A real-time query method, system, medium and product for dynamic information of a container cargo

Abstract

A kind of real-time query method, system, medium and product of box cargo dynamic information, involve port logistics management field, this method includes: receiving the yard image data and laser ranging data collected by unmanned aerial vehicle, constructs three-dimensional digital model;According to yard image data, the box number information of each container is identified, and box cargo position data is generated;Collect the electronic tag information of each container, generate real-time box cargo information;Identify the surface feature of each container, and compare the surface feature with the damage feature library, generate damage detection result;Damage detection result is associated with real-time box cargo information, and the box cargo information database indicating the state of container is updated;Receive query request, extract the target box cargo information corresponding to query request from box cargo information database, and generate query report based on target box cargo information.The application can improve the efficiency of container inspection, and ensure the timely update of box dynamic information.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of port logistics management, and in particular to a real-time query method and system for container cargo dynamic information, a medium and a product. BACKGROUND

[0002] With the rapid development of international trade, the throughput of port containers continues to grow, and real-time access to container cargo dynamic information has become a basic requirement for port operations. Container cargo dynamic information includes real-time location, loading and unloading status, damage condition and other data of containers, which are of great significance to improving port operation efficiency, ensuring cargo safety and optimizing resource allocation.

[0003] In related technologies, port logistics management mainly uses fixed video monitoring systems for container cargo tracking. Fixed cameras are installed around the container yard, and the yard conditions are monitored in real time by the video monitoring center. The operator records the changes in the location of the container by viewing the monitoring screen. In terms of damage detection, the inspector regularly inspects the yard to record and manually fill out forms for any container damage found.

[0004] However, in the monitoring scheme of fixed cameras, due to the limitations of the angle of view and coverage range, there are many blind spots for monitoring of stacked containers, and the data of containers in some areas is difficult to collect and update. Although manual inspection can make up for the monitoring blind spots, the inspection efficiency is limited by human resources, and there is a situation of delayed information update. SUMMARY

[0005] The present application provides a real-time query method and system for container cargo dynamic information, a medium and a product for improving container inspection efficiency and ensuring timely update of container dynamic information.

[0006] In a first aspect, the present application provides a real-time query method for container cargo dynamic information, applied to a logistics management system, which comprises: receiving yard image data and laser ranging data collected by a UAV, constructing a three-dimensional digital model of the container yard to determine the spatial coordinate information of each container; identifying the container number information of each container according to the yard image data, and associating the container number information with the spatial coordinate information to generate container cargo location data; collecting electronic tag information of each container based on the RFID reading device of the UAV, matching and verifying the electronic tag information with the container cargo location data to generate real-time container cargo information; performing image segmentation processing on the yard image data to identify the container cargo surface features of each container, and comparing the container cargo surface features with a damage feature library to generate a damage detection result; associating the damage detection result with the real-time container cargo information to update the container cargo information database representing the state of the container; receiving a query request, extracting target container cargo information corresponding to the query request from the container cargo information database, and generating a query report based on the target container cargo information.

[0007] In the above embodiments, the logistics management system uses drones to collect yard images and laser ranging data, combined with RFID electronic tag reading, to achieve precise positioning and real-time monitoring of container location and status. The system analyzes container surface features using a deep learning model, automatically identifies damage, and updates the database with real-time container and cargo information, significantly improving the efficiency of port logistics management and solving the problems of monitoring blind spots and untimely information updates inherent in traditional manual inspections.

[0008] In conjunction with some embodiments of the first aspect, in some embodiments, the steps of performing image segmentation processing on the yard image data, identifying the surface features of each container, and comparing the surface features with a damage feature library to generate a damage detection result specifically include: inputting the yard image data into a deep learning model, extracting texture features, color features, and edge features of the container surface to obtain a container surface feature vector; obtaining feature templates corresponding to preset damage types from the damage feature library; the feature templates include dent features, crack features, corrosion features, and leakage features; calculating the similarity score between the container surface feature vector and each feature template; and determining the damage type corresponding to the feature template whose similarity score exceeds a preset similarity threshold as the damage detection result.

[0009] In the above embodiments, the logistics management system extracts features from the surface of the cargo box using a deep learning model, obtaining texture, color, and edge features, and compares them with a preset damage feature template. By setting a similarity threshold, the system accurately identifies damage types such as dents, cracks, rust, and leaks in the box, significantly improving the accuracy and efficiency of detection and avoiding the subjectivity and omissions of manual inspection.

[0010] In conjunction with some embodiments of the first aspect, in some embodiments, after calculating the similarity scores of the container surface feature vector and each feature template, the method further includes: performing image segmentation on the yard image data based on the container surface feature vector to determine suspected damaged areas; extracting historical image blocks corresponding to the suspected damaged areas from historical image data, and calculating the image difference between the suspected damaged areas and the historical image blocks; when the image difference is greater than a preset difference threshold, marking the suspected damaged areas as newly added damaged points; and performing an evaluation and classification on the newly added damaged points, including damage type and damage degree, to generate a damage detection report containing damage location, damage type, and damage degree.

[0011] In the above embodiment, the logistics management system can automatically identify newly added damage points and intelligently evaluate and classify the damage types and degrees through comparative analysis of historical image data. By generating a detection report containing damage location, type and degree, reliable data support is provided for subsequent container maintenance and responsibility determination, improving the standardization and scientificity of port container management.

[0012] In combination with some embodiments of the first aspect, in some embodiments, the receiving the query request, extracting the target container cargo information corresponding to the query request from the container cargo information database, and generating the query report based on the target container cargo information specifically includes: receiving a query request containing a query condition, retrieving the target container cargo information from the container cargo information database based on the query condition; the query condition includes at least one of container number information, time range, location area and damage state; structurally processing the target container cargo information according to a preset data template to generate a data report containing container cargo basic information, location trajectory information and damage record, and highlighting abnormal data in the data report; determining accessible report fields according to user permission configuration information, and filtering the data report based on the accessible report fields to generate a query report meeting different user needs.

[0013] In the above embodiment, the logistics management system can quickly generate a report containing container cargo basic information, location trajectory and damage record through flexible query condition setting and data structuring. At the same time, the report field is filtered according to the user permission to ensure the security of data access and meet the individualized query needs of different users.

[0014] In combination with some embodiments of the first aspect, in some embodiments, after the step of associating the damage detection result with the real-time container cargo information and updating the information database representing the state of the container, the method further includes: after receiving a container cargo addition request or a container cargo displacement request, determining the request location of the requester container cargo based on the yard monitoring image; generating a patrol path for the unmanned aerial vehicle to arrive at the request location based on the three-dimensional digital model; and issuing the patrol path to the unmanned aerial vehicle to trigger the unmanned aerial vehicle to perform a new round of container cargo information collection task.

[0015] In the above embodiment, the logistics management system can respond to container cargo addition or displacement requests in a timely manner and automatically plan the unmanned aerial vehicle patrol path through real-time monitoring of yard changes. This intelligent patrol task scheduling mechanism ensures the timely updating of container cargo information and improves the efficiency of port operations.

[0016] In some embodiments of the first aspect, the generating, based on the three-dimensional digital model, an inspection path of the UAV to the arrival request location specifically comprises: obtaining real-time wind speed data and real-time wind direction data collected by a port weather station of the port where the container yard is located; calculating a wind resistance coefficient of each passage suitable for the UAV based on the stacking layout of the containers in the three-dimensional digital model; determining a wind-affected windward area and a wind-protected area formed by the container stacking according to the wind resistance coefficient, the real-time wind speed data and the real-time wind direction data; and planning and generating the inspection path according to the distribution characteristics of the windward area and the wind-protected area.

[0017] In the above embodiments, the logistics management system integrates the port weather station data and the yard layout information, and can calculate the passage wind resistance coefficient, identify the windward and wind-protected areas, and ensure the safety and reliability of the UAV inspection operation.

[0018] In some embodiments of the first aspect, the planning and generating the inspection path according to the distribution characteristics of the windward area and the wind-protected area specifically comprises: calculating a short-range path distance through the windward area and a detour path distance through the wind-protected area in a path sub-area containing the windward area and the wind-protected area; calculating a wind-protected path consumption corresponding to the detour path distance; calculating an additional energy consumption value of the UAV against the wind force when passing through the windward area, and determining a windward path consumption according to the additional energy consumption value and a path length consumption of the short-range path distance; determining a path with the minimum consumption value as an inspection sub-path of the path sub-area according to the wind-protected path consumption and the windward path consumption; generating the inspection path based on a plurality of inspection sub-paths, and generating a path execution instruction containing a flight height, a speed and an obstacle avoidance parameter.

[0019] In the above embodiments, the logistics management system comprehensively considers the influence of wind force, can intelligently calculate the energy consumption of different paths, and selects the optimal inspection path. By generating detailed path execution instructions, the UAV can safely and efficiently complete the inspection task, and the practicability of the system is improved.

[0020] In the second aspect, the embodiments of the present application provide a logistics management system, which comprises one or more processors and a memory; the memory is coupled to the one or more processors, and is used to store computer program codes, the computer program codes comprising computer instructions, and the one or more processors invoke the computer instructions to enable the logistics management system to perform the method described in the first aspect and any possible implementation manner of the first aspect.

[0021] In a third aspect, the embodiments of the present application provide a computer program product comprising instructions which, when executed on a logistics management system, cause the logistics management system to carry out the method according to the first aspect and any possible implementation of the first aspect.

[0022] In a fourth aspect, the embodiments of the present application provide a computer-readable storage medium comprising instructions which, when executed on a logistics management system, cause the logistics management system to carry out the method according to the first aspect and any possible implementation of the first aspect.

[0023] It can be understood that the logistics management system provided by the second aspect, the computer program product provided by the third aspect and the computer storage medium provided by the fourth aspect are all used to execute the method provided by the embodiments of the present application. Therefore, the beneficial effects that can be achieved thereby can refer to the beneficial effects in the corresponding method, which will not be described here again.

[0024] The one or more technical solutions provided in the embodiments of the present application have at least the following technical effects or advantages:

[0025] 1. Since the unmanned aerial vehicle is used to automatically collect the yard data and construct a three-dimensional digital model, combined with RFID electronic tag identification and deep learning damage detection technology, the container position and state are automatically collected and intelligently analyzed, effectively solving the problems of monitoring blind area of fixed camera, low efficiency and strong subjectivity of manual inspection in the prior art, and further realizing accurate positioning, real-time monitoring and automatic updating of dynamic information of the container cargo. Through the strong maneuverability of the unmanned aerial vehicle, the visual angle limitation of the traditional fixed monitoring equipment is broken through, and effective coverage of the dead angle of the yard is realized. At the same time, the application of RFID technology ensures the accuracy of the container number information identification, and the introduction of the deep learning model makes the damage detection more objective and reliable.

[0026] 2. Since the container cargo surface feature extraction technology based on deep learning is used, and a damage feature library containing multiple types of damage such as concave, crack, rust and leakage is established, the accurate identification of the damage type can be realized through the feature vector similarity calculation, effectively solving the problems of strong subjectivity, non-uniform standard and easy omission of manual detection in the prior art, and further realizing the standardization, automation and quantification of damage detection. The system extracts the texture, color and edge features of the container cargo surface to construct a comprehensive feature representation system. Through intelligent matching with the preset feature template, not only the detection accuracy is improved, but also the traceability and comparability of the detection result are realized.

[0027] 3、Due to the adoption of intelligent path planning technology based on three-dimensional digital model, combined with real-time container status monitoring and automatic task triggering mechanism, the optimal inspection path can be planned in time when the container status change is detected, effectively solving the problem of low efficiency caused by the response lag of inspection task and unreasonable path planning in the prior art, and further realizing intelligent scheduling and automatic execution of the inspection task. The system can discover the container addition or displacement in the first time through real-time monitoring of the yard change, and immediately start the inspection task. The path planning based on three-dimensional digital model fully considers the actual scene constraints, ensures that the unmanned aerial vehicle can safely and efficiently complete the inspection task, and greatly improves the intelligent level of port logistics management. BRIEF DESCRIPTION OF DRAWINGS

[0028] Figure 1 is a flowchart of a real-time query method of container dynamic information in the embodiment of the application;

[0029] Figure 2 is another flowchart of a real-time query method of container dynamic information in the embodiment of the application;

[0030] Figure 3 is a schematic diagram of an entity device structure of a logistics management system in the embodiment of the application. DETAILED DESCRIPTION

[0031] The terms used in the following embodiments of the present application are only for the purpose of describing specific embodiments, and are not intended to be limiting to the present application. As used in the specification of the present application, the singular expression "one", "a", "the", "said" and "this" are intended to include the plural expression, unless there is clear indication to the contrary in the context. It should also be understood that the term "and / or" used in the present application means any or all possible combinations of one or more listed items.

[0032] Hereinafter, the terms "first", "second" are only used for the purpose of description, and cannot be understood as implying or suggesting relative importance or implicitly indicating the number of indicated technical features. Therefore, the features defined with "first", "second" can explicitly or implicitly include one or more of the features, and in the description of the embodiments of the present application, the meaning of "multiple" is two or more, unless otherwise specified.

[0033] In order to facilitate understanding, the application scenario of the embodiments of the present application is introduced as follows.

[0034] In a certain large port, tens of thousands of containers need to be handled for inbound and outbound operations every day. Traditional container management mainly relies on manual inspection and fixed camera monitoring, but due to the large area of the yard and the complex stacking of containers, manual inspection is inefficient and prone to omissions. Especially in the detection of container damage state, manual detection is highly subjective and it is difficult to ensure accuracy and consistency. At the same time, due to environmental factors, fixed cameras have monitoring blind spots and cannot timely detect changes in container state. This leads to outdated container state information, affecting port operation efficiency and safety management.

[0035] In related technologies, fixed cameras and ground inspection robots can be used to realize automatic monitoring of container yards. Fixed cameras are installed at key locations in the yard for real-time monitoring, and ground inspection robots follow a preset route for regular inspection, with image recognition and RFID reading to collect container information. The following describes a scenario using the real-time query method of container dynamic information in related technologies.

[0036] A certain port has introduced a monitoring system based on fixed cameras and ground inspection robots. Fixed cameras are installed at key locations in the yard to monitor container status in real time through image recognition technology. Ground inspection robots follow a preset route for inspection, collecting container images and RFID information. However, due to the fixed angle of the cameras, it is difficult to cover the dead corners of the yard; ground robots are limited by ground access and cannot obtain state information from high containers. At the same time, in adverse weather conditions, the reliability of the equipment decreases, affecting the monitoring effect. This solution, although partially automated, still has monitoring blind spots and efficiency bottlenecks.

[0037] The real-time query method of container dynamic information in the embodiments of the present application uses multi-sensor data fusion and deep learning technology to achieve comprehensive monitoring and intelligent analysis of container status, not only solving the blind spot problem of traditional fixed monitoring equipment, but also improving inspection efficiency through adaptive path planning, and achieving objective and accurate evaluation of damage detection. The following describes a scenario using the real-time query method of container dynamic information in the present application.

[0038] After adopting this solution, the port deployed a UAV system equipped with high-definition cameras, laser radars, and RFID reading devices. The UAV can flexibly maneuver and collect container information from multiple angles, effectively solving the problem of monitoring blind spots. Through a deep learning model for damage detection, objective and accurate state evaluation is achieved. The system can also automatically plan the optimal inspection path based on wind conditions to ensure safe and efficient task execution. When new or displaced containers are found, the system can trigger targeted inspection tasks in a timely manner to update container information in real time. This solution significantly improves the automation and intelligence level of port management.

[0039] It can be seen that the real-time query method of container dynamic information in the embodiments of the present application can realize automatic container information collection, effectively solve the problems of monitoring blind area, strong subjectivity of detection and untimely information updating in the traditional scheme, and further realize intelligent upgrading of port logistics management.

[0040] For ease of understanding, the method provided by the present embodiment is described in the following flow in combination with the above scenario. Please refer to Figure 1 , a flowchart of the real-time query method of container dynamic information in the embodiments of the present application.

[0041] S101, receiving the yard image data and laser ranging data collected by the unmanned aerial vehicle, and constructing a three-dimensional digital model of the container yard to determine the spatial coordinate information of each container.

[0042] The yard image data represents the image information obtained by the high-definition camera carried by the unmanned aerial vehicle when taking pictures of the container yard, including visible light images and infrared images; the laser ranging data refers to the point cloud data obtained by scanning the laser radar carried by the unmanned aerial vehicle, which is used to represent the three-dimensional spatial information of the objects in the yard; the three-dimensional digital model represents a virtual yard environment constructed by computer graphics technology, which contains the geometric shape, position and spatial relationship of the containers; the spatial coordinate information is used to represent the specific position of each container in the yard, including longitude, latitude and height values.

[0043] When the logistics management system receives the yard inspection task instruction, it will trigger the execution of this step. Specifically, the logistics management system first controls the unmanned aerial vehicle to cruise the yard according to the preset route, and at the same time starts the camera and laser radar device for data collection. The obtained image data is preprocessed to generate an orthographic image, and the laser ranging data is formed into a three-dimensional point cloud model through point cloud registration. Then, the logistics management system fuses the two kinds of data, constructs an accurate three-dimensional digital model of the yard through feature matching and three-dimensional reconstruction algorithm. Finally, through spatial analysis algorithm, the position coordinates of each container are calculated, and the coordinate index is established.

[0044] In some embodiments, the three-dimensional modeling of the yard and the positioning of the containers can be realized in various ways: alternatively, the logistics management system can use the SFM algorithm based on multi-view geometry, realize three-dimensional reconstruction through image feature point matching and bundle adjustment, and finally determine the position of the container by combining the RANSAC algorithm for plane segmentation and boundary extraction; alternatively, the logistics management system can directly use the laser point cloud data, perform plane clustering through region growing algorithm, identify the container contour by combining height jump detection, and finally determine the spatial coordinates through the minimum circumscribed rectangle algorithm. It can be understood that other three-dimensional reconstruction and target positioning methods can also be used to realize this function, which is not limited here.

[0045] In practical applications, there may be problems of poor image quality or missing point cloud data due to weather conditions. In this regard, the logistics management system adopts a multi-source data complementary solution: when the visible light image quality is poor, the infrared image is preferentially used for processing; when there is local point cloud data missing, the point cloud data of the adjacent area is interpolated to supplement, and the image information is combined for verification and correction. At the same time, the system also establishes a data quality evaluation mechanism, when the collected data quality is lower than the preset threshold, the collection parameters are automatically adjusted or the supplementary collection path is planned.

[0046] S102, according to the yard image data, identifying the box number information of each container, and associating the box number information with the spatial coordinate information, generating the box cargo position data.

[0047] Among them, the box number information represents the unique identification code of the container, including the owner code, the equipment category identifier, the registration number and the check digit; the yard image data refers to the preprocessed high-resolution orthophoto, which is used to represent the visual information of the container surface; the spatial coordinate information represents the position of the container in the three-dimensional space, including the X, Y and Z axis coordinate values; the box cargo position data is used to represent the correspondence between the box number and the spatial position, containing the box number, the coordinate, the yard area number and other information.

[0048] After completing the three-dimensional modeling of the yard, the logistics management system immediately starts the box number recognition process. Specifically, the logistics management system first performs image enhancement and distortion correction on the yard image to improve the image clarity. Then, the target detection algorithm is used to locate the box number area in the image, and the optical character recognition (OCR) technology is used to extract the box number character information. For the recognized box number, the system performs ISO 6346 standard format verification to ensure the validity of the box number. Then, the logistics management system establishes a mapping relationship between the verified box number and the previously obtained spatial coordinate information, forming a structured box cargo position data record.

[0049] In some embodiments, box number recognition and position association can be achieved in various ways: optionally, the logistics management system can use a deep learning method, locate the box number area through the YOLOv5 target detection model, use the improved ResNet-OCR network to recognize the box number characters, and finally establish the correspondence between the box number and the coordinate through spatial projection transformation; optionally, the logistics management system can use a traditional image processing method, locate the box number area through edge detection and rectangle extraction, recognize the box number through template matching and character segmentation, and then associate the position information through spatial geometric transformation. It can be understood that other image recognition and data association methods can also be used to achieve this function, which is not limited here.

[0050] In practical applications, the box number may be blocked or contaminated, causing difficulties in identification. To this end, the logistics management system adopts a multi-angle verification solution: by adjusting the position of the unmanned aerial vehicle to obtain box number images from different angles, the identification results from multiple angles are integrated to improve accuracy. At the same time, the system establishes a box number identification confidence evaluation mechanism. When the confidence of the identification result is lower than the threshold, it is automatically marked as a to-be-verified state, and the manual review process is triggered. For the box number that cannot be identified at all, the system records its spatial position and on-site image to facilitate subsequent manual processing.

[0051] S103, the RFID reading device of the unmanned aerial vehicle collects the electronic tag information of each container, matches and verifies the electronic tag information with the box cargo position data, and generates real-time box cargo information.

[0052] Among them, the RFID reading device represents the radio frequency identification reader carried by the unmanned aerial vehicle, which is used to collect container electronic tag information; the electronic tag information refers to the digital information stored in the container RFID tag, including box number, box type, weight and other attribute data; the box cargo position data represents the mapping relationship between the box number and the spatial coordinates generated previously; the real-time box cargo information is used to represent the complete container state data verified, including tag information, position information and timestamp.

[0053] After completing the box number identification and position association, the logistics management system starts the RFID information collection process. Specifically, the logistics management system first controls the unmanned aerial vehicle to approach the target container at the best reading distance, activates the RFID reading device to send an inquiry signal. After receiving the response signal of the container electronic tag, the system parses and obtains the box cargo information stored in the tag. Subsequently, the logistics management system cross- verifies the parsed electronic tag information with the generated box cargo position data to ensure consistency of the box number. After verification, the system integrates the tag information and position data to generate real-time box cargo information records with timestamps.

[0054] In some embodiments, RFID information collection and data matching can be achieved in various ways: optionally, the logistics management system can use adaptive power control technology to dynamically adjust the RFID reading power according to the distance between the unmanned aerial vehicle and the container, ensure data accuracy through multiple readings and signal strength analysis, and then use hash index to quickly match position data; optionally, the logistics management system can use multi-antenna array technology to read multiple electronic tag information at the same time, determine the signal source combined with the spatial positioning algorithm, and ensure information consistency through database transaction processing. It can be understood that other RFID reading and data verification methods can also be used to achieve this function, which is not limited here.

[0055] In practical applications, problems such as electronic tag signal interference or data inconsistency may be encountered. To this end, the logistics management system adopts multiple safeguard mechanisms: first, a conflict avoidance algorithm is set to avoid signal interference caused by multiple tags responding at the same time; second, a comparison mechanism between tag data and visual recognition results is established, and when inconsistencies are found, the system will record the exception and trigger the re-acquisition process; finally, for cases where continuous reading is not possible or there are obvious data conflicts, the system will automatically generate an exception report while keeping the original data unchanged, waiting for manual confirmation before updating.

[0056] S104, image segmentation processing is performed on the yard image data, the surface features of each container are identified, and the surface features are compared with the damage feature library to generate a damage detection result.

[0057] The image segmentation processing represents a process of dividing the yard image into regions, which is used to separate the image area of a single container; the surface features of the container refer to the visual feature information of the container surface, including texture, color, edge and other feature parameters; the damage feature library represents a pre-established database of container damage feature templates of various types, containing damage samples of different types and degrees; the damage detection result is used to represent the damage condition evaluation information of each container, including damage type, location and degree.

[0058] After the logistics management system obtains real-time container information, it begins to execute the damage detection process. Specifically, the logistics management system first applies a semantic segmentation algorithm to the yard image to accurately extract the surface area of each container. Then, the system extracts features from the single container image obtained by segmentation to construct a feature vector containing texture, color and edge information. Next, the logistics management system calculates the similarity between the extracted feature vector and the standard template in the damage feature library to determine whether there is damage by setting a threshold. For the detected damage area, the system further analyzes the damage type and degree to generate a detailed damage detection report.

[0059] In some embodiments, damage detection and evaluation can be achieved in various ways: alternatively, the logistics management system can use a deep learning segmentation network to extract the container area, use a multi-scale feature pyramid to extract local and global features, highlight key areas through an attention mechanism, and finally use a multi-classifier to determine the damage type and degree; alternatively, the logistics management system can use traditional image processing methods, use edge detection and region growing algorithms to segment the target area, extract texture features using a gray level co-occurrence matrix, and implement damage classification through a support vector machine. It can be understood that other image analysis and pattern recognition methods can also be used to achieve this function, which is not limited here.

[0060] In practical applications, the problem of decreased accuracy of damage detection caused by changes in lighting conditions may be encountered. To this end, the logistics management system adopts an adaptive detection strategy: first, image preprocessing techniques are used to standardize images under different lighting conditions, reducing the influence of environmental factors; second, a time series-based damage change tracking mechanism is established, which compares the detection results of the same container at different time points to improve the reliability of the judgment; finally, the system also integrates a fusion analysis function based on multispectral data, which can assist in judging damage when the quality of visible light images is poor.

[0061] S105, associate the damage detection result with the real-time container information, and update the container information database representing the state of the container.

[0062] Among them, the damage detection result represents the evaluation data of the surface damage of the container, including damage type, location coordinates and severity; the real-time container information refers to the comprehensive information containing the container number, location and label data; the container information database represents a structured database storing container state data, used to record the state changes of the container throughout its life cycle; the association update refers to the process of matching and updating new damage information with existing container records.

[0063] After completing the damage detection, the logistics management system immediately starts the database update process. Specifically, the logistics management system first associates the damage detection result with the real-time container information according to the container number, forming a complete state update data package. Then, the system checks the historical damage records of the container and compares and analyzes the changes of the new damage points and existing damage. Next, the logistics management system formats the update data package according to the predefined data model to generate standardized database update instructions. Finally, the system ensures the atomicity and consistency of data updates through transaction processing mechanisms to complete the real-time update of container state information.

[0064] In some embodiments, data association and update can be achieved in various ways: optionally, the logistics management system can use an incremental update strategy to manage data updates through timestamp and version control mechanisms, establish damage history change trajectories, use difference comparison algorithms to identify state changes, and ensure the safety of concurrent updates through transaction locks; optionally, the logistics management system can use distributed database technology to shard container state information, ensure data consistency through asynchronous replication, and improve query efficiency using caching mechanisms. It can be understood that other data management and update methods can also be used to achieve this function, which is not limited here.

[0065] In practical applications, high concurrency data updates may cause system response delays. To address this issue, the logistics management system adopts a multi-level cache strategy: first, hot data is stored in memory cache to reduce database access pressure; second, an update queue based on priority is implemented to ensure that important updates are processed first; finally, the system also establishes an automatic retry mechanism for data update failures and ensures data update traceability through log recording to quickly locate and recover when an exception occurs.

[0066] S106, receiving a query request, extracting target container cargo information corresponding to the query request from the container cargo information database, and generating a query report based on the target container cargo information.

[0067] Among them, the query request represents a container cargo information retrieval instruction initiated by the user, including query conditions and output requirements; the target container cargo information refers to the container state record that meets the query conditions; the container cargo information database represents a central database that stores complete container cargo state data; the query report is used to represent a container cargo information display document organized in a specific format, including basic information, location trajectory, and damage record.

[0068] After receiving the query request of the user, the logistics management system starts the report generation process. Specifically, the logistics management system first parses the query request to extract the query conditions and output parameters. Then, the system performs access control verification according to the user's permission level to determine the queryable data range. Next, the logistics management system constructs an optimized SQL query statement to retrieve records that meet the conditions from the container cargo information database. After obtaining the data, the system formats the data according to the preset report template to generate a structured report document containing statistical analysis results.

[0069] In some embodiments, data query and report generation can be implemented in various ways: optionally, the logistics management system can use multi-dimensional data analysis technology to pre-calculate common statistical indicators by establishing data cubes, use a dynamic SQL generator to construct query statements, combine data visualization components to generate charts, and finally use a template engine to render and output reports; optionally, the logistics management system can use a real-time computing framework to accelerate query response through in-memory data grid technology, use a rule engine for data filtering and conversion, and use a report engine to support multiple formats for output. It can be understood that other data query and report generation methods can also be used to implement this function, which is not limited here.

[0070] In practical applications, large data queries may cause report generation delays. To address this issue, the logistics management system employs query optimization strategies: first, it improves query efficiency by establishing multi-level indexes and optimizing indexes for frequently used query conditions; second, it implements a paging query mechanism to process large amounts of data in batches, ensuring timely system responses; finally, the system provides asynchronous report generation functionality for complex statistical analysis tasks, allowing report generation to be placed in a background queue for processing, and notifying users of the availability of reports for download through message pushing. Additionally, to handle sudden high-concurrency query requests, the system establishes a query result caching mechanism that returns cached data for queries with the same conditions, significantly improving query performance.

[0071] In the above embodiment, the method of box cargo information collection and damage detection based on unmanned aerial vehicles is mainly described. In practical applications, system functions can be expanded according to specific scene requirements, such as adding multi-vehicle coordination scheduling, predictive maintenance, intelligent early warning, and other functional modules. The scene of this embodiment is supplemented as follows.

[0072] After the scheme is applied in depth, the system further integrates port weather, operation scheduling, and other multi-source data to achieve more intelligent task management. For example, the system can adjust the inspection plan in advance according to weather forecasts to avoid adverse weather; based on historical data analysis of box cargo movement patterns, it can predict hotspots and optimize inspection frequency; multiple unmanned aerial vehicles can work together, automatically assigning tasks based on task priority and energy status. At the same time, the system continuously learns and optimizes the damage detection model, constantly improving recognition accuracy. These optimization measures further improve the system's intelligence level and operational efficiency.

[0073] After combining the above scenarios, the method provided in this embodiment is further described in more detail. Please refer to Figure 2 , another flowchart of the real-time query method of box cargo dynamic information in the embodiment of the present application.

[0074] S201, receive the image data and laser ranging data collected by the unmanned aerial vehicle, and construct a three-dimensional digital model of the container yard to determine the spatial coordinate information of each container.

[0075] Referring to step S101, the logistics management system collects data through the unmanned aerial vehicle and constructs a three-dimensional model of the yard.

[0076] S202, according to the yard image data, identify the box number information of each container, and associate the box number information with the spatial coordinate information to generate box cargo position data.

[0077] Referring to step S102, the logistics management system identifies the container box number and associates it with the position information.

[0078] S203, the electronic tag information of each container is collected by the UAV-based RFID reading device, and the electronic tag information is matched and verified with the container cargo position data to generate real-time container cargo information.

[0079] Referring to step S103, the logistics management system collects the electronic tag information and verifies and matches the position data.

[0080] S204, input the yard image data into the deep learning model, extract the texture features, color features and edge features of the container cargo surface, and obtain the container cargo surface feature vector.

[0081] Wherein, the deep learning model represents a pre-trained neural network model for automatically extracting image features; the texture feature refers to the texture structure information of the container surface, including roughness, regularity and other features; the color feature represents the color distribution information of the container surface; the edge feature is used to represent the contour and boundary information of the container surface; the feature vector is a numerical representation of the multi-dimensional feature information after encoding.

[0082] After the logistics management system obtains the yard image, it starts the feature extraction process. Specifically, the logistics management system first normalizes the image and enhances the data to ensure the quality of the input data. Then, the system inputs the preprocessed image into the deep learning model and extracts features layer by layer through a multi-layer convolutional network. Different layers of the model are responsible for extracting local texture features, regional color features and global edge features. Finally, the system fuses and reduces the dimensions of the features to generate a fixed-dimensional feature vector.

[0083] In some embodiments, feature extraction can be achieved in various ways: optionally, the logistics management system can use an improved VGG network structure to extract multi-level features through a multi-scale feature pyramid, combine attention mechanisms to highlight key regional features, and finally generate a compact feature representation through a feature aggregation module; optionally, the logistics management system can use a lightweight MobileNet architecture, reduce the amount of calculation through depthwise separable convolution, while using residual connections to maintain feature integrity, and obtain a fixed-dimensional feature vector through a global pooling layer. It can be understood that other deep learning architectures can also be used to implement feature extraction, which is not limited here.

[0084] In practical applications, there may be problems of unstable feature extraction. To this end, the logistics management system adopts a feature enhancement strategy: through multi-view feature fusion technology, the image features of different angles are integrated; a feature quality evaluation mechanism is established, and when the feature quality does not meet the requirements, automatic reacquisition is triggered; at the same time, a contrast learning method is introduced to improve the discriminability and robustness of the features.

[0085] S205, obtain the feature template corresponding to the preset damage type from the damage feature library.

[0086] wherein the residual damage feature library represents a database storing standard residual damage sample features, including various residual damage types such as dents, cracks, rust, etc.; the preset residual damage type refers to a container common damage category defined by the system in advance; the feature template is used to represent the standard feature representation of each residual damage type, including a feature vector and description information.

[0087] After the surface feature extraction is completed, the logistics management system starts the template matching process. Specifically, the logistics management system first retrieves the corresponding feature template set from the residual damage feature library according to the preset residual damage type list. Then, the system pre-processes the feature template, including feature standardization and noise reduction processing. Next, the logistics management system loads the processed feature template into the memory to build an efficient feature index structure, preparing for subsequent similarity calculation.

[0088] In some embodiments, feature template management can be implemented in various ways: optionally, the logistics management system can use a hierarchical index structure to organize different types of residual damage feature templates into a tree structure, optimize retrieval efficiency through feature clustering, maintain template usage frequency statistics, and implement intelligent cache management; optionally, the logistics management system can use a distributed storage architecture to store feature templates in multiple nodes, improve access speed through parallel retrieval, and establish a template version control mechanism to ensure data consistency. It can be understood that other data organization and management methods can also be used to implement this function, which is not limited here.

[0089] In actual application, there may be a problem of low retrieval efficiency due to a large number of feature templates. To this end, the logistics management system adopts a template optimization strategy: first, reduce the template storage space through feature dimension reduction technology; second, implement a scenario-based template preloading mechanism to preload highly relevant templates according to the characteristics of the current detection task; finally, the system also establishes a template update mechanism to regularly clean up low-usage templates and continuously optimize the template library through online learning.

[0090] S206, calculate the similarity score of the box cargo surface feature vector and each feature template.

[0091] wherein the similarity score represents the matching degree of the feature to be detected and the template feature, with a value range of 0-1; the feature vector refers to the multi-dimensional feature information extracted from the surface of the box cargo; the feature template represents the feature representation of the standard residual damage sample; the calculation process represents the distance or similarity measurement method between feature vectors.

[0092] After obtaining the feature template, the logistics management system starts the similarity calculation process. Specifically, the logistics management system first aligns the dimensions and normalizes the surface feature vector of the container cargo and the template feature. Then, the system uses cosine similarity and other measurement methods to calculate the similarity between the detected feature and each template feature. Next, the logistics management system sorts the calculation results to obtain a matching score list for each type of damage template. Finally, the system compares the score list with the preset confidence threshold to filter out high-confidence matching results.

[0093] In some embodiments, similarity calculation can be achieved in various ways: optionally, the logistics management system can use a multi-feature fusion strategy to calculate the similarity of texture, color, and edge features respectively, and obtain a comprehensive score through weighted fusion, with the weight coefficient being adjusted adaptively through machine learning methods; optionally, the logistics management system can use deep metric learning methods to learn distance metrics in the feature space through twin networks to achieve more accurate similarity calculation. It can be understood that other similarity calculation methods can also be used to achieve this function, which is not limited here.

[0094] In actual application, there may be problems of unstable similarity calculation results. For this, the logistics management system uses a multiple verification strategy: first, multiple similarity measurement methods are integrated to reduce the deviation of a single algorithm; second, a local feature matching mechanism is introduced to consider the local similarity of the damage area; finally, the system also establishes a similarity threshold adaptive mechanism to dynamically adjust the judgment standard based on historical data.

[0095] In some embodiments, the logistics management system will perform intelligent damage detection analysis and evaluation on the surface of the container, that is, the logistics management system will perform image segmentation on the yard image data based on the surface feature vector of the container cargo to determine the suspected damage area; extract the historical image block corresponding to the suspected damage area from the historical image data, and calculate the image difference degree of the suspected damage area and the historical image block; when the image difference degree is greater than a preset difference threshold, mark the suspected damage area as a new damage point; evaluate and classify the new damage point including the damage type and the damage degree to generate a damage detection report containing the damage location, the damage type, and the damage degree.

[0096] Among them, the surface feature vector of the container cargo represents the multi-dimensional feature data of the container surface, including texture, color, and edge information; the suspected damage area refers to the area that is preliminarily determined to have damage through image analysis; the historical image block represents the reference image collected at the same position before; the image difference degree is used to represent the degree of change between two image areas; the new damage point refers to the newly detected damage location; the damage detection report is used to represent the complete container damage evaluation result.

[0097] After completing feature extraction, the logistics management system initiates the damage analysis process. Specifically, the system first uses the surface feature vectors of the cargo box for image segmentation and then uses anomaly detection algorithms to mark suspected damaged areas that differ significantly from normal areas. Next, the system retrieves historical images corresponding to these areas from the historical database, aligns them using image registration technology, and calculates a structural similarity index. Then, the system compares the calculated difference with a preset threshold to identify newly added damage points. Finally, the system performs in-depth analysis on each newly added damage point, using a multi-level classifier to evaluate the damage type and severity, and generates a standardized inspection report.

[0098] In some embodiments, damage analysis and assessment can be implemented in multiple ways: Optionally, the logistics management system can use an attention-based segmentation network to locate suspected areas, calculate image changes through structural similarity indices and texture feature differences, classify damage features using hierarchical clustering methods, and finally assess the degree of damage using an expert rule system; Optionally, the logistics management system can use a region growing algorithm to segment abnormal regions, analyze image spectral differences through Fourier transform, implement damage classification using a support vector machine, and assess the degree of damage using a fuzzy inference system. It is understood that other image analysis and assessment methods can also be used to achieve this function, and no limitation is made here.

[0099] In practical applications, changes in ambient lighting may lead to inaccurate image difference calculations. To address this, the logistics management system employs a robustness enhancement strategy: First, image enhancement preprocessing eliminates the influence of lighting; second, a multi-scale analysis mechanism is introduced to verify damage features at different resolutions; finally, the system also establishes a time-series-based verification mechanism to improve the reliability of judgments through consistency analysis of multiple consecutive detection results.

[0100] S207. Determine the damage type corresponding to the feature template whose similarity score exceeds the preset similarity threshold, and use it as the damage detection result.

[0101] Among them, the preset similarity threshold represents the system's predefined similarity judgment standard, which is used to distinguish whether there is damage; the similarity score refers to the calculation result of feature matching; the damage type represents the preset category of box damage, such as dents, cracks, etc.; the damage detection result is used to represent the final identified box damage situation.

[0102] After obtaining the similarity scores, the logistics management system initiates the damage determination process. Specifically, the logistics management system first sorts the similarity scores of each feature template and selects a number of candidate results with the highest scores. Then, the system compares the scores of the candidate results with a pre-set similarity threshold to filter out matching items that exceed the threshold. Next, the logistics management system performs conflict detection on the filtered results to handle possible multiple type overlaps. Finally, the system generates a detection report containing the damage type, location, and confidence level.

[0103] In some embodiments, damage determination can be achieved in various ways. Optionally, the logistics management system can use a multi-threshold grading strategy, set differentiated determination criteria for different damage types, and use a decision tree model to integrate the determination results of multiple features to ultimately determine the damage type. Optionally, the logistics management system can use a probabilistic graph model, consider the correlation between damage types, and use a conditional random field algorithm to optimize the overall determination result. It can be understood that other determination methods can also be used to achieve this function, which is not limited here.

[0104] In actual applications, there may be misjudgment and missed judgment problems. To this end, the logistics management system uses a result optimization strategy: first, a damage degree evaluation model is established to quantify the severity of the detection results; second, time sequence information analysis is introduced to compare the detection results of the same container at different time points to improve the reliability of the determination; finally, the system also implements a manual review interface for the detection results, which can trigger a manual review process for results with low confidence.

[0105] S208, associate the damage detection result with the real-time container cargo information, and update the container cargo information database representing the state of the container.

[0106] Referring to step S105, the logistics management system updates the damage detection result to the container cargo information database.

[0107] S209, after receiving a container cargo addition request or a container cargo relocation request, determine the request position of the request container cargo based on the yard monitoring image.

[0108] Among them, the container cargo addition request represents the information collection demand when the new container enters the yard; the container cargo relocation request is the update demand after the position of the container is changed; the request position represents the target position coordinates that need to be collected information; the yard monitoring image is used to represent the real-time acquired site state image.

[0109] After receiving the operation request, the logistics management system initiates the position confirmation process. Specifically, the logistics management system first parses the request information, identifies the request type and target box cargo information. Then, the system performs scene analysis through real-time image monitoring, and locates the specific position of the target box cargo in combination with image recognition technology. Next, the logistics management system matches the identified position information with the yard map to generate standardized spatial coordinates. Finally, the system verifies the accessibility of the position to ensure that the UAV can safely reach the collection position.

[0110] In some embodiments, position confirmation can be achieved in various ways: optionally, the logistics management system can use multi-source data fusion technology to comprehensively utilize visible light images, infrared images, and laser ranging data, accurately locate the position of the box cargo through target detection algorithms, and verify the rationality of the position in combination with historical trajectory data; optionally, the logistics management system can use real-time three-dimensional reconstruction technology to quickly construct a local scene model based on multi-view images and determine the optimal collection position through spatial analysis. It can be understood that other position confirmation methods can also be used to achieve this function, which is not limited here.

[0111] In actual application, the problem of position confirmation difficulty caused by scene occlusion may be encountered. For this purpose, the logistics management system adopts an intelligent positioning strategy: first, a complete scene visibility map is constructed through multi-angle view synthesis technology; second, position estimation based on prediction is realized, and the possible target position is predicted using the movement law of the box cargo; finally, the system also establishes alternative solutions for position confirmation, which can be switched to other positioning methods when the main method fails.

[0112] S210, based on the three-dimensional digital model, generating a patrol path of the UAV reaching the request position.

[0113] Among them, the three-dimensional digital model represents the virtual space representation of the yard, containing terrain, obstacle and channel information; the patrol path refers to the flight trajectory of the UAV from the current position to the target position; the path generation process is used to represent the flight planning method considering various constraint conditions.

[0114] After determining the target position, the logistics management system initiates the path planning process. Specifically, the logistics management system first loads the three-dimensional digital model to construct a constraint map containing static obstacles and dynamic work areas. Then, the system calculates the spatial relationship between the current position and the target position of the UAV, and sets the flight control points. Next, the logistics management system considers safety distance, energy efficiency and work requirements, and generates the optimal flight trajectory through the path planning algorithm. Finally, the system verifies the feasibility of the planned path to ensure that it meets various flight constraints.

[0115] In some embodiments, path planning can be achieved in various ways: optionally, the logistics management system can use an improved A* algorithm to generate a smooth flight path by introducing dynamic weights and heuristic functions, taking into account yard channel width, turning radius and height restrictions, while optimizing energy consumption; optionally, the logistics management system can use a rapidly expanding random tree algorithm to quickly explore feasible paths in three-dimensional space, and smooth and simplify the initial path through the path optimization module. It can be understood that other path planning methods can also be used to achieve this function, which is not limited here.

[0116] In practical applications, dynamic obstacles may interfere with path planning. For this purpose, the logistics management system adopts an adaptive planning strategy: first, a dynamic obstacle avoidance mechanism is established through real-time scene monitoring; second, priority-based dynamic path adjustment is implemented, which can update the flight trajectory in real time according to the on-site situation; finally, the system also establishes an emergency avoidance plan, which can quickly generate a backup path when potential collision risks are detected.

[0117] In some embodiments, the logistics management system optimizes the UAV inspection path planning based on weather data, that is, the logistics management system obtains real-time wind speed data and real-time wind direction data collected by the port weather station at the container yard; based on the stacking layout of containers in the three-dimensional digital model, the wind resistance coefficient of each passageway suitable for UAVs is calculated; according to the wind resistance coefficient, real-time wind speed data and real-time wind direction data, the wind-affected area and the wind-sheltered area formed by the container stacking are determined; according to the distribution characteristics of the wind-affected area and the wind-sheltered area, the inspection path is planned and generated.

[0118] Among them, the port weather station represents a professional weather monitoring device within the port; real-time wind speed data and wind direction data refer to wind power condition parameters of the current environment; the wind resistance coefficient is used to represent the blocking effect of the passageway on the wind; the wind-affected area represents the spatial area directly affected by the wind; the wind-sheltered area refers to the area with less wind power formed by the container shelter; the inspection path is used to represent the flight trajectory of the UAV performing the task.

[0119] Before planning the inspection path, the logistics management system starts the environmental analysis process. Specifically, the logistics management system first obtains real-time wind data from the port weather station, including wind speed, wind direction and other meteorological parameters. Then, the system analyzes the spatial layout of the container stacking based on the three-dimensional digital model, and calculates the wind influence factors of each passageway. Next, the logistics management system combines wind data and wind resistance coefficients to determine the distribution characteristics of the wind-affected area through flow field analysis. Finally, the system considers safety and efficiency comprehensively, and plans the optimal inspection path based on the regional characteristics.

[0120] In some embodiments, the environmental analysis and path planning can be implemented in various ways: optionally, the logistics management system can use a computational fluid dynamics model to simulate the wind field distribution, calculate the local wind blocking effect through grid division, divide the area level combined with the risk assessment model, and finally generate a safe path based on the dynamic programming algorithm; optionally, the logistics management system can use a simplified wind field model to estimate the influence of wind, identify wind-sheltered areas through spatial clustering methods, optimize the path scheme using heuristic algorithms, and ensure path feasibility through safety checks. It can be understood that other environmental analysis and path planning methods can also be used to achieve this function, which is not limited here.

[0121] In practical applications, the problem of planned path failure caused by sudden changes in wind field may be encountered. For this purpose, the logistics management system adopts a dynamic adjustment strategy: first, a wind field early warning mechanism is established to monitor the trend of wind changes in real time; second, the path dynamic optimization function is realized, which can adjust the flight path according to real-time wind field data; finally, the system also designs an emergency wind-avoiding plan, which can quickly plan a safe path to the nearest wind-sheltered area when detecting dangerous wind conditions.

[0122] In some embodiments, the logistics management system selects the optimal inspection path by considering the influence of wind, i.e., the logistics management system calculates the short-range path distance through the wind area and the detour path distance through the wind-sheltered area in the path sub-area containing the wind area and the wind-sheltered area; calculates the wind-sheltered path consumption corresponding to the detour path distance; calculates the additional energy consumption value of the UAV resisting wind when passing through the wind area, and determines the wind path consumption according to the additional energy consumption value and the path length consumption of the short-range path distance; determines the path with the minimum consumption value as the inspection sub-path of the path sub-area according to the wind-sheltered path consumption and the wind path consumption; generates an inspection path based on multiple inspection sub-paths, and generates a path execution instruction containing flight height, speed and obstacle avoidance parameters.

[0123] Wherein, the path sub-area represents a local spatial area that needs to be selected for path selection; the short-range path distance refers to the shortest path length through the wind area; the detour path distance is used to represent the longer path length through the wind-sheltered area; the additional energy consumption value represents the additional energy consumption required to resist wind; the path length consumption refers to the energy loss of the basic flight distance; the inspection sub-path is used to represent the optimal flight path of the local area; the path execution instruction represents the control command containing specific flight parameters.

[0124] After completing the environmental analysis, the logistics management system initiates the route optimization process. Specifically, the system first divides the planned area into multiple sub-areas and calculates the path distances for both direct flight and detour routes. Then, based on a wind model, the system estimates the additional energy consumption for windward flight and calculates the total energy consumption for the detour route. Next, the system selects the optimal sub-path by comparing energy consumption and connects multiple sub-paths to form a complete inspection path. Finally, the system generates detailed flight control parameters based on the path characteristics to ensure the drone can perform its mission safely and efficiently.

[0125] In some embodiments, path optimization and command generation can be implemented in multiple ways: Optionally, the logistics management system can employ a multi-objective optimization algorithm to comprehensively consider energy consumption, time, and safety, solve for the local optimal path using dynamic programming, ensure path continuity through smoothing, and finally generate precise control commands through trajectory planning; alternatively, the logistics management system can use an improved ant colony algorithm to search for the energy-optimal path, optimize the path shape using Bézier curves, generate adaptive flight parameters using model predictive control, and ensure command executability through simulation verification. It is understood that other path optimization and command generation methods can also be used to achieve this function, and no limitation is made here.

[0126] In practical applications, inaccurate energy consumption estimates may lead to incorrect route selection. To address this, the logistics management system employs an energy consumption optimization strategy: First, it improves estimation accuracy by establishing a precise energy consumption model that considers the impact of wind. Second, it implements an energy consumption correction mechanism based on historical data to continuously optimize the energy consumption prediction model. Finally, the system also establishes an energy early warning mechanism, dynamically adjusting flight parameters during execution by monitoring remaining battery power in real time to ensure safe mission completion. Simultaneously, the system also implements multi-drone collaborative scheduling, allowing for the rational allocation of multiple drones to perform inspection tasks based on mission requirements and energy availability.

[0127] S211. The inspection route is sent to the drone to trigger the drone to perform a new round of cargo information collection tasks.

[0128] Among them, the inspection path represents the planned flight trajectory data of the UAV, including waypoint coordinates and control parameters; the issuance process refers to the communication mechanism that transmits path instructions to the UAV; and the new round of data collection task represents the information acquisition process for the target cargo.

[0129] After completing the path planning, the logistics management system starts the task triggering process. Specifically, the logistics management system first converts the inspection path into a standardized flight instruction format, including parameters such as waypoint sequence, flight altitude, and speed. Then, the system sends the instructions to the UAV control system through a secure communication channel. Next, the logistics management system monitors the task execution status of the UAV and receives real-time feedback. Finally, the system configures the corresponding data collection parameters according to the task type to ensure that the next round of collection tasks can obtain the required box cargo information.

[0130] In some embodiments, task triggering can be achieved in various ways: optionally, the logistics management system can use a distributed task scheduling mechanism to manage the task allocation of multiple UAVs through a message queue, implement task priority sorting and load balancing, and establish task execution monitoring and exception handling mechanisms; optionally, the logistics management system can use a real-time control protocol to implement instruction issuance and state feedback through a bidirectional data channel, support real-time adjustment of task parameters and interruption recovery. It can be understood that other task management methods can also be used to achieve this function, which is not limited here.

[0131] In actual application, the problem of task interruption may be encountered. For this, the logistics management system adopts a task guarantee strategy: first, a task checkpoint mechanism is established to save the task execution status regularly; second, the breakpoint resume function is implemented to support the continuation of the task from the interruption point; finally, the system also establishes a task retry mechanism that can automatically trigger the retry process when an execution exception is detected. At the same time, in order to improve the efficiency of task execution, the system implements the task parallel processing function, which can simultaneously dispatch multiple UAVs to cooperate in completing data collection work.

[0132] S212, receiving a query request, extracting target box cargo information corresponding to the query request from the box cargo information database, and generating a query report based on the target box cargo information.

[0133] Referring to step S106, the logistics management system will generate a corresponding box cargo information report according to the query conditions.

[0134] In some embodiments, the logistics management system will generate a customized query report according to user permissions, i.e., the logistics management system will receive a query request containing query conditions, retrieve target box cargo information from the box cargo information database based on the query conditions; the query conditions include at least one of the box number information, the time range, the location area, and the damage state; the target box cargo information is structured according to a preset data template to generate a data report containing box cargo basic information, location trajectory information, and damage records, and abnormal data in the data report is highlighted; the accessible report fields are determined according to the user permission configuration information, and the data report is filtered based on the accessible report fields to generate a query report that meets the needs of different users.

[0135] wherein the query condition represents user-specified information retrieval parameters, containing multiple filtering dimensions; the target container information refers to container records that meet the query condition; the preset data template is used to represent a standardized data organization format; the abnormal data represents data items deviating from the normal range; the user permission configuration information refers to the definition of the data range accessible by different users; and the report field is used to represent specific data items in the data report.

[0136] After receiving the user query request, the logistics management system starts the data retrieval and report generation process. Specifically, the logistics management system first parses each parameter in the query condition to construct an optimized database query statement. Then, the system retrieves matching records from the container information database and reorganizes the data according to the standardized template. Next, the logistics management system performs statistical analysis on the data, identifies abnormal values and adds special markers. Finally, the system filters sensitive fields according to user permission configuration and generates personalized query reports for different user roles.

[0137] In some embodiments, data retrieval and report generation can be implemented in various ways: optionally, the logistics management system can use multi-dimensional indexing technology to accelerate data retrieval, precompute common statistical indicators through data cubes, use anomaly detection algorithms to identify data anomalies, and finally generate rich-text format reports through a template engine; optionally, the logistics management system can use a distributed query engine to handle massive data, implement real-time statistical analysis through stream computing, use a rule engine for anomaly marking, and generate interactive reports through a visualization component. It can be understood that other data processing and report generation methods can also be used to implement this function, which is not limited here.

[0138] In practical applications, large data volume queries may cause response delays. To address this issue, the logistics management system uses query optimization strategies: first, composite indexes are established to improve query efficiency; second, a query result caching mechanism is implemented, which can directly return cached data for similar query conditions; finally, the system also supports asynchronous query mode, and for complex statistical analysis tasks, uses a background processing method to complete the task and notifies the user through message pushing.

[0139] In the embodiments of the present application, the all-around three-dimensional monitoring and intelligent management of the container yard can be realized by using the unmanned aerial vehicle-based multi-sensor data acquisition system and deep learning analysis platform in combination with intelligent path planning and environment adaptive technology. The flexible mobility of the unmanned aerial vehicle and the multi-sensor fusion technology can effectively solve the problems of blind area of the traditional fixed monitoring equipment, low efficiency and strong subjectivity of manual inspection. The deep learning model realizes objective and accurate evaluation of damage detection, the environment perception and adaptive path planning ensure the safe and efficient execution of the inspection task, the real-time data analysis and multi-dimensional report function support intelligent decision-making, thereby realizing the overall upgrade of the port logistics management and significantly improving the operation efficiency and management level. Meanwhile, the scalability of the system provides a good foundation for future functional upgrade and intelligent development.

[0140] The logistics management system in the embodiments of the present application will be described from the perspective of hardware processing. Please refer to Figure 3 , which is a schematic diagram of an entity device structure of the logistics management system in the embodiments of the present application.

[0141] It should be noted that Figure 3 The structure of the logistics management system shown is only an example and should not impose any limitation on the functions and use range of the embodiments of the present application.

[0142] As Figure 3 shown, the logistics management system includes a CPU 301 which can perform various appropriate actions and processes, such as the methods described in the above embodiments, according to programs stored in a ROM 302 or loaded from a storage portion 308 to a RAM 303. Various programs and data required for system operation are also stored in the RAM 303. The CPU 301, the ROM 302, and the RAM 303 are connected to each other through a bus 304. An I / O interface 305 is also connected to the bus 304.

[0143] The following components are connected to the I / O interface 305: an input portion 306 including an audio input device, a push button switch, and the like; an output portion 307 including a liquid crystal display (LCD), an audio output device, an indicator, and the like; a storage portion 308 including a hard disk and the like; and a communication portion 309 including a network interface card such as a LAN (Local Area Network) card, a modem, and the like. The communication portion 309 performs communication processing via a network such as the Internet. A drive 310 is also connected to the I / O interface 305 as needed. A removable medium 311 such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, and the like is mounted on the drive 310 as needed, so that a computer program read therefrom is installed in the storage portion 308 as needed.

[0144] In particular, the processes described above with reference to the flow charts can be implemented as a computer software program according to embodiments of the present application. For example, embodiments of the present application include a computer program product comprising a computer program carried on a computer readable medium, the computer program comprising computer programs for executing the methods illustrated by the flow charts. In such embodiments, the computer program can be downloaded and installed from a network via the communication section 309, and / or installed from the removable medium 311. When the computer program is executed by the CPU 301, various functions defined in the present application are executed.

[0145] The flow charts and block diagrams in the drawings are schematic illustrations of possible architectures, functions and operations of systems, methods and computer program products in accordance with various embodiments of the present application. In this regard, each block in the flow charts or block diagrams can represent a module, a segment, or a portion of code, which comprises one or more executable instructions for implementing the specified logical functions. It should also be noted that in some alternative implementations, the functions noted in the blocks can occur out of the order noted in the figures.

[0146] In particular, the logistics management system of the embodiments includes a processor and a memory, and the memory stores a computer program which, when executed by the processor, implements the real-time query method of the dynamic information of the container cargo provided by the above embodiments.

[0147] As another aspect, the present application also provides a computer readable storage medium, which can be included in the logistics management system described in the above embodiments, or can exist separately without being assembled into the logistics management system. The storage medium carries one or more computer programs, which, when executed by a processor of the logistics management system, enable the logistics management system to implement the real-time query method of the dynamic information of the container cargo provided by the above embodiments.

[0148] The above embodiments are only used to illustrate the technical solutions of the present application, but not limit the present application; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacements to some technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present application.

[0149] As used in the above embodiments, the term "when" can be interpreted to mean "if" or "after" or "in response to determining" or "in response to detecting" depending on the context. Similarly, the phrase "on determining" or "if detecting (a stated condition or event)" can be interpreted to mean "if determined" or "in response to determining" or "on detecting (a stated condition or event)" or "in response to detecting (a stated condition or event)" depending on the context.

Claims

1. A method for real-time query of cargo dynamic information, characterized in that, Applied to a logistics management system, the method includes: Receive yard image data and laser ranging data collected by drones, construct a three-dimensional digital model of the container yard, and determine the spatial coordinate information of each container; Based on the yard image data, the container number information of each container is identified, and the container number information is associated with the spatial coordinate information to generate container location data; The RFID reading device of the drone collects the electronic tag information of each container, matches and verifies the electronic tag information with the container and cargo location data, and generates real-time container and cargo information. The image data of the container yard is processed by image segmentation to identify the surface features of each container, and the surface features of the container are compared with the damage feature library to generate damage detection results; The damage detection results are correlated with the real-time container and cargo information to update the container and cargo information database that represents the container status; Receive a query request, extract the target cargo information corresponding to the query request from the cargo information database, and generate a query report based on the target cargo information.

2. The method according to claim 1, characterized in that, The steps of performing image segmentation processing on the yard image data, identifying the surface features of each container, and comparing the surface features with a damage feature database to generate damage detection results specifically include: The image data of the storage yard is input into a deep learning model to extract the texture features, color features and edge features of the container surface to obtain the container surface feature vector; Obtain feature templates corresponding to preset damage types from the damage feature library; the feature templates include dent features, crack features, corrosion features, and leakage features; Calculate the similarity score between the surface feature vector of the container and each of the feature templates; The damage type corresponding to the feature template whose similarity score exceeds a preset similarity threshold is determined as the damage detection result.

3. The method according to claim 2, characterized in that, After the step of calculating the similarity scores of the surface feature vector of the container and each of the feature templates, the method further includes: Based on the surface feature vector of the container, the yard image data is segmented to identify suspected damaged areas; Extract historical image blocks corresponding to the suspected damaged area from historical image data, and calculate the image difference between the suspected damaged area and the historical image blocks; When the image difference exceeds a preset difference threshold, the suspected damaged area is marked as a newly added damaged point; The newly added damaged points are evaluated and classified according to the type and degree of damage, and a damage inspection report containing the location, type and degree of damage is generated.

4. The method according to claim 1, characterized in that, The steps of receiving a query request, extracting the target cargo information corresponding to the query request from the cargo information database, and generating a query report based on the target cargo information specifically include: Receive a query request containing query conditions, and retrieve target cargo information from the cargo information database based on the query conditions; the query conditions include at least one of the following: container number information, time range, location area, and damage status; The target cargo information is structured according to a preset data template to generate a data report containing basic cargo information, location trajectory information and damage records, and abnormal data in the data report is highlighted. The accessible report fields are determined based on the user permission configuration information, and the data report is filtered based on the accessible report fields to generate query reports that meet the needs of different users.

5. The method according to claim 1, characterized in that, After the step of associating the damage detection results with the real-time container cargo information and updating the information database characterizing the container status, the method further includes: Upon receiving a request to add or move a container cargo, the requested location of the container cargo is determined based on the yard monitoring images. Based on the three-dimensional digital model, an inspection path is generated for the UAV to reach the requested location; The inspection path is sent to the drone to trigger the drone to perform a new round of cargo information collection tasks.

6. The method according to claim 5, characterized in that, The step of generating an inspection path for the UAV to reach the requested location based on the three-dimensional digital model specifically includes: Obtain real-time wind speed and real-time wind direction data collected by the port meteorological station at the port where the container stacking site is located; Based on the stacking layout of the containers in the three-dimensional digital model, the wind resistance coefficient of each passageway applicable to the UAV is calculated. Based on the wind resistance coefficient, the real-time wind speed data, and the real-time wind direction data, determine the windward area affected by wind force and the sheltered area formed by container stacking; Based on the distribution characteristics of the windward and sheltered areas, an inspection path is planned and generated.

7. The method according to claim 6, characterized in that, The step of planning and generating inspection paths based on the distribution characteristics of the windward and sheltered areas specifically includes: In the path sub-region that includes the windward area and the sheltered area, calculate the short path distance through the windward area and the detour path distance through the sheltered area, respectively; Calculate the windbreak path consumption corresponding to the detour path distance; Calculate the additional energy consumption of the UAV when resisting wind force when passing through the windward area, and determine the windward path consumption based on the additional energy consumption and the path length consumption of the short-range path distance; The path with the lowest consumption value is determined based on the consumption of the sheltered path and the consumption of the windward path, and is used as the inspection sub-path of the path sub-region. An inspection path is generated based on multiple inspection sub-paths, and a path execution command containing flight altitude, speed, and obstacle avoidance parameters is generated.

8. A logistics management system, characterized in that, The logistics management system includes: one or more processors and a memory; the memory is coupled to the one or more processors, the memory is used to store computer program code, the computer program code including computer instructions, and the one or more processors call the computer instructions to cause the logistics management system to perform the method as described in any one of claims 1-7.

9. A computer-readable storage medium comprising instructions, characterized in that, When the instruction is executed on the logistics management system, the logistics management system performs the method as described in any one of claims 1-7.

10. A computer program product, characterized in that, When the computer program product is run on the logistics management system, the logistics management system performs the method as described in any one of claims 1-7.

Images