature propagation characteristics in the actual scene.

[0120] For example, by multiplying the pixel probability value of the ventilation grille area in the continuous probability distribution by 1.3, the cargo area by 0.7, and keeping the side wall area the original value, the resulting spatial heat map shows the ventilation grille area as a deeper red (high anomaly probability) and the cargo area as a lighter yellow (low anomaly probability), accurately reflecting the cold chain anomaly risk in different areas of the carriage.

[0121] In this embodiment of the invention, a mapping relationship between temperature sensor coordinates and image data is established, allowing abstract temperature data to "land" in a specific image, avoiding data disconnect from actual location; at the same time, dispersed anomaly probabilities are "marked" on the image, forming intuitive preliminary markings; a spatial heat map is obtained by correcting the continuous probability distribution based on texture features, and the results are optimized by combining the "visual cues" of the image itself, reducing misjudgments, so that the heat map not only relies on temperature data, but also incorporates the actual visual state of the goods, making risk assessment more in line with the real situation and reducing the error of relying solely on sensor data.

[0122] The disclosed devices, systems, and media can be implemented in other ways. For example, the system embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and other division methods may be used in actual implementation.

[0123] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0124] Furthermore, the functional modules in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or in the form of hardware plus software functional modules.

[0125] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0126] Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be embraced within the invention. No appended diagram markings in the claims should be construed as limiting the scope of the claims.

[0127] The embodiments of this application can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence (AI) refers to the theories, systems, technologies, and applications that use digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.

[0128] Furthermore, it is clear that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. Multiple units or systems described in the system embodiments may also be implemented by a single unit or system through software or hardware. Terms such as "first," "second," etc., are used to indicate names and do not indicate any specific order.

[0129] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A food safety system with dynamic message queue and composite encryption network control transmission, characterized in that, The system comprises an acquisition configuration module, a division monitoring module, an adjustment module, an encryption and decryption module, an analysis module and a detection module, wherein: The acquisition configuration module is configured to acquire multi-source monitoring data in a transport carriage, configure risk weights of the multi-source monitoring data, and obtain data risk weights; The division monitoring module is configured to divide transmission priorities of the multi-source monitoring data according to the data risk weights, and monitor data flow characteristics of a priority queue; The adjustment module is configured to adjust queue throughput rates of the multi-source monitoring data according to monitoring results and the transmission priorities, and obtain adjusted structured data sets; The encryption and decryption module is configured to perform composite encryption on the adjusted structured data sets, perform composite decryption on the encrypted data, and obtain decrypted structured data; The analysis module is configured to perform abnormal probability analysis on temperature data in the decrypted structured data, and obtain cold chain abnormal probabilities; The detection module is configured to perform double-flow detection on image data in the decrypted structured data and the cold chain abnormal probabilities, and obtain deterioration area detection results.

2. The dynamic message queue and composite encryption network control transmission food safety system of claim 1, wherein, When the acquisition configuration module performs risk weight configuration on the multi-source monitoring data and obtains data risk weights, the acquisition configuration module comprises: A multi-dimensional risk factor matrix is constructed according to the multi-source monitoring data, and invalid factors in the multi-dimensional risk factor matrix are screened to obtain an initial factor data set; The dispersion degree of each factor in the initial factor data set is determined through preset historical data, and the initial dynamic weight of each factor is calculated through the dispersion degree; Regional risk characteristics in transportation are acquired, the perishable weight in the initial dynamic weight is adjusted according to the regional risk characteristics, and an adjusted factor weight is obtained; The data risk weights are generated according to the adjusted factor weight and the multi-source monitoring data.

3. The dynamic message queue and composite encryption network control transmission food safety system of claim 1, wherein, When the division monitoring module divides transmission priorities of the multi-source monitoring data according to the data risk weights, the division monitoring module comprises: The multi-source monitoring data is initially sorted in data priority according to the data risk weights, and a basic transmission priority queue is obtained; Dynamic transmission condition configuration is performed on sub-priority data in the basic transmission priority queue, and a dynamic transmission condition of the sub-priority data is obtained; The packet loss rate and the delay gradient change of the basic transmission priority queue are monitored in real time according to the dynamic transmission condition; The data priority of the basic transmission priority queue is dynamically adjusted through the packet loss rate and the delay gradient change, and the transmission priorities of the multi-source monitoring data are obtained.

4. The dynamic message queue and composite encryption network control transmission food safety system of claim 1, wherein, When the adjustment module adjusts queue throughput rates of the multi-source monitoring data according to monitoring results and the transmission priorities, and obtains adjusted structured data sets, the adjustment module comprises: The multi-source monitoring data is assigned with a basic throughput rate according to monitoring results and the transmission priorities, and an initial basic throughput rate is obtained; The round-trip delay of a data transmission link probe packet is acquired, and a network quality index is calculated based on the round-trip delay; A throughput rate dynamic adjustment condition is constructed according to the initial basic throughput rate, the network quality index and the data risk weights; According to the throughput rate dynamic adjustment condition, the multi-source monitoring data is subjected to queue throughput rate adjustment, and an adjusted structured data set is obtained.

5. The dynamic message queue and composite encryption network control transmission food safety system of claim 1, wherein, The encryption and decryption module comprises the following steps when performing composite decryption on the encrypted data to obtain decrypted structured data: Hardware layer decryption is performed on the encrypted data to obtain hardware layer decrypted data; Transmission layer decryption is performed on the hardware layer decrypted data to obtain hardware transmission decrypted data; Application layer decryption is performed on the hardware transmission decrypted data, and decryption fuse authentication is performed on the decryption process; Decrypted structured data is generated according to the application layer decryption result and the authentication result.

6. The dynamic message queue and composite encryption network control transmission food safety system of claim 1, wherein, The analysis module comprises the following steps when performing abnormal probability analysis on the temperature data in the decrypted structured data to obtain a cold chain abnormal probability: Wave characteristic analysis is performed on the temperature data in the decrypted structured data to obtain temperature data wave characteristics; Mutation signal screening is performed on the temperature data according to the temperature data wave characteristics to obtain frequency domain mutation characteristics of the temperature data; Vehicle vibration intensity data is extracted from the decrypted structured data; A comprehensive risk assessment dimension is constructed through the vehicle vibration intensity data and the frequency domain mutation characteristics; A cold chain abnormal probability is calculated according to the comprehensive risk assessment dimension.

7. The dynamic message queue and composite encryption network control transmission food safety system of claim 6, wherein, The analysis module comprises the following steps when calculating a cold chain abnormal probability according to the comprehensive risk assessment dimension: A core index of the comprehensive risk assessment dimension is determined; Characteristic values of the core index are extracted to obtain index characteristic values; A mapping rule of characteristic intervals and abnormal probabilities is established according to a preset historical characteristic abnormality corresponding relationship; An abnormal probability matching is performed on the index characteristic values according to the mapping rule to obtain an index abnormal probability; A cold chain abnormal probability is generated according to the index abnormal probability and the risk weight of the core index.

8. The dynamic message queue and composite encryption network control transmission food safety system of claim 1, wherein, The detection module comprises the following steps when performing double-flow detection on the image data in the decrypted structured data and the cold chain abnormal probability to obtain a deterioration area detection result: The cold chain abnormal probability is converted into a spatial thermal map; Original feature extraction is performed on the image data in the decrypted structured data to obtain image original features; The image original features are weighted and enhanced according to the spatial thermal map to obtain enhanced features; A deterioration area boundary and a deterioration confidence are located according to the enhanced features, and the deterioration area boundary and the deterioration confidence correspond to the probability distribution of the spatial thermal map; A deterioration area detection result is determined according to the corresponding result.

9. The dynamic message queue and composite encryption network control transmission food safety system of claim 8, wherein, The detection module comprises the following steps when converting the cold chain abnormal probability into a spatial thermal map: A mapping relationship between temperature sensor coordinates in a transport vehicle cabin and image data in the decrypted structured data is established; The cold chain abnormal probability is loaded to a target pixel area of the cabin image according to the mapping relationship to obtain a discrete distribution of abnormal probability pixel points; Spatial interpolation calculation is performed on the probability values of adjacent pixel points in the abnormal probability pixel point set to obtain a continuous probability distribution; Texture features of the image data are extracted, and the continuous probability distribution is corrected according to the texture features to obtain a spatial thermal map.

10. The dynamic message queue and composite encryption network control transmission food safety system of claim 9, wherein, The detection module comprises the following steps when performing correction on the continuous probability distribution according to the texture feature to obtain a spatial heat map: determining a physical region corresponding to the texture feature; analyzing an influence law of temperature distribution according to the texture feature and the physical region, and determining a region correction rule of the continuous probability distribution according to the influence law; adjusting a region probability value of the continuous probability distribution according to the region correction rule to obtain a spatial heat map.

Images