A smart cold chain transportation information management method, system, device and storage medium

Abstract

The present application provides a smart cold chain transportation information management method, system, device and storage medium, including: identifying the type of transported goods and environmental requirements through a task identification module, and determining and configuring the target sensor of the cold chain vehicle through a sensor configuration module. The target sensor collects environmental data, the on-board GPS module collects location information and driving status, the RFID tag records the cargo loading status, and all data is sent to the central management platform. The platform constructs a multi-dimensional feature vector, compresses it into a low-dimensional feature representation through an encoder, reconstructs it using a decoder to obtain a reconstructed data vector, and calculates the reconstruction error. By analyzing the reconstruction error, the platform generates an analysis result, and automatically adjusts the environmental control equipment in the vehicle according to the result to ensure the stability of cargo transportation.

Description

Technical Field

[0001] The present application relates to the field of intelligent cold chain information technology, and in particular to an intelligent cold chain transportation information management method, system, device and storage medium. Background Art

[0002] As the global supply chain becomes increasingly complex, cold chain logistics has become a key link in ensuring food safety, drug effectiveness, and the quality of biological products. Cold chain transportation needs to be carried out under strict temperature and humidity control to ensure that the goods are not affected by the external environment during the entire transportation process. However, traditional cold chain transportation management relies on manual monitoring, which has problems such as poor real-time performance, incomplete data collection, and untimely fault response, making it difficult to effectively ensure the safety of goods during transportation.

[0003] During cold chain transportation, many factors such as ambient temperature, humidity, vehicle driving status, and cargo sealing will affect the final quality of the goods. For example, if the temperature of food fluctuates too much during transportation, it may cause spoilage; medicines and vaccines need to be stored within a specific temperature range, otherwise their efficacy will be greatly reduced. Therefore, cold chain transportation places higher requirements on temperature control accuracy, real-time monitoring and management.

[0004] Some cold chain transportation management systems on the market currently focus on temperature monitoring and early warning, but lack comprehensive information management functions and cannot achieve full-process and intelligent management of goods, transportation equipment, and transportation routes. Since cold chain transportation involves multiple links, such as transportation vehicles, refrigeration equipment, sensors, and information management systems, existing technologies are difficult to achieve effective coordination between various links. In addition, when cold chain transportation equipment fails, existing manual or semi-automatic management systems are difficult to respond quickly, resulting in damage or even scrapping of the transported goods, causing huge economic losses.

[0005] With the rapid development of technologies such as the Internet of Things, big data, and artificial intelligence, intelligent cold chain transportation information management methods based on sensor networks, data analysis, and intelligent control have gradually become the key development direction of future cold chain logistics. By integrating multi-dimensional information such as temperature, humidity, vehicle location, and cargo status into an intelligent management platform, enterprises can achieve real-time monitoring, analysis, and decision-making of the entire cold chain transportation process, thereby improving the stability and efficiency of cold chain transportation and reducing the risks brought by human operations.

[0006] To sum up, the existing cold chain transportation information management technology has problems such as limited monitoring range, insufficient automation, and untimely emergency response. Therefore, there is an urgent need for a solution that can provide all-round intelligent management of cold chain transportation to meet the increasingly stringent temperature control needs and cargo quality assurance requirements of the cold chain logistics industry. Summary of the invention

[0007] In order to solve the above technical problems, the present application provides a smart cold chain transportation information management method, system, device and storage medium.

[0008] The technical solution provided in this application is described below:

[0009] The first aspect of the present application provides a smart cold chain transportation information management method, the method comprising:

[0010] Identify the type of goods currently being transported, transportation needs and environmental requirements through the task identification module;

[0011] According to the task identification result, the sensor configuration module dynamically determines the target sensor among the sensors of the cold chain transport vehicle, and configures the target sensor;

[0012] Collecting real-time environmental data through the target sensor;

[0013] The location information and driving status data of cold chain transport vehicles are collected through the vehicle-mounted GPS module, and the driving route is constructed;

[0014] Collect the loading status information of the goods through the RFID tags set on the goods;

[0015] Sending the environmental data, loading status information and the driving path to a central management platform;

[0016] The central management platform constructs a multi-dimensional feature vector based on the environmental data, loading status information and the driving path;

[0017] The central management platform inputs the multi-dimensional feature vector into a pre-configured encoder for compression to obtain a low-dimensional feature representation;

[0018] The central management platform reconstructs the low-dimensional features using a preconfigured decoder to obtain a reconstructed data vector;

[0019] The central management platform calculates a reconstruction error between the multidimensional feature vector and the reconstructed data vector;

[0020] The central management platform analyzes the reconstruction error and generates an analysis result;

[0021] The central management platform automatically adjusts the environmental control equipment in the cold chain transport vehicle according to the analysis results.

[0022] Optionally, dynamically determining a target sensor among sensors of the cold chain transport vehicle through a sensor configuration module according to the task identification result, and configuring the target sensor includes:

[0023] Based on the task identification result, the sensor configuration module reads the preconfigured mapping table to determine the target sensor among the sensors of the cold chain transportation vehicle;

[0024] The sensor configuration module reads a sensor database and configures the target sensor based on the sensor database, the type of the transported goods, transport requirements and environmental requirements.

[0025] Optionally, the central management platform constructs a multidimensional feature vector based on the environmental data, the loading status information and the driving path, including:

[0026] calculating statistical features in the environmental data;

[0027] The type of cargo currently being transported generates a binary cargo feature;

[0028] Count the weight and volume characteristics of the goods currently being transported;

[0029] Extract the changes in longitude and latitude from GPS data and calculate the distance and time characteristics of travel;

[0030] Generate context features based on the geographical location and weather information of the driving route;

[0031] The statistical features, binary cargo features, weight features, volume features, distance features, time features and context features are integrated into a multi-dimensional feature vector.

[0032] Optionally, the central management platform inputs the multi-dimensional feature vector into a preconfigured encoder for compression to obtain a low-dimensional feature representation, including:

[0033] Define an m*n weight matrix W, where m is the latitude of the input feature, m=7, and n is the desired feature latitude of the compression;

[0034] The first output value Z is calculated by the following formula:

[0035] ;

[0036] Where X represents the input multidimensional feature vector, and b represents the bias term;

[0037] The first output value Z is passed into the following activation function:

[0038]

[0039] A represents the compressed eigenvector, and e is the base of a natural number.

[0040] Optionally, the central management platform calculates a reconstruction error between the multidimensional feature vector and the reconstructed data vector, including:

[0041] The mean square error (MSE) is calculated as follows:

[0042] ;

[0043] in, represents the multi-dimensional feature vector, represents the reconstructed data vector;

[0044] The central management platform analyzes the reconstruction error and generates analysis results including:

[0045] Determine whether the value of the MSE exceeds a preset error threshold;

[0046] If it exceeds, it is determined that an abnormality exists.

[0047] Optionally, the method further includes:

[0048] According to the analysis results, the Dijkstra algorithm is called to generate the optimal driving path for the cold chain transport vehicle.

[0049] Optionally, calling the Dijkstra algorithm to generate an optimal driving path for the cold chain transport vehicle according to the analysis result includes:

[0050] The transportation route of cold chain vehicles is modeled as a graph structure, where the nodes represent locations, the edges represent road sections, and the weights of the edges represent the transportation costs of each road section.

[0051] Set the current cold chain vehicle position as the starting node;

[0052] Set the destination location as the target node;

[0053] Initialize a distance value for each node, set the distance value of the starting node to 0, and set the distances of other nodes to infinity;

[0054] Use a priority queue or minimum heap to record the order in which nodes are visited, and start visiting from the starting node;

[0055] For the current node, check all its adjacent nodes and calculate the potential distance to the adjacent nodes through the current node;

[0056] If the path through the current node is shorter than the known shortest path, the shortest path of the adjacent node is updated and the adjacent node is added to the priority queue;

[0057] Repeat the above steps until the shortest path to the target node is determined or the priority queue is empty.

[0058] A second aspect of the present application provides a smart cold chain transportation information management system, the system comprising:

[0059] An identification unit, used to identify the type of goods currently being transported, transportation requirements and environmental requirements through a task identification module;

[0060] A sensor configuration unit, used to dynamically determine a target sensor among sensors of the cold chain transport vehicle through a sensor configuration module according to a task identification result, and configure the target sensor;

[0061] A real-time collection unit, used to collect real-time environmental data through the target sensor;

[0062] A path construction unit, used to collect the location information and driving status data of the cold chain transport vehicle through the vehicle-mounted GPS module, and to construct a driving path;

[0063] An RFID unit, used to collect loading status information of the goods through an RFID tag set on the goods;

[0064] A sending unit, used for sending the environmental data, loading status information and the driving path to a central management platform;

[0065] A multidimensional feature construction unit, configured to construct a multidimensional feature vector based on the environmental data, the loading status information and the driving path;

[0066] An encoder unit, used for inputting the multi-dimensional feature vector into a pre-configured encoder for compression to obtain a low-dimensional feature representation;

[0067] A decoder unit, configured to reconstruct the low-dimensional features using a preconfigured decoder to obtain a reconstructed data vector;

[0068] an error calculation unit, used to calculate a reconstruction error between the multidimensional feature vector and the reconstructed data vector;

[0069] An analysis unit, used for analyzing the reconstruction error and generating an analysis result;

[0070] A control unit is used to automatically adjust the environmental control equipment in the cold chain transportation vehicle according to the analysis result. The third aspect of the present application provides a smart cold chain transportation information management device, including:

[0071] Processor, memory, input-output unit, and bus;

[0072] The processor is connected to the memory, the input and output unit, and the bus;

[0073] The memory stores a program, and the processor calls the program to execute the first aspect and any optional method in the first aspect.

[0074] A fourth aspect of the present application provides a computer-readable storage medium, on which a program is stored. When the program is executed on a computer, the program executes the first aspect and any optional method in the first aspect.

[0075] It can be seen from the above technical solutions that this application has the following advantages:

[0076] 1. Through the sensor configuration module, the target sensor can be dynamically selected and configured according to different cargo types and transportation requirements to ensure the pertinence and accuracy of environmental monitoring, reduce unnecessary sensor resource consumption, and improve the adaptability of cold chain transportation.

[0077] 2. The central management platform uses the encoder to compress the multi-dimensional feature vector to obtain a low-dimensional feature representation. This process reduces the burden of data transmission and storage while retaining core information, which helps to perform rapid data processing in cold chain transportation environments with high real-time requirements.

[0078] 3. By reconstructing the data vector through the decoder and analyzing the reconstruction error, anomalies in environmental data or cargo status can be detected in a timely manner. For example, when the reconstruction error exceeds the threshold, the system can identify the abnormal situation and provide support for possible risk warnings in cold chain transportation.

[0079] 4. The analysis results can be used to adjust the driving route to ensure that cold chain transport vehicles choose the optimal route in logistics transportation, thereby saving time and fuel costs, and realizing intelligent risk avoidance of the route in special circumstances.

[0080] 5. Automatically adjust environmental control equipment (such as temperature control devices, humidity controllers, etc.) according to the analysis results to achieve closed-loop control of the cold chain transportation environment, ensure that parameters such as temperature and humidity of the goods during the entire transportation process remain in the optimal range, and improve the preservation effect and transportation quality of the goods.

[0081] 6. The system collects environmental data, loading status information and driving routes, and constructs multi-dimensional feature vectors, which are analyzed and managed on the platform to form detailed transportation data records, facilitating the traceability of the transportation process and the optimization of logistics processes. BRIEF DESCRIPTION OF THE DRAWINGS

[0082] In order to more clearly illustrate the technical solution in the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0083] Figure 1 A flowchart of an embodiment of the intelligent cold chain transportation information management method provided in this application;

[0084] Figure 2 A schematic diagram of an embodiment of a process for generating a driving path for a cold chain transport vehicle in the intelligent cold chain transport information management method provided in this application;

[0085] Figure 3 This is a flow chart of an embodiment of the smart cold chain transportation information management system provided in this application;

[0086] Figure 4 A flow chart of an embodiment of the intelligent cold chain transportation information management device provided in this application. DETAILED DESCRIPTION

[0087] It should be noted that the method provided in this application can be applied to a terminal, a system, or a server. For example, the terminal can be a smart phone or a computer, a tablet computer, a smart TV, a smart watch, a portable computer terminal, or a fixed terminal such as a desktop computer. For the convenience of explanation, this application uses the terminal as an example for explanation.

[0088] See also Figure 1 The present application first provides an embodiment of a method, which comprises:

[0089] S101, identifying the type of goods currently being transported, transportation requirements and environmental requirements through a task identification module;

[0090] In step S101, the task identification module identifies the type of goods in the cold chain transportation task, such as food, medicine, fresh produce, etc. These types of goods have different transportation needs and environmental requirements, so the task identification module determines the key transportation conditions (such as temperature, humidity, light, air quality, etc.) based on the specific type of goods being transported, as well as the environmental parameters required during transportation to ensure that the cold chain environment is suitable for the preservation or safety needs of the goods.

[0091] Specifically, when the transport task is initialized, the system obtains basic information about the goods from the order management module or user input. This information may include the name, type, quantity, volume, quality, etc. of the goods. If the order contains preset transport requirements and environmental requirements, this information can be directly extracted.

[0092] Based on the collected cargo information, the task identification module matches the cargo type with the cargo category information in the database. The database can contain predefined transport types (such as food, medicine, fresh produce, etc.) and classify and label the typical transport conditions for each type. Through type matching, the system automatically identifies the specific transport type of the cargo.

[0093] After the matching is completed, the task identification module retrieves the key transportation parameters of the corresponding cargo type from the environmental requirements database. For example, food and fresh produce generally require a frozen environment (low temperature, high humidity), medicines require a constant temperature, low humidity environment, and specific medicines may require certain light protection or a vibration-free environment. The database lists a series of reference environmental requirements for each type of cargo, covering parameter ranges such as temperature, humidity, air quality, and light.

[0094] The task identification module uses the extracted transportation conditions as the initial configuration target and makes appropriate adjustments based on the needs of the specific task. For example, if the cold chain transportation time is long, the module will consider the shelf life of the goods and add strict temperature and humidity conditions; if it is a short-distance transportation, the environmental requirements can be moderately relaxed.

[0095] After identifying the key transportation conditions suitable for the goods, the task identification module passes these conditions to the sensor configuration module and the control unit, and provides a reference for subsequent transportation management. These environmental parameters are used to set the sensor range and acquisition frequency in the cold chain vehicle to ensure that the environment meets the transportation requirements.

[0096] S102, according to the task identification result, dynamically determine the target sensor among the sensors of the cold chain transport vehicle through the sensor configuration module, and configure the target sensor;

[0097] In step S102, the sensor configuration module dynamically selects and configures the target sensors on the cold chain transport vehicle based on the results of the task identification in S101. For example, if the task identification result indicates that the goods are food, the temperature and humidity sensors are activated; if they are medicines, the air quality sensor may also need to be enabled. The sensor configuration module will apply the set rules or algorithms to determine the target sensors that need to be activated, and adjust the collection frequency or data threshold according to environmental requirements to improve the accuracy and efficiency of the collection.

[0098] Specifically, in this step, according to the cargo type (such as food, medicine or fresh produce) identified in S101, the sensor configuration module maps cargo requirements to the required sensor type. The module searches for the required sensor type through a preset rule base or database. For example:

[0099] Food cargo: requires low temperature and moderate humidity, so activate the temperature and humidity sensor.

[0100] Pharmaceutical goods: Temperature, humidity and air quality need to be precisely controlled, so an air quality sensor is added on the basis of activating the temperature and humidity sensor.

[0101] Fresh goods: There are requirements for lighting, and the module will activate the light sensor.

[0102] The module applies set rules or dynamic configuration algorithms to determine the target sensors that need to be activated. Configuration rules can be simple conditional judgment structures (such as "if-else") or more advanced decision trees and logic tables. The module performs the following logic:

[0103] First, activate the necessary sensors based on the environmental requirements of the cargo type;

[0104] If the mission has additional requirements (such as time constraints for long-distance transportation or high preservation requirements for perishable items), enable more sensors or increase the collection frequency.

[0105] Collection frequency and data threshold adjustment;

[0106] The task identification module dynamically sets the sensor's acquisition frequency based on the specific needs of the goods. For example:

[0107] For temperature-sensitive goods, set a higher sampling frequency to monitor temperature fluctuations in real time.

[0108] For goods that are sensitive to air quality, a lower sampling frequency is set, and data is collected only when the environment changes significantly to save energy. At the same time, the module can dynamically set data thresholds, such as setting a threshold range suitable for drug transportation in the air quality sensor (such as volatile organic compounds below a certain concentration).

[0109] The sensor configuration module sends a startup or configuration command to each target sensor to start real-time data collection. The module also communicates with the control system in the vehicle to monitor the working status of the sensor and regularly calibrates the sensor to ensure the accuracy and reliability of the data.

[0110] The sensor configuration module will regularly receive data analysis feedback from the central management platform and dynamically adjust the collection frequency, threshold and collection method of each sensor to cope with sudden environmental changes during transportation and further optimize resource utilization efficiency.

[0111] S103, collecting real-time environmental data through the target sensor;

[0112] In step S103, the target sensors in the vehicle start to collect data according to the configuration, collecting real-time environmental data in the current cold chain vehicle, such as temperature, humidity, air quality, light intensity, etc. These data will be uploaded to the central management platform in real time through the network equipment in the vehicle to monitor the environmental conditions of cold chain transportation.

[0113] In this step, the target sensor starts working according to the acquisition frequency and data threshold set in S102. The sensor performs self-calibration when it is first started to ensure the accuracy of the collected data. For example, the temperature sensor will perform zero point and range calibration to ensure measurement accuracy.

[0114] Each sensor collects data at a configured frequency and interval, for example:

[0115] The temperature and humidity sensors collect and record the temperature and humidity information inside the cold chain vehicle every few seconds or minutes.

[0116] Air quality sensors detect pollutants or specific gas concentrations in the air, such as carbon dioxide or volatile organic compounds, and typically operate at low sampling rates unless significant fluctuations are detected.

[0117] Light sensors are used in food or medicine transportation to regularly collect light intensity to ensure that the ambient light meets the requirements of the goods.

[0118] The collected data is transmitted to the vehicle’s central control module through the sensor nodes. The module can integrate and temporarily store the data, and then send the data to the central management platform in real time through the in-vehicle network (such as Wi-Fi, LTE or 5G network).

[0119] Transmitted data usually uses standard communication protocols (such as MQTT, HTTP, or WebSocket) and is formatted as JSON or XML for easy parsing and processing.

[0120] In order to adapt to the limited bandwidth of the vehicle communication module, the data transmission module will package and compress multiple data before uploading to reduce bandwidth consumption.

[0121] S104, collecting location information and driving status data of cold chain transport vehicles through the vehicle-mounted GPS module, and constructing a driving route;

[0122] In step S104, the vehicle-mounted GPS module will periodically obtain the real-time location information of the vehicle and record the vehicle's driving speed, acceleration and other status data. Based on this data, the central management platform can construct the vehicle's driving path to optimize the driving route and monitor key factors such as transportation time and transportation environment.

[0123] In this embodiment, the GPS module on the cold chain transport vehicle is automatically started when the vehicle is started, and establishes a connection with the satellite system for initial positioning. The GPS module regularly receives and records the vehicle's geographic coordinates (longitude and latitude), altitude, and timestamp information according to a set sampling frequency.

[0124] In addition to location information, the GPS module or the acceleration sensor in the vehicle collects the vehicle's driving status data, including:

[0125] Driving speed: Calculate the speed through GPS displacement, or obtain speed data from the vehicle's onboard computer.

[0126] Acceleration and turning status: Monitor whether the vehicle has sudden deceleration, acceleration or turning during driving, so as to record the vehicle's driving stability and transportation conditions.

[0127] The collected location information and driving status data are regularly sent to the vehicle's central control module, which can integrate the data and temporarily store it in local memory, and then upload it to the central management platform in real time through the in-vehicle network (such as LTE or 5G).

[0128] Before uploading the data, the control module can compress the data to save bandwidth and encrypt it to protect data privacy. After receiving the location information, the central management platform uses an algorithm to connect these continuous geographic coordinate points into a path. Path construction includes the following processes:

[0129] A sequence of geographic coordinates of the vehicle is generated based on the received position information.

[0130] The path point sequence is smoothed to remove the error caused by positioning drift and make the driving path more accurate. The path will be dynamically updated during the vehicle's driving process to reflect the vehicle's latest driving trajectory.

[0131] After the path is constructed, the central management platform can further optimize the driving route. By combining map data and real-time traffic information, the platform can apply path optimization algorithms (such as Dijkstra or A* algorithms) to find the fastest, shortest or most energy-efficient route.

[0132] The platform calculates the total driving time of the vehicle and the estimated time from the current destination to monitor the transportation progress.

[0133] S105, collecting loading status information of the goods through RFID tags set on the goods;

[0134] In step S105, the system obtains loading status information through the RFID tags on the goods. This information usually includes the loading location, quantity, status (such as whether there is any change, tilt), etc. of the goods, which helps to monitor the safety and stability of the goods during transportation in real time, and records this information on the central management platform to trace the entire process of the cold chain.

[0135] When loading goods, a unique RFID tag is attached to each cargo unit (such as a pallet or box). The RFID tag is pre-written with relevant information about the goods, such as the cargo ID, category, loading location and quantity, etc. The initialization of the RFID tag can be completed through the central management platform when the goods are loaded or stored.

[0136] Multiple RFID readers are deployed in the cargo compartment of cold chain transport vehicles, covering the area where the goods are loaded. These readers are configured to scan at regular intervals or start when triggered by specific events (such as vehicle start or arrival at a certain checkpoint). The readers are connected to the on-board control system wirelessly or wired to transmit the collected data in real time.

[0137] When the vehicle is started or in transit, the RFID reader automatically scans the RFID tag on the cargo and obtains the following information:

[0138] Cargo location: The unique ID of each tag and the relative position of the reader can infer the loading position of each cargo within the carriage.

[0139] Quantity of goods: By counting the number of RFID tags that have been read, the platform can calculate the total number of goods currently loaded.

[0140] Goods status: The system can detect the movement or disappearance of the tag. If the tag is abnormal, such as position shift or temporary disappearance, the system will identify the possible tilt, movement or loss of the goods.

[0141] The RFID reader transmits the acquired loading status data (including cargo ID, location, quantity, etc.) to the central management platform through the vehicle control system. The platform records this information and establishes a real-time monitoring model so that the specific loading status of the cargo can be checked at any time.

[0142] S106, sending the environmental data, loading status information and the driving path to a central management platform;

[0143] In step S106, the environmental data, loading status information, and driving path data are sent to the central management platform through the data transmission network. The central management platform is responsible for processing, analyzing, and storing these data to ensure that the transportation environment always meets the storage requirements of the goods.

[0144] Specifically, sensors in cold chain transport vehicles (such as temperature, humidity, light, air quality, etc.) collect environmental data in real time. The on-board RFID reader obtains the loading status information of the goods, including location, quantity, stability, etc. The on-board GPS module collects the vehicle's location information and driving status to generate driving path data. These data are aggregated through the interface of the on-board control system to form a complete data package containing environmental data, loading status information and driving path.

[0145] S107, the central management platform constructs a multi-dimensional feature vector based on the environmental data, loading status information and the driving path;

[0146] In step S107, the central management platform integrates the environmental data, loading status information and driving path data obtained in S106 and extracts features. The platform integrates these feature data into a multi-dimensional feature vector as a comprehensive description of the cold chain transportation environment. These features include temperature, humidity, location information, cargo status, etc., which will be used for further data analysis.

[0147] Specifically, the central management platform receives and analyzes environmental data (such as temperature and humidity), loading status information (such as cargo location, quantity, tilt status), and driving path data (such as GPS location information, driving speed, acceleration, etc.) uploaded from vehicles. Each data type corresponds to a dimension, and environmental data, loading status, and path data are analyzed and preprocessed in different modules and converted into integrated numerical or symbolic feature representations.

[0148] The platform performs feature extraction on each data type to form a comprehensive description of the cold chain transportation status:

[0149] Environmental characteristics: Extract average value, fluctuation range (maximum value, minimum value), change rate and other characteristics from temperature, humidity, light and other data to indicate the environmental stability during transportation.

[0150] Loading status characteristics: including cargo position, loading status, etc., extracting inclination, center of gravity position, etc., to ensure the stability of cargo during transportation.

[0151] Path features: Extract features such as driving speed, acceleration, current location information, etc. based on GPS location information, and generate time series features such as average speed, driving time, etc. based on the path.

[0152] To ensure that each feature dimension has a consistent dimension and scale in the vector, the platform standardizes the extracted features, such as normalizing or standardizing values ​​such as temperature, humidity, and location information, to facilitate subsequent model processing or analysis.

[0153] The standardization method may include Z-score standardization or normalization to the interval [0,1] to make the range of eigenvalues ​​in each dimension consistent and improve the calculation accuracy of vectors in multi-dimensional space.

[0154] The standardized features are integrated into a vector to form a set of unified numerical vectors. Each feature represents the dimension of a vector, which together describe the environmental conditions and states during transportation.

[0155] The feature vector can be expressed as: V = [Tmean, Trange, Hmean, Hrange, Plat, Plong, Stilt, Vavg, Amax, ...], where Tmean represents the mean temperature, Hmean represents the mean humidity, Plat and Plong represent the latitude and longitude of the current location, Stilt represents the loading inclination, Vavg represents the average driving speed, Amax represents the maximum acceleration, etc. The final generated feature vector V is stored in the platform database as a complete description of the cold chain transportation status at that moment.

[0156] In an optional embodiment, the central management platform constructs a multi-dimensional feature vector based on the environmental data, the loading status information and the driving path, including:

[0157] calculating statistical features in the environmental data;

[0158] The type of cargo currently being transported generates a binary cargo feature;

[0159] Count the weight and volume characteristics of the goods currently being transported;

[0160] Extract the changes in longitude and latitude from GPS data and calculate the distance and time characteristics of travel;

[0161] Generate context features based on the geographical location and weather information of the driving route;

[0162] The statistical features, binary cargo features, weight features, volume features, distance features, time features and context features are integrated into a multi-dimensional feature vector.

[0163] In this optional embodiment, statistical features are first extracted from the real-time environmental data, including the mean, maximum, minimum, standard deviation, etc. of temperature and humidity. These statistical features reflect the changes in the environment during transportation, for example:

[0164] Mean: represents the average level of the overall environment;

[0165] Maximum / minimum values: help identify extreme environments that may affect cargo;

[0166] Standard Deviation: A measure of environmental volatility.

[0167] These features help monitor the stability of the environment and ensure that the environmental conditions are suitable for the storage needs of the cargo.

[0168] Then, based on the type of transported goods (such as food, fresh produce, medicine, etc.), a binary vector is generated to represent the type of transported goods. For example:

[0169] Food type can be represented as [1,0,0], medicine type can be represented as [0,1,0], and so on.

[0170] These binary feature vectors can facilitate the subsequent identification of transportation requirements of different goods in the model.

[0171] The weight and volume information of the cargo is obtained from the RFID tag or cargo sensor system as numerical features.

[0172] Weight and volume characteristics help the platform determine the space utilization of cold chain vehicles and further decide whether environmental control parameters need to be adjusted (for example, optimizing cold air circulation based on the total weight of the goods).

[0173] It is also possible to extract longitude and latitude changes from GPS data and calculate the distance and time characteristics of travel. The vehicle-mounted GPS module can be used to regularly record the longitude and latitude of the vehicle, calculate the longitude and latitude changes between two consecutive positions, and thus calculate the distance traveled. At the same time, the time characteristics of travel (such as total travel time, average speed, etc.) are calculated based on the recorded timestamp. Distance and time characteristics are not only used for path tracking, but also provide basic data for subsequent path optimization.

[0174] You can also obtain geographic information and weather conditions of the area passed by the route, such as humidity, precipitation, temperature, etc., through a third-party interface to generate context features. Geographic and weather information helps to determine the potential impact of the external environment on the cold chain environment (such as high temperature or heavy rain may increase the cold chain load), providing a decision-making basis for environmental control.

[0175] Finally, the statistical features, binary cargo features, weight features, volume features, distance features, time features and context features extracted above are integrated into a multidimensional feature vector to form a global description of the current transportation environment.

[0176] S108, the central management platform inputs the multi-dimensional feature vector into a pre-configured encoder for compression to obtain a low-dimensional feature representation;

[0177] In step S108, the central management platform inputs the multi-dimensional feature vector into the pre-trained encoder model for compression. The encoder generates a low-dimensional feature representation by reducing the dimensionality of the feature space of the input data. This process retains the key information in the feature vector and removes redundant or unimportant information, which facilitates subsequent analysis and reduces the consumption of computing resources.

[0178] In this embodiment, first, it is ensured that the multi-dimensional feature vector has been successfully constructed in the previous step (such as step S107). The multi-dimensional feature vector includes multiple features such as environmental data, loading status information and driving path.

[0179] Next, a pre-trained encoder model is used, which is usually based on an autoencoder or a neural network. The model should be trained to effectively capture the main features of the input data and retain the importance of the information. The constructed multi-dimensional feature vector is passed as input to the encoder model. The encoder converts the input multi-dimensional feature vector into a low-dimensional feature representation. The specific implementation method can be as follows:

[0180] Forward propagation: The input vector is passed through each layer of the encoder for forward propagation calculation.

[0181] Activation function: Apply activation functions (such as ReLU, Sigmoid, etc.) at each layer to introduce nonlinearity so that the model can better learn the characteristics of the data.

[0182] Dimensionality reduction: At the output layer of the encoder, a low-dimensional feature representation is generated. The dimension of the output will be much smaller than the dimension of the input to achieve the purpose of compression.

[0183] The encoder generates a low-dimensional feature representation as output and stores it in the central management platform. This low-dimensional representation contains a compact representation of the most critical information in the multi-dimensional feature vector and can be represented as a smaller vector, for example:

[0184] Vlow=[v1,v2,…,vk].

[0185] The following is a specific example for further description:

[0186] First, define an m*n weight matrix W, where m is the latitude of the input feature, m=7, and n is the desired feature latitude of the compression;

[0187] The first output value Z is calculated by the following formula:

[0188] ;

[0189] Where X represents the input multidimensional feature vector, and b represents the bias term;

[0190] The first output value Z is passed into the following activation function:

[0191]

[0192] A represents the compressed eigenvector, and e is the base of a natural number.

[0193] Okay, let's describe this process in detail and use a specific example to show how to use the weight matrix and activation function to generate a compressed feature vector.

[0194] Assume there is a weight matrix W and an input feature vector X:

[0195] Weight matrix W: This is a 7 times n matrix, where m = 7 represents the dimension of the input feature and n is the feature dimension that you want to compress.

[0196] In the above example, assuming that we want to compress the features to n=3, the weight matrix W can be expressed as:

[0197] W=

[0198] w_{11}w_{12}w_{13}

[0199] w_{21}w_{22}w_{23}

[0200] w_{31}w_{32}w_{33}

[0201] w_{41}w_{42}w_{43}

[0202] w_{51}w_{52}w_{53}

[0203] w_{61}w_{62}w_{63}

[0204] w_{71}w_{72}w_{73};

[0205] Input feature vector X: This is a 7x1 column vector containing the features we extracted from the cold chain transportation data.

[0206] For example:

[0207] X=

[0208] T_{mean}

[0209] T_{max}

[0210] T_{std}

[0211] H_{mean}

[0212] H_{std}

[0213] text{weight}

[0214] text{volume};

[0215] The formula for calculating the first output value Z is:

[0216] ;

[0217] Where X represents the input multidimensional feature vector, and b represents the bias term;

[0218] b can be a vector of ntimes1:

[0219] b=

[0220] b_1

[0221] b_2

[0222] b_3;

[0223] So the process of calculating Z is as follows:

[0224] First, calculate the product of the transpose of the weight matrix and the input feature vector:

[0225] WTcdotX=

[0226] w_{11}w_{21}w_{31}w_{41}w_{51}w_{61}w_{71}

[0227] w_{12}w_{22}w_{32}w_{42}w_{52}w_{62}w_{72}

[0228] w_{13}w_{23}w_{33}w_{43}w_{53}w_{63}w_{73}

[0229] cdot

[0230] T_{mean}

[0231] T_{max}

[0232] T_{std}

[0233] H_{mean}

[0234] H_{std}

[0235] text{weight}

[0236] text{volume};

[0237] This will produce a 3x1 vector Z. In this example, cdot represents a multiplication operation, which is used to identify the multiplication of matrix or vector elements.

[0238] Second, add the bias term b to the result of the matrix product:

[0239] Z=

[0240] z_1

[0241] z_2

[0242] z_3;

[0243] w_{11}cdotT_{mean}+w_{21}cdotT_{max}+w_{31}cdotT_{std}+w_{41}cdotH_{mean}+w_{51}cdotH_{std}+w_{61}cdottext{weight}+w_{71}cdottext{volume}+b_1;

[0244] w_{12}cdotT_{mean}+w_{22}cdotT_{max}+w_{32}cdotT_{std}+w_{42}cdotH_{mean}+w_{52}cdotH_{std}+w_{62}cdottext{weight}+w_{72}cdottext{volume}+b_2;

[0245] w_{13}cdotT_{mean}+w_{23}cdotT_{max}+w_{33}cdotT_{std}+w_{43}cdotH_{mean}+w_{53}cdotH_{std}+w_{63}cdottext{weight}+w_{73}cdottext{volume}+b_3;

[0246] Next, the first output value Z is input into the activation function to obtain the compressed feature vector A. In this embodiment, the sigmoid activation function that can be used is as follows:

[0247] A=frac{1}{1+e^{-Z}}

[0248] This formula means that for each element z_i (i=1,2,3) of Z, the following calculation is performed:

[0249] A_i=frac{1}{1+e^{-z_i}},i=1,2,3;

[0250] Finally, the compressed feature vector (A) is obtained:

[0251] A=

[0252] A_1

[0253] A_2

[0254] A_3

[0255] Through the above process, the original multi-dimensional feature vector X can be compressed into a low-dimensional feature vector (A), which is helpful for subsequent data analysis and modeling. This process also retains important information in the data and removes redundant parts, laying the foundation for further processing of cold chain transportation data.

[0256] S109, the central management platform uses a preconfigured decoder to reconstruct the low-dimensional features to obtain a reconstructed data vector;

[0257] In step S109, the central management platform inputs the low-dimensional feature representation into the decoder, and the decoder reconstructs a reconstructed data vector as close to the original data as possible. This reconstructed data vector is compared with the original multi-dimensional feature vector to calculate the reconstruction error, which is used to determine whether there is any abnormality in the data.

[0258] In this step, the decoder structure corresponds to the encoder and can be a fully connected neural network. The input dimension of the decoder is equal to the low-dimensional feature dimension after compression by the encoder, and the output dimension is consistent with the dimension of the original multi-dimensional feature vector. The network weights and bias items of the decoder have been optimized during the model training phase to "restore" the low-dimensional features to a form close to the original input.

[0259] First, the low-dimensional feature vector A is input into the decoder. Assume that A is a three-dimensional vector (A_1, A_2, A_3), which represents the features compressed by the encoder.

[0260] In the decoder, the low-dimensional feature vector A is processed by the weight matrix and the bias term to generate the reconstructed data vector X^, which is calculated as follows:

[0261] X^=W′·A+b′;

[0262] Where W′ is the weight matrix of the decoder, which is used to map the low-dimensional vector A back to the high-dimensional feature space.

[0263] b′ is the bias term of the decoder. After calculation, the reconstructed vector X^ contains features similar to the original data.

[0264] You can also further use activation functions (such as sigmoid or ReLU) to transform each element of X^ to ensure that the reconstruction result is within a reasonable range and enhance the nonlinear expression ability of the model. For example, when using the sigmoid activation function:

[0265] X^i=1\(1+e−(W′·A+b′))

[0266] Through the above steps, the reconstructed vector X^ generated by the decoder is as close as possible to the original multi-dimensional feature vector X, retaining important features such as environmental data, loading status and path information.

[0267] S110, the central management platform calculates a reconstruction error between the multidimensional feature vector and the reconstructed data vector;

[0268] In step S110, the central management platform calculates the reconstruction error between the original multidimensional feature vector and the reconstructed data vector. If the reconstruction error is too large, it indicates that the current environment or state may be abnormal; if the error is small, the system determines that the transportation environment and cargo state are normal. The reconstruction error is a key indicator for judging the transportation environment and is used to discover potential abnormalities.

[0269] In an optional embodiment, in order to determine whether the data is abnormal, the reconstruction error between the reconstructed data vector X^ and the original feature vector X is calculated. The error can be calculated using the mean square error MSE formula:

[0270] ;

[0271] in, represents the multi-dimensional feature vector, Denotes the reconstructed data vector.

[0272] Further, it can be determined whether the value of the MSE exceeds a preset error threshold;

[0273] If it exceeds, it is determined that there is an abnormality. If it does not exceed, it means there is no abnormality.

[0274] In this embodiment, the error threshold Threshold can be determined through historical data analysis, statistical distribution or empirical value, reflecting a normal error range.

[0275] If MSE ≤ Threshold, it means that the reconstructed data is consistent with the original data, and the system believes that the current transportation environment and cargo status are normal.

[0276] If MSE>Threshold, it means that the current environment or status may be abnormal. The system can further record the abnormal situation or trigger an alarm mechanism.

[0277] S111, the central management platform analyzes the reconstruction error and generates an analysis result;

[0278] In step S111, the central management platform will conduct an in-depth analysis of the calculated reconstruction error. By setting the reconstruction error threshold, it is determined whether there is a potential risk. If the error exceeds the preset threshold, the corresponding analysis result will be generated to indicate abnormal risk; if it is lower than the threshold, it is judged to be normal. The analysis result is used for system decision-making and environmental control.

[0279] S112. The central management platform automatically adjusts the environmental control equipment in the cold chain transport vehicle according to the analysis results.

[0280] In step S112, the central management platform automatically adjusts the environmental control equipment of the cold chain transport vehicle based on the analysis results. If the temperature or humidity deviates from the cargo requirements, the central management platform will control the air conditioner, humidifier and other equipment to adjust the environmental parameters to an appropriate range to ensure the safety of the cargo.

[0281] This embodiment has the following beneficial effects:

[0282] 1. Through the sensor configuration module, the target sensor can be dynamically selected and configured according to different cargo types and transportation requirements to ensure the pertinence and accuracy of environmental monitoring, reduce unnecessary sensor resource consumption, and improve the adaptability of cold chain transportation.

[0283] 2. The central management platform uses the encoder to compress the multi-dimensional feature vector to obtain a low-dimensional feature representation. This process reduces the burden of data transmission and storage while retaining core information, which helps to perform rapid data processing in cold chain transportation environments with high real-time requirements.

[0284] 3. By reconstructing the data vector through the decoder and analyzing the reconstruction error, anomalies in environmental data or cargo status can be detected in a timely manner. For example, when the reconstruction error exceeds the threshold, the system can identify the abnormal situation and provide support for possible risk warnings in cold chain transportation.

[0285] 4. The analysis results can be used to adjust the driving route to ensure that cold chain transport vehicles choose the optimal route in logistics transportation, thereby saving time and fuel costs, and realizing intelligent risk avoidance of the route in special circumstances.

[0286] 5. Automatically adjust environmental control equipment (such as temperature control devices, humidity controllers, etc.) according to the analysis results to achieve closed-loop control of the cold chain transportation environment, ensure that parameters such as temperature and humidity of the goods during the entire transportation process remain in the optimal range, and improve the preservation effect and transportation quality of the goods.

[0287] 6. The system collects environmental data, loading status information and driving routes, and constructs multi-dimensional feature vectors, which are analyzed and managed on the platform to form detailed transportation data records, facilitating the traceability of the transportation process and the optimization of logistics processes.

[0288] In another optional embodiment, after obtaining the analysis results, the Dijkstra algorithm may be further called to generate an optimal driving path for the cold chain transport vehicle based on the analysis results.

[0289] This embodiment specifically includes the following steps:

[0290] Step 1: Model the cold chain vehicle transportation route into a graph structure, where the nodes represent locations, the edges represent road sections, and the edge weights represent the transportation cost of each road section;

[0291] Step 2: Set the current cold chain vehicle location as the starting node;

[0292] Step 3: Set the destination location as the target node;

[0293] Step 4: Initialize a distance value for each node, set the distance value of the starting node to 0, and set the distances of other nodes to infinity;

[0294] Step 5: Use a priority queue or minimum heap to record the order of visiting nodes and start visiting from the starting node;

[0295] Step 6: For the current node, check all its adjacent nodes and calculate the potential distance to the adjacent nodes through the current node;

[0296] Step 7: If the path through the current node is shorter than the known shortest path, update the shortest path of the adjacent node and add the adjacent node to the priority queue;

[0297] Step 8: Repeat steps 6 to 7 above until the shortest path to the target node is determined or the priority queue is empty.

[0298] In this embodiment, the cold chain transportation route is modeled as a graph structure, where:

[0299] Vertices: represent different geographical locations or stations along the way.

[0300] Edges: represent the road segments connecting these locations.

[0301] Weights: Assign a weight to each edge. The weight represents the transportation cost of each road section, such as travel time, fuel consumption, road condition impact, etc.

[0302] The value of the weight can be based on different cost factors, such as transportation time, fuel consumption, temperature impact, real-time road conditions and other information.

[0303] The current location of the cold chain vehicle is set as the starting node in the graph, representing the current geographical location of the vehicle.

[0304] Set the destination location of the cold chain vehicle as the target node in the graph. The target node can be the final delivery location of the transportation.

[0305] Set a distance value for each node, which represents the shortest known path distance from the starting node to the node.

[0306] Initialize the distance value of the starting node to 0, because the distance from the starting node to itself is 0.

[0307] The distance values ​​of all other nodes are initially set to infinity, indicating that the path to the node has not yet been calculated.

[0308] Use priority queues or min-heaps to manage node access

[0309] Priority queue or minimum heap: Use a priority queue (such as a minimum heap) to manage and record the order of nodes to be visited. The priority queue can dynamically select the node with the smallest distance from the unvisited nodes to ensure the efficiency of the algorithm. Add the starting node to the priority queue and start visiting.

[0310] Then, take the node with the smallest distance value from the priority queue as the current node. Traverse all adjacent nodes of the current node (i.e. nodes that have direct edge connections with the current node).

[0311] For each adjacent node, calculate the potential distance from the starting node through the current node to reach the adjacent node. The calculation method is as follows:

[0312] Potential distance = distance of current node + weight of current node to adjacent node;

[0313] Compare the shortest path distance currently recorded by the adjacent node with the potential path distance. If the path distance through the current node is shorter, update the shortest path distance of the adjacent node to the potential path distance, and add the adjacent node to the priority queue. Reorder the nodes through the priority queue so that the nodes with the shortest distance are visited first.

[0314] Repeat steps 6 and 7, visit nodes step by step, and update the shortest path distance until the target node is visited or the priority queue is empty. When the target node is reached, the shortest path from the start node to the target node has been determined. Based on the recorded shortest path information, an optimal path is generated as the vehicle's driving route. If an abnormality occurs or the route needs to be adjusted, the central management platform can call the Dijkstra algorithm again to replan the route. The driving environment of cold chain vehicles may be affected by factors such as road conditions and weather changes. Real-time data (such as traffic flow, accident information, etc.) can be dynamically updated to the weights to regenerate a better driving path.

[0315] The above embodiment describes in detail the intelligent cold chain transportation information management method provided in this application. The following describes the intelligent cold chain transportation information management system provided in this application:

[0316] See also Figure 3 The present application provides an embodiment of a smart cold chain transportation information management system, and the embodiment of the system includes:

[0317] An identification unit 301 is used to identify the type of goods currently being transported, transportation requirements and environmental requirements through a task identification module;

[0318] The sensor configuration unit 302 is used to dynamically determine the target sensor among the sensors of the cold chain transport vehicle through the sensor configuration module according to the task identification result, and configure the target sensor;

[0319] A real-time collection unit 303, used to collect real-time environmental data through the target sensor;

[0320] A path construction unit 304 is used to collect the location information and driving status data of the cold chain transport vehicle through the vehicle-mounted GPS module and construct a driving path;

[0321] The RFID unit 305 is used to collect the loading status information of the goods through the RFID tags set on the goods;

[0322] A sending unit 306, configured to send the environmental data, loading status information and the driving path to a central management platform;

[0323] A multi-dimensional feature construction unit 307, configured to construct a multi-dimensional feature vector based on the environmental data, the loading status information and the driving path;

[0324] The encoder unit 308 is used to input the multi-dimensional feature vector into a pre-configured encoder for compression to obtain a low-dimensional feature representation;

[0325] A decoder unit 309, configured to reconstruct the low-dimensional features using a preconfigured decoder to obtain a reconstructed data vector;

[0326] An error calculation unit 310, used to calculate a reconstruction error between the multidimensional feature vector and the reconstructed data vector;

[0327] An analysis unit 311 is used to analyze the reconstruction error and generate an analysis result;

[0328] The control unit 312 is used to automatically adjust the environmental control equipment in the cold chain transport vehicle according to the analysis results.

[0329] Optionally, the sensor configuration unit 302 is specifically configured to:

[0330] Based on the task identification result, the sensor configuration module reads the preconfigured mapping table to determine the target sensor among the sensors of the cold chain transportation vehicle;

[0331] The sensor configuration module reads a sensor database and configures the target sensor based on the sensor database, the type of the transported goods, transport requirements and environmental requirements.

[0332] Optionally, the multi-dimensional feature construction unit 307 is specifically used for:

[0333] calculating statistical features in the environmental data;

[0334] The type of cargo currently being transported generates a binary cargo feature;

[0335] Count the weight and volume characteristics of the goods currently being transported;

[0336] Extract the changes in longitude and latitude from GPS data and calculate the distance and time characteristics of travel;

[0337] Generate context features based on the geographical location and weather information of the driving route;

[0338] The statistical features, binary cargo features, weight features, volume features, distance features, time features and context features are integrated into a multi-dimensional feature vector.

[0339] Optionally, the encoder unit 308 is specifically configured to:

[0340] Define an m*n weight matrix W, where m is the latitude of the input feature, m=7, and n is the desired feature latitude of the compression;

[0341] The first output value Z is calculated by the following formula:

[0342] ;

[0343] Where X represents the input multidimensional feature vector, and b represents the bias term;

[0344] The first output value Z is passed into the following activation function:

[0345]

[0346] A represents the compressed eigenvector, and e is the base of a natural number.

[0347] Optionally, the error calculation unit 310 is specifically used for:

[0348] The mean square error (MSE) is calculated as follows:

[0349] ;

[0350] in, represents the multi-dimensional feature vector, represents the reconstructed data vector;

[0351] The analysis unit 311 is specifically used for:

[0352] Determine whether the value of the MSE exceeds a preset error threshold;

[0353] If it exceeds, it is determined that an abnormality exists.

[0354] Optionally, the method further includes a path generating unit 313, configured to:

[0355] According to the analysis results, the Dijkstra algorithm is called to generate the optimal driving path for the cold chain transport vehicle.

[0356] Optionally, the path generating unit 313 is specifically configured to:

[0357] The transportation route of cold chain vehicles is modeled as a graph structure, where the nodes represent locations, the edges represent road sections, and the weights of the edges represent the transportation costs of each road section.

[0358] Set the current cold chain vehicle position as the starting node;

[0359] Set the destination location as the target node;

[0360] Initialize a distance value for each node, set the distance value of the starting node to 0, and set the distances of other nodes to infinity;

[0361] Use a priority queue or minimum heap to record the order in which nodes are visited, and start visiting from the starting node;

[0362] For the current node, check all its adjacent nodes and calculate the potential distance to the adjacent nodes through the current node;

[0363] If the path through the current node is shorter than the known shortest path, the shortest path of the adjacent node is updated and the adjacent node is added to the priority queue;

[0364] Repeat the above steps until the shortest path to the target node is determined or the priority queue is empty.

[0365] See also Figure 4 , the present application also provides a smart cold chain transportation information management device, including:

[0366] Processor 401, memory 402, input and output unit 403, bus 404;

[0367] The processor 401 is connected to the memory 402, the input and output unit 403 and the bus 404;

[0368] The memory 402 stores a program, and the processor 401 calls the program to execute any of the above methods.

[0369] The present application also relates to a computer-readable storage medium, on which a program is stored. When the program is run on a computer, the computer executes any of the above methods.

[0370] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0371] In the several embodiments provided in the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or units, which can be electrical, mechanical or other forms.

[0372] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0373] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0374] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, read-only memory), random access memory (RAM, random access memory), disk or optical disk and other media that can store program code.

Claims

1. A smart cold chain transportation information management method, characterized in that: The method comprises: Identify the type of goods currently being transported, transportation needs and environmental requirements through the task identification module; According to the task identification result, the sensor configuration module dynamically determines the target sensor among the sensors of the cold chain transport vehicle, and configures the target sensor; Collecting real-time environmental data through the target sensor; The location information and driving status data of cold chain transport vehicles are collected through the vehicle-mounted GPS module, and the driving route is constructed; Collect the loading status information of the goods through the RFID tags set on the goods; Sending the environmental data, loading status information and the driving path to a central management platform; The central management platform constructs a multi-dimensional feature vector based on the environmental data, loading status information and the driving path; The central management platform inputs the multi-dimensional feature vector into a pre-configured encoder for compression to obtain a low-dimensional feature representation; The central management platform reconstructs the low-dimensional features using a preconfigured decoder to obtain a reconstructed data vector; The central management platform calculates a reconstruction error between the multidimensional feature vector and the reconstructed data vector; The central management platform analyzes the reconstruction error and generates an analysis result; The central management platform automatically adjusts the environmental control equipment in the cold chain transport vehicle according to the analysis results; The central management platform constructs a multidimensional feature vector based on the environmental data, the loading status information and the driving path, including: calculating statistical features in the environmental data; The type of cargo currently being transported generates a binary cargo feature; Count the weight and volume characteristics of the goods currently being transported; Extract the changes in longitude and latitude from GPS data and calculate the travel distance and time characteristics; Generate context features based on the geographical locations and weather information of the driving route; integrating the statistical features, binary cargo features, weight features, volume features, distance features, time features, and context features into a multidimensional feature vector; The central management platform inputs the multi-dimensional feature vector into a pre-configured encoder for compression to obtain a low-dimensional feature representation including: Define an m*n weight matrix W, where m is the latitude of the input feature, m=7, and n is the desired feature latitude of the compression; The first output value Z is calculated by the following formula: ； Where X represents the input multidimensional feature vector, and b represents the bias term; The first output value Z is passed into the following activation function: A represents the compressed eigenvector, and e is the base of a natural number.

2. The intelligent cold chain transportation information management method according to claim 1 is characterized in that: The step of dynamically determining a target sensor from among sensors of the cold chain transport vehicle through a sensor configuration module according to the task identification result and configuring the target sensor includes: Based on the task identification result, the sensor configuration module reads the preconfigured mapping table to determine the target sensor among the sensors of the cold chain transportation vehicle; The sensor configuration module reads a sensor database and configures the target sensor based on the sensor database, the type of the transported goods, transport requirements and environmental requirements.

3. According to the intelligent cold chain transportation information management method described in claim 1, it is characterized in that: The central management platform calculates the reconstruction error between the multidimensional feature vector and the reconstructed data vector including: The mean square error (MSE) is calculated as follows: ； in, represents the multi-dimensional feature vector, represents the reconstructed data vector; The central management platform analyzes the reconstruction error and generates analysis results including: Determine whether the value of the MSE exceeds a preset error threshold; If it exceeds, it is determined that an abnormality exists.

4. The intelligent cold chain transportation information management method according to claim 1 is characterized in that: The method further comprises: According to the analysis results, the Dijkstra algorithm is called to generate the optimal driving path for the cold chain transport vehicle.

5. The intelligent cold chain transportation information management method according to claim 4 is characterized in that: According to the analysis results, calling the Dijkstra algorithm to generate an optimal driving path for the cold chain transport vehicle includes: The transportation route of cold chain vehicles is modeled as a graph structure, where the nodes represent locations, the edges represent road sections, and the weights of the edges represent the transportation costs of each road section. Set the current cold chain vehicle position as the starting node; Set the destination location as the target node; Initialize a distance value for each node, set the distance value of the starting node to 0, and set the distances of other nodes to infinity; Use a priority queue or minimum heap to record the order in which nodes are visited, and start visiting from the starting node; For the current node, check all its adjacent nodes and calculate the potential distance to the adjacent nodes through the current node; If the path through the current node is shorter than the known shortest path, the shortest path of the adjacent node is updated and the adjacent node is added to the priority queue; Repeat the above steps until the shortest path to the target node is determined or the priority queue is empty.

6. A smart cold chain transportation information management system, characterized in that: The system comprises: An identification unit, used to identify the type of goods currently being transported, transportation requirements and environmental requirements through a task identification module; A sensor configuration unit, used to dynamically determine a target sensor among sensors of the cold chain transport vehicle through a sensor configuration module according to a task identification result, and configure the target sensor; A real-time collection unit, used to collect real-time environmental data through the target sensor; A path construction unit, used to collect the location information and driving status data of the cold chain transport vehicle through the vehicle-mounted GPS module, and to construct a driving path; An RFID unit, used to collect loading status information of the goods through an RFID tag set on the goods; A sending unit, used for sending the environmental data, loading status information and the driving path to a central management platform; A multidimensional feature construction unit, configured to construct a multidimensional feature vector based on the environmental data, the loading status information and the driving path; An encoder unit, used for inputting the multi-dimensional feature vector into a pre-configured encoder for compression to obtain a low-dimensional feature representation; A decoder unit, configured to reconstruct the low-dimensional features using a preconfigured decoder to obtain a reconstructed data vector; an error calculation unit, used to calculate a reconstruction error between the multidimensional feature vector and the reconstructed data vector; An analysis unit, used for analyzing the reconstruction error and generating an analysis result; A control unit, configured to automatically adjust the environmental control equipment in the cold chain transport vehicle according to the analysis result; The multi-dimensional feature construction unit is specifically used for: calculating statistical features in the environmental data; The type of cargo currently being transported generates a binary cargo feature; Count the weight and volume characteristics of the goods currently being transported; Extract the changes in longitude and latitude from GPS data and calculate the travel distance and time characteristics; Generate context features based on the geographical locations and weather information of the driving route; integrating the statistical features, binary cargo features, weight features, volume features, distance features, time features, and context features into a multidimensional feature vector; The encoder unit is specifically used for: Define an m*n weight matrix W, where m is the latitude of the input feature, m=7, and n is the desired feature latitude of the compression; The first output value Z is calculated by the following formula: ； Where X represents the input multidimensional feature vector, and b represents the bias term; The first output value Z is passed into the following activation function: A represents the compressed eigenvector, and e is the base of a natural number.

7. A smart cold chain transportation information management device, characterized in that: include: Processor, memory, input-output unit, and bus; The processor is connected to the memory, the input and output unit, and the bus; The memory stores a program, and the processor calls the program to execute the method according to any one of claims 1 to 5.

8. A computer-readable storage medium having a program stored thereon, wherein the program, when executed on a computer, performs the method according to any one of claims 1 to 5.

Images