Bridge operation and maintenance collaborative scheduling system based on cloud platform

Abstract

The invention relates to the technical field of bridge operation and maintenance, in particular to a bridge operation and maintenance collaborative scheduling system based on a cloud platform, which comprises a node data collection module, a state feature analysis module, a rhythm scheduling generation module, an operation execution feedback module, a global scheduling optimization module and a monitoring sensor deployed based on a bridge structure, the uploaded data flow is analyzed through the cloud platform, and sensor load changes are distinguished node by node. According to the method, a node operation time sequence-oriented collection mechanism is adopted, collected bridge state data are continuously mapped to a cloud platform, real-time linkage of node states and resource changes is realized, and the real-time linkage of the node states and the resource changes is realized by synchronously judging node load trends, resource pressure and queue dynamics, generating targeted scheduling decisions and combining node operation to execute whole-process feedback. A node trend analysis result is dynamically applied to scheduling sequence optimization, balanced adjustment of task flows among multiple nodes is achieved, and stable promotion of collaborative operation and maintenance in a complex bridge scene is guaranteed.

Description

A cloud-based bridge operation and maintenance collaborative scheduling system Technical Field

[0001] This invention relates to the field of bridge operation and maintenance technology, and in particular to a cloud-based bridge operation and maintenance collaborative scheduling system. Background Technology

[0002] Bridge operation and maintenance (O&M) involves the daily upkeep, inspection and evaluation, defect diagnosis, reinforcement and repair, and O&M management of bridge structures. It is a crucial support for ensuring the safe operation of bridges and extending their service life. Traditional bridge O&M collaborative scheduling systems are information systems used to achieve unified management of bridge O&M tasks, coordinated resource allocation, and task execution. They are primarily used to address issues such as fragmented O&M tasks, low scheduling efficiency, and untimely response during bridge maintenance.

[0003] The existing system is slow to respond to real-time changes in node status, relies on periodic data aggregation, and the node resource status is not synchronized with the actual load. Task allocation is mostly in a fixed pattern, the scheduling between nodes is not tightly connected, nodes are both idle and overloaded, task execution status is difficult to report and affect subsequent decisions, scheduling strategies lack adaptive adjustment capabilities, bridge group management often suffers from resource configuration deviations and node collaboration delays, the operation and maintenance response chain is broken, resulting in fluctuations in operation and maintenance efficiency, and the overall system and scenario adaptability are limited. Summary of the Invention

[0004] To address the technical problems existing in the prior art, this invention provides a cloud-based bridge operation and maintenance collaborative scheduling system. The technical solution is as follows: On the one hand, a cloud-based bridge operation and maintenance collaborative scheduling system is provided, including: a node data collection module that, based on monitoring sensors deployed on the bridge structure, determines load changes, monitors current fluctuations, records task execution cycle sorting, integrates node communication delays, and uploads them to the cloud platform to obtain a bridge node runtime sequence set; a state feature analysis module that determines the load change trend of each bridge node in the bridge node runtime sequence set, combines waiting time fluctuations, compares state changes to determine resource pressure, integrates schedulable features, and obtains resource residual comparison data; a rhythm scheduling generation module that, based on the resource residual comparison data, adjusts the scheduling order of the work arms according to node trends, adjusts the rhythm if the load is released, and postpones the allocation if the pressure increases, and integrates node numbers and rhythm parameters to obtain task injection timing instructions; a work execution feedback module that, based on the task injection timing instructions, allocates inspection robots through the cloud platform, starts the work maintenance cycle, records work load and period information, and creates feedback data to obtain bridge work feedback results; and a global scheduling optimization module that, based on the bridge work feedback results, analyzes queue fluctuations and state trends, identifies nodes with consistent performance, compares communication response speeds to determine the optimal node, updates the scheduling order, and obtains the bridge task scheduling configuration.

[0005] On the other hand, the bridge node runtime sequence set includes collection time identifier, data synchronization status and node validity; the resource residual comparison data includes remaining scheduling capacity, task waiting period and resource adaptability index; the task injection timing instructions include instruction issuance sequence, node priority parameters and rhythm adjustment parameters; the bridge operation feedback results include operation completion identifier, load change characteristics and operation time period record; and the bridge task scheduling configuration includes node scheduling identifier, group sorting parameters and UAV response order.

[0006] On the other hand, the node data collection module includes: a data stream receiving submodule, based on monitoring sensors deployed on the bridge structure, analyzes the uploaded mechanical state, electrical parameters, structural deformation, and temperature and humidity data through a cloud platform, determines the data synchronization status and the completeness of node identification, classifies the collected data according to the node order, identifies valid data uploaded through the communication link, and obtains a node synchronization data sequence set; a current fluctuation monitoring submodule, based on the node synchronization data sequence set, compares the current sampling data of each node, arranges the current information according to time by node number, determines the trend and stage of current change during continuous monitoring of each node, divides the working state during operation, and obtains a node current change sequence group; a task cycle sorting submodule, based on the node current change sequence group, calculates the reflected node task start and end characteristics, combines the data upload time order of each node, analyzes the state distribution and start and end order of each node during operation, adjusts the node task arrangement data, and obtains a bridge node running sequence set.

[0007] On the other hand, the state feature parsing module includes: a load trend analysis submodule, which analyzes the periodic load monitoring data of each node based on the bridge node runtime sequence set, identifies the load increase / decrease trend in combination with the sampling period sequence, determines the direction of load change for each node in each period, classifies and statistically analyzes the segments of continuous load increase and decrease, and obtains the node load change trend vector; a state sequence comparison submodule, which compares the node load change trend vector with the node task queue waiting time sequence, analyzes the change in waiting time of nodes in different periods, determines the relationship between load trend and queue fluctuation, identifies nodes with changing running states based on period index, and obtains the node running state comparison results; and a schedulable feature extraction submodule, which, based on the node running state comparison results, determines the distribution of continuous non-task states of nodes, analyzes the joint performance of queue fluctuation amplitude and state occupancy ratio of each node, screens nodes with scheduling response potential, and obtains resource residual comparison data.

[0008] On the other hand, the rhythm scheduling generation module includes: a queue state parsing submodule, which, based on the resource residual comparison data and the periodic change sequence comparison of the node task waiting status, determines whether the node is currently in a load release or pressure increase phase, filters nodes whose load changes are consistent with the queue fluctuation direction, optimizes the node order and classifies the node status to obtain a node trend joint feature group; an arm scheduling adjustment submodule, based on the node trend joint feature group, determines the scheduling priority of each node, analyzes the node scheduling order under the load release state, adjusts the task allocation order if the node is in a pressure increase trend, organizes the scheduling relationship with the node number, merges the scheduling parameters of each node, and obtains a node priority sorting index; and a task rhythm injection submodule, based on the node priority sorting index, adjusts the scheduling rhythm parameters of each node, constructs a scheduling instruction chain, forms a scheduling data stream with the node number and allocation order, summarizes the node task allocation information and rhythm control elements, outputs the corresponding instruction set, and obtains the task injection timing instruction.

[0009] On the other hand, the operation execution feedback module includes: a robot scheduling and execution submodule that, based on the task injection timing instructions, sends operation start instructions to the inspection robot through the cloud platform, optimizes the task allocation order under the operation maintenance process, monitors the robot's state changes at each stage of task execution, identifies abnormal state nodes, and obtains the operation maintenance process trajectory set; an operation information recording submodule that, based on the operation maintenance process trajectory set, records the load changes and task continuity status of the robot during operation, analyzes the timing of the operation start and end nodes, compares the distribution characteristics of the operation load at each stage, integrates operation information, and obtains operation stage load distribution data; and a feedback data upload submodule that, based on the operation stage load distribution data, determines the correlation of each task data, analyzes the correspondence between operation time, nodes, and load information, optimizes the data transmission process of the on-site communication interface, merges the data packets of the entire operation process, and verifies the index parameters to obtain the bridge operation feedback result.

[0010] On the other hand, the global scheduling optimization module includes: a node status clustering submodule, which analyzes the fluctuation of the node queue and the periodic status trend based on the bridge operation feedback results, judges the change characteristics of the feedback data of each node within the same period, identifies the node combination with the same resource fluctuation trend direction, and obtains the node behavior synchronous distribution data; a response performance calculation submodule, which compares the communication response performance of the inspection UAVs to which each node belongs based on the node behavior synchronous distribution data, analyzes the communication delay distribution and command reception stability during the node feedback process, judges the response order of each node, and obtains the response node index information; and a scheduling order update submodule, which adjusts the resource occupation and communication performance of the nodes based on the response node index information, analyzes the scheduling priority parameters, optimizes the task allocation and sorting of the bridge group, and obtains the bridge task scheduling configuration.

[0011] On the other hand, the load change refers to the dynamic changes in the mechanical load, stress, strain, etc., of each sensor-monitored node over time, reflecting the stress state of the bridge, while the current fluctuation refers to the trend of the electrical current used by the maintenance equipment during operation over time.

[0012] On the other hand, the waiting time fluctuation refers to the fluctuation of the waiting time from when a task enters the queue to when it is actually executed, as the running status changes during the task scheduling and execution process. The resource pressure refers to the degree to which the available resources of a node are occupied by the current task. High pressure indicates that resources are in short supply, while low pressure indicates that resources are idle.

[0013] On the other hand, the node trend refers to the direction and rate of change of node load or resource occupancy over time, and the delayed allocation refers to the postponement of task allocation time when node pressure increases or resources are scarce.

[0014] The beneficial effects of the technical solution provided by the embodiments of the present invention include at least the following: by adopting a node-oriented runtime sequence collection mechanism, the collected bridge status data is continuously mapped to the cloud platform to achieve real-time linkage between node status and resource changes; by synchronously judging node load trends, resource pressure and queue dynamics, targeted scheduling decisions are generated; and by combining the feedback of the entire node job execution process, the node trend analysis results are dynamically applied to the scheduling order optimization to achieve balanced adjustment of task flow among multiple nodes, improve the adaptability of resource allocation and the overall scheduling continuity of the system, and ensure the stable progress of collaborative operation and maintenance in complex bridge scenarios. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 is a schematic diagram of the system of the present invention; Figure 2 is a schematic diagram of the system framework of the present invention; Figure 3 is a flowchart of the node data collection module of the present invention; Figure 4 is a flowchart of the state feature parsing module of the present invention; Figure 5 is a flowchart of the rhythm scheduling generation module of the present invention; Figure 6 is a flowchart of the job execution feedback module of the present invention; Figure 7 is a flowchart of the global scheduling optimization module of the present invention. Detailed Implementation

[0017] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0018] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0019] This invention provides a cloud-based bridge operation and maintenance collaborative scheduling system, as shown in Figure 1. The system includes: a node data collection module that, based on monitoring sensors deployed on the bridge structure, analyzes the uploaded data stream through the cloud platform, identifies sensor load changes node by node, continuously monitors current fluctuations, records the order of task execution cycles for each node, integrates node communication delays, and uploads the data via network interface according to the collection order to obtain a bridge node runtime sequence set; a state feature analysis module that determines the load change trend of each bridge node in the bridge node runtime sequence set, combines the fluctuation of task queue waiting time, compares the changes in node status within each cycle, determines resource pressure and idle level, and integrates the schedulable features corresponding to node queue fluctuations to obtain resource residual comparison data; and a rhythm scheduling generation module that, based on the resource residual comparison data, analyzes the load and queue status of each node one by one, and dynamically schedules according to node trends. The scheduling order of the electrically driven booms is adjusted. If the node load is released, the task allocation rhythm is adjusted; if the pressure increases, the allocation is postponed. The scheduling node number and task rhythm parameters are then integrated to obtain the task injection timing command. Based on the task injection timing command, the operation execution feedback module allocates inspection robots through the cloud platform, starts the operation maintenance cycle according to the control parameters, and records the operation load and period information in real time after the maintenance operation is completed. The data is uploaded to the cloud platform through the field communication interface to create feedback data for the entire bridge operation process and obtain the bridge operation feedback result. Based on the bridge operation feedback result, the global scheduling optimization module analyzes the queue fluctuation and status trend of node feedback within the same cycle. For resource fluctuations, nodes with consistent feedback performance are identified. The communication response speed of the inspection drone cluster is compared, and the optimal response node is determined as the priority allocation object. The bridge group scheduling order is updated to obtain the bridge task scheduling configuration.

[0020] The bridge node runtime sequence set includes the collection time identifier, data synchronization status, and node validity; the resource residual comparison data includes the remaining scheduling capacity, task waiting period, and resource adaptability indicators; the task injection timing instructions include the instruction issuance sequence, node priority parameters, and rhythm adjustment parameters; the bridge operation feedback results include the operation completion identifier, load change characteristics, and operation time period records; and the bridge task scheduling configuration includes the node scheduling identifier, group sorting parameters, and UAV response order.

[0021] In the node data collection module, monitoring sensors refer to various data acquisition devices installed on the bridge structure, used to monitor the bridge's mechanical state, electrical parameters, structural deformation, temperature and humidity in real time (such as strain gauges, accelerometers, current sensors, temperature and humidity sensors, etc.); data stream refers to the continuous raw or pre-processed digital signals generated by the monitoring sensors and uploaded to the cloud platform in real time via wireless or wired means, including periodic sampling data from various sensors; sensor load change refers to the dynamic changes in mechanical load, stress, strain, etc., on the monitored nodes (such as piers, main beams, etc.) over time, reflecting the stress state of the bridge; current fluctuation refers to the trend of the current consumption of maintenance equipment (such as inspection robots, monitoring terminals, etc.) changing over time, used to assess the equipment's operating status and energy consumption; node communication delay refers to the time delay that occurs during data transmission between each bridge monitoring node and the cloud platform server, usually affected by factors such as network bandwidth and link quality.

[0022] In the status feature analysis module, each bridge node refers to a physical location unit on the bridge structure that has independent monitoring and control capabilities, such as pier nodes, main beam nodes, and cable tower nodes. Each node can be equipped with independent sensing and communication equipment. The fluctuation of waiting time refers to the fluctuation of the waiting time from when a task enters the queue to when it is actually executed, which varies with the change of the operating status during task scheduling and execution, reflecting the scheduling and response efficiency. The change of node status refers to the time change process of key parameters (such as load, stress, current, and operating status) monitored by the node in a continuous period, reflecting the dynamic changes in node operation and task completion. Resource pressure refers to the degree to which the available resources (such as computing power, bandwidth, and power) of the node are occupied by the current task. High pressure indicates that resources are in short supply, while low pressure indicates that resources are idle. Idleness refers to the proportion or duration of a node in a non-task state, used to determine whether the node has the ability to undertake new tasks within a certain period of time. Scheduleable characteristics refer to comprehensive characteristic parameters that reflect whether a node is suitable for being prioritized for new maintenance tasks, which are combined with dynamic factors such as node idleness, load, and historical completion status for comprehensive judgment.

[0023] In the rhythm scheduling generation module, node load refers to the equipment or system load status caused by the current tasks undertaken by a single bridge node, usually reflected by indicators such as processor utilization and equipment operating rate; node trend refers to the direction and rate of change of node load or resource utilization over time, such as continuous increase, decrease, or fluctuation, used to determine the subsequent scheduling direction of the node; electric drive operation arm refers to the electric robotic arm equipped at the bridge maintenance site, used to automatically complete inspection, maintenance and other operations, with real-time monitoring and remote scheduling capabilities for position, action, and working condition; task allocation rhythm refers to the time interval and order in which the cloud platform assigns maintenance tasks to each node, a fast rhythm indicates dense allocation, and a slow rhythm indicates a longer interval; delayed allocation means that when the node pressure increases or resources are scarce, the task allocation time point is postponed to avoid node overload, and the allocation rhythm is automatically adjusted to a more relaxed state.

[0024] In the job execution feedback module, control parameters refer to the specific task scheduling parameters received by the robot from the cloud platform, which typically include the operation object, task type, execution start and end cycle, job priority, etc., to guide specific job actions; job maintenance cycle refers to the time period during which a single robot completes a maintenance task, from receiving the instruction to the completion of the task, including the entire process of movement, operation, and data transmission; job load and period information refer to the operating load (such as power consumption and computing pressure) generated by the robot during the operation and the time sequence information such as the start and end time and duration of task execution, for subsequent data archiving and analysis; communication interface refers to the data exchange channel between the robot or node terminal and the cloud platform, which may be wireless (4G / 5G, WiFi) or wired (Ethernet), to realize real-time communication of task issuance and feedback information.

[0025] In the global scheduling optimization module, resource fluctuation refers to the dynamic changes in available resources and task pressure of nodes within the bridge group over a period of time. Large resource fluctuations indicate uneven distribution of maintenance pressure, while small fluctuations indicate stable resource allocation. Nodes with consistent performance refer to multiple nodes whose key characteristics such as resource pressure, idle time, and queue status show similar or identical trends in feedback information, facilitating unified scheduling optimization. Communication response speed refers to the response time required for each node or inspection drone to receive instructions from the cloud platform or report task feedback, used to evaluate communication link quality and priority ranking. Optimal response node refers to the bridge node or inspection terminal most suitable for undertaking priority task allocation, determined by comparing communication response speed and current task completion status. Bridge group scheduling order refers to the overall task execution and allocation order optimized for all nodes in the current bridge group based on resource status, response speed, etc.

[0026] As shown in Figures 2 and 3, the node data collection module includes: a data stream receiving submodule based on monitoring sensors deployed on the bridge structure, which analyzes uploaded mechanical state, electrical parameters, structural deformation, and temperature and humidity data through a cloud platform to determine data synchronization and node identification completeness. The module categorizes the collected data according to node sequence, identifies valid data uploaded via communication links, and obtains a node synchronization data sequence set. Data is collected through strain gauges, accelerometers, current sensors, and temperature and humidity sensors deployed at different structural locations on the bridge. For example, strain gauges collect the main beam strain value once per second, accelerometers collect the lateral vibration value of the cable tower every 2 seconds, current sensors record the current fluctuations of the operating equipment in real time and upload data once per second, and temperature and humidity sensors upload temperature and humidity environmental status information once per minute. The data is uploaded to the cloud platform via wired or wireless links. In the cloud, each data entry includes the sampling time and node number. Timestamps are compared; for example, any 10 nodes are selected... For each node's data set, upload times within the same sampling period are extracted. If the maximum time difference is less than 0.5 seconds, data synchronization is considered successful; otherwise, it is marked as asynchronous. The node number for each data entry is then compared. If the number is not in the node configuration table, the data is marked as an invalid node. After number verification and time synchronization, the platform sorts all verified data by node number, and then arranges multiple data entries from the same node in ascending order of upload time, forming the node's data stream. The number of data entries uploaded by each node per unit time is then counted. For example, if a node uploads 50 data entries within 60 seconds, its upload rate is 50 entries per minute. A stable communication threshold of 30 entries per minute is set. If a node's upload rate is below 30 entries per minute for two consecutive minutes, or if the data upload interval exceeds two minutes, the node is identified as an unstable node, and its uploaded data is removed. The remaining data constitutes a well-structured, numbered, and time-ordered node synchronization data sequence set.

[0027] The current fluctuation monitoring submodule compares the current sampling data of each node based on the node synchronization data sequence set. It arranges the current information by time according to the node number, determines the trend and stage of current change during continuous monitoring, and classifies the working status during operation to obtain node current change sequence groups. The current data uploaded by each node is arranged in chronological order. For example, node 1 collects 600 data points from 09:00 to 09:10, processing them in windows of 10 data points. First, the average current and fluctuation range within each window are calculated. Then, the current value of each subsequent sampling point is compared with the average value of the previous window. If the current value of three consecutive sampling points increases by more than 0.2 amperes compared to the previous average, this data segment is considered a current rise phase. If the current remains stable above 1.5 amperes for more than 30 seconds, this time period is marked. During the operation phase, if the current continuously decreases and stabilizes below 0.5 amps, the task is considered complete. The criteria are as follows: the standby current is stable between 0.1 and 0.5 amps with fluctuations not exceeding 0.05 amps; the current rises above 0.2 amps for three consecutive times during the startup phase; the current remains between 1.3 and 2.0 amps for at least 30 seconds during the operation phase; and the current shows a clear downward trend and remains below 0.5 amps for more than 20 seconds after the drop during the termination phase. For example, if the current at node 2 rises from 0.3 amps to 1.6 amps at 09:02, remains at 1.6 amps at 09:04, begins to decrease at 09:05, and drops to 0.3 amps and remains stable at 09:06, this period is divided into three phases: startup, execution, and termination. All nodes are processed accordingly, generating their own current change phase sequences, forming a node current change sequence group.

[0028] The task cycle sorting submodule calculates the node task start and end characteristics based on the node current change sequence group. Combining this with the data upload time sequence of each node, it analyzes the state distribution and start / end sequence of each node during operation, adjusts the node task arrangement data, and obtains the bridge node runtime sequence set. It reads the task start and end time points marked for each node in the current change sequence. For example, if node 3 is marked as starting at 09:10 and ending at 09:20, the task cycle is 10 minutes. The task cycle information of this node is summarized along with the data of its nodes. For example, the task cycle of node 1 is from 08:55 to 09:05, and that of node 2 is from 09:00 to 09:15. The module then sorts the task start times of all nodes, first… The tasks are arranged in chronological order of start time as Node 1, Node 2, and Node 3. Then, the task durations are compared: Node 1 lasts 10 minutes, Node 2 lasts 15 minutes, and Node 3 lasts 10 minutes. If there are similar start times in the sorting, such as Node 4 and Node 5 both starting at 09:10, their task durations are compared. If they are the same, they are sorted in ascending order according to their node numbers. All nodes are arranged in the order of priority: task start time, task duration, and node number, to obtain the complete node execution sequence, for example, Node 1, Node 2, Node 4, Node 5, and Node 3. The task time window of each node is marked one by one, forming the bridge node execution sequence set.

[0029] As shown in Figures 2 and 4, the state feature analysis module includes: a load trend analysis submodule, which analyzes the periodic load monitoring data of each node based on the bridge node runtime sequence set, identifies the load increase / decrease trend by combining the sampling period sequence, determines the direction of load change for each node in each period, classifies and statistically analyzes the segments of continuous load increase and continuous decrease, and obtains the node load change trend vector; the load values ​​in each period are arranged in the sampling order, and the direction of the difference between sampling time points is used as the basis for judging the load trend. By statistically analyzing the load increment between every two adjacent sampling time points, it is determined whether the load change of the node in the corresponding period is increasing or decreasing. For example, in one period of a certain bridge pier node N5, the sampling values ​​are 400, 410, 428, 432, and 435 Newtons respectively. The load difference is calculated as +10, +18, +4, and +3 Newtons respectively, indicating that the period is a continuous upward trend of load. If the sampling values ​​in the next period become 435, 422, 410, 398, and 39... If the load difference is 0 Newtons, then the difference is -13, -12, -12, or -8 Newtons, indicating a continuous downward trend. The same judgment process is performed for all cycles of each node. If a node's load continues to rise for three consecutive cycles and the sum of the differences in a single cycle is greater than 50 Newtons, then the node is marked as a continuous growth segment in that stage. If a node's load continues to fall for two cycles and the sum of the absolute values ​​of the differences is greater than 40 Newtons, then it is marked as a continuous decline segment. To avoid misjudgment, the load change judgment threshold is set to an average difference of no less than 5 Newtons per cycle and a continuous judgment segment of no less than two cycles. All nodes are classified by number and their index positions for growth, decline, and unchanged segments are statistically analyzed, and a load change trend vector arranged in chronological order is generated. For example, node N5 is in an upward segment from 08:00 to 08:10 and from 08:10 to 08:20, and in a downward segment from 08:20 to 08:30, recorded as "++−". Similar trend vectors are generated for the other nodes.

[0030] The state sequence comparison submodule compares the node load change trend vector with the node task queue waiting time sequence, analyzes the changes in waiting time of nodes in different cycles, determines the relationship between load trends and queue fluctuations, and identifies nodes with changing operating states based on cycle indices to obtain node operating state comparison results. First, it extracts the average waiting time of each node in each cycle. For example, the waiting times of node N3 in four consecutive cycles are 20 seconds, 25 seconds, 28 seconds, and 23 seconds, respectively. These are listed as a state sequence in cycle order and then analyzed in parallel with the load change trend vector. For example, if the trend corresponding to this node is "++−−", the direction of waiting time change is calculated cycle by cycle as "++−". Then, it determines whether the change directions of the two sets of sequences are consistent, i.e., whether the waiting time increases when the load increases and decreases when the load decreases. If the trend directions are consistent... If there is a correlation, it is marked as positive; if the correlation is negative, it is marked as negative; if there is no change, it is marked as uncorrelated. This correlation determination is embedded in the period index to identify which periods each node experiences abnormal increases or decreases in waiting time or trend reversals. For example, if node N3's waiting time increases in periods 2 and 3 but the load trend changes from rising to falling, then period 3 is marked as a state change node. The same operation is performed on all nodes. For the portion of waiting time change in each period that exceeds 10% of the previous period, a weighted judgment is made. If the previous period was 20 seconds and the current period exceeds 22 seconds, it is judged as an increasing period. The portion of fluctuation change less than 5% is marked as a stable state. If a node in the node queue has a waiting time increase of more than 15% for two consecutive periods and the load trend changes to falling, it is marked as a sudden change point in the running state. The time index of all changing nodes will be recorded to form the node running state comparison results.

[0031] The schedulable feature extraction submodule, based on the node running status comparison results, determines the distribution of consecutive non-task states of nodes, analyzes the joint performance of queue fluctuation amplitude and state occupancy ratio of each node, and filters nodes with scheduling response potential, obtaining resource residual comparison data. Based on the state change node index recorded in the node running status comparison results, it statistically analyzes the time periods of each node in the non-task state, extracting the number of consecutive non-task cycles and the proportion of non-task states in the total cycles. For example, if node N7 has no task scheduling in 6 of the last 10 cycles, including 3 consecutive idle cycles, it first determines whether the node has more than 2 consecutive idle cycles; if so, it marks the node as potentially schedulable. Then, it analyzes its task queue... The system checks for fluctuations within idle periods. For example, if the number of queued tasks drops from 5 to 2 within a period, it is considered high fluctuation. If the drop is less than 1 task, it is considered low fluctuation. The fluctuation threshold is set to 2 tasks. If the number of high fluctuation periods accounts for more than 50% of the idle periods and the proportion of non-task states is greater than 40% of the total periods, the node is marked as a schedulable response node. In addition, the system summarizes the distribution of the running status of each node and compares the proportion of the time period in the "idle" state with the "occupied" state. If the total idle time exceeds 60 minutes and the number of tasks is 0 for 3 consecutive periods, the node is assigned the resource available label. Finally, all nodes that meet the criteria of high idle ratio, large task fluctuation, and many consecutive periods without tasks will be selected as schedulable candidates, and resource residual comparison data will be obtained.

[0032] As shown in Figures 2 and 5, the rhythm scheduling generation module includes: a queue state analysis submodule that, based on resource residual comparison data and combined with the periodic change sequence comparison of node task waiting status, determines whether a node is currently in a load release or pressure increase phase, filters nodes whose load changes are consistent with the direction of queue fluctuations, optimizes the node sequence and classifies the node status, and obtains a joint feature group of node trends; maps the correspondence between indicators and the historical periodic arrangement order of nodes, expands nodes sequentially according to period number, extracts the queue task length and change trend of each node in each period, and then analyzes it in correspondence with node load data to determine whether the task queue length shows an increasing or decreasing trend, and determines whether the node is currently in a load release or pressure increase phase. For example, if a node's task queue length in the last four periods is 6, 5, 3, and 2 respectively, and its load value decreases from 560 Newtons to 480 Newtons, the task queue length decreases in each period. If a node's load decreases by more than 15 Newtons, it is marked as being in a load release state. Conversely, if another node's queue length increases from 3 to 6 queues while its load increases from 430 Newtons to 490 Newtons, that node is judged as being in a pressure increase state. Based on this, the load change direction of each node is judged to be consistent with the queue fluctuation direction. That is, if the load increases and the queue tasks increase, it is considered a same-direction increase; if the load decreases and the queue shrinks, it is considered a same-direction decrease. Only when both trends are in the same direction are the node selected as a node with consistent trend direction. The selected nodes are sorted by period index number and grouped into three types: same-direction increase, same-direction decrease, and inconsistent. A multi-dimensional status identifier is constructed by combining data such as the number of task completion cycles, the proportion of non-task states, and resource adaptability scores of the node in the scheduling. Nodes in the same state are grouped into a unified category to facilitate subsequent scheduling parameter organization and form a joint feature group of node trends.

[0033] The robotic arm scheduling adjustment submodule determines the scheduling priority of each node based on node trend joint feature groups, analyzes the node scheduling order under load release status, and adjusts the task allocation order if a node is in a pressure increase trend. It also organizes scheduling relationships by combining node numbers, merges scheduling parameters for each node, and obtains a node priority ranking index. The module extracts the status labels of all nodes, adds nodes in the load release state to the priority scheduling list, and sorts them according to their idle time percentage and resource adaptability score. For example, nodes with an idle time percentage greater than 60% and a resource adaptability score higher than 70 are prioritized. Conversely, for nodes in a pressure increase state, their current ranking is adjusted backward by one or more cycles to avoid scheduling pressure accumulation. During the ranking adjustment, if a node number is small and the scheduling frequency is high... If a node has low resource availability but high resource adaptability, it is moved up one position in the priority ranking. The ranking priority score weights are set as follows: resource score 50%, idle ratio 30%, and number order 20%. Each dimension score is set to a maximum of 100. The node score is the sum of the scores of each dimension multiplied by their weights. For example, node N4 has a resource score of 80, an idle ratio score of 70, and a number score of 90. Its weighted score is calculated as 80×0.5+70×0.3+90×0.2=79 points. After sorting the nodes by score from high to low, the scheduling priority is rearranged. The number, status category, scheduling order, and score are all summarized. The original scheduling cycle corresponding to the node with increased pressure is postponed by 1-2 cycles to release resource load. The scheduling parameters of all nodes are summarized according to the updated priority to obtain the node priority ranking index.

[0034] The task rhythm injection submodule, based on the node priority sorting index, adjusts the scheduling rhythm parameters of each node, constructs a scheduling instruction chain, combines node number and allocation order to form a scheduling data stream, summarizes node task allocation information and rhythm control elements, outputs the corresponding instruction set, and obtains the task injection timing instructions; the scheduling rhythm parameters of each node are adjusted using the formula: The system calculates node scheduling rhythm adjustment parameters, constructs a scheduling instruction chain, combines node numbers and allocation order to form a scheduling data stream, summarizes node task allocation information and rhythm control elements, and obtains task injection timing instructions. Representing the The first one within the scheduling cycle Adjust the node scheduling rhythm parameter values. Representing the The first one within the scheduling cycle Node task cadence load parameter values, Representing the The first one within the scheduling cycle Reference balanced values ​​for node task rhythm load parameters. Representing the The first one within the scheduling cycle The rhythm correction parameter value corresponding to the node's priority ranking weight. Representing the The first one within the scheduling cycle The current task allocation sequence parameter of the node. Represents the first [number] in the previous scheduling cycle The node's historical task allocation sequence parameter; the scheduling rhythm adjustment parameter value reflects how much the task scheduling rhythm (i.e., the frequency and timing of new task allocation) of a certain node needs to be adjusted within the current scheduling cycle; the larger this parameter value, the faster or more maintenance tasks will be allocated to the node under the combined effect of factors such as the current task load status and sorting weight; the smaller the parameter value, the slower or less the rhythm of task allocation for the node will be.

[0035] Get the The first one within the scheduling cycle The raw task cadence load data of the node is derived from the ratio between the actual task size processed by the node and the node's available processing capacity within the scheduling cycle. For example, the node... In the cycle The cumulative number of tasks received and entered into the execution queue is 180 task units. The maximum available processing capacity of a node within the same period is 300 task units, resulting in an original load ratio of 180 / 300. This original ratio is then processed using an interval proportional normalization method to obtain the normalized task rhythm load parameter. Meanwhile, historical scheduling data for the same node under the same resource type and scheduling strategy were collected. The original load ratios for the most recent five scheduling cycles were selected as 150 / 300, 165 / 300, 180 / 300, 195 / 300, and 210 / 300. After normalizing these original load ratios using the same interval ratio normalization method, the normalized results were 0.5, 0.55, 0.6, 0.65, and 0.7, respectively. Subsequently, an arithmetic mean was calculated for this normalized sequence. The calculation process is as follows: Thus, the first The first one within the scheduling cycle Reference balancing value for node task cadence load parameters Subsequently and The process of performing the difference and absolute value operation is as follows: ; as the numerical result of the task rhythm offset of the current node; then introduce the first The first one within the scheduling cycle The rhythm correction parameter corresponding to the node's priority weight The original value of this parameter comes from the node's priority score in the global task allocation sequence, assuming the node... The original priority score in the scheduling system is 75 points, with a score range of 0–100 points. After linear normalization of this original score, a normalized priority weight value of 0.75 is obtained. Combined with the preset rhythm mapping coefficient range of 0–1 in the scheduling system, the mapping calculation process is as follows: Thus, the normalized rhythm correction parameters are obtained. Then read the first The first one within the scheduling cycle The current task allocation sequence parameter of the node and the historical task allocation sequence parameters within the previous scheduling cycle. The original value of the ordinal parameter is an integer sorting position, such as the node in the current cycle. The original sequence position is the 4th position, and the original sequence position in the previous period is the 1st position. After processing this type of sequence position data using the maximum sequence position value normalization method, the following results are obtained: , The difference and absolute value operation is performed on the two, and the calculation process is as follows: After all the above parameters have been standardized in terms of dimensions, they are substituted into the scheduling rhythm adjustment formula in sequence. First, the numerator is calculated: Then calculate the denominator: Finally, perform the division operation: This numerical result serves as the first... The first one within the scheduling cycle The node's scheduling rhythm adjustment parameter value is written into the scheduling instruction chain and sequentially concatenated with the node number and the current allocation order to form a corresponding scheduling data stream segment. Subsequently, this scheduling data stream is summarized with the node task allocation information and rhythm control elements to obtain timing instruction data for subsequent task injection.

[0036] The scheduling strategy response level is divided into three segments, where: if This is considered a period of slight fluctuation in scheduling rhythm, corresponding to a relatively stable task rhythm at the corresponding node, and the system instruction module maintains the original scheduling rhythm; if This period is considered a minor adjustment range in the scheduling rhythm. While the task allocation order of the corresponding nodes is somewhat disturbed, no significant load shift occurs. Within this range, the system will scale the node scheduling interval to a limited extent, and the adjustment range of the rhythm control submodule will remain within a controllable range. If the task suddenly increases or its priority changes significantly, it falls under the category of a sudden change in scheduling rhythm. This corresponds to a situation where the number of tasks suddenly increases or its priority changes significantly. In this case, the node needs to be stressed and the task sequence needs to be reallocated.

[0037] The result satisfy Therefore, it belongs to the period of slight adjustment in scheduling rhythm, and this result indicates that the node Within the current scheduling cycle, a limited acceleration adjustment to the task rhythm is needed. Based on this parameter, the original injection interval is adjusted from 10ms to: The above values ​​indicate that the system increases the injection frequency of the current node by approximately 5.77%, without entering a drastic control mechanism, and continues to operate under a gradual adjustment strategy.

[0038] As shown in Figures 2 and 6, the job execution feedback module includes: a robot scheduling and execution submodule that, based on task-injected timing instructions, sends job start instructions to the inspection robot through the cloud platform, optimizes the task allocation sequence under the job maintenance process, monitors the robot's state changes at each stage of task execution, identifies abnormal state nodes, and obtains the job maintenance process trajectory set; it matches each job task with the corresponding inspection robot number according to the time sequence, and sends a job start instruction to the designated robot through the cloud platform instruction interface. The instruction includes the job type, target node, expected execution cycle, and scheduling number. After receiving the instruction, the robot enters a standby state, monitors its navigation path from the starting position to the target node and records the time, and uploads its status information to the cloud platform in real time after starting the job. In the cloud, indicators such as current location, job status (e.g., preparing, working, abnormal, completed), device power consumption, motor current, and execution progress are recorded. For example, in a task, robot R2 receives instructions at 09:00, arrives at bridge node N5 at 09:03, starts working at 09:04 and continues until 09:11. The entire job status is recorded once per minute, with the statuses being preparing, working, working, and completed. If a node records an abnormal status at two consecutive time points during the job and the device current is higher than 5 amps, it is marked as an abnormal status node. Throughout the entire job execution process, the start and end times of the task, the stage status, and the location of the abnormal point are recorded in real time, and a complete job process log chain is constructed according to the task number to obtain the job maintenance process trajectory set.

[0039] The job information recording submodule, based on the job maintenance process trajectory set, records the load changes and task continuity status of the robot during operation, analyzes the timing of the start and end nodes of the job, compares the distribution characteristics of the job load in each stage, integrates job information, and obtains job stage load distribution data. It also analyzes the entire process information of each job task, extracts the robot's load change curve during operation, and reflects the dynamic fluctuations of the job load through the current value and internal transmission device power value every minute. The task is divided into a start-up stage, an execution stage, and an end-of-job stage. The duration and average load value of each stage are statistically analyzed. For example, if a task lasts 12 minutes at node N3, the start-up stage... The average current over 2 minutes is 1.2 amps, the average current over 8 minutes during the execution phase is 2.4 amps, and the average current over 2 minutes during the final phase is 1.0 amps. The data is organized into a task load record table according to the phases. The timing relationship between the start and end nodes is determined. The task boundary is constructed using the nodes with the earliest start time and the latest end time. In the load change curves between nodes, the distribution of peak and valley points is identified. If the load fluctuation in a certain phase is greater than 30%, it is marked as a high fluctuation segment; otherwise, it is a stable segment if it is less than 10%. The load fluctuation types of the entire task process are classified and summarized into a job load distribution record according to the phases. Combined with the task timeline and robot job number for unified coding, the job phase load distribution data is constructed.

[0040] The feedback data upload submodule, based on the load distribution data of the operation phase, determines the correlation of each task data, analyzes the correspondence between operation time, nodes, and load information, optimizes the data transmission process of the on-site communication interface, merges data packets of the entire operation process, and verifies index parameters to obtain bridge operation feedback results. It compares the time information, node identifiers, and load records in each task data item by item, reads the task start and end times, compares the corresponding node numbers, extracts the load record curve of that node within the corresponding time period, and matches it with the operation status identifier recorded in the robot status information. If the time period is aligned and the status sequence conforms to the normal task flow order, the task is marked as data consistent. If there are problems such as misaligned time sequences or state jumps, it is marked as suspected abnormal data. Correlation statistics are performed on all tasks, and each task must correspond to a node. The system requires a point number, a load change curve, a start time, and an end time, with a complete status sequence. If any element is missing, it is marked as a missing record. During the data upload phase, based on the bandwidth and latency of the on-site communication interface, the batch size for data upload is set to a maximum of 50 data entries per batch, with each data entry not exceeding 1MB. If the robot uploads data too frequently and the communication interface bandwidth is less than 2Mbps, the upload frequency will be delayed to once per minute, and data fragments will be cached locally until communication is restored before being uploaded in a unified manner. During the data aggregation process, the task number is used as the primary key to associate the operation information, load data, and status records into a complete data package. At the same time, an index check is performed on each field, including the field order, the number of missing fields, and the content type matching. After the check is correct, the data is aggregated into the bridge operation feedback record database, and the bridge operation feedback results are output.

[0041] As shown in Figures 2 and 7, the global scheduling optimization module includes: a node status clustering submodule, which analyzes node queue fluctuations and periodic status trends based on bridge operation feedback results, determines the change characteristics of each node's feedback data within the same period, identifies node combinations with the same resource fluctuation trend direction, and obtains synchronous distribution data of node behavior; it plots the task queue data of each node into a discrete change sequence according to the periodic order, and combines the load change direction and task completion indicator of the node within the same period to classify the status into three types: load increase - task pending execution, load decrease - task completed, and load stable - idle; after constructing the periodic status trend sequence, it performs a horizontal comparison of all nodes according to the period to identify node groups with consistent status trends within the same period, for example, in the period numbered 15, nodes N1 and N3... If nodes N4, N7, and N8 both exhibit a continuous increase in load, with two more task queues added and the task status as "not executed," they are classified into the same trend group. If the resource usage of the three nodes in this group increases by more than 30% and the trends are the same for two consecutive periods, they are merged into a resource fluctuation synchronization node group. The fluctuation trend judgment threshold is set as follows: if the resource usage changes by more than 20% in adjacent periods, the queue direction is consistent, and the number of queues changes by no less than two, then the trend is considered consistent. If more than three nodes meet this condition, they are merged into a synchronization behavior group. In an actual test, in period 20, nodes N4, N7, and N8 all entered an idle state after task execution, the task queue length decreased by more than three queues, and the load decreased by more than 25%. They were identified as synchronization release type nodes and assigned the synchronization distribution label "CR," generating node behavior synchronization distribution data.

[0042] The response performance calculation submodule compares the communication response performance of the inspection drones belonging to each node based on the node behavior synchronization distribution data, analyzes the communication delay distribution and command reception stability during the node feedback process, determines the response order of each node, and obtains the response node index information; it extracts the bound inspection drone number and calls the communication response log of the drone during the task execution period to record the command issuance time, reception time, and feedback time difference, and calculates the single communication response latency. For example, if drone U1 bound to node N3 receives the task command at 10:03:05 in period 15 and the first response time is 10:03:08, the response latency is 3 seconds. It accumulates all response latency data for each node and calculates the average. If the average communication response latency of a node exceeds 5 seconds, or there are more than 3 records of single latency exceeding 10 seconds, it is marked as an unstable communication response node. The stability threshold is set to a single latency not exceeding 10 seconds. Nodes with a latency of 8 seconds and a fluctuation range of no more than ±3 seconds are classified as stable nodes, and those with a latency exceeding this range are classified as fluctuating nodes. The response data of all nodes is then sorted, with the node with the shortest latency ranked first and those with longer latency or greater fluctuation ranked last. Furthermore, each node is examined for data packet loss or duplicate instruction reception during the task reception phase. If there are more than two instances of duplicate transmissions or more than three instances of missing data, the node is marked as an unstable instruction node. Based on a combination of latency data and instruction stability, each node is assigned a communication response score, with a maximum score of 100, where latency accounts for 60 points and stability accounts for 40 points. Nodes with scores greater than 80 are considered priority nodes, those between 60 and 80 are considered medium priority nodes, and those less than 60 are considered low priority nodes. For example, node N2 has an average latency of 3 seconds and no duplicate transmissions, so its score is 92. Node N6 has a latency of 8 seconds and two instances of duplicate instruction reception, so its score is 65. The response node index information is then formed by sorting all node scores.

[0043] The scheduling order update submodule, based on the response node index information, adjusts the resource usage and communication performance of nodes, analyzes scheduling priority parameters, optimizes the task allocation and sorting of the bridge group, and obtains the bridge task scheduling configuration. It also statistically analyzes the number of tasks, cumulative operation duration, and resource usage frequency of each node over the past 10 periods, defining nodes with an operation frequency higher than 5 times or an average resource utilization rate exceeding 80% as high-utilization nodes, and nodes with an operation frequency less than 3 times and resource utilization less than 40% as low-utilization nodes. Combined with node communication scores, it adjusts the current scheduling priority. If a node... If the score is higher than 85 and the current occupancy rate is low, the node is prioritized and moved up 2 positions. If the score is lower than 65 and the occupancy rate is higher than 70%, the node is moved down 2 positions. The nodes with higher scores are given priority in the next inspection task. During the sorting process, the score thresholds are set as follows: 90 points and above is priority level A, 75 to 89 is level B, 60 to 74 is level C, and below 60 is level D. The node allocation strategy is sorted and updated by combining the sorting level, node number and current job load balancing factor. The output includes the bridge task scheduling configuration containing the node number, priority level, scheduling cycle order and the expected number of task instructions.

[0044] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of protection of the described technical solutions.

Claims

1. A cloud-based bridge operation and maintenance collaborative scheduling system, characterized in that, The system includes: a node data collection module that uses monitoring sensors deployed on the bridge structure to identify load changes, monitor current fluctuations, record task execution cycle order, integrate node communication delays, and upload them to the cloud platform to obtain a bridge node runtime sequence set; a state feature analysis module that judges the load change trend of each bridge node in the bridge node runtime sequence set, combines waiting time fluctuations, compares state changes to determine resource pressure, integrates schedulable features, and obtains resource residual comparison data; a rhythm scheduling generation module that adjusts the scheduling order of work arms based on the resource residual comparison data and node trends, adjusting the rhythm if the load is released and delaying allocation if the pressure increases, and integrates node numbers and rhythm parameters to obtain task injection timing instructions; a work execution feedback module that allocates inspection robots through the cloud platform based on the task injection timing instructions, starts work maintenance cycles, records work load and period information, and creates feedback data to obtain bridge work feedback results; and a global scheduling optimization module that analyzes queue fluctuations and state trends based on the bridge work feedback results, identifies nodes with consistent performance, compares communication response speeds to determine the optimal node, updates the scheduling order, and obtains the bridge task scheduling configuration.

2. The bridge operation and maintenance collaborative scheduling system based on a cloud platform according to claim 1, characterized in that, The bridge node runtime sequence set includes collection time identifier, data synchronization status, and node validity. The resource residual comparison data includes remaining scheduling capacity, task waiting period, and resource adaptability index. The task injection timing instructions include instruction issuance sequence, node priority parameters, and rhythm adjustment parameters. The bridge operation feedback results include operation completion identifier, load change characteristics, and operation time period records. The bridge task scheduling configuration includes node scheduling identifier, group sorting parameters, and UAV response order.

3. The bridge operation and maintenance collaborative scheduling system based on a cloud platform according to claim 1, characterized in that, The node data collection module includes: a data stream receiving submodule, which analyzes the uploaded mechanical state, electrical parameters, structural deformation, and temperature and humidity data through a cloud platform based on monitoring sensors deployed on the bridge structure, determines the data synchronization status and the completeness of node identification, classifies the collected data according to the node order, identifies valid data uploaded through the communication link, and obtains a node synchronization data sequence set; a current fluctuation monitoring submodule, based on the node synchronization data sequence set, compares the current sampling data of each node, arranges the current information according to time by node number, determines the trend and stage of current change during continuous monitoring of each node, classifies the working state during operation, and obtains a node current change sequence group; and a task cycle sorting submodule, based on the node current change sequence group, calculates the reflected node task start and end characteristics, combines the data upload time order of each node, analyzes the state distribution and start and end order of each node during operation, adjusts the node task arrangement data, and obtains a bridge node running sequence set.

4. The bridge operation and maintenance collaborative scheduling system based on a cloud platform according to claim 1, characterized in that, The state feature parsing module includes: a load trend analysis submodule, which analyzes the periodic load monitoring data of each node based on the bridge node runtime sequence set, identifies the load increase / decrease trend by combining the sampling period sequence, determines the direction of load change for each node in each period, classifies and statistically analyzes the segments of continuous load increase and decrease, and obtains the node load change trend vector; a state sequence comparison submodule, which compares the node load change trend vector with the node task queue waiting time sequence, analyzes the changes in waiting time of nodes in different periods, determines the relationship between load trend and queue fluctuation, identifies nodes with changing running states based on period index, and obtains the node running state comparison results; and a schedulable feature extraction submodule, which, based on the node running state comparison results, determines the distribution of continuous non-task states of nodes, analyzes the joint performance of queue fluctuation amplitude and state occupancy ratio of each node, screens nodes with scheduling response potential, and obtains resource residual comparison data.

5. The bridge operation and maintenance collaborative scheduling system based on a cloud platform according to claim 1, characterized in that, The rhythm scheduling generation module includes: a queue state parsing submodule, which, based on the resource residual comparison data and the periodic change sequence comparison of node task waiting status, determines whether a node is currently in a load release or pressure increase phase, filters nodes whose load changes are consistent with the queue fluctuation direction, optimizes the node order, and classifies the node status to obtain a node trend joint feature group; a work arm scheduling adjustment submodule, which, based on the node trend joint feature group, determines the scheduling priority of each node, analyzes the node scheduling order under the load release state, adjusts the task allocation order if a node is in a pressure increase trend, organizes the scheduling relationship based on the node number, merges the scheduling parameters of each node, and obtains a node priority sorting index; and a task rhythm injection submodule, which, based on the node priority sorting index, adjusts the scheduling rhythm parameters of each node, constructs a scheduling instruction chain, forms a scheduling data stream based on the node number and allocation order, summarizes the node task allocation information and rhythm control elements, outputs the corresponding instruction set, and obtains the task injection timing instruction.

6. The bridge operation and maintenance collaborative scheduling system based on a cloud platform according to claim 1, characterized in that, The operation execution feedback module includes: a robot scheduling and execution submodule, which, based on the task injection timing instructions, sends operation start instructions to the inspection robot through the cloud platform, optimizes the task allocation order under the operation maintenance process, monitors the robot's status changes at each stage of task execution, identifies abnormal status nodes, and obtains the operation maintenance process trajectory set; an operation information recording submodule, based on the operation maintenance process trajectory set, records the load changes and task continuity status of the robot during operation, analyzes the timing of operation start and end nodes, compares the distribution characteristics of operation load at each stage, integrates operation information, and obtains operation stage load distribution data; and a feedback data upload submodule, based on the operation stage load distribution data, determines the correlation of each task data, analyzes the correspondence between operation time, nodes, and load information, optimizes the data transmission process of the on-site communication interface, merges data packets of the entire operation process, verifies index parameters, and obtains bridge operation feedback results.

7. The bridge operation and maintenance collaborative scheduling system based on a cloud platform according to claim 1, characterized in that, The global scheduling optimization module includes: a node status clustering submodule, which analyzes the fluctuation of node queues and periodic status trends based on the bridge operation feedback results, determines the change characteristics of each node's feedback data within the same period, identifies node combinations with the same resource fluctuation trend direction, and obtains node behavior synchronous distribution data; a response performance calculation submodule, which compares the communication response performance of the inspection drones to which each node belongs based on the node behavior synchronous distribution data, analyzes the communication delay distribution and command reception stability during the node feedback process, determines the response order of each node, and obtains response node index information; and a scheduling order update submodule, which adjusts the resource occupancy and communication performance of nodes based on the response node index information, analyzes scheduling priority parameters, optimizes the task allocation and sorting of the bridge group, and obtains the bridge task scheduling configuration.

8. The bridge operation and maintenance collaborative scheduling system based on a cloud platform according to claim 1, characterized in that, The load change refers to the dynamic changes in mechanical load, stress, strain, etc., at each sensor-monitored node over time, reflecting the stress state of the bridge. The current fluctuation refers to the trend of the electrical current used by the maintenance equipment during operation over time.

9. The bridge operation and maintenance collaborative scheduling system based on a cloud platform according to claim 1, characterized in that, The waiting time fluctuation refers to the fluctuation in the waiting time from when a task enters the queue to when it is actually executed, which varies with the running status during task scheduling and execution. The resource pressure refers to the degree to which the available resources of a node are occupied by the current task. High pressure indicates that resources are in short supply, while low pressure indicates that resources are idle.

10. The bridge operation and maintenance collaborative scheduling system based on a cloud platform according to claim 1, characterized in that, The node trend refers to the direction and rate of change in node load or resource usage over time, and the delayed allocation refers to the postponement of task allocation time when node pressure increases or resources are scarce.

Images