Intelligent water conservancy monitoring system and method based on air and underwater cooperation

Abstract

The invention discloses an intelligent water conservancy monitoring system and method based on air and underwater cooperation, and relates to the technical field of water conservancy projects, and functional modules of the system comprise an air platform, an underwater platform and a command center; comprising the following steps of: modeling a dynamic task demand based on multi-dimensional environment collaborative awareness; an air-horizontal platform task allocation mechanism based on collaborative path-load optimization; a multi-source data space-time coding and platform embedded sensing guide sampling mechanism; edge node multi-source observation collaborative fusion and local anomaly recognition; carrying out risk trend prediction and regional priority reconstruction based on a dynamic evolution network; and generating an intelligent response strategy based on risk dynamic reconstruction. According to the method, a cooperative working mechanism of an air platform and an underwater platform is constructed, and technologies of multi-source data fusion, a multi-platform linkage response strategy and the like are combined, so that systematic enhancement of a water conservancy project area in multiple dimensions of full space coverage, high time frequency, strong risk identification, response schedulability and the like is realized.

Description

Technical Field

[0001] The present invention relates to the field of water conservancy engineering technology, and specifically to an intelligent water conservancy monitoring system and method based on aerial and underwater collaboration. Background Art

[0002] With the rapid development of smart water conservancy and digital twin technologies, the demand for real-time, accurate, and global monitoring of water conservancy projects and the environment is becoming increasingly urgent. Traditional monitoring methods, which primarily rely on fixed sensor networks and manual inspections, suffer from issues such as insufficient data collection coverage, low update frequency, and delayed response times, making them inadequate for addressing the challenges of sudden flood disasters, project safety risks, and ecological and environmental changes. Simultaneously, the rapid maturation of drone and underwater robot technologies has provided a new means for monitoring. Drones utilize high-precision sensors such as multispectral cameras and lidar to conduct full-time, dynamic inspections of the water surface and surrounding environment. Underwater robots, on the other hand, employ acoustic and ultrasonic imaging technologies to accurately collect data on water quality, flow velocity, and structural deformation in complex underwater environments. By seamlessly integrating aerial and underwater platforms, an "air-water" collaborative monitoring system can be established, enabling real-time collection, rapid integration, and intelligent early warning of multi-source data. This provides scientific and timely data support for smart water conservancy safety management, disaster warning, and ecological and environmental protection. Summary of the Invention

[0003] In response to the shortcomings of the existing technology, the present invention provides a smart water conservancy monitoring system and method based on aerial and underwater collaboration to solve the problems raised in the above background technology.

[0004] To achieve the above object, the present invention provides the following technical solutions: In a first aspect, an embodiment of the present invention provides a smart water conservancy monitoring method based on aerial and underwater collaboration, comprising the following steps: S1. Based on the environmental parameters collected by the aerial and underwater platforms, a task requirement function is constructed that integrates the multi-dimensional environmental perception information of the aerial and underwater platforms. S2. Using the task demand function as input, a collaborative optimization scheduling model for joint execution of tasks by air and underwater platforms is constructed, and the execution task set of each platform is output; S3, based on the output execution task set, determines the multi-source data sampling method of each platform at the specific task point and outputs the encoded data set; S4. Using the encoded data set output by each platform, perform edge data fusion and anomaly detection on the data of each task point and construct a fusion responsiveness function; S5. Using the output of the fusion responsiveness function as input, a regional priority model is constructed to predict potential risk trends over several future cycles. S6. Based on the output of the risk trend prediction model, a multi-level response model is constructed according to the task completion status and the remaining platform capabilities, and a scheduling instruction set is output.

[0005] To further optimize this technical solution, in step S1, the task requirement function is set represents the monitoring priority level of the i-th area, and the task demand function is modeled as:

[0006] in, : The environmental complexity score of the area obtained by the aerial platform; : underwater structure risk index; : Regional communication connectivity score; : weight of historical disaster records; : Weight factor of the current monitoring period; : Weight parameter, satisfying .

[0007] To further optimize this technical solution, in step S2, in the collaborative optimization scheduling model, the aerial platform set , underwater platform collection , the task of each region i is performed by the subset platform; For each platform , whose execution task set is , the collaborative optimization scheduling model is constructed as follows:

[0008] in, : The task demand function of the i-th region, input by step S1; :platform residual mission capability; :platform The path cost of executing task i; : Platform type penalty factor; This model allocates high-priority tasks to platforms with low path costs and high task capabilities as much as possible, ensuring a balance between the rationality of task scheduling and energy efficiency.

[0009] To further optimize this technical solution, in step S3, a space-time-feature joint compression mechanism is introduced to guide each platform to autonomously adjust the data sampling granularity based on the spatiotemporal characteristics of specific task points and the changing trends of key sensor indicators, and set the following model: Set task points ,platform The data compression coding ratio sampled at this task point is ,but:

[0010] in, : The task demand function of the i-th region, input by step S1; : the temporal change rate of the sensor observation value at the task point i of the platform; : Actual communication delay between the platform and the command center, in ms; : The reference communication delay between the system-preset platform and the command center, in ms; : System control constant, which determines the global compression strength; Output encoded dataset Contains data compression coding ratio and summary of original observation data , used for edge data fusion and anomaly detection in subsequent steps.

[0011] To further optimize this technical solution, in step S4, the edge computing node is deployed near the aerial platform mother machine or the shore-based receiving station. The node receives the data summary uploaded by the aerial and underwater platforms and constructs a fusion responsiveness function for each region i:

[0012] in, : fusion feature response value; : The set of platforms that perform tasks at task point i; : importance coefficient of data provided by platform j at task point i; : The data encoding compression ratio calculated in step S3; : A weighted indicator of platform data packet arrival delay and transmission jitter; Used to express the data response intensity and potential abnormal probability of region i in the current cycle. Once a region Above the set threshold , it is marked as a "local high-risk candidate area" and enters the risk trend prediction step in the subsequent step.

[0013] Further optimize this technical solution, in step S5, in the regional priority model, a dynamic regional risk propagation network is constructed , where the node set Represents the task area, edge set Represents the risk transmission relationship between regions at different times; For each region i:

[0014] in, : The predicted risk priority of region i at time t+1; : The fusion feature response value of region i at the current time t in step S4; : the set of adjacent regions of region i; : risk diffusion weight from region k to i; : Control the weight coefficient of the current anomaly and the neighborhood transmission, satisfying ; The model propagates the abnormal trend of high-risk areas to adjacent areas, constructs the risk trend forecast in the future multiple periods, and outputs Used for dynamic sorting of global monitoring areas.

[0015] Further optimize this technical solution, in step S6, according to the output of the regional priority model Real-time update results and platform status, design a multi-level response model; Each platform Corresponding instructions It includes specific contents such as rearrangement of inspection points, increase of data frequency, and temporary return. At the same time, the output scheduling instruction set will be fed back to each platform and command center.

[0016] To further optimize this technical solution, the multi-level response model is as follows:

[0017] in, :platform Reachable area subset; : The node set representing the task area in step S5; : The current load status indicator of the platform; : The path cost for the platform to reach area i.

[0018] Further optimization of this technical solution, during continuous operation, including specific river sections, dam bottoms, and old culverts, may trigger multiple times in different time periods and , introduced an embedded memory enhancement revisit mechanism. By introducing an abnormal memory index table in the data scheduling layer of the command center, all areas marked as high-risk are archived and annotated at the event level, and the following information is recorded: Abnormal event timestamp and level; Associate the platform data upload time and platform number; Continuous before the abnormality occurs Change curve; Whether there are deviations or errors in the platform equipment status; Whether the current anomaly is a repeat of the previous anomaly area.

[0019] A smart water conservancy monitoring system based on aerial and underwater collaboration is constructed based on the above-mentioned smart water conservancy monitoring method. The functional modules of the system include: an aerial platform, an underwater platform, and a command center; The aerial platform consists of a multi-rotor UAV or a fixed-wing UAV, equipped with a lidar, high-precision GPS, infrared thermal imager, multispectral imaging equipment, and video acquisition terminal, and integrates an intelligent path planning module and a comprehensive environmental perception module. Through the onboard intelligent controller, it can realize autonomous planning and obstacle avoidance of flight paths, data collection targets, and flight dynamics. The underwater platform is mainly composed of an adaptive underwater detection robot, deployable underwater sensor nodes and acoustic communication modules. It integrates a multi-dimensional environmental monitoring module, including a water quality probe, a flow meter, and a structural deformation sensor, and is equipped with an inertial navigation system and an acoustic positioning system to achieve precise positioning. The command center is the central control unit of the system, consisting of edge computing servers, data fusion and processing modules, risk assessment and strategy generation engines, scheduling consoles, and three-dimensional visualization platforms; the command center is equipped with standardized data interfaces to support seamless connection with aerial and underwater platforms.

[0020] In a second aspect, an embodiment of the present invention provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program instructions are executed by the processor, the steps of a smart water conservancy monitoring system and method based on aerial and underwater collaboration as described in the first aspect of the present invention are implemented.

[0021] In a third aspect, an embodiment of the present invention provides a computer-readable storage medium on which a computer program is stored, wherein: when the computer program instructions are executed by a processor, the steps of a smart water conservancy monitoring system and method based on aerial and underwater collaboration as described in the first aspect of the present invention are implemented.

[0022] Compared with the existing technology, the present invention provides a smart water conservancy monitoring system and method based on aerial and underwater collaboration, involving machine learning and deep learning technologies, and has the following beneficial effects: This intelligent water conservancy monitoring system and method, based on aerial and underwater collaboration, achieves systematic enhancements to water conservancy project areas in multiple dimensions, including full spatial coverage, high temporal frequency, strong risk identification, and schedulable responses, by building a collaborative working mechanism between aerial and underwater platforms and combining multi-source data fusion, collaborative optimization scheduling, adaptive risk analysis, and multi-platform linkage response strategies. This system overcomes the limitations of traditional water conservancy monitoring methods and achieves a full-process technological upgrade from multi-source perception to intelligent decision-making, and from static deployment to dynamic collaboration. It boasts the combined advantages of high monitoring accuracy, fast response, accurate risk identification, and flexible deployment, significantly improving the intelligent operation and maintenance level and disaster prevention and control capabilities of water conservancy facilities, and has broad engineering application value and promotion prospects. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0024] Figure 1 This is a flow chart of a smart water conservancy monitoring method based on aerial and underwater collaboration proposed by the present invention; Figure 2 This is a schematic diagram of the composition of an intelligent water conservancy monitoring system based on aerial and underwater collaboration proposed by the present invention. DETAILED DESCRIPTION

[0025] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0026] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0027] Secondly, the term "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in various places throughout this specification does not necessarily refer to the same embodiment, nor does it designate a separate or selective embodiment that is mutually exclusive with other embodiments.

[0028] Example 1: Reference Figure 1 , which is the first embodiment of the present invention, provides a smart water conservancy monitoring method based on aerial and underwater collaboration, including the following steps: S1. Based on the environmental parameters collected by the aerial and underwater platforms, a task requirement function is constructed that integrates the multi-dimensional environmental perception information of the aerial and underwater platforms. To achieve efficient water conservancy monitoring and dispatch, it is necessary to first build a task requirement model that integrates multi-dimensional environmental perception information from the air and underwater. This is based on environmental parameters collected by aerial platforms (lidar, infrared, optical, etc.) and underwater platforms (acoustic flow meters, water quality monitoring, structural inspection, etc.).

[0029] Traditional water conservancy monitoring and dispatching strategies are mostly based on single dimensions or static indicators (such as regular inspection plans and historical fault frequencies), lacking comprehensive consideration of real-time environmental conditions, spatial coordination factors, and historical disaster potential, resulting in delayed dispatching responses and uneven distribution of monitoring resources. The multi-dimensional environment collaborative perception modeling mechanism proposed in this step integrates heterogeneous perception data from aerial and underwater platforms for the first time, and uses a unified task requirement function to achieve the desired effect. Realize dynamic quantitative expression of monitoring priority.

[0030] Setting up task requirement functions represents the monitoring priority level (grade or degree) of the i-th area, and the task demand function is modeled as:

[0031] in, Environmental complexity score (0-1) obtained by the aerial platform, taking into account factors such as vegetation obstruction and lighting conditions. The aerial platform (lidar, infrared thermal imager, and multispectral camera) collects real-time surface image data during patrol. Image analysis algorithms identify vegetation coverage, terrain relief, and lighting conditions. A deep learning classification model is then used to assign a score to the environmental complexity, ranging from 0 (simple) to 1 (extremely complex). This score reflects the degree of sensory interference encountered by the aerial platform during mission execution, such as dense vegetation obstruction and strong reflective glare. A higher score indicates a more challenging area to monitor.

[0032] : Underwater structure risk index. Real-time data is obtained by the underwater platform using acoustic Doppler flowmeters, structural deformation monitoring sensors, water quality sensors and other equipment. By calculating the structural anomaly amplitude, water flow disturbance intensity and pollution index volatility, and then normalizing them, a risk score is output. It is used to determine whether underwater projects (such as dams and culverts) have structural hazards such as leakage, cracks or scour, and is an important basis for task scheduling. Even if an area has a high underwater structure risk index (such as severe cracks), it may be located in an open area, making the aerial environment less complex, and vice versa. Therefore, the model does not directly couple the two values, but instead weights and integrates them as independent dimensional items with independent physical meanings in the task requirement function.

[0033] :Regional communication connectivity score. Based on the communication link quality statistics of the aerial platform and the underwater platform during the execution of historical missions, combined with the current regional wireless signal strength (underwater acoustic communication strength, 4G / 5G coverage), the score is evaluated and the link stability and transmission success rate are recorded and updated in real time. This score measures the data transmission capability of the air-water platform in a certain area. Areas with poor communication conditions need to prioritize high-power or low-frequency communication nodes. Even if a region has a poor communication connectivity score (i.e., low ), and its airborne observation conditions may also be better (i.e., low ), and vice versa. Therefore, the model does not directly couple the two values, but instead integrates them as weighted dimensions with independent physical meanings in the task requirement function.

[0034] : Weight of historical disaster records. The command center calls the natural disaster records of the area in the historical database (such as floods, dam breaches, waterlogging, etc.), and weights them according to the frequency and level of the events. The system automatically matches historical events through geocoding. This parameter reflects the vulnerability of the area to disasters. High-weighted areas need to be included in monitoring tasks more frequently to identify potential risks in advance. Even if a region has a high frequency of historical disasters (i.e., high ), there may also be good airborne observation conditions (i.e. ), so the two are modeled and treated separately as independent risk and observation dimensions in the model.

[0035] : Weight factor for the current monitoring period. Automatically assigned by the system based on the current time (day / night), seasonal information (rainy / dry season), and meteorological data (rainfall warnings, typhoon information). This weight can also be adjusted by user policy settings. Used to express the time urgency of the task. For example, the weight should be significantly increased on the eve of a rainstorm or in a typhoon path forecast area to enable early resource deployment. High May occur during the day or in extreme weather conditions, and It can take high or low values in any time period. Therefore, the model does not directly couple the two values, but instead integrates them as independent physical dimensions in the task requirement function. The weight factor of the current monitoring period is The value range is usually set between 0.8 and 1.5. In stable or low-risk periods (such as sunny days, nights, and non-flood seasons), It can be set to 0.8-1.0. In high-risk periods such as severe convective weather, red alert for heavy rain, and typhoon landing, Improved to 1.2-1.5 to enhance the task requirement function Sensitivity to changes in urgency.

[0036] : Weight parameter, satisfying It can be determined by an optimization algorithm or expert system based on historical scheduling results, so that The output is more application-accurate and responsive.

[0037] In actual application, after completing the first collaborative perception of the air and underwater platforms, the system will automatically start the mission requirement modeling module, extract the above parameters and uniformly construct .

[0038] S2. Using the task demand function as input, a collaborative optimization scheduling model for joint execution of tasks by air and underwater platforms is constructed, and the execution task set of each platform is output; Traditional monitoring platform task allocation methods often rely on independent scheduling of a single platform or an average round-robin dispatch mechanism. These often fail to consider the synergy benefits of multiple platforms, platform differences (such as load variations, communication constraints, and power limits), path energy consumption, and task priority. This can lead to problems such as insufficient resources for high-priority tasks and duplicate execution of low-priority tasks. In contrast, the collaborative optimization scheduling model, for the first time, achieves unified modeling of collaborative task decomposition and energy-efficient scheduling across heterogeneous air and water platforms in a water conservancy monitoring system, ensuring that platform capabilities match task requirements while minimizing execution costs.

[0039] In the collaborative optimization scheduling model, set up the aerial platform collection , underwater platform collection , the task of each region i is performed by the subset platform; For each platform , whose execution task set is , the collaborative optimization scheduling model is constructed as follows:

[0040] in, : The task demand function of the i-th region, input by step S1.

[0041] :platform The system measures the platform's remaining mission capacity. This includes, but is not limited to, the current remaining battery percentage, available sensor payload capacity, and remaining bandwidth available for communication modules. The system aggregates these three resources using standardized weights to calculate the platform's current mission-assignable capacity. A higher value indicates a platform's suitability for continued missions.

[0042] :platform The path cost for executing mission i. For aerial platforms, a trajectory planning algorithm (such as A* or Dijkstra's algorithm) is used to calculate the shortest flight path from the current location to the i-th area, correcting for wind speed and obstacles. For underwater platforms, a terrain database and a water flow distribution model are used to calculate the optimal path distance, accounting for speed changes due to the influence of water flow. This path length (in meters or kilometers) is multiplied by a unit energy consumption weight and used as the cost input to measure the relative "cost" of executing the mission.

[0043] : Platform type penalty factor. This parameter is a static value preset by the system, reflecting the relative weight of resource scarcity of different platforms: For aerial platforms: set a higher value (such as ), indicating that its task allocation needs to be more cautious; for underwater platforms: set a lower value (such as ) to encourage the priority use of platforms with richer resources. This factor is used to weight path costs in the model, reflecting the system's preference for platform utilization in task allocation.

[0044] The optimizer (such as genetic algorithm or integer programming) calculates the optimal task assignment set with the goal of platform load balancing and minimizing the overall path cost. .

[0045] Each platform will eventually be assigned a set of mission areas , as a checklist for subsequent actual inspection or detection tasks, and continue to optimize the path.

[0046] This model allocates high-priority tasks to platforms with low path costs and high task capabilities as much as possible, ensuring a balance between the rationality of task scheduling and energy efficiency.

[0047] S3, based on the output execution task set, determines the multi-source data sampling method of each platform at the specific task point and outputs the encoded data set; Traditional water conservancy monitoring systems typically use a fixed cycle and compression rate during data collection, failing to adapt to the urgency of the task, the dynamic nature of the observed data, and the status of the communication link. This mechanism often results in over-compression of high-priority task data, leading to information loss, or redundant transmission of low-value task data, resulting in a waste of system resources.

[0048] Considering the energy and communication constraints of the platform, a space-time-feature joint compression mechanism is introduced to guide each platform to autonomously adjust the data sampling granularity based on the spatiotemporal characteristics of specific task points and the changing trends of key sensor indicators. The following model is set: Set task points ,platform The data compression coding ratio sampled at this task point is ,but:

[0049] in, : The task demand function of the i-th region, input by step S1; : The temporal change rate of the sensor observation value at the task point i (such as the difference between image frames, temperature fluctuation rate), which is obtained by calculating the change gradient of adjacent observation frames or continuous sensor data; : Actual communication delay between the platform and the command center, in ms; : The reference communication delay between the platform and the command center preset by the system, in milliseconds, for example, the average transmission delay allowed by the system (such as 100 ms); : System control constant (experience value or strategy training output) determines the global compression strength; The model ensures that in areas of abnormally rapid change ( In areas with poor communication links or stable task point changes, the compression ratio is automatically increased to save bandwidth and storage.

[0050] The model calculates the compression ratio corresponding to each task point i , and then guide the platform to adopt different data sampling periods and compression algorithm levels at this point; like , then the compression is smaller, the sampling frequency is increased, and more information is retained; like , the compression rate is improved, which is suitable for scenarios where data changes smoothly or transmission is limited.

[0051] Output encoded dataset Contains data compression coding ratio and summary of original observation data , used for edge data fusion and anomaly detection in subsequent steps.

[0052] S4. Using the encoded data set output by each platform, perform edge data fusion and anomaly detection on the data of each task point and construct a fusion responsiveness function; Existing water environment monitoring systems often use a centralized architecture, uploading observation data from multiple platforms to a remote server for centralized processing. This approach suffers from slow response times, bandwidth bottlenecks, and difficulties integrating heterogeneous data. This approach is particularly problematic when it comes to providing early warnings for sudden regional hydrological anomalies (such as short-term runoff surges and pollution leaks).

[0053] The edge computing node is deployed near the aerial platform mothership or shore-based receiving station. The node receives the data summary uploaded by the aerial and underwater platforms and constructs a fusion responsiveness function for each region i:

[0054] in, : fusion feature response value; : The set of platforms that perform tasks at task point i; : The importance coefficient of the data provided by platform j at task point i is determined by the type and weight of the sensors carried by the platform; : The data encoding compression ratio calculated in step S3 represents the current information granularity; : A weighted indicator of platform data packet arrival delay and transmission jitter (the higher the delay, the lower the trust level); Used to express the data response intensity and potential abnormal probability of region i in the current cycle. Once a region Above the set threshold (Usually based on statistical analysis of historical monitoring data, in the initial stage, a large number of normal and abnormal areas are collected If a sample is distributed, its mean and standard deviation are determined, and a threshold is set according to a set confidence level, such as 95% or 99%, it will be marked as a "local high-risk candidate area" and enter the risk trend prediction step in the subsequent step.

[0055] S5. Using the output of the fusion responsiveness function as input, a regional priority model is constructed to predict potential risk trends over several future cycles. Most current water environment monitoring systems rely on static rule matching or single-point prediction models based on time series for risk warnings, making it difficult to fully capture the spatiotemporal spread of risks across regions. For example, some methods only consider the changing trends of local historical outliers, while ignoring the spatial diffusion of risk events and inter-regional correlations. This results in delayed responses or significant prediction bias when addressing continuous risks (such as pollution spread and cascading hydrological anomalies).

[0056] Constructing a dynamic regional risk propagation network in the regional priority model , where the node set Represents the task area, edge set Represents the risk transmission relationship between regions at different times; For each region i:

[0057] in, : The predicted risk priority of region i at time t+1.

[0058] : The fused feature response value of region i at the current time t in step S4.

[0059] : The set of adjacent regions of region i.

[0060] : The risk diffusion weight from region k to i (usually set between 0 and 1) is calculated by combining the following three items: geographic proximity (based on topological distance or water flow connectivity); The co-occurrence probability of historical disasters (the frequency of simultaneous anomalies in two regions); Structural or functional similarity (such as water body type, vegetation cover, etc.).

[0061] : Control the weight coefficient of the current anomaly and the neighborhood transmission, satisfying .

[0062] The model propagates abnormal trends in high-risk areas to adjacent areas, and can be iterated multiple times as needed to construct risk trend forecasts for multiple periods in the future. Used for dynamic sorting of global monitoring areas.

[0063] S6. Based on the output of the risk trend prediction model, a multi-level response model is constructed according to the task completion status and the remaining platform capabilities, and a scheduling instruction set is output.

[0064] Traditional water environment monitoring response and scheduling mechanisms typically issue tasks based on preset rules or scheduled polling, lacking a coordinated assessment of risk severity and platform capabilities. This "static, one-way" instruction model often leads to wasted resources (e.g., platforms repeatedly operating in low-risk areas) or delayed responses (e.g., high-risk areas not receiving priority). Especially in multi-platform collaborative tasks, current technologies rarely consider the integrated assessment of platform capabilities, load status, and real-time path costs, making it difficult to achieve multi-objective optimal scheduling.

[0065] Output from the regional priority model Design a multi-level response model based on the real-time update results and platform status (such as task completion, remaining power, and communication status); Each platform Corresponding instructions It includes specific contents such as rearrangement of inspection points, increase of data frequency, and temporary return. At the same time, the output scheduling instruction set will be fed back to each platform and command center.

[0066] The multi-level response model is as follows:

[0067] in, :platform A subset of reachable areas.

[0068] : Indicator of the current platform load status (such as the number of tasks executed, remaining battery power, data cache ratio, etc.). The larger the value, the busier the platform.

[0069] : The path cost for the platform to reach area i can be calculated by the path planning algorithm given above (such as A* or Dijkstra).

[0070] For each platform ,The system first obtains its current status (such as whether it is running, the amount of tasks remaining, and the battery status); Construct its reachable area set , and based on the , the platform's own status , path time Calculate the above-mentioned benefit ratio; Evaluate all candidate task area combinations and find the optimal combination that maximizes the above objective function; Output instruction set and sends it to the platform execution layer and uploads it to the command center for manual intervention.

[0071] In this embodiment, during continuous operation, the area including a specific river section, the bottom of the dam, and the old culvert may be triggered multiple times in different time periods. and , introducing an embedded memory-enhanced revisit mechanism, embedded between steps S5 and S6, after risk trend prediction and before policy response. By introducing the Anomaly Memory Index Table (AMIT) in the command center's data scheduling layer, all areas marked as high-risk are archived and annotated at the event level, recording the following information: Abnormal event timestamp and level; Associate the platform data upload time and platform number; Continuous before the abnormality occurs Change curve; Whether there are deviations or errors in the platform equipment status; Whether the current anomaly is a repeat of the previous anomaly area.

[0072] Once an area is marked as "medium-high level abnormality" three times in a row in different cycles, AMIT will set a "priority revisit tag" for the area, even if If the high-risk threshold is not reached, some platform resources will be allocated as a suboptimal response target for low-frequency tracking and inspection until the abnormality level drops significantly or the system detection and stability maintenance exceeds the set time window.

[0073] The key advantages of this mechanism are: Breaking through the limitations of single-cycle response: By accumulating and strengthening historical abnormal trajectories, the system is equipped with "scenario memory capabilities"; Improve resource utilization accuracy: Ensure that platform resources are more effectively used in areas sensitive to structural evolution, avoiding blind spots that are ignored simply based on current thresholds. Enhance the foresight of early warning: intervene in advance when risks have not yet increased significantly, and improve predictive capabilities.

[0074] AMIT can reside in the data management layer of the command center as an independent module and interact with the historical database of the cloud platform processing layer; In step S6, the priority return label will affect the path cost function Parameters are used to increase the response probability of the marked area by lowering the cost factor.

[0075] Example 2: Reference Figure 2, which is the second embodiment of the present invention, provides a smart water conservancy monitoring system based on aerial and underwater collaboration, which is constructed based on the smart water conservancy monitoring method described in Example 1. The functional modules of the system include: an aerial platform, an underwater platform and a command center.

[0076] The aerial platform consists of a multi-rotor UAV or a fixed-wing UAV, equipped with a lidar, high-precision GPS, an infrared thermal imager, a multi-spectral imaging device and a video acquisition terminal, and integrates an intelligent path planning module and a comprehensive environmental perception module; through the onboard intelligent controller, it can realize autonomous planning and obstacle avoidance of flight paths, data collection targets and flight dynamics.

[0077] The underwater platform is mainly composed of an adaptive underwater detection robot, deployable underwater sensor nodes and an acoustic communication module; it integrates a multi-dimensional environmental monitoring module, including water quality probes, flow meters, structural deformation sensors, and is equipped with an inertial navigation system and an acoustic positioning system to achieve precise positioning.

[0078] The command center is the central control unit of the system, consisting of edge computing servers, data fusion and processing modules, risk assessment and strategy generation engines, scheduling consoles, and three-dimensional visualization platforms; the command center is equipped with standardized data interfaces to support seamless connection with aerial and underwater platforms.

[0079] In summary, the present system and method have significant advantages in the following aspects: Air-water collaboration improves monitoring coverage and integrity: The aerial platform has a large-scale, rapid-response inspection capability, while the underwater platform can penetrate deep into complex structures or key underwater sections for microscopic inspection. The collaboration of the two can achieve three-dimensional monitoring of water conservancy projects from upstream to downstream, and from the surface to the interior of the structure.

[0080] The risk factor identification accuracy based on joint feature fusion is higher: the system adopts the task requirement function and jointly models the aerial and underwater sensor data, which significantly improves the recognition accuracy of key indicators such as structural anomalies, water quality mutations, and flow rate fluctuations, avoiding misjudgments and omissions that may be caused by a single data dimension.

[0081] Spatial partitioning and dynamic task matching enhance system adaptability: By building a collaborative optimization scheduling model based on importance and task weight, dynamically dividing the detection area based on platform status, data density and response capability, and realizing intelligent allocation of tasks among different platforms, the problems of rigid task scheduling and uneven resource utilization in the existing system are solved.

[0082] Time synchronization and data consistency enhance dynamic monitoring capabilities: The method introduces a space-time-feature joint compression mechanism to ensure temporal consistency in high-frequency dynamic acquisition, providing a data basis for subsequent multi-temporal analysis and trend prediction, and significantly improving the continuity and accuracy of data processing.

[0083] The risk assessment and trend prediction mechanism under multi-factor coupling is more forward-looking: by constructing a fusion response function and regional priority model, the system can quantitatively assess the current risk status of the monitored area, predict future evolution trends, issue early warning information, and enhance the disaster prevention and control capabilities of water conservancy facilities.

[0084] Intelligent scheduling optimizes platform response paths and efficiency: A multi-level response model is built to achieve intelligent command issuance and global optimization of platform execution paths, significantly reducing response delays and resource waste caused by repeated flights / cruise.

[0085] It has extremely high system scalability and environmental adaptability: the system structure can be deployed modularly, supports platform addition and deletion, and scenario migration, and is suitable for a variety of typical water conservancy scenarios such as reservoirs, dams, rivers, culverts, and pumping stations; at the same time, the system's core algorithm has adaptive capabilities and can adjust monitoring strategies in real time according to environmental changes.

[0086] Example 3: This embodiment also provides a computer device, which is suitable for a smart water conservancy monitoring system and method based on aerial and underwater collaboration, including a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute computer-executable instructions to implement a smart water conservancy monitoring system and method based on aerial and underwater collaboration as proposed in the above embodiment.

[0087] This embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, an intelligent water conservancy monitoring system and method based on aerial and underwater collaboration as proposed in the above embodiment is implemented.

[0088] The computer device may be a terminal, comprising a processor, memory, a communication interface, a display, and an input device connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores an operating system and computer programs. The internal memory provides an environment for the operating system and computer programs stored in the non-volatile storage media. The communication interface of the computer device is used to communicate with external terminals via wired or wireless communication. Wireless communication may be achieved via Wi-Fi, a carrier network, NFC (near-field communication), or other technologies. The display of the computer device may be a liquid crystal display or an electronic ink display. The input device may be a touchscreen overlay on the display, buttons, a trackball, or a touchpad on the computer device housing, or an external keyboard, touchpad, or mouse.

[0089] If a function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the portion that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the various embodiments of the method of the present invention. The aforementioned storage medium includes various media that can store program code, such as USB flash drives, mobile hard drives, read-only memories (ROMs), random access memories (RAMs), magnetic disks, or optical disks.

[0090] The logic and / or steps represented in the flowcharts or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing the logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (e.g., a computer-based system, a system including a processor, or other system that can fetch and execute instructions from an instruction execution system, apparatus, or device). For purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transport a program for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0091] More specific examples (a non-exhaustive list) of computer-readable media include the following: an electrical connection with one or more wires (electronic devices), a portable computer disk cartridge (magnetic devices), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), a fiber optic device, and a portable compact disc read-only memory (CDROM). In addition, the computer-readable medium may even be paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium and then editing, interpreting, or processing it in another suitable manner as necessary, and then storing it in a computer memory.

[0092] It should be understood that various parts of the present invention can be implemented using hardware, software, firmware, or a combination thereof. In the above-described embodiments, multiple steps or methods can be implemented using software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented using hardware, as in another embodiment, any one of the following technologies known in the art or a combination thereof can be used: a discrete logic circuit having logic gate circuits for implementing logic functions on data signals, an application-specific integrated circuit having suitable combinational logic gate circuits, a programmable gate array (PGA), a field-programmable gate array (FPGA), etc.

[0093] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A smart water conservancy monitoring method based on aerial and underwater collaboration, characterized in that: The following steps are involved: S1. Based on the environmental parameters collected by the aerial and underwater platforms, a task requirement function is constructed that integrates the multi-dimensional environmental perception information of the aerial and underwater platforms. S2. Using the task demand function as input, a collaborative optimization scheduling model for joint execution of tasks by air and underwater platforms is constructed, and the execution task set of each platform is output; S3, based on the output execution task set, determines the multi-source data sampling method of each platform at the specific task point and outputs the encoded data set; S4. Using the encoded data set output by each platform, perform edge data fusion and anomaly detection on the data of each task point and construct a fusion responsiveness function; S5. Using the output of the fusion responsiveness function as input, a regional priority model is constructed to predict potential risk trends over several future cycles. S6. Based on the output of the risk trend prediction model, a multi-level response model is constructed according to the task completion status and the remaining platform capabilities, and a scheduling instruction set is output.

2. The intelligent water conservancy monitoring method based on aerial and underwater collaboration according to claim 1 is characterized in that: In step S1, the task requirement function is set represents the monitoring priority level of the i-th area, and the task demand function is modeled as: ； in, : The environmental complexity score of the area obtained by the aerial platform; : underwater structure risk index; : Regional communication connectivity score; : weight of historical disaster records; : Weight factor of the current monitoring period; : Weight parameter, satisfying .

3. The intelligent water conservancy monitoring method based on aerial and underwater collaboration according to claim 1 is characterized in that: In step S2, in the collaborative optimization scheduling model, the aerial platform set is set , underwater platform collection , the task of each region i is performed by the subset platform; For each platform , whose execution task set is , the collaborative optimization scheduling model is constructed as follows: ； in, : The task demand function of the i-th region, input by step S1; :platform residual mission capability; :platform The path cost of executing task i; : Platform type penalty factor; This model allocates high-priority tasks to platforms with low path costs and high task capabilities as much as possible, ensuring a balance between the rationality of task scheduling and energy efficiency.

4. The intelligent water conservancy monitoring method based on aerial and underwater collaboration according to claim 1 is characterized in that: In step S3, a space-time-feature joint compression mechanism is introduced to guide each platform to autonomously adjust the data sampling granularity according to the spatiotemporal characteristics of specific task points and the changing trends of key sensor indicators, and set the following model: Set task points ,platform The data compression coding ratio sampled at this task point is ,but: ； in, : The task demand function of the i-th region, input by step S1; : the temporal change rate of the sensor observation value at the task point i of the platform; : Actual communication delay between the platform and the command center, in ms; : The reference communication delay between the system-preset platform and the command center, in ms; : System control constant, which determines the global compression strength; Output encoded dataset Contains data compression coding ratio and summary of original observation data , used for edge data fusion and anomaly detection in subsequent steps.

5. The intelligent water conservancy monitoring method based on aerial and underwater collaboration according to claim 1 is characterized in that: In step S4, the edge computing node is deployed near the aerial platform mother machine or the shore-based receiving station. The node receives the data summary uploaded by the aerial and underwater platforms and constructs a fusion responsiveness function for each region i: ； in, : fusion feature response value; : The set of platforms that perform tasks at task point i; : importance coefficient of data provided by platform j at task point i; : The data encoding compression ratio calculated in step S3; : A weighted indicator of platform data packet arrival delay and transmission jitter; Used to express the data response intensity and potential abnormal probability of region i in the current cycle. Once a region Above the set threshold , it is marked as a "local high-risk candidate area" and enters the risk trend prediction step in the subsequent step.

6. The intelligent water conservancy monitoring method based on aerial and underwater collaboration according to claim 1 is characterized in that: In step S5, a dynamic regional risk propagation network is constructed in the regional priority model. , where the node set Represents the task area, edge set Represents the risk transmission relationship between regions at different times; For each region i: ； in, : The predicted risk priority of region i at time t+1; : The fusion feature response value of region i at the current time t in step S4; : the set of adjacent regions of region i; : risk diffusion weight from region k to i; : Control the weight coefficient of the current anomaly and the neighborhood transmission, satisfying ; The model propagates the abnormal trend of high-risk areas to adjacent areas, constructs the risk trend forecast in the future multiple periods, and outputs Used for dynamic sorting of global monitoring areas.

7. The intelligent water conservancy monitoring method based on aerial and underwater collaboration according to claim 1 is characterized in that: In step S6, the output of the regional priority model is Real-time update results and platform status, design a multi-level response model; Each platform Corresponding instructions It includes specific contents such as rearrangement of inspection points, increase of data frequency, and temporary return. At the same time, the output scheduling instruction set will be fed back to each platform and command center.

8. The intelligent water conservancy monitoring method based on aerial and underwater collaboration according to claim 7 is characterized in that: The multi-level response model is as follows: ； in, :platform Reachable area subset; : The node set representing the task area in step S5; : The current load status indicator of the platform; : The path cost for the platform to reach area i.

9. The intelligent water conservancy monitoring method based on aerial and underwater collaboration according to claim 6 is characterized in that: During continuous operation, the triggering may occur multiple times in different time periods in areas including specific river sections, dam bottoms, and old culverts. and , introduced an embedded memory enhancement revisit mechanism. By introducing an abnormal memory index table in the data scheduling layer of the command center, all areas marked as high-risk are archived and annotated at the event level, and the following information is recorded: Abnormal event timestamp and level; Associate the platform data upload time and platform number; Continuous before the abnormality occurs Change curve; Whether there are deviations or errors in the platform equipment status; Whether the current anomaly is a repeat of the previous anomaly area.

10. A smart water conservancy monitoring system based on aerial and underwater collaboration, constructed based on the smart water conservancy monitoring method according to any one of claims 1 to 9, characterized in that: The system's functional modules include: air platform, underwater platform and command center; The aerial platform consists of a multi-rotor UAV or a fixed-wing UAV, equipped with a lidar, high-precision GPS, infrared thermal imager, multispectral imaging equipment, and video acquisition terminal, and integrates an intelligent path planning module and a comprehensive environmental perception module. Through the onboard intelligent controller, it can realize autonomous planning and obstacle avoidance of flight paths, data collection targets, and flight dynamics. The underwater platform is mainly composed of an adaptive underwater detection robot, deployable underwater sensor nodes and acoustic communication modules. It integrates a multi-dimensional environmental monitoring module, including a water quality probe, a flow meter, and a structural deformation sensor, and is equipped with an inertial navigation system and an acoustic positioning system to achieve precise positioning. The command center is the central control unit of the system, consisting of edge computing servers, data fusion and processing modules, risk assessment and strategy generation engines, scheduling consoles, and three-dimensional visualization platforms; the command center is equipped with standardized data interfaces to support seamless connection with aerial and underwater platforms.

Images