A method and system for bridge detection using a drone swarm

Abstract

The embodiment of the application provides a kind of method and system for bridge detection using unmanned aerial vehicle group, belong to bridge structure detection technical field.The method comprises: forming the joint section matrix for measurement task generation based on initialization acquisition data;Based on each target time scale, generate a multi-time scale measurement task set;According to the multi-time scale measurement task set, execute unmanned aerial vehicle distribution, determine the flight window of each unmanned aerial vehicle under each time scale and the corresponding target joint section set, obtain the detection task;Based on the detection task, control each unmanned aerial vehicle to collect the measurement data of the laser of target joint section in the corresponding flight window, to execute the joint size extraction of corresponding joint section, and update the joint size to the size life cycle curve of corresponding joint section.The application scheme effectively solves the problems of insufficient joint section size data acquisition accuracy, cross-time scale information fragmentation, etc., and improves the integrity and dynamics of bridge joint section size evolution monitoring.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of bridge structure detection, in particular to a method for bridge detection using a UAV group and a system for bridge detection using a UAV group. BACKGROUND

[0002] As the core infrastructure in modern transportation networks, the operation safety of bridges is directly related to the lifeline of the national economy and the safety of public life and property. With the extension of the service life of bridges, problems such as structural aging, stress accumulation, and load fatigue gradually appear. The joint structure area in the bridge structure (such as expansion joints, joints, and construction joints) often becomes the concentrated part of structural diseases. Due to the combined effects of long-term exposure to environmental loads, temperature and humidity changes, and vehicle impact, the joint area is prone to problems such as abnormal opening of the joint, closure lag, crack expansion, or filling peeling, and in severe cases, it may even affect the structural integrity and performance of the bridge.

[0003] Traditional bridge detection techniques rely on manual inspection, static monitoring point layout, or periodic instrument measurement, which have the disadvantages of low monitoring frequency, limited detection distribution, and data update lag. In particular, it is difficult to cover the typical characteristics of the joint segment, such as dispersed position distribution, complex scale variation, and significant evolution rate difference. Therefore, how to establish a repeatable and quantifiable physical size measurement mechanism for the bridge joint area has become one of the important problems to be solved in current bridge operation monitoring.

[0004] Among all the detection indicators related to the state of the joint segment, the physical size of the joint (especially the joint width) is a key quantity that directly represents the structural stress release, deformation behavior, and disease evolution trend, with high structural sensitivity and monitoring value. The accurate extraction of joint size can not only be used to quantify the relative displacement between bridge components, but also provide data basis for structural safety warning. However, the field measurement of joint size faces many challenges in actual operation, such as complex joint morphology, serious reflection disturbance, irregular distribution of measurement area, and high-altitude or hidden operation risks in some areas, which greatly limits the acquisition of high-precision and high-frequency physical size.

[0005] Therefore, there is an urgent need for a technical means that can measure the physical size of the bridge joint segment with high resolution, spatial continuity, and temporal multi-scale, to realize dynamic perception and time management of the joint size, serving the operation evaluation and risk control of the bridge structure. SUMMARY

[0006] The purpose of the embodiments of the present application is to provide a method and system for bridge detection using a UAV group to at least solve the problem of difficulty in high-precision, continuous collection and unified management of bridge joint physical size in multiple time scales.

[0007] To achieve the above object, the first aspect of the present application provides a method for bridge detection using a group of unmanned aerial vehicles, the method comprising: collecting initialization collection data of the group of unmanned aerial vehicles, and performing joint type calibration and spatial position partitioning of a target bridge based on the initialization collection data to form a joint matrix for generating a measurement task; defining a measurement time window and parameter configuration of each joint in the joint matrix based on each target time scale to generate a multi-time scale measurement task set; performing unmanned aerial vehicle allocation according to the multi-time scale measurement task set to determine a flight window and a corresponding target joint set of each unmanned aerial vehicle at each time scale to obtain a detection task; controlling each unmanned aerial vehicle to collect measurement data of laser light of the target joint within the corresponding flight window based on the detection task to perform joint size extraction of the corresponding joint, and updating the joint size to a size life cycle curve of the corresponding joint.

[0008] Optionally, the initialization collection data of the group of unmanned aerial vehicles is collected, and the joint type calibration and spatial position partitioning of the target bridge are performed based on the initialization collection data to form a joint matrix for generating a measurement task, comprising: controlling each unmanned aerial vehicle to perform bridge deck image sequence, unmanned aerial vehicle position information and joint edge point cloud data collection during cruising in the range of the target bridge; performing joint type classification based on the image feature matching results of the joint edge point cloud data and the bridge deck image sequence; wherein the joint type is any one of a bridge deck expansion joint, a guardrail expansion joint, a bridge pier expansion joint, a sidewalk expansion joint, a hanging basket track structure joint and a cable-stayed anchor plate joint; performing spatial paragraph division of each joint based on the position information of each unmanned aerial vehicle; outputting a joint type identifier and a spatial position index based on the joint type and the spatial paragraph of each joint to construct a corresponding joint matrix based on the joint type identifier and the spatial position index.

[0009] Optionally, the measurement time window and parameter configuration of each joint in the joint matrix are defined based on each target time scale to generate a multi-time scale measurement task set, comprising: obtaining time control parameters of a daily variation scale, a social cycle scale and a seasonal cycle scale from a preset time scale configuration set respectively; setting the measurement start and end time, the sampling frequency and the response delay threshold value at the corresponding time scale as the collection control parameters based on the time control parameters of each time scale; mapping and combining the collection control parameters with each joint number in the joint matrix one by one, and constructing the multi-time scale measurement task set based on each mapping combination.

[0010] Optionally, each task in the multi-time scale measurement task set comprises: a joint number, a joint type, a time scale label, a task trigger condition, an image collection resolution requirement and a tolerance; wherein the time scale label is used to calibrate the task execution priority; the task trigger condition is used to define an automatic scheduling strategy; and the image collection resolution requirement is used to guide the unmanned aerial vehicle to configure an adaptive flight height and camera focal length.

[0011] Optionally, the UAV allocation is performed according to the multi-time scale measurement task set, comprising: determining a corresponding scheduling parameter for each task in the multi-time scale measurement task set; calculating the accessibility score and the task load balancing score of each UAV in the corresponding time window of the task based on the remaining power, flight capability, historical task load condition and image resolution adaptation capability of each UAV; under the premise of meeting the task triggering condition, performing scheduling score calculation by comprehensively considering the accessibility score, the task load balancing score and the task priority corresponding to the time scale label, determining the UAV with the highest task scheduling score, and constructing the binding relationship between the corresponding UAV and the corresponding task; wherein each task is only bound to one UAV.

[0012] Optionally, the flight window and the corresponding target seam set of each UAV at each time scale are determined, and a detection task is obtained, comprising: based on the binding relationship between each task and the UAV formed by the highest scheduling score, all task items bound by each UAV are counted; extracting the seam number from each task item and summarizing it as the target seam set of the UAV; according to the measurement time window of the bound task, combining the flight height required by the image acquisition resolution requirement and the heading stability time, calculating the flight start time, the flight end time and the path planning result of the UAV at each time scale to form the flight window; pairing the target seam set with the flight window and storing it as the detection task of the UAV.

[0013] Optionally, based on the detection task, the measurement data of the laser collected by each UAV in the corresponding flight window is controlled to perform seam size extraction of the corresponding seam, comprising: controlling the UAV to stably hover at the seam position in the flight window to shoot seam area images at a preset pitch angle; performing edge recognition based on seam boundary line extraction algorithm in seam area images to perform target seam positioning; according to the target seam positioning result, performing laser measurement on the target seam to obtain the seam size information of the target seam.

[0014] Optionally, the size life cycle curve of the seam is indexed by the seam number, and records the time-varying sequence of the measured seam size of the seam at all time scales; wherein the curve node of the size life cycle curve is composed of a timestamp and a corresponding size value.

[0015] Optionally, the seam size is updated to the size life cycle curve of the corresponding seam, comprising: performing time axis standardization on the measurement results at different time scales, and respectively entering the data with different measurement frequency and time length characteristics into the corresponding size life cycle curve according to the seam number, and performing cross-scale difference analysis and trend alignment processing based on the overlapping segments between time scales to realize the synchronization of the monitoring data at each time scale.

[0016] The second aspect of the present application provides a system for bridge detection using a UAV group, the system comprising: a collection unit configured to collect initialization collection data of the UAV group, and to perform target bridge joint segment type calibration and spatial position partitioning based on the initialization collection data, to form a joint segment matrix for generating a measurement task; a processing unit configured to define a measurement time window and parameter configuration of each joint segment in the joint segment matrix based on each target time scale, to generate a multi-time scale measurement task set; a task generation unit configured to perform UAV allocation according to the multi-time scale measurement task set, to determine a flight window and a corresponding target joint segment set of each UAV at each time scale, to obtain a detection task; and a detection unit configured to control each UAV to collect measurement data of laser light of the target joint segment within the corresponding flight window based on the detection task, to perform joint opening size extraction of the corresponding joint segment, and to update the joint opening size to a size life cycle curve of the corresponding joint segment.

[0017] Through the above technical solution, the present application constructs a joint segment matrix to perform type calibration and spatial partitioning on the bridge joint body area, realizes structured modeling of the detection object, and guarantees the comprehensiveness and traceability of subsequent monitoring; combines multi-time scale definition of measurement task windows and parameter configuration, supports continuous tracking of the bridge structure at different time periods such as daily changes and seasonal responses; performs UAV allocation mechanism driven by execution capability and task characteristics, realizes efficient matching of flight tasks and detection tasks; finally, extracts the joint opening physical size in a laser measurement mode, and updates the joint opening physical size to the size life cycle curve, effectively solves the problems of insufficient joint segment size data collection accuracy and cross-time scale information fragmentation, and improves the integrity and dynamics of bridge joint segment size evolution monitoring.

[0018] Other features and advantages of the present application will be described in detail in the following detailed description. BRIEF DESCRIPTION OF DRAWINGS

[0019] The accompanying drawings are included to provide a further understanding of the present application and constitute a part of the specification, and are used together with the following detailed description to explain the present application, but do not constitute a limitation of the present application. In the drawings:

[0020] Figure 1 is a step flowchart of a method for bridge detection using a UAV group provided by an embodiment of the present application;

[0021] Figure 2 is a system structure diagram of a system for bridge detection using a UAV group provided by an embodiment of the present application. DETAILED DESCRIPTION

[0022] The specific embodiments of the present application are described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are only used to illustrate and explain the present application, and are not used to limit the present application.

[0023] Figure 1 is a step flow chart of a method for bridge detection using a UAV group provided by an embodiment of the present application. As shown in Figure 1 , the embodiment of the present application provides a method for bridge detection using a UAV group, which comprises:

[0024] Step S10: Collecting initialization collection data of the UAV group, and based on the initialization collection data, performing joint type calibration and spatial position partitioning of the target bridge to form a joint matrix generated for measurement task.

[0025] Specifically, the control of each UAV performs bridge deck image sequence, UAV position information and joint edge point cloud data collection during the cruising process in the target bridge range; based on the image feature matching result of the joint edge point cloud data and the bridge deck image sequence, the joint type classification is performed; wherein the joint type is any one of bridge deck expansion joint, guardrail expansion joint, bridge pier expansion joint, bridge head expansion joint, hanging basket track structure joint and inclined cable anchor plate joint; based on the position information of each UAV, the spatial paragraph division of each joint is performed; based on the joint type and spatial paragraph of each joint, the joint type identifier and spatial position index are outputted, and the corresponding joint matrix is constructed based on the joint type identifier and spatial position index.

[0026] In the embodiment of the present application, in the actual application scene of bridge structure detection, before the physical size measurement task of different types of joints is carried out, the position and structural characteristics of all joints in the target area of the bridge must be comprehensively and standardized identified and organized. Therefore, the embodiment sets a complete set of joint information collection and structured expression process to ensure the pertinence and integrity of the subsequent measurement task.

[0027] Specifically, first, the initialization collection task of the UAV group is performed, and each UAV is controlled to complete the cruising operation according to the preset flight route in the designated monitoring area of the bridge, and simultaneously performs the collection of three types of data during the cruising process:

[0028] 1) Bridge deck image sequence, used to capture the surface visual information and boundary texture features of the joint area.

[0029] 2) UAV position information itself, including time-synchronized spatial coordinates and flight attitude angle, used for spatial positioning and angle correction of the collected image.

[0030] 3) Joint edge point cloud data collected by laser radar or structured light module, used to construct the geometric boundary and three-dimensional spatial distribution of the joint.

[0031] After the basic data acquisition is completed, the image sequence and the edge point cloud data are registered and fused. First, the boundary features of the joint section area are identified through an image feature point extraction algorithm (such as SIFT, SURF or ORB, etc.); then, feature matching is performed with the edge change in the point cloud to form an image-point cloud joint feature set. On this basis, a joint section type classification operation is performed. The classification process can be completed based on a supervised learning model or a rule matching model constructed based on an existing structure sample library, and different types of joint sections are identified according to the differences in visual features, edge topology and typical morphology, and finally each identified joint section is labeled as any one of the following types: bridge expansion joint, guardrail expansion joint, bridge pier expansion joint, sidewalk expansion joint, bridge head plate joint gap, hanging basket track structure joint or cable anchor plate joint.

[0032] To further support the allocation and path planning of tasks, after the type identification is completed, the spatial position of the joint section also needs to be standardized described. The position information of each unmanned aerial vehicle during collection is used to divide and label the identified joint sections of various types in the bridge panoramic space according to their projection positions, actual distribution and relative distance. Usually, the bridge axis direction is divided into multiple space paragraphs, the arrangement order, length coverage and adjacent relationship of the joint sections in each paragraph are recorded, and a space paragraph index is generated.

[0033] Finally, combined with the type identification of each joint section and the space paragraph index to which it belongs, a structured description format of “joint section type identification plus space position index” is output to construct a joint section matrix. The joint section matrix organizes all detectable joint sections in a row-column manner, abstracts the joint distribution of the bridge structure into a data structure with statistical boundaries and attribute labels, and supports various operations such as subsequent task generation, scheduling planning and data archiving.

[0034] Through the above processing procedure, not only the unified expression of the joint section type information is ensured, but also the coordinate-level modeling of the spatial distribution is realized, which lays a foundation for multiple unmanned aerial vehicles to cooperatively carry out high-precision physical size measurement tasks in different regions. In terms of technical effects, this method effectively solves the problems of manual dependence of bridge joint section identification, non-uniform spatial positioning and incomplete data structure, improves the organization efficiency of monitoring objects and the adaptability of scheduling algorithms, and has good engineering implementability and platform portability.

[0035] Step S20: Based on each target time scale, define the measurement time window and parameter configuration of each joint section in the joint section matrix to generate a multi-time scale measurement task set.

[0036] Specifically, the time control parameters of the daily variation scale, the social cycle scale and the seasonal cycle scale are respectively obtained from a preset time scale configuration set; the measurement start and end time, the sampling frequency and the response delay threshold under the corresponding time scale are respectively set based on the time control parameters of each time scale as the acquisition control parameters; the acquisition control parameters are mapped and combined with each joint segment number in the joint segment matrix one by one, and a multi-time scale measurement task set is constructed based on each mapping combination.

[0037] Further, each task in the multi-time scale measurement task set includes a joint segment number, a joint segment type, a time scale label, a task trigger condition, an image acquisition resolution requirement and a repetition tolerance; wherein the time scale label is used to mark the task execution priority; the task trigger condition is used to define an automatic scheduling strategy; and the image acquisition resolution requirement is used to guide the flight height and camera focal length of the unmanned aerial vehicle to be configured and adapted.

[0038] In the embodiment of the application, in the process of monitoring the structure of the bridge joint segment, in order to effectively capture the change trend of the joint size under different time periods, a measurement task set with multi-time scale control capability needs to be constructed to support monitoring targets of different granularities from minutes to quarters. The premise of constructing such a task set is to reasonably define the measurement time window and corresponding parameter configuration of each type of joint segment under different time sequences according to the multi-source change mode exhibited during the service of the bridge.

[0039] Specifically, first, three types of target time scales are extracted from a preset time scale configuration set: daily variation scale, social cycle scale and seasonal cycle scale. The daily variation scale is mainly used to depict the regular response of the bridge structure within a 24-hour cycle, such as thermal expansion and contraction, traffic load fluctuation, etc., for example, sudden temperature rise or drop, dynamic change of joint width caused by morning and evening peak commuting load.

[0040] Further, the social cycle scale is used to monitor the structural response of the bridge during the social activity cycle, for example, the periodic disturbance of the joint width caused by the difference in traffic mode between weekdays and holidays, and also includes the local stress concentration on the bridge deck caused by large-scale activities or sudden events in the short term.

[0041] Further, the seasonal cycle scale is mainly used to evaluate the joint evolution characteristics caused by structural deformation, material aging, concrete creep and other factors caused by long-term climate change.

[0042] For each target time scale, four control indicators, including measurement start time, measurement end time, sampling frequency and response delay threshold, are extracted from its time control parameters, which are collectively referred to as acquisition control parameters. Among them, the measurement start and end time are used to define the boundary of the observation interval at this scale; the sampling frequency represents the number of images required to be collected within this time period, which directly affects the periodicity density of the task; the response delay threshold is used to limit the maximum tolerance time of the task execution delay, which is used to distinguish between tasks with strong and weak real-time requirements.

[0043] Subsequently, each set of acquisition control parameters is mapped and combined with each joint number in the joint matrix. The numbers recorded in the joint matrix are the basic joint units classified in the process of bridge joint type calibration and spatial position indexing, including bridge expansion joints, guardrail expansion joints, pier expansion joints, sidewalk expansion joints, bridge head slab gaps, basket track structure joints and cable anchor plate joints, etc. The combination of each joint number and a set of acquisition control parameters constitutes a monitoring task to be executed.

[0044] Based on the above joint number and control parameter combination, a multi-time scale measurement task set is constructed. Each task exists as an independently recorded task unit, which includes the following six key fields in its structure:

[0045] 1) Joint number: used to identify the target joint corresponding to the task, ensuring that the measurement task is positioned to a unique spatial target.

[0046] 2) Joint type: indicates the structure type of the target joint, providing a basis for subsequent selection of appropriate processing models and task priority strategies.

[0047] 3) Time scale label: represents the time scale type to which the task belongs (e.g. "daily scale", "social cycle scale", "seasonal scale"), used to guide the UAV scheduling system to prioritize tasks.

[0048] 4) Task trigger condition: used to define the automatic scheduling logic, including fixed period trigger, event-driven trigger (such as previous period measurement anomaly) or hybrid trigger strategy, supporting dynamic adaptation of task execution frequency.

[0049] 5) Image acquisition resolution requirement: set according to the joint structure size, surface feature complexity and crack boundary resolution requirement, which indirectly determines the flight height, camera focal length, stabilization time, etc. of the UAV when executing the task.

[0050] 6) Repetition tolerance: used to measure the acceptable execution redundancy of the task, i.e. the tolerance threshold that allows part of the task to be delayed or merged in consecutive multiple measurements, which plays an important role in flight path compression and multi-task merging.

[0051] Each task record is organized in a structured manner, facilitating subsequent sorting, filtering, binding and scheduling of the task set at each time scale, and enabling fine-grained task strategy arrangement through multi-condition combination of task fields. For example, on the same joint segment, there may be tasks with "daily scale, morning peak sampling, real-time response" and tasks with "seasonal scale, end-of-month sampling, tolerate two-day delay", and their task execution methods and resource matching requirements are different.

[0052] This process has high universality and scalability. On the one hand, through multi-time scale task configuration, the multi-dimensional coverage ability of the bridge structure response law is significantly improved, so that the short-period fluctuation, periodic behavior and long-term evolution process of the structure change can be recorded and tracked at the same time. On the other hand, the task set is constructed in an independent field manner, which is convenient for cross-platform adaptation and supports flexible calling in subsequent flight path optimization, unmanned aerial vehicle capability scheduling, image data standardization and other processes.

[0053] The scheme effectively solves the problems of single monitoring frequency, rough period division and uncontrollable task parameter configuration in the traditional bridge monitoring scheme, especially for the special target of bridge joint segment which has strong structure locality, uneven spatial distribution and diverse response behavior. Through the double mapping of time scale and spatial segmentation, highly fine-grained task configuration is realized. Further, since the task parameters have completeness and clear attribution, the intelligent scheduling efficiency of subsequent unmanned aerial vehicle allocation process and the time consistency of detection data can be significantly improved, laying a solid foundation for the construction of bridge joint segment size life cycle curve.

[0054] Step S30: performing unmanned aerial vehicle allocation according to the multi-time scale measurement task set, determining the flight window of each unmanned aerial vehicle at each time scale and the corresponding target joint segment set, and obtaining the detection task.

[0055] Specifically, performing unmanned aerial vehicle allocation according to the multi-time scale measurement task set comprises: determining the corresponding scheduling parameter for each task in the multi-time scale measurement task set; calculating the accessibility score and task load balancing score of each unmanned aerial vehicle in the corresponding time window of the task based on the remaining power, flight capability, historical task load condition and image resolution adaptation capability of each unmanned aerial vehicle; under the premise of meeting the task triggering condition, performing scheduling score calculation by comprehensively considering the accessibility score, task load balancing score and task priority corresponding to the time scale label, determining the unmanned aerial vehicle with the highest task scheduling score for each task, and constructing the binding relationship between the corresponding unmanned aerial vehicle and the corresponding task; wherein, each task is only bound to one unmanned aerial vehicle.

[0056] Further, the flight window of each unmanned aerial vehicle at each time scale and the corresponding target seam section set are determined, and a detection task is obtained, including: based on the highest scheduling score, the binding relationship between each task and the unmanned aerial vehicle is formed, and all task items bound by each unmanned aerial vehicle are counted; the seam section number is extracted from each task item and is summarized as the target seam section set of the unmanned aerial vehicle; the flight start time, the flight end time and the path planning result of the unmanned aerial vehicle at each time scale are calculated according to the measurement time window of the bound task, combined with the flight height required by the image acquisition resolution requirement and the heading stability time, to form a flight window; the target seam section set and the flight window are paired and stored as the detection task of the unmanned aerial vehicle.

[0057] In the embodiment of the application, in the multi-time scale detection process of the bridge seam section, in order to guarantee the integrity, accuracy and scheduling efficiency of task execution, an intelligent allocation mechanism based on task set, taking into account the differences in unmanned aerial vehicle capabilities and time requirements, is needed. Especially in the complex background involving multiple unmanned aerial vehicles, multiple seam sections and multiple time dimensions, a single allocation method cannot meet the comprehensive requirements of task refinement, task collaboration and maximum utilization of resources. Therefore, the present embodiment proposes a multi-time scale measurement task set based unmanned aerial vehicle allocation rule, and based on this, the flight window of each unmanned aerial vehicle at each time scale and the corresponding seam section set are determined, and finally a detection task list with high matching degree and high execution efficiency is output.

[0058] Specifically, first, the multi-time scale measurement task set constructed in the early stage is taken as the input basis. Each task in the task set contains fields such as seam section number, seam section type, time scale label, task trigger condition, image acquisition resolution requirement and repetition tolerance. In order to achieve optimal adaptation of unmanned aerial vehicles and tasks, scheduling parameters need to be extracted from each task, including but not limited to the measurement time window corresponding to the task, the target resolution level, the execution priority corresponding to the time scale label, and the trigger condition.

[0059] On the basis of clear scheduling parameters, the execution capability information of each unmanned aerial vehicle participating in scheduling needs to be obtained. The execution capability information at least includes the following four types:

[0060] 1) Remaining power: the remaining power of the current unmanned aerial vehicle and the calculated endurance time of the standard flight power, which directly affects whether it has execution capability within the target time window.

[0061] 2) Flight capability: including maximum flight speed, maximum range, height adjustment range and wind resistance level, etc., which is used to determine whether it can meet the flight height and path coverage requirements specified by the task.

[0062] 3) Historical task load condition: the number, distribution density and execution frequency of tasks that the UAV has undertaken in the current task period, used for task balancing scheduling.

[0063] 4) Image resolution adaptation capability: whether the actual imaging capability of the camera carried by the UAV meets the image acquisition resolution requirement of the specified task, which needs to be matched with the "image acquisition resolution requirement" in the task field.

[0064] Based on the above execution capability parameters, the reachability score and task load balancing score of each UAV in the corresponding time window of each task are calculated. The calculation method of reachability score is to consider the flight path cost (distance, path complexity), arrival time and matching degree of time window of the UAV from the current location to the target section under the premise of ensuring sufficient remaining power to complete the entire task range, and to generate a reachability score. The task load balancing score is based on the overall distribution of the current task load of the UAV and the task queue, to build a task density function, and to preferentially allocate new tasks to UAVs with lower current load, thereby avoiding excessive concentration of local resources.

[0065] Further, under the premise of meeting the task triggering condition, the above scoring results and task priority are included in the scheduling score function together to perform scheduling score calculation.

[0066] Specifically, the scheduling score = a x reachability score + b x task load balancing score + g x priority score, where a, b, g are adjustable weight factors, and the priority score is generated by time scale label mapping (e.g. low for seasonal scale, medium for daily scale, high for early morning or evening rush hour). For each task, by traversing all available UAVs, the scheduling score is calculated, and the UAV with the highest score is selected to bind with the task.

[0067] It should be noted that in order to ensure the uniqueness and integrity of task execution, each task is only allowed to establish a binding relationship with one UAV after scheduling is completed, ensuring that the task is not executed multiple times, and facilitating unified management of task paths and resources. When all tasks are initially allocated, based on the formed "task-UAV" binding relationship, the complete detection task list for each UAV is further generated.

[0068] Specifically, all allocated tasks are traversed, and the binding results are respectively attributed to the corresponding UAV ID to form a preliminary task list of each UAV; the seam section numbers corresponding to all tasks are extracted from the task list, and the target seam section set that the UAV needs to perform detection operation is formed by de-duplication and summarization; for each task, the shortest path planning from the starting point to the target seam section is calculated according to the measurement time window, image acquisition resolution requirement and flight height requirement; considering the time window overlap of multiple tasks, path compression and path splicing between tasks are performed to obtain a continuous and efficient flight route; for each continuous task combination path, flight control parameters such as start time, end time, shortest parking time, turning radius and stable shooting interval are calculated to finally form a complete flight window. The target seam section set and its corresponding flight window information are paired, and the task number, corresponding seam section number, execution time, flight parameter, image acquisition parameter and other contents are recorded to finally form the formal detection task record of the UAV.

[0069] Through the above process, it is ensured that each UAV can match appropriate tasks according to its own ability, and continuous flight and multi-seam section collaborative detection can be realized under different time scales. Since the task scheduling follows the principle of "score optimization plus unique binding", the optimal distribution of tasks can also be realized in the scenario where the number of UAVs is limited.

[0070] Step S40: Based on the detection task, control each UAV to collect the measurement data of the laser of the target seam section in the corresponding flight window to perform seam size extraction of the corresponding seam section, and update the seam size to the size life cycle curve of the corresponding seam section.

[0071] Specifically, based on the detection task, control each UAV to collect the measurement data of the laser of the target seam section in the corresponding flight window to perform seam size extraction of the corresponding seam section, including: controlling the UAV to stably hover at the seam section position in the flight window to shoot a seam area image at a preset pitch angle; performing edge recognition based on a seam boundary line extraction algorithm in the seam area image to perform target seam positioning; performing laser measurement on the target seam according to the target seam positioning result to obtain seam size information of the target seam.

[0072] Further, the size life cycle curve of the seam section takes the seam section number as an index to record the sequence of the seam size measured at all time scales changing with time; wherein, the curve node of the size life cycle curve is composed of a time stamp and a corresponding size value.

[0073] Further, the joint size is updated to the size life cycle curve corresponding to the joint section, including: time axis standardization of measurement results at different time scales, respectively, data with different measurement frequency and time length characteristics are classified into corresponding size life cycle curves according to joint section number, and cross-scale difference analysis and trend alignment processing are performed based on the overlapping fragments between time scales to realize synchronization of monitoring data at each time scale.

[0074] In the embodiment of the application, in order to realize long-term, stable and repeatable monitoring of the physical size of the joint section in the bridge structure, each unmanned aerial vehicle needs to be controlled to fly based on the generated detection task list, so that it collects laser measurement data of the target joint section within the flight window corresponding to the task, and obtains joint size information based on image processing and measurement extraction algorithm, and further classifies the obtained joint size result into the size life cycle curve of the joint section in time sequence to realize continuous recording of structural evolution trend.

[0075] Specifically, based on the flight window and the target joint section set contained in the detection task, each unmanned aerial vehicle is controlled to enter each flight window in turn, perform the take-off, hovering, shooting, return and other actions set in the task path planning, and complete image acquisition and size measurement at the target joint section. In the flight window control stage, priority is given to ensuring that the unmanned aerial vehicle has stable hovering capability, especially in the case of wind load disturbance, narrow area of bridge structure or high-low span partition, the unmanned aerial vehicle is kept stable within the preset height range through attitude control to ensure the consistency of image acquisition angle and laser ranging path.

[0076] In the image acquisition process, the image acquisition operation is performed on the joint area by using the vertical shooting mode with a downward angle. In actual operation, the downward angle of this shooting mode can be set to be between 40° and 70° relative to the horizontal plane, which not only ensures that the laser ranging path is perpendicular to the joint edge, but also provides sufficient field of view to cover the structural features around the joint. The acquired image needs to meet the minimum resolution standard set in the image acquisition resolution requirement, for example, the resolution is less than 1 mm / pixel, to ensure the accurate segmentation of the joint edge by the subsequent boundary extraction algorithm.

[0077] Further, based on the acquired image, joint positioning and boundary recognition operation is performed. In this step, the boundary search range is first narrowed according to the spatial index range preset for the joint section number, and initial edge detection is performed using edge gradient features, texture variability and gray distribution in the image. The edge extraction algorithm preferably uses Canny edge detection and edge fusion strategy, and combines with the color features of the joint surface in the image (such as concrete color difference, metal plate joint edge) to perform region segmentation and obtain the joint boundary line.

[0078] After positioning, laser measurement is performed on the target seam section using the image boundary as the measurement reference. The laser ranging module performs multi-point distance measurement on the selected measurement points in the target seam section boundary line. The measurement path is mapped with the image calibration path to obtain physical width data of the seam in multiple directions. To improve accuracy, a bidirectional laser cross-ranging path is preferably configured, and pixel value conversion to actual distance is achieved through image pixel size calibration. The final output is the seam size value of the target seam section at the current time, including but not limited to average width, maximum opening value, edge irregularity, etc.

[0079] The obtained seam size value is the result of the current measurement period and needs to be updated in the size life cycle curve. Each size life cycle curve is uniquely indexed by the seam number and records the size measurement values of the corresponding seam at all observation times since the system was deployed. The core structure is a set of curve nodes composed of timestamps and corresponding size values. New measurement data needs to be preprocessed for consistency, including timestamp standardization, duplicate data removal, and outlier removal operations.

[0080] Considering the significant difference in data collection frequency at different time scales, further normalization and standardization processing of the time axis is required. For example, at the daily scale, data is collected at hourly intervals, while at the seasonal scale, data is collected at weekly or monthly intervals. To ensure curve continuity and consistent comparison, a unified time axis unit (e.g., day) is used to map all measurement nodes. For time periods not covered by measurements, use compensation values or interpolation methods in adjacent time scales to fill in the gaps.

[0081] In addition, to ensure the fusibility of data between different time scales, cross-scale difference analysis and trend alignment processing are also required. This processing includes calculating the seam size change rate at different time scales, identifying its change period characteristics, matching and analyzing the change amplitude and direction based on overlapping time segments, and performing trend line fitting and correction on measurement values that deviate significantly, thereby forming a unified data sequence reflecting the change law of the seam section over multiple time dimensions.

[0082] The seam section size life cycle curve constructed by the present scheme not only reflects the structural state of the bridge at a specific time point, but more importantly, records the entire evolution process over time in a continuous and comparable form. This method significantly improves the accuracy and dimensionality of structural monitoring, avoiding the monitoring blind spots and misjudgments caused by traditional periodic manual sampling detection. At the same time, the data obtained through the combination of optical images and laser measurement has stronger reviewability and measurement robustness, providing high-quality data support for bridge structure maintenance, evaluation, and early warning.

[0083] Further, based on the size life cycle curve, a threshold alarm mechanism, change trend discrimination rule and periodic abnormal behavior recognition model can be set for a certain joint section, promoting the upgrade of bridge health monitoring from static interpretation to dynamic prediction, and thus comprehensively supporting the implementation of the intelligent maintenance concept of "observable, traceable and predictable" for bridges.

[0084] Embodiment:

[0085] Taking a city trunk road bridge as the target object, a bridge joint section physical size monitoring task based on a UAV group is carried out, aiming to build a size life cycle curve of the joint section and realize the evolution state perception and data archiving of the bridge joint body at multiple time scales. In this embodiment, a laser ranging and optical image fusion method is adopted to perform the joint size collection task, and a plurality of UAVs are cooperatively used to complete the monitoring operation with balanced time distribution and spatial distribution.

[0086] In the task initialization stage, firstly, the deployed plurality of UAVs are controlled to cruise over the bridge structure in turn, and the flight path covers the positions of all possible structural joint sections such as the bridge deck, guardrail, pier, sidewalk, bridge head apron, sling track and cable anchor plate. During this cruising process, each UAV synchronously collects the bridge deck image sequence, position information (GNSS+IMU combined navigation) and three-dimensional point cloud data of the structural edge. The point cloud data is collected by a rotating laser radar, and the optical image is collected by a 50 million pixel full-frame camera at a top-down angle to ensure that the image and the point cloud data are splicable.

[0087] After the point cloud data and the image sequence are fused, the image feature point matching + point cloud boundary extraction algorithm is used to identify the joint section boundary, and the joint section type classification is performed based on the boundary morphology and the bridge structure template library. The classification labels include bridge deck joint, guardrail joint, pier joint, sidewalk joint, bridge head apron joint, sling track structure joint and cable anchor plate joint, etc. Based on the recognition result, each joint section is assigned a "joint section number", and the spatial paragraph position of each joint section on the bridge structure axis is indexed through the position information, and finally a joint section matrix with "joint section number-joint section type-spatial position" as the key value structure is generated, providing a standard task target set for subsequent scheduling and monitoring.

[0088] Subsequently, the monitoring requirements of three typical time scales are extracted from the preset time scale configuration set: the "daily variation scale" (such as the evaluation of traffic load influence during morning and evening peak hours) at the minute to hour level, the "social activity cycle scale" (such as load changes during holidays, events and exhibitions) at the day to week level, and the "seasonal variation scale" (thermal expansion and contraction, freeze-thaw cycle, etc.) at the month to quarter level. The measurement start and end time, sampling frequency (such as 30 minutes for the minute level, and 7 days for the seasonal level) and response delay threshold (tolerance interval for controlling data redundancy or lag response) are defined for each type of time scale.

[0089] The above time scale parameters are mapped one by one to each seam number in the seam matrix to generate a complete multi-time scale measurement task set. Each task contains seam number, seam type, time scale label, task trigger condition (such as "collect when temperature change exceeds 5°C"), image acquisition resolution requirement (e.g. higher than 1 mm / pixel) and repetition tolerance (minimum time interval allowed for repeated collection).

[0090] In the task allocation stage, for each task item in the task set, the scheduling parameters are extracted and the schedulability is analyzed. For each UAV, based on its remaining power, maximum range, historical task execution (such as the execution density of the last round), and laser measurement and camera parameters, the accessibility score and load balancing score of each task are calculated. Combined with the time scale label (priority) and whether the task trigger condition is met, the scheduling score of each task for each UAV is calculated comprehensively, and the UAV with the highest scheduling score is selected to build a task binding relationship. Each task is only bound to one UAV to avoid task conflict or resource duplication.

[0091] After completing the binding relationship, the UAV binds all task items, extracts the corresponding seam number from the task set, and forms the target seam set for the UAV. According to the time window set by each task, combined with the ranging height requirement (such as vertical distance of 3-5 meters) of the UAV laser device and the shooting posture adjustment time (heading adjustment time ≥ 3 seconds), the UAV is planned to take off, the flight route, the operation time of each seam, and the return time, and the flight window is generated. Finally, the flight window and the seam set are bound to form the detection task list of the UAV.

[0092] In the detection task execution stage, each UAV enters the flight window under the corresponding time scale, automatically takes off and cruises to the specified seam area according to the task list, and starts the laser measurement device after stable hovering. The image of the seam is taken at a preset depression angle (typical depression angle is 60°), and the laser ranging is synchronized to obtain multiple physical size parameters such as the maximum width and average opening value of the seam. After aligning the image data and laser measurement data, the outliers are removed, and the legal data is taken as a measurement result node.

[0093] The measurement data is then written into the "size life cycle curve" indexed by seam number, which records the timestamp of each collection and the corresponding size value. For data collected at different time scales, uniform time axis standardization (such as all converted to daily units) is performed, and trend alignment and difference correction are carried out in time segments with overlapping data to ensure consistency between different time scale data.

[0094] The size life cycle curve can be used for subsequent structural health diagnosis, change trend prediction, alarm threshold setting and other high-order analysis processes, and meanwhile realizes multi-scale continuous and structured management of the physical size of the bridge joint section. The whole process does not need frequent manual intervention, and has high practicability and engineering popularization value.

[0095] Figure 2 is a system structure diagram of a system for bridge detection using a UAV group provided by an embodiment of the present application. As shown in Figure 2 The system for bridge detection using a UAV group provided by the embodiment of the present application comprises: a collection unit configured to collect initialization collection data of the UAV group, and perform joint type calibration and spatial position partitioning of a target bridge based on the initialization collection data to form a joint matrix for measurement task generation; a processing unit configured to define a measurement time window and parameter configuration of each joint in the joint matrix based on each target time scale to generate a multi-time scale measurement task set; a task generation unit configured to perform UAV distribution according to the multi-time scale measurement task set, determine a flight window and a corresponding target joint set of each UAV under each time scale, and obtain a detection task; and a detection unit configured to control each UAV to collect measurement data of laser of a target joint in a corresponding flight window based on the detection task, perform joint size extraction of the corresponding joint, and update the joint size to a size life cycle curve of the corresponding joint.

[0096] Those skilled in the art can understand that all or part of the steps in the method for implementing the above-mentioned embodiments can be completed by a program instructing related hardware, the program is stored in a storage medium, and the program includes a plurality of instructions for causing a single-chip microcomputer, a chip or a processor to execute all or part of the steps of the method described in each embodiment of the present application. The foregoing storage medium includes a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk and various storage medium capable of storing program codes.

[0097] The optional embodiments of the present application are described in detail above in combination with the drawings, but the embodiments of the present application are not limited to the specific details in the above-mentioned embodiments. Within the technical concept range of the embodiments of the present application, the technical solutions of the embodiments of the present application can be subjected to various simple modifications, and these simple modifications all belong to the protection range of the embodiments of the present application. In addition, it should be noted that each specific technical feature described in the above-mentioned specific embodiments can be combined in any appropriate manner without contradiction. In order to avoid unnecessary repetition, the embodiments of the present application will not further describe various possible combination manners.

[0098] Besides, the various embodiments of the present application can be combined arbitrarily, as long as it does not violate the idea of the embodiments of the present application, which should be considered as the disclosed content of the embodiments of the present application.

Claims

1. A method for bridge inspection using a fleet of unmanned aerial vehicles, the method comprising: The method comprises: ​ Collecting initialization data of the UAV group, and based on the initialization data, calibrating the joint type of the target bridge and partitioning the space position to form a joint matrix for generating a measurement task; including: Controlling each UAV to execute bridge deck image sequence, UAV position information and joint edge point cloud data collection during cruising in the target bridge range; based on the joint edge point cloud data and the image feature matching result of the bridge deck image sequence, executing joint type classification; wherein the joint type is any one of bridge deck expansion joint, guardrail expansion joint, bridge pier expansion joint, sidewalk expansion joint, bridge head apron gap, hanging basket track structure joint and cable-stayed anchor plate joint; based on the position information of each UAV, dividing the space paragraphs of each joint; based on the joint type and the space paragraph of each joint, outputting joint type identification and space position index to construct a corresponding joint matrix based on the joint type identification and the space position index; Based on each target time scale, defining the measurement time window and parameter configuration of each joint in the joint matrix to generate a multi-time scale measurement task set; including: Obtaining time control parameters of daily variation scale, social cycle scale and seasonal cycle scale from a preset time scale configuration set respectively; based on the time control parameters of each time scale, setting the measurement start and end time, sampling frequency and response delay threshold under the corresponding time scale as the collection control parameters; mapping and combining the collection control parameters with each joint number in the joint matrix one by one, and constructing a multi-time scale measurement task set based on each mapping combination; wherein, The social cycle scale is used to monitor the structural response of the bridge in the social activity cycle; According to the multi-time scale measurement task set, performing UAV distribution to determine the flight window and the corresponding target joint set of each UAV under each time scale to obtain a detection task; wherein, For each detection task, according to its measurement time window, image acquisition resolution requirement and flight height requirement, calculating the shortest path planning from the starting point to the target joint, considering the time window overlap of multiple tasks, and performing path compression and path splicing between tasks; Based on the detection task, controlling each UAV to collect the measurement data of the target joint laser in the corresponding flight window to execute the joint size extraction of the corresponding joint, and updating the joint size to the size life cycle curve of the corresponding joint; wherein, Updating the joint size to the size life cycle curve of the corresponding joint includes: time axis standardization of the measurement results under different time scales, respectively assigning data with different measurement frequency and time length characteristics to the corresponding size life cycle curve according to the joint number, and performing cross-scale difference analysis and trend alignment processing based on the overlapping segments between time scales to realize the synchronization of monitoring data of each time scale. 2.The method of claim 1, wherein, Each task in the multi-time scale measurement task set includes: Joint number, joint type, time scale label, task trigger condition, image acquisition resolution requirement and repetition tolerance; wherein, The time scale label is used to calibrate the task execution priority; The task trigger condition is used to define an automatic scheduling strategy; The image acquisition resolution requirement is used to guide the flight height and camera focal length of the unmanned aerial vehicle configuration adaptation. 3.The method of claim 1, wherein, The unmanned aerial vehicle allocation is performed according to the multi-time scale measurement task set, including: A corresponding scheduling parameter is determined for each task in the multi-time scale measurement task set; The reachability score and the task load balancing score of each unmanned aerial vehicle in the corresponding time window of the task are respectively calculated based on the remaining power, flight capability, historical task load condition and image resolution adaptation capability of each unmanned aerial vehicle; Under the premise of meeting the task triggering condition, the scheduling score calculation is performed by comprehensively considering the reachability score, the task load balancing score and the task priority corresponding to the time scale label, the unmanned aerial vehicle with the highest task scheduling score is determined, and the binding relationship between the corresponding unmanned aerial vehicle and the corresponding task is constructed; wherein, Each task is only bound to one unmanned aerial vehicle. 4.The method of claim 3, wherein, The flight window and the corresponding target seam set of each unmanned aerial vehicle at each time scale are determined, and the detection task is obtained, including: Based on the binding relationship between each task and the unmanned aerial vehicle formed by the highest scheduling score, all task items bound to each unmanned aerial vehicle are counted; The seam number is extracted from each task item and is summarized as the target seam set of the unmanned aerial vehicle; According to the measurement time window of the bound task, the flight height and the heading stability time required by the image acquisition resolution requirement are combined to calculate the flight start time, the flight end time and the path planning result of the unmanned aerial vehicle at each time scale, and the flight window is formed; The target seam set and the flight window are paired and stored as the detection task of the unmanned aerial vehicle. 5.The method of claim 1, wherein, Based on the detection task, the measurement data of the laser collected by each unmanned aerial vehicle in the corresponding flight window is controlled to perform the seam size extraction of the corresponding seam, including: The unmanned aerial vehicle is controlled to stably hover at the seam position in the flight window to shoot the seam area image at a preset pitch angle; Edge recognition is performed based on the seam boundary line extraction algorithm in the seam area image to perform target seam positioning; According to the target seam positioning result, laser measurement is performed on the target seam to obtain the seam size information of the target seam. 6.The method of claim 1, wherein, The size life cycle curve of the seam is indexed by the seam number, and records the time-varying sequence of the measured seam size of the seam at all time scales; wherein, The curve node of the size life cycle curve is composed of a timestamp and a corresponding size value.

7. A system for bridge inspection using a swarm of unmanned aerial vehicles (UAVs), characterized in that, The system is used to perform the method for bridge detection by using a group of unmanned aerial vehicles according to any one of claims 1-6, and the system includes: An acquisition unit is configured to acquire initialization acquisition data of the group of unmanned aerial vehicles, and perform seam type labeling and spatial position partitioning of a target bridge based on the initialization acquisition data to form a seam matrix for measurement task generation; A processing unit is configured to define a measurement time window and parameter configuration of each seam in the seam matrix based on each target time scale to generate a multi-time scale measurement task set; A task generation unit is configured to perform unmanned aerial vehicle allocation according to the multi-time scale measurement task set, determine a flight window and a corresponding target seam set of each unmanned aerial vehicle at each time scale, and obtain a detection task. The detection unit is used for controlling each unmanned aerial vehicle to collect the measurement data of the laser of the target seam section in the corresponding flight window based on the detection task, so as to perform the seam size extraction of the corresponding seam section, and update the seam size to the size life curve of the corresponding seam section.

Images