Goods full-process tracking method for logistics transportation

Abstract

The invention discloses a cargo full-process tracking method for logistics transportation, and relates to the technical field of logistics transportation tracking, and the method comprises the following steps: collecting cargo state data and transportation route data in real time through a sensor and a camera which are installed on a transportation vehicle and a cargo; multi-type sensors and cameras are deployed on transport vehicles and goods, real-time acquisition of goods state and transport route data is realized, a multi-dimensional prediction model is constructed through a long-short-term memory network and a convolutional neural network, and anomaly identification is performed, so that the real-time prediction of the goods state and the transport route data is realized. According to the invention, accurate detection of goods theft, deviation of a vehicle from a predetermined route, abnormal temperature and humidity, abnormal pressure and the like is realized, automatic alarm and intelligent processing of abnormal conditions are realized through a preset hierarchical response mechanism, the problems of untimely monitoring, poor abnormal identification capability and the like in logistics transportation are effectively solved, and the logistics transportation safety is improved. And the safety, controllability and management efficiency of cargo transportation are obviously improved.

Description

Technical Field

[0001] The present invention relates to the technical field of logistics and transportation tracking, and in particular to a method for tracking the entire process of goods used in logistics and transportation. Background Art

[0002] Logistics and transportation refers to the physical flow of goods from the point of supply to the point of receipt. By organically integrating basic functions such as transportation, storage, loading and unloading, handling, packaging, distribution, and information processing, it achieves spatial and time-efficient transfer of goods. During the logistics and transportation process, cargo tracking is a core component of ensuring supply chain transparency, improving transportation safety, and enhancing management efficiency. By acquiring real-time information such as cargo location, status, and environmental data, it ensures traceability at every stage of transportation, allowing for timely detection and intervention to prevent loss, damage, or delays. It also provides data support for logistics companies to optimize route planning, resource scheduling, and dispute resolution, ultimately achieving the goal of reducing operational risks and enhancing customer trust.

[0003] According to patent application number 202410120982.1, a Beidou-based logistics tracking and query method is disclosed, including the following steps: obtaining multiple data information when the Beidou satellite positioning terminal performs logistics tracking and query, wherein the multiple data information includes the communication quality index between the Beidou satellite positioning terminal and the Beidou satellite system and the receiving signal characteristics of the satellite signal during the transmission and reception process; after obtaining, the communication quality index and the receiving signal characteristics are analyzed and processed.

[0004] The above scheme can be used to control the positioning accuracy of logistics tracking and inquiry by Beidou satellite positioning terminal in real time. However, the existing technology still has certain defects in application. During the logistics transportation process, the monitoring of cargo status and transportation routes mainly relies on manual inspection, simple GPS positioning or Beidou positioning. This method has problems such as untimely monitoring and poor ability to identify abnormal situations. It is difficult to accurately grasp the full-process transportation status of cargo in real time, resulting in abnormal situations such as cargo theft and vehicle deviation from the scheduled route. It cannot be discovered and handled in time, which brings great safety risks and management difficulties to logistics transportation. Therefore, it is of great significance to develop a full-process tracking method for cargo in logistics transportation. Summary of the Invention

[0005] The purpose of the present invention is to make up for the shortcomings of the existing technology and provide a full-process tracking method for goods used in logistics transportation. It can deploy multiple types of sensors and cameras on transport vehicles and goods to realize real-time collection of cargo status and transportation route data, build a multi-dimensional prediction model and perform anomaly identification, and realize accurate detection of cargo theft, vehicle deviation from the planned route, abnormal temperature and humidity, abnormal pressure, etc. Through a preset hierarchical response mechanism, automatic alarm and intelligent processing of abnormal situations are realized.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions: a method for tracking the entire process of goods for logistics transportation, the method comprising the following steps:

[0007] Utilize sensors and cameras installed on transport vehicles and cargo to collect cargo status data and transport route data in real time. Cargo status data includes cargo location, movement status, temperature, humidity, pressure, and appearance. Transport route data includes vehicle real-time location, speed, direction, and planned route.

[0008] The collected data is transmitted to the cloud server and logistics management center database in real time through the wireless communication module;

[0009] Preprocess the transmitted data, including data cleaning, data conversion and data normalization;

[0010] Use a long short-term memory network to perform real-time analysis on pre-processed data, establish a normal cargo transportation status model and a normal transportation route model based on historical data, compare the current data with the normal model, and identify abnormal cargo status and transportation routes. This includes calculating the deviation between the current status and the normal model based on a spatiotemporal correlation formula;

[0011] When an abnormal situation is identified, an alarm is issued through sound alarms, SMS notifications, and APP push notifications, and processing measures are automatically taken according to preset rules. At the same time, the collected data and abnormal processing records are stored for a long time for query, statistics, and analysis.

[0012] Furthermore, the sensors include a GPS sensor, an acceleration sensor, a temperature sensor, a humidity sensor, and a pressure sensor. The camera is used to capture the appearance of the goods to obtain image data. The image data is preprocessed and input into a convolutional neural network model. Abnormal features in the image are extracted based on multi-scale feature fusion to detect damage to the appearance of the goods and abnormal packaging.

[0013] Furthermore, the wireless communication module uses an encryption protocol to encrypt data during data transmission, and the encryption protocol includes an SSL / TLS encryption algorithm.

[0014] Furthermore, the data cleaning includes removing noise from time series data through a sliding window filtering algorithm, eliminating erroneous data using data consistency verification rules, and identifying and deleting duplicate data using a hash deduplication algorithm. The data conversion includes converting analog signal data output by different sensors into digital signal data, and unifying the formats of heterogeneous data sources. The data normalization uses a minimum-maximum normalization method to convert data of different dimensions into values ​​within the range of [0, 1].

[0015] Furthermore, the long short-term memory network model constructs a multi-dimensional normal state prediction model containing time series features through supervised learning of historical normal transportation data. The model input layer receives preprocessed multi-dimensional data, captures the temporal dependency through the LSTM layer, and outputs the deviation value between the current state and the normal model through the fully connected layer. Based on the preset threshold, it is determined whether to trigger abnormal recognition. The deviation value is calculated based on the spatiotemporal correlation formula: Among them, S(t) is the spatiotemporal correlation score at time t, x i (t) is the measured value of the i-th state parameter at time t, μ i and σ i are the historical mean and standard deviation of the i-th state parameter, w i is the weight coefficient of the i-th state parameter, d j (t) is the measured value of the jth spatial parameter at time t, λ j and ρ j are the historical mean and standard deviation of the j-th spatial parameter, v j is the weight coefficient of the jth spatial parameter. The weight coefficient is automatically determined by analyzing the importance of the features of normal and abnormal states in historical transportation data and training the model using the gradient descent optimization algorithm to maximize the model's recognition accuracy for abnormal states. α and β are the balance coefficients of the time and space dimensions determined by cross-validation, and α + β = 1.

[0016] Furthermore, the abnormal conditions include cargo theft, vehicle deviation from the planned route, abnormal cargo temperature and humidity, and abnormal cargo pressure;

[0017] The identification rule for cargo theft is: when the cargo's stationary time exceeds a preset threshold and its distance from the planned route exceeds a preset distance threshold, and the acceleration sensor data shows no continuous motion signal;

[0018] The identification rule of the vehicle deviating from the predetermined route is: the angle between the vehicle's driving direction and the predetermined route direction exceeds a preset angle threshold and the duration exceeds a preset time length;

[0019] The identification rule for abnormal cargo temperature and humidity is: the fluctuation coefficient calculated based on the temperature and humidity fluctuation coefficient formula exceeds the preset threshold. The temperature and humidity fluctuation coefficient formula is:

[0020] Among them, K TH is the temperature and humidity fluctuation coefficient, T(t) and H(t) are the measured values ​​of temperature and humidity at time t, respectively. ref and H ref are the reference values ​​of temperature and humidity, ΔT and ΔH are the allowable fluctuation ranges of temperature and humidity, respectively.

[0021] Furthermore, the preset processing measures include:

[0022] When cargo theft is detected, the vehicle's GPS lock function is automatically triggered and an alarm message is sent to the police. At the same time, an electronic report containing the vehicle's location and cargo information is generated.

[0023] When a vehicle deviates from the planned route, a voice reminder and an app pop-up window will be sent to the driver. If the deviation time exceeds the secondary threshold, a manual intervention request will be sent to the logistics management center.

[0024] When the temperature and humidity of the cargo are abnormal, the temperature and humidity control system in the cargo hold of the transport vehicle will be automatically adjusted;

[0025] When the cargo pressure is abnormal, the cargo stacking method and tightening degree are adjusted based on the pressure distribution balance formula. The pressure distribution balance formula is: Among them, P eq is the pressure balance value, P k is the measured value of the kth pressure sensor, w k is the weight coefficient of the kth pressure sensor, and its value is related to the position of the sensor on the cargo carrying surface.

[0026] Furthermore, the data storage adopts a distributed database architecture, and the distributed database includes HBase and Cassandra. The storage content includes original collected data, pre-processed data, abnormality identification results, processing measure execution records and timestamps. It supports rapid retrieval of historical data through keyword search, time range search, and cargo number search, and the data storage period is not less than 5 years.

[0027] Compared with the existing technology, this method for tracking the whole process of goods for logistics transportation has the following advantages:

[0028] Beneficial effects:

[0029] The present invention realizes real-time collection of cargo status and transportation route data by deploying multiple types of sensors and cameras on transport vehicles and cargo, utilizes wireless communication technology combined with encryption protocols to ensure the security and integrity of data transmission, constructs a multi-dimensional prediction model through long-short-term memory networks and convolutional neural networks, and performs anomaly identification to achieve accurate detection of cargo theft, vehicle deviation from the planned route, abnormal temperature and humidity, abnormal pressure, etc. Through a preset hierarchical response mechanism, automatic alarm and intelligent processing of abnormal situations are realized, effectively solving the problems of untimely monitoring and poor anomaly identification capabilities in logistics transportation, and significantly improving the safety, controllability and management efficiency of cargo transportation.

[0030] Other advantages, objects and features of the present invention will be described in part in the following description and, in part, will be apparent to those skilled in the art based on an examination of the following or may be learned from the practice of the invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] To more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. Those skilled in the art can also derive other drawings based on these drawings without inventive effort.

[0032] Figure 1 This is a flow chart of a method for tracking the entire process of goods used in logistics transportation;

[0033] Figure 2 The present invention is a flowchart of a method for tracking the entire process of goods used in logistics transportation. DETAILED DESCRIPTION

[0034] In order to further illustrate the technical means and effects adopted by the present invention to achieve the predetermined purpose of the invention, the specific implementation methods, structures, features and effects of the present invention are described in detail below in conjunction with the accompanying drawings and preferred embodiments.

[0035] Example 1

[0036] This embodiment is applied to the fresh product transportation scenario of a cold chain logistics company, targeting the full-process tracking requirements of beef products from slaughterhouse to retail terminal. Figure 1 and Figure 2 By deploying a variety of sensors and high-definition cameras on refrigerated transport vehicles and cargo packaging, multi-dimensional data such as cargo location, movement status, temperature and humidity environment, pressure distribution and appearance status are collected in real time, and transmitted to the cloud server through an encrypted wireless communication module. The long short-term memory network LSTM and the graph convolutional network GCN are used to build an anomaly detection model. Combined with the spatiotemporal correlation formula, the temperature and humidity fluctuation coefficient formula and the pressure distribution balance formula, accurate identification and graded response to scenarios such as cargo theft, vehicle deviation from the route, temperature and humidity abnormalities, and pressure abnormalities can be achieved.

[0037] A GPS sensor is installed on the top of the refrigerated transport vehicle's cab to obtain real-time vehicle location data; multiple temperature sensors and humidity sensors are evenly distributed on the inner wall of the vehicle compartment to monitor the temperature and humidity environment in different areas of the cargo hold; multiple pressure sensors are embedded in the bottom of the cargo pallet, and a thin film structure is used to adhere to the pallet surface to monitor the pressure distribution of the stacked cargo. The sensor coverage area matches the key stress points on the cargo bearing surface, and the weight coefficient is set according to the distance between the sensor position and the center of gravity of the cargo. The weight coefficient of the sensor close to the center of gravity is higher.

[0038] Multiple high-definition cameras are installed at the top corners of the carriage, with their lenses facing the cargo stacking area, to collect image data of the cargo's appearance. Miniature acceleration sensors are affixed to the outside of the beef product boxes to monitor the vibration state of the cargo during transportation; miniature temperature and humidity sensors are placed inside the boxes to form dual monitoring with external sensors to ensure the accuracy of environmental data.

[0039] Cargo status data includes cargo location, movement status, temperature, humidity, pressure, appearance status, etc. Among them, pressure data is the weighted average of multiple sensors at the bottom of the pallet, and the weight coefficient is associated with the position parameter through the pressure distribution balance formula. Transportation route data includes the vehicle's real-time location, driving speed, driving direction, and scheduled route (electronic map path pre-imported through the logistics management system, including key node coordinates).

[0040] All sensor data is collected synchronously through the on-board gateway, and the collection frequency is set to a reasonable interval based on the parameter characteristics. The camera image data is pre-processed by the edge computing device, and the key frames are extracted and transmitted at a reasonable frequency.

[0041] The on-board gateway integrates a wireless communication module and supports general communication protocols. The data transmission interval is dynamically adjusted according to the network status and automatically switches to real-time transmission mode under abnormal conditions. The encryption protocol is used to encapsulate the transmitted data to ensure that the data is not stolen or tampered with during transmission.

[0042] The cloud server is deployed on the mainstream cloud service platform, configured with a load balancer and security gateway. The logistics management center database adopts a cluster architecture and supports read-write separation and fault switching.

[0043] The raw sensor data is packaged by the on-board gateway and sent to the cloud server access layer through the wireless communication network. After security filtering and protocol conversion, it is stored in the temporary storage area of ​​the distributed file system. The camera image data is compressed and format converted by the edge computing device and pushed to the cloud image processing server through the real-time communication protocol.

[0044] For time series data such as temperature, humidity, and pressure, a sliding window filtering algorithm is used to remove random noise. The window size is reasonably set according to the data fluctuation characteristics. The data consistency verification rules are used to check whether the temperature data is within the cold chain transportation standard range, and outliers that exceed the range are eliminated; the GPS location data is checked to see if it is logically consistent with the driving speed, and contradictory data is eliminated. The hash deduplication algorithm is used to identify the acceleration sensor and pressure sensor data, and records collected repeatedly at the same time are deleted. The analog signal output by the sensor is converted into a digital signal through an analog-to-digital converter. The conversion formula is determined according to the sensor characteristics, and data from different sources and formats are uniformly converted into a format suitable for subsequent analysis and processing to facilitate data integration and analysis. The minimum-maximum normalization method is used to convert data of different dimensions into values ​​within the range of [0, 1].

[0045] The normal transportation data of the cold chain enterprise in the past period of time, including temperature, humidity, pressure, location, speed and other data accumulated over multiple hours for multiple trips, are collected as historical training data to construct a multi-layer LSTM network. The dimension of the input layer is determined according to the number of collected features, and the number of hidden layer neurons is reasonably set according to the complexity of the model. The output layer outputs the deviation value between the current state and the normal model. Appropriate loss function and optimizer are used, and parameters such as learning rate, batch size, and training cycle are dynamically adjusted according to the training effect. Through supervised learning, the model can capture the temporal dependencies in time series data.

[0046] Based on the trained LSTM model, the deviation between the current data and the normal model is calculated in real time, and the time-space correlation formula is used to integrate the time and space dimension features:

[0047] Among them, α and β are balance coefficients (α+β=1), n ​​is the number of time series parameters, m is the number of spatial parameters, and w i With v j is the weight coefficient of each parameter, μ i , σ i is the historical mean and standard deviation of the time parameter, λ j , ρ j is the historical mean and standard deviation of the spatial parameter.

[0048] The images of the cargo appearance captured by the camera are first preprocessed and then input into the CNN model for anomaly detection. The model adopts a suitable architecture, and the input size is reasonably set according to the image resolution. The output layer is classified into categories such as "normal", "damaged", and "packaging abnormality" through the softmax function. The training data uses a large number of manually annotated images (including a certain proportion of abnormal samples). Parameters such as the loss function, learning rate, and training cycle are optimized according to the training effect. Through multi-scale feature fusion, the detection accuracy of subtle damage is improved.

[0049] When the cargo's stationary time exceeds a preset threshold and its distance from the scheduled route exceeds a preset distance threshold, and the acceleration sensor data shows no continuous motion signal, it is determined to be suspected of being stolen, and the vehicle positioning and locking function is automatically triggered. An alarm message containing the vehicle's real-time location, cargo information, etc. is sent to the police, and an electronic report is generated and pushed to the logistics management center and the owner's mobile phone APP.

[0050] The angle between the vehicle's direction of travel and the direction of the planned route is calculated. If it exceeds the preset angle threshold and lasts for more than the preset time, it is determined to be a route deviation. First, a voice reminder and an APP pop-up prompt are sent to the driver. If the deviation time exceeds the secondary threshold, a manual intervention request is sent to the logistics management center. The dispatch center remotely plans a new route and sends it to the on-board navigation device.

[0051] Calculate the real-time fluctuation coefficient based on the temperature and humidity fluctuation coefficient formula:

[0052] Among them, T ref is the temperature reference value, ΔT is the allowable temperature fluctuation range, H ref is the humidity reference value, ΔH is the allowable humidity fluctuation range. When the fluctuation coefficient exceeds the preset threshold, it is determined that the temperature and humidity are abnormal, and the temperature and humidity control system in the refrigerated truck cargo compartment is automatically adjusted. If it does not return to normal within a period of time, an early warning message is sent to the logistics management center, indicating that unloading and inspection may be required midway.

[0053] Calculate the pressure equalization value: Where p is the number of pressure sensors, w k According to the sensor position setting, when the pressure balance value exceeds a reasonable multiple standard deviation of the historical mean, it is judged as abnormal pressure distribution. The on-board robotic arm system automatically adjusts the cargo stacking method and tightening degree, and recalculates the pressure balance value after adjustment until the standard is met.

[0054] A distributed database is used as the core storage engine, and a composite primary key is constructed based on the cargo number and timestamp to support high-concurrency reading and writing. An appropriate database is used to assist in storing image data, and data sharding is achieved through a distributed hash table to ensure image retrieval efficiency. The storage content includes original collected data, pre-processed data, anomaly recognition results, processing measure execution records and timestamps. The storage period is set to a reasonable number of years. It supports retrieval of relevant records by cargo name, transport order number, anomaly type, etc. The start and end time can be selected through the calendar control to query all transportation data and abnormal events within the time period. Enter the unique cargo number to obtain the entire process data from loading to receipt, including trajectory maps, environmental data curves, anomaly handling reports, etc.

[0055] To sum up, this embodiment integrates multiple types of sensors, intelligent algorithms and automatic response mechanisms to build a full-process cargo tracking system covering data collection, transmission, processing, analysis and response. The data collection and anomaly recognition speeds meet the real-time monitoring requirements, and the response speed is greatly improved compared to the traditional manual inspection method. Various emergency abnormal situations can be discovered in a timely manner. The anomaly detection solution based on multi-model fusion significantly reduces the false alarm rate, improves the detection accuracy, and effectively reduces the cost of manual misjudgment. The preset hierarchical response mechanism realizes the automatic processing of most abnormal scenarios, reduces the frequency of manual intervention, and improves transportation management efficiency. The rapid response function in emergency scenarios such as cargo theft significantly improves the cargo security capability, effectively solves the core issues of monitoring and management in cold chain logistics, and significantly improves the safety, controllability and management efficiency of cargo transportation. It has broad industry application value and promotion prospects.

[0056] Example 2

[0057] This embodiment is applied to the electronic product transportation scenario of a cross-border e-commerce company. To meet the tracking requirements of smartwatch components from factory to warehouse, a full-process tracking system integrating satellite positioning, environmental perception, image recognition and blockchain technology is constructed. Figure 1 and Figure 2 By deploying multiple types of sensors, smart tags and high-definition cameras on containers, pallets and cargo units, data such as geographic location, vibration and impact, temperature and humidity, light intensity, and packaging integrity are collected in real time and transmitted to blockchain nodes through a hybrid communication network. A multimodal anomaly detection model is constructed using the long short-term memory network LSTM and the graph convolutional network GCN. Combined with spatiotemporal association rules, the impact index formula and the packaging damage probability model, accurate identification of anomalies such as cargo loss, transportation route deviation, environmental parameter exceeding the limit, and packaging damage can be achieved.

[0058] A satellite positioning module is installed on the top of the container to support real-time positioning worldwide; an electronic seal sensor is installed on the inside of the container door to monitor the seal status in real time; multi-axis acceleration sensors are installed at the four corners of the container to monitor vibration and impact during transportation; temperature and humidity sensors and light intensity sensors are installed on the top and side walls to monitor changes in environmental parameters; an RFID tag is affixed to each pallet to store information such as the cargo list and batch number; a pressure sensor array is embedded in the bottom of the pallet to monitor the pressure distribution of the cargo stacking, and the sensor weight is dynamically adjusted according to the carrying capacity of the pallet area.

[0059] Each smartwatch component packaging box has a built-in micro impact sensor to capture abnormal impacts during loading and unloading; a visual identification label containing the product's unique identification and traceability information is affixed to the box surface; industrial-grade cameras are installed inside the container and on the sorting lines of cross-border logistics hubs to read the cargo label information through optical character recognition (OCR) technology and detect abnormalities in the packaging appearance through image recognition algorithms.

[0060] The container's real-time location, travel speed, heading angle, scheduled route nodes (multimodal transport routes imported through the cross-border logistics management system, including segmented routes), temperature and humidity, light intensity, vibration and impact parameters, pressure distribution (weighted average of pallet sensors), electronic seal status, RFID tag reading status, visual tag recognition results, and positioning data are dynamically adjusted according to the transportation stage; environmental data is collected at high frequency; and image data is collected in real time at key nodes.

[0061] Adaptable communication modules are used to transmit data under different transportation modes, with key data transmitted first; encryption algorithms are used to encrypt sensor data end-to-end, and low-power transmission is achieved through lightweight protocols; image data is compressed through edge computing and pushed to the cloud.

[0062] Blockchain nodes are deployed in cross-border e-commerce companies, logistics service providers, customs, commodity inspection and other institutions to build a consortium chain network. Key information such as sensor raw data hash values, RFID read and write records, electronic seal status changes, customs clearance documents, etc. are uploaded to the chain in real time. A consensus mechanism is used to ensure that the data cannot be tampered with. Customs clearance trigger contracts and abnormal alarm contracts are preset to achieve key process automation.

[0063] A noise reduction algorithm is used on the acceleration data to remove high-frequency noise and retain effective impact signals. Outliers are eliminated through statistical methods, and the rationality of the data is verified in combination with the transportation stage logic. Data from different sources and formats are uniformly converted into a format that is easy to analyze, facilitating subsequent fusion analysis.

[0064] Statistical features are extracted from temperature, humidity, and vibration data to calculate the deviation between the container location and the planned route, the time required for segmented transportation, etc. The damage features of the packaging images are extracted through a convolutional neural network, and the feature vector is output to construct a logistics node-container-cargo graph structure. The node and edge attributes contain multi-dimensional information.

[0065] A bidirectional LSTM network is used to process time series data, capture the temporal dependencies of environmental parameters, and output a temporal anomaly score. A GCN network is used to analyze the transportation paths between logistics nodes, identify spatial anomalies such as route deviation and detention timeout, and output a path anomaly score. The temporal score and graph structure score are fused through an attention mechanism, combined with the image damage probability, to generate a comprehensive anomaly index E(t). The formula is: E(t) = γ·S LSTM (t)+δ·S GCN (t)+∈·P CNN (t) Where γ+δ+∈=1, and the weight is determined through historical data training. When the number of RFID tags read does not match the number in the list, and the electronic seal status is abnormal, and the graph model detects a path abnormality, it is determined to be suspected loss.

[0066] Abnormal responses and smart contract execution are as follows: if the electronic seal is opened abnormally, the temperature and humidity exceed the product tolerance limit, or the impact index exceeds the limit, the smart contract will be automatically triggered to freeze the logistics fee payment and send an early warning message to the relevant parties; mandatory unpacking inspection and upload the report to the blockchain; if the route deviation exceeds the preset range, the RFID number difference, or the probability of package damage exceeds the threshold, an alert will be pushed to the operations team through the logistics app with abnormal information and historical solutions, and an on-site verification will be initiated to confirm the situation.

[0067] When the container arrives at the port, the smart contract automatically extracts the documents in the blockchain, generates and submits a declaration form that complies with the local customs format, shortening the customs clearance time. If the consignee reports that the goods are damaged, the entire process data can be traced through the blockchain, and the responsible party can be automatically identified based on the records, triggering the insurance claim process.

[0068] A distributed database is used to store real-time data for a recent period of time, supporting high-concurrency reading and writing, and partitioned storage according to multi-dimensional conditions, which facilitates quick query of the status of goods in transit. Historical data is archived to a distributed file system, combined with a search engine to achieve full-text retrieval, and supports multi-dimensional retrieval of similar events for transportation plan optimization. In the logistics management interface, the container location, transportation route, and estimated arrival time are displayed in real time on a map, with superimposed environmental parameter curves and abnormal event markers. Reports containing abnormal type distribution, high-incidence periods, and responsible party statistics are automatically generated to provide data support for enterprises to optimize logistics strategies.

[0069] In summary, this embodiment combines multimodal data fusion, blockchain technology, and smart contracts to build a full-process tracking system that adapts to the complex scenarios of cross-border e-commerce logistics. Through route optimization and abnormality prevention, transportation timeliness is improved, cargo damage rate is reduced, and significant cost savings are achieved. This solution provides a replicable technical paradigm for the intelligent and compliant development of cross-border e-commerce logistics, and is particularly suitable for cross-border transportation scenarios of high-value and fragile goods, and has significant industry demonstration significance.

[0070] The above description is merely a preferred embodiment of the present invention and does not constitute any form of limitation to the present invention. Although the present invention has been disclosed as above in terms of a preferred embodiment, it is not intended to limit the present invention. Any person skilled in the art can, without departing from the scope of the technical solution of the present invention, make some changes or modifications to equivalent embodiments using the technical contents disclosed above. However, any brief modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solution of the present invention are still within the scope of the technical solution of the present invention.

Claims

1. A method for tracking the entire process of goods for logistics transportation, characterized in that: The method comprises the following steps: Utilize sensors and cameras installed on transport vehicles and cargo to collect cargo status data and transport route data in real time. Cargo status data includes cargo location, movement status, temperature, humidity, pressure, and appearance. Transport route data includes vehicle real-time location, speed, direction, and planned route. The collected data is transmitted to the cloud server and logistics management center database in real time through the wireless communication module; Preprocess the transmitted data, including data cleaning, data conversion and data normalization; Use a long short-term memory network to perform real-time analysis on pre-processed data, establish a normal cargo transportation status model and a normal transportation route model based on historical data, compare the current data with the normal model, and identify abnormal cargo status and transportation routes. This includes calculating the deviation between the current status and the normal model based on a spatiotemporal correlation formula; When an abnormal situation is identified, an alarm is issued through sound alarms, SMS notifications, and APP push notifications, and processing measures are automatically taken according to preset rules. At the same time, the collected data and abnormal processing records are stored for a long time.

2. A method for tracking the entire process of goods for logistics transportation according to claim 1, characterized in that: The sensors include a GPS sensor, an acceleration sensor, a temperature sensor, a humidity sensor, and a pressure sensor. The camera is used to capture the appearance of the goods to obtain image data. The image data is preprocessed and input into a convolutional neural network model. Abnormal features in the image are extracted based on multi-scale feature fusion to detect damage to the appearance of the goods and abnormal packaging.

3. A method for tracking the entire process of goods for logistics transportation according to claim 1, characterized in that: The wireless communication module uses an encryption protocol to encrypt data during data transmission, and the encryption protocol includes an SSL / TLS encryption algorithm.

4. A method for tracking the entire process of goods for logistics transportation according to claim 1, characterized in that: The data cleaning includes removing noise from time series data through a sliding window filtering algorithm, eliminating erroneous data using data consistency verification rules, and identifying and deleting duplicate data using a hash deduplication algorithm. The data conversion includes converting analog signal data output by different sensors into digital signal data, and unifying the formats of heterogeneous data sources. The data normalization uses a minimum-maximum normalization method to convert data of different dimensions into values ​​within the range of [0, 1].

5. A method for tracking the entire process of goods for logistics transportation according to claim 1, characterized in that: The long short-term memory network model constructs a multi-dimensional normal state prediction model containing time series features through supervised learning of historical normal transportation data. The model input layer receives pre-processed multi-dimensional data, captures the temporal dependency through the LSTM layer, and outputs the deviation value between the current state and the normal model through the fully connected layer. Based on the preset threshold, it is determined whether to trigger abnormal recognition. The deviation value is calculated based on the spatiotemporal correlation formula: Among them, S(t) is the spatiotemporal correlation score at time t, x i (t) is the measured value of the i-th state parameter at time t, μ i and σ i are the historical mean and standard deviation of the i-th state parameter, w i is the weight coefficient of the i-th state parameter, d j (t) is the measured value of the jth spatial parameter at time t, λ j and ρ j are the historical mean and standard deviation of the j-th spatial parameter, v j is the weight coefficient of the jth spatial parameter, α and β are the balance coefficients of time and space dimensions, and α+β=1.

6. A method for tracking the entire process of goods for logistics transportation according to claim 1, characterized in that: Abnormal conditions include cargo theft, vehicle deviation from the planned route, abnormal cargo temperature and humidity, and abnormal cargo pressure; The identification rule for cargo theft is: when the cargo's stationary time exceeds a preset threshold and its distance from the planned route exceeds a preset distance threshold, and the acceleration sensor data shows no continuous motion signal; The identification rule of the vehicle deviating from the predetermined route is: the angle between the vehicle's driving direction and the predetermined route direction exceeds a preset angle threshold and the duration exceeds a preset time length; The identification rule for abnormal cargo temperature and humidity is: the fluctuation coefficient calculated based on the temperature and humidity fluctuation coefficient formula exceeds the preset threshold. The temperature and humidity fluctuation coefficient formula is: Among them, KT H is the temperature and humidity fluctuation coefficient, T(t) and H(t) are the measured values ​​of temperature and humidity at time t, respectively. ref and H ref are the reference values ​​of temperature and humidity, ΔT and ΔH are the allowable fluctuation ranges of temperature and humidity, respectively.

7. A method for tracking the entire process of goods for logistics transportation according to claim 1, characterized in that: The preset processing measures include: When cargo theft is detected, the vehicle's GPS lock function is automatically triggered and an alarm message is sent to the police. At the same time, an electronic report containing the vehicle's location and cargo information is generated. When a vehicle deviates from the planned route, a voice reminder and an app pop-up window will be sent to the driver. If the deviation time exceeds the secondary threshold, a manual intervention request will be sent to the logistics management center. When the temperature and humidity of the cargo are abnormal, the temperature and humidity control system in the cargo hold of the transport vehicle will be automatically adjusted; When the cargo pressure is abnormal, the cargo stacking method and tightening degree are adjusted based on the pressure distribution balance formula. The pressure distribution balance formula is: Among them, P eq is the pressure balance value, P k is the measured value of the kth pressure sensor, W k is the weight coefficient of the kth pressure sensor, and its value is related to the position of the sensor on the cargo carrying surface.

8. A method for tracking the entire process of goods for logistics transportation according to claim 1, characterized in that: The data storage adopts a distributed database architecture, and the storage content includes original collected data, pre-processed data, abnormality recognition results, processing measure execution records and timestamps, and supports rapid retrieval of historical data through keyword search, time range search, and cargo number search.

Images