An unmanned aerial vehicle-based ecological flow intelligent monitoring method and system

Abstract

The application discloses an ecological flow intelligent monitoring method and system based on a UAV, and belongs to the technical field of intelligent monitoring of water conservancy projects. The method comprises the following steps: UAV multi-sensor cooperative data acquisition and adaptive cruise control, intelligent positioning of a discharge opening and water flow pattern identification, multi-scene adaptive ecological flow calculation of a hydraulic model coupling, dynamic evaluation of a water surface of a water-reducing section, and ecological flow standard research and judgment and closed-loop feedback of collected parameters. The application realizes non-contact multi-scene adaptive flow checking by loading multiple types of sensors on a UAV, and evaluates dynamic changes of a water surface of a water-reducing section in combination with multi-spectrum cooperative technology, so that a complete closed-loop monitoring system is formed.

Description

Technical Field

[0001] This invention relates to the field of intelligent monitoring and ecological protection technology for water conservancy projects, and in particular to an intelligent monitoring method and system for ecological flow based on unmanned aerial vehicles (UAVs). Background Technology

[0002] Ecological flow is the minimum water volume required to maintain the health and functional integrity of a river's aquatic ecosystem, and it plays a crucial regulatory role in the operation and management of water conservancy projects. Currently, my country has incorporated the ecological flow compliance rate into its water conservancy performance evaluation system, requiring hydropower stations and water conservancy projects to continuously and stably release ecological flow according to approved standards. Water administrative departments at all levels supervise this process through a combination of regular ecological flow inspections and online monitoring. However, in actual supervision, the on-site calibration and verification of ecological flow still face numerous technical challenges, especially in small and medium-sized hydropower station clusters with diverse release facilities and complex flow patterns, where efficient and accurate flow verification methods remain scarce.

[0003] Traditional ecological flow monitoring primarily relies on manual on-site measurements, requiring professionals to carry equipment such as flow meters and water level gauges into dangerous areas downstream of dams and sluice gates for contact measurements. Under high-flow conditions such as full-pipe outflow and high-speed sluice gate releases, the water flow is turbulent and splashing is severe, making manual work not only inefficient but also posing significant safety risks. Furthermore, the large number of small hydropower stations, often located in remote mountainous areas, makes station-by-station manual inspections unsuitable for large-scale, routine monitoring in terms of both time and manpower. Regarding monitoring technologies, traditional solutions typically employ a single type of flow measurement equipment, which is ill-suited to various significantly different release conditions and lacks the ability to adaptively switch measurement methods based on different flow characteristics. In recent years, unmanned aerial vehicle (UAV) remote sensing technology has shown broad application prospects in the field of hydrological and water resource monitoring. Its advantages include mobility, rapid deployment, and the ability to carry multiple sensors, making it particularly suitable for the inspection and monitoring of small hydropower station clusters in mountainous areas with complex terrain and inconvenient transportation. However, a mature multi-scenario adaptive monitoring technology system for ecological flow based on drones has not yet been formed. The technical solution that integrates drone remote sensing data acquisition, intelligent water flow pattern recognition, multi-scenario hydraulic model coupling calculation, and ecological assessment of dewatered river sections into an integrated monitoring solution is still lacking.

[0004] Chinese patent CN 118863275 A discloses a method for determining the ecological flow of tidal river sections. This method determines the change in river channel volume by establishing a river topographic grid and a river water level grid, and corrects the net discharge based on the principle of water balance to determine the ecological flow of the tidal river section. This technical solution proposes an effective correction strategy to address the problem of data distortion caused by changes in river channel volume due to tidal cycles in tidal river sections. However, this solution is essentially still within the scope of fixed-section, long-term fixed-point monitoring, relying on the density of water level monitoring stations along the river and lacking the ability to quickly calibrate on-site for various discharge flow patterns. For various typical discharge scenarios such as full-pipe discharge, gate discharge, and horizontal discharge from the pipe outlet, this solution cannot adaptively select a hydraulic calculation model for flow verification based on different flow characteristics.

[0005] Furthermore, existing ecological flow monitoring methods generally focus on judging whether the flow at the outlet meets the standards, while lacking effective areal assessment capabilities for the ecological effects on the downstream dewatering sections. The ultimate goal of ecological flow is to ensure that the downstream river channel maintains a certain water surface width and ecological water use; flow data from the outlet alone cannot fully reflect the actual aquatic ecological status of the river channel. Dewatering sections may experience ecological problems such as large-scale riverbed exposure and degradation of aquatic habitats due to insufficient flow, and current technologies have not yet established an integrated monitoring link from outlet flow verification to assessment of the water surface coverage status of the downstream river section.

[0006] In view of the shortcomings of the existing technologies, it is necessary to provide an intelligent monitoring method and system that can use drones for non-contact, multi-scenario adaptive rapid monitoring of ecological flow and simultaneously assess the dynamic changes of the water surface in the dehydrated river section, so as to comprehensively improve the efficiency, safety and coverage of ecological flow supervision. Summary of the Invention

[0007] To address the aforementioned technical issues, this invention provides an intelligent monitoring method and system for ecological flow based on unmanned aerial vehicles (UAVs). This method utilizes UAVs equipped with multiple types of sensors for non-contact remote data acquisition. It combines intelligent water flow morphology recognition and multi-scenario adaptive hydraulic model coupling calculation to achieve rapid verification of ecological flow under various typical discharge conditions. Simultaneously, it uses multispectral and thermal infrared collaborative technologies to assess the dynamic changes of the water surface in the dewatered river section. Finally, it continuously optimizes the cruise strategy through a closed-loop feedback mechanism driven by compliance status.

[0008] This invention provides an intelligent monitoring method for ecological traffic based on unmanned aerial vehicles (UAVs), the method comprising the following steps:

[0009] Step S1, UAV multi-sensor collaborative data acquisition and adaptive cruise control: Based on the geographical coordinates of the water conservancy hub to be monitored and the distribution information of the discharge facilities, a UAV cruise route is generated. The UAV equipped with a high-definition visible light camera, a laser ranging module, and a multispectral imager and / or thermal infrared imager is controlled to fly along the cruise route. The UAV performs hovering acquisition operations at the preset acquisition points of each discharge facility to acquire visible light image sequences, laser ranging data, multispectral images or infrared images of the discharge outlet and its surrounding area. The UAV's spatial position coordinates and flight attitude parameters at each acquisition point are recorded simultaneously.

[0010] Step S2, Intelligent positioning and water flow pattern recognition of the vent: Water flow region detection and vent positioning processing are performed on the visible light image sequence, the structural geometric parameters of the vent are extracted, and the water flow pattern is classified and identified based on the water flow texture features and spatial motion pattern information to determine the current water flow type of the vent. The water flow type includes full pipe outflow, gate outlet outflow, and horizontal discharge outflow from the pipe opening.

[0011] Step S3, Ecological flow calculation of multi-scenario adaptive hydraulic model coupling: According to the water flow type, the hydraulic calculation model matching the current water flow type is called from the preset multi-scenario hydraulic model library. The structural geometric parameters and the head parameters obtained by laser ranging data are substituted into the hydraulic calculation model to calculate the ecological discharge flow value of the current discharge outlet.

[0012] Step S4, Dynamic Assessment of Water Surface in the Dewatered River Section: Water surface coverage is identified in the dewatered river section downstream of the outlet using multispectral or thermal infrared images. Water surface areas are extracted by constructing normalized difference water body index or ISO clustering index features, the water surface coverage rate is calculated, and the results of water surface change assessment are obtained by comparing with the historical baseline water surface coverage rate.

[0013] Step S5, Ecological flow compliance assessment and closed-loop feedback of collected parameters: The ecological discharge flow value is compared with the preset ecological flow threshold, and the water surface coverage rate is compared with the preset lower limit of water surface coverage rate. The ecological flow compliance status is comprehensively determined, a monitoring report is generated, and a cruise parameter adjustment instruction is generated and fed back to step S1 based on the compliance status to adjust the collection frequency of the corresponding discharge outlet in the subsequent cruise route.

[0014] Accordingly, the present invention also provides an intelligent monitoring system for ecological flow based on unmanned aerial vehicles (UAVs), including a multi-sensor collaborative data acquisition and adaptive cruise control module, an intelligent outlet positioning and water flow pattern recognition module, a multi-scenario adaptive hydraulic model coupling calculation module, a dewatering river section water surface dynamic assessment module, and a compliance assessment and closed-loop feedback module. Each module corresponds one-to-one with the above-mentioned method steps, forming a closed-loop collaborative monitoring system from data acquisition to compliance assessment and then to the feedback optimization of acquisition parameters.

[0015] The beneficial effects of this invention are as follows: It enables non-contact remote data acquisition via drones equipped with multiple types of sensors, significantly improving operational safety and efficiency; it supports rapid flow verification for various typical discharge scenarios, such as full-pipe outflow, gate outflow, and horizontal discharge from pipe openings, through intelligent water flow pattern recognition and adaptive coupling with multi-scenario hydraulic models; it achieves point-to-area ecological effect assessment of dewatered river sections through multispectral and thermal infrared collaborative water surface identification technology; and it continuously optimizes patrol strategies and improves the efficiency of monitoring resource allocation through a closed-loop feedback mechanism driven by compliance status. Compared with traditional manual measurement methods, this invention reduces single-station inspection time from several hours to just over ten minutes, achieving a flow verification accuracy of over 85%, meeting the engineering accuracy requirements for ecological flow calibration. Compared with existing fixed-section monitoring schemes, this invention does not rely on the dense deployment of riverside water level monitoring stations, offering advantages such as flexible deployment, wide coverage, and fast response speed. Attached Figure Description

[0016] Figure 1 This is a flowchart illustrating an intelligent monitoring method for ecological flow based on unmanned aerial vehicles (UAVs) provided in an embodiment of the present invention.

[0017] Figure 2 This is a schematic diagram of the structure of an intelligent ecological flow monitoring system based on unmanned aerial vehicles (UAVs) provided in an embodiment of the present invention. Detailed Implementation

[0018] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. It should be noted that the specific parameter values ​​in the following embodiments are merely illustrative and do not constitute a limitation on the scope of protection of the present invention. Those skilled in the art can make adaptive adjustments according to actual application scenarios.

[0019] Reference Figure 1 As shown in the figure, this invention provides an intelligent monitoring method for ecological flow based on unmanned aerial vehicles (UAVs), applied to the ecological flow supervision scenario of hydropower stations and water conservancy projects. This method uses UAVs equipped with multiple types of sensors for non-contact data collection, combines a multi-scenario adaptive hydraulic model to achieve rapid verification of ecological flow, and simultaneously assesses the dynamic changes of the water surface in the dewatered river section, forming a complete closed loop from data collection and target assessment to parameter feedback optimization. The method specifically includes the following steps.

[0020] It should be noted that the data processing flow involved in the method described in this embodiment of the invention follows the following technical route: Step S1 is responsible for multi-source collaborative acquisition of raw data, and the output visible light image sequence is transmitted to Step S2 for outlet location and water flow pattern identification. Simultaneously, laser ranging data is transmitted to Step S3 for head parameter calculation, and multispectral and thermal infrared images are transmitted to Step S4 for water surface coverage identification. The flow calculation results of Step S3 and the water surface assessment results of Step S4 are converged in Step S5 for comprehensive analysis. The feedback command from Step S5 is then transmitted back to Step S1 to update the cruise parameters, forming a complete data processing closed loop. The data transmission relationship between the above steps reflects the technical architecture features of multi-sensor data fusion and multi-step collaboration in the technical solution of this invention.

[0021] Step S1: Multi-sensor collaborative data acquisition and adaptive cruise control of the UAV. In one embodiment of the present invention, a UAV cruise route is first generated based on the geographical coordinates of the water conservancy hub to be monitored and the distribution information of the discharge facilities. Specifically, the latitude and longitude coordinates, facility type labels, and spatial layout relationships of all discharge facilities in the target water conservancy hub are obtained. Based on the above distribution information, a single-station fixed-point acquisition route is generated according to the latitude and longitude coordinates of each discharge facility. After the operator drives the vehicle to the vicinity of the discharge outlet, the UAV takes off within a range of 1km to 2km around the discharge outlet and flies along the preset route to perform the acquisition task above the discharge outlet. The cruise route sets preset acquisition points for each discharge facility. Preferably, at least 3 acquisition points are set for each discharge outlet, located directly above the discharge outlet, above the water surface of the upstream reservoir area, and above the downstream outflow area, respectively. The distance between adjacent acquisition points is determined according to the spatial scale of the discharge facility, usually within the range of 20m to 80m.

[0022] In a preferred embodiment of the present invention, the sensor components carried by the UAV include a high-definition visible light camera, a laser ranging module, and a multispectral imager or a thermal infrared imager. The high-definition visible light camera has a resolution of at least 4000×3000 pixels and a frame rate of at least 30fps, used to acquire visible light image sequences of the vent and its surrounding area. The laser ranging module uses a pulsed laser ranging sensor with a measurement range of 0.5m to 200m and a ranging accuracy of 0.1m to 0.2m, maintaining a vertically downward lens for ground ranging during hovering acquisition. The multispectral imager is configured with at least five band channels, covering at least the green light band (center wavelength approximately 560nm) and the near-infrared band (center wavelength approximately 840nm), with a spatial resolution of at least 5cm per pixel (at a flight altitude of 50m). The thermal infrared imager operates in the 8μm to 14μm band and has a temperature resolution of at least 0.05K, used to acquire information on surface temperature distribution.

[0023] Furthermore, the adaptive cruise control includes dynamically setting the collection priority. The initial collection priority is determined based on the importance level and historical compliance records of each discharge facility. Preferably, the collection frequency is determined based on the regulatory level of the discharge facility, with higher sampling collection ratios set for hydropower stations on major rivers or those listed in the key regulatory directory, and regular collection ratios set for small and medium-sized hydropower stations. Discharge outlets that fail to meet standards three or more times consecutively in the historical compliance record are automatically upgraded by one priority level. When the UAV performs hovering collection operations at each collection point, it simultaneously records the spatial coordinates (including longitude, latitude, and altitude) and flight attitude parameters (including pitch angle, roll angle, and heading angle) of the current point. These coordinates and attitude data are used for spatial geometric calculations in subsequent steps.

[0024] In one embodiment of the present invention, the laser ranging module performs vertical ranging on the water surface at a hovering point upstream to obtain the upstream distance to the ground. The distance from the downstream water surface to the ground is obtained by vertical distance measurement of the water surface at the downstream hovering point. Since the drone maintains a constant flight altitude during its cruise, the elevation difference between the upstream and downstream water surfaces can be obtained using the following formula: ,in: The difference in water surface elevation between upstream and downstream is expressed in meters (m). The value is the vertical distance measured by the UAV from its upstream hovering point relative to the water surface, in meters. The value is the vertical distance measured by the UAV to the water surface at the downstream hovering point, in meters. This calculation method assumes that the UAV maintains a constant flight altitude during the two distance measurements, with a distance measurement error within the range of 0.1m to 0.2m. This error accuracy meets the engineering accuracy requirements for ecological flow calibration. In a preferred embodiment of the invention, to further improve the reliability of the water surface elevation difference measurement, at least five consecutive distance measurements are performed at each hovering measurement point, and the median value is taken as the final distance measurement value to eliminate the influence of water surface fluctuations and laser reflection noise. Simultaneously, the change in the UAV's flight altitude is monitored in real time during data acquisition. When the altitude fluctuation exceeds 0.3m, a re-hovering stabilization is automatically triggered before resuming the distance measurement operation.

[0025] In one embodiment of the invention, the acquisition of multispectral images needs to be carried out during a period of relatively stable lighting conditions. Preferably, the acquisition period is between 10:00 and 14:00 local time and the weather is sunny or partly cloudy. The acquisition of thermal infrared images is not limited by lighting conditions, but acquisition should be avoided within 1 hour after rainfall, as residual moisture on the ground will affect the representativeness of the temperature distribution.

[0026] The output data of step S1 includes visible light image sequences, laser ranging data, multispectral images or thermal infrared images, as well as the spatial position coordinates and flight attitude parameters of the UAV. All of the above data are used as inputs for subsequent steps S2 to S5.

[0027] Step S2: Intelligent localization of the vent and recognition of water flow patterns. In one embodiment of the present invention, step S2 receives the visible light image sequence output in step S1 and performs water flow region detection and vent localization processing on it. Specifically, firstly, the visible light image is converted to a color space from RGB to HSV. Preliminary segmentation is performed using the specific responses of the water flow region in the hue and saturation channels to obtain candidate water flow regions. Subsequently, the structural boundary contour of the vent is extracted within the candidate water flow regions using the Canny edge detection algorithm, and geometric fitting of circular or rectangular structures is performed using Hough transform, thereby achieving accurate localization of the vent.

[0028] Preferably, the structural geometric parameters of the vent are obtained by combining laser dot measurement or image measurement with calibration of known reference dimensions. For pipe-type vents, the inner diameter of the pipe outlet section is extracted. When a reference object of known dimensions exists on-site (such as standard railing spacing, maintenance platform width, etc.), calibration and conversion are performed using the ratio between pixel dimensions and actual dimensions. Specifically, the pixel coordinates of both ends of the reference object are calibrated in the visible light image, the pixel length of the reference object is calculated, and this is divided by the known actual length to obtain the actual distance per pixel, i.e., the spatial resolution. This spatial resolution is then used to convert the pixel distance of the pipe cross-section edge into the actual inner diameter value. For gate-type discharge outlets, the gate width is extracted. and gate opening The gate opening is determined by the distance between the bottom edge of the gate and the bottom plate in the image. When a reliable reference point is unavailable, a UAV laser ranging module can be used to assist in spatial scale calibration, i.e., the ground sampling distance can be calculated from the known flight altitude and camera focal length parameters. Pipe length. Distance measurement is performed by using drones to mark points.

[0029] Further, the classification and recognition process of the water flow pattern is as follows: Gray-level co-occurrence matrix (GLCM) texture features of the water flow region in the visible light image are extracted, including four texture statistics: energy, contrast, correlation, and homogeneity. Simultaneously, optical flow field calculations are performed on consecutive frame images to extract the directional angle distribution and velocity distribution statistics of the water flow motion vector. The above texture features and motion features are combined to construct a water flow pattern feature vector with a dimension of 12. This water flow pattern feature vector is input into a pre-trained random forest classifier. During the training phase, this classifier is trained using no fewer than 500 sets of labeled samples, with 100 decision trees. The classification output is one of three water flow types: full-pipe outflow, gate outflow, or horizontal outflow from the pipe opening. In a preferred embodiment of the present invention, the discrimination characteristics of full-pipe outflow are: the outlet has a circular cross-section and the water flow continuously fills the pipe cross-section, the water flow texture is uniform, and the optical flow field direction is consistent; the discrimination characteristics of gate outflow are: the outlet has a rectangular cross-section and the water flow diffuses downstream from the bottom of the gate, with a high degree of texture disorder; the discrimination characteristics of horizontal ejection from the pipe opening are: the water flow exits horizontally from the pipe opening and falls in a parabolic trajectory, and the optical flow field has obvious horizontal components and increasing vertical components. The water flow type classification results and structural geometric parameters output in step S2 are passed to step S3 as the basis for hydraulic model selection and parameter assignment.

[0030] Step S3: Ecological flow calculation using multi-scenario adaptive hydraulic model coupling. In one embodiment of the invention, step S3 adaptively calls a hydraulic calculation model matching the current flow type from a preset multi-scenario hydraulic model library based on the flow type classification result output in step S2. The multi-scenario hydraulic model library includes three hydraulic calculation models: full pipe outflow model, gate outflow model, and horizontal discharge outflow model, corresponding to three typical hydropower station ecological discharge scenarios. The model calling process adopts a strategy pattern, that is, directly indexing the corresponding calculation model instance based on the flow type label, avoiding the use of multi-layer conditional branching structures, thereby improving computational efficiency and scalability. In a preferred embodiment of the invention, when the classification confidence of step S2 is lower than a preset confidence threshold (usually 0.75), the system will prompt the operator to manually confirm the flow type before executing the model call, in order to avoid selecting the wrong hydraulic calculation model due to misclassification. Meanwhile, each hydraulic calculation model will perform a reasonableness check on the input parameters before performing flow calculations. For example, it will check whether the inner diameter of the pipe is within a reasonable range of 0.02m to 1.0m, whether the head difference is positive and does not exceed 50m, and whether the gate opening is less than the water depth in front of the gate. When the input parameters exceed the reasonable range, the system will output an abnormal alarm and suspend the calculation, requesting the operator to verify the data.

[0031] Scenario 1: Pipe flow full-pipe outflow model.

[0032] When step S2 determines that the water flow type is full-pipe outflow, the pressurized pipe flow hydraulic calculation model is invoked. This scenario is applicable to hydropower stations that release ecological flow through pipeline facilities such as spillways, spill valves, or siphons. Typically, the pipe inner diameter is between 0.05m and 2.0m, the flow rate is large, and manual measurement is difficult. In one embodiment of the invention, the ecological release flow rate of full-pipe outflow is calculated using the following formula: ,in: The ecological discharge flow rate is the full outflow from the pipe, expressed in m³ / s. The flow coefficient is a dimensionless coefficient, typically ranging from 0.15 to 0.85, with the specific value determined by the pipe's geometric parameters and head loss characteristics. This refers to the outflow cross-sectional area of ​​the pipe, expressed in m². For a circular cross-section pipe, ,in The inner diameter of the pipe extracted in step S2; The acceleration due to gravity is taken as 9.81 m / s². The effective head includes the approach velocity head, and the unit is meters (m). Preferably, since the approach velocity head is usually relatively small, it can be ignored in actual calculations. When the discharge outlet is in a free outflow state, Water head from the water surface in front of the dam to the center elevation of the pipeline When the discharge outlet is in a submerged outflow state, Take the difference in water surface elevation between upstream and downstream The That is, the result obtained from the laser ranging data in step S1. .

[0033] Furthermore, the pipeline flow coefficient Determined by the following formula: ,in: The calculated length of the pipeline section, in meters, is obtained by measuring the difference between the start and end coordinates along the pipeline axis using a drone. This refers to the inner diameter of the pipe, in meters (m). is the head loss coefficient along the friction, and is a dimensionless coefficient; This is the sum of the local head loss coefficients in the calculated section of the pipeline. It is a dimensionless coefficient and includes local losses at the inlet (usually taken as 1.0), bend losses (usually taken as 0.5 for 90-degree bends), etc., summed according to the actual pipeline layout. Preferably, the friction head loss coefficient is... Determined according to the following relationship: ,in: The Chezy coefficient is based on the pipe roughness coefficient. and hydraulic radius Determined by the correlation of Manning's formula, i.e. ; For pipe roughness coefficient, a dimensionless coefficient, it is usually taken as 0.012 to 0.014 for concrete pipes and 0.010 to 0.012 for steel pipes. The hydraulic radius is expressed in meters (m). For a full-bore circular pipe, Through the aforementioned nested coefficient calculation chain, a complete solution from pipeline physical parameters to flow coefficients and then to discharge flow is achieved. In a verification case of this invention, the inner diameter of the ecological discharge pipe of a hydropower station is... The length of the pipe is 0.08m. The length is 15m, and the head difference between the upstream and downstream sides is 15m. The roughness coefficient is 10m. Taking a value of 0.012 and assuming one bend in the pipe, the pipe flow coefficient is calculated using the above model. Approximately 0.311, corresponding to ecological release flow. It is approximately 0.022 m³ / s, with a deviation from the actual measured value within 15%.

[0034] Scenario 2: Gate outflow model.

[0035] When step S2 determines that the water flow type is gate outlet, the gate discharge hydraulic calculation model is invoked. This scenario is applicable to situations where ecological flow is controlled by a gate. The drone can calculate the flow velocity and flow rate by acquiring the water depth upstream of the gate and the gate opening. In one embodiment of the present invention, the ecological discharge flow rate of the gate outlet is calculated according to the following formula: ,in: The ecological discharge flow rate at the gate outlet is expressed in m³ / s. is the gate orifice flow coefficient, and is a dimensionless coefficient; is the submergence coefficient of the gate outlet, and is a dimensionless coefficient with a value ranging from 0 to 1; This refers to the width of the gate opening, in meters (m). The gate opening is expressed in meters (m). The acceleration due to gravity is taken as 9.81 m / s². The head above the weir, including the approach velocity head, is measured in meters (m). The approach velocity head can be ignored when it is small. The water depth before the gate is opened is also considered. .

[0036] Preferably, the gate flow coefficient Determined using empirical formulas from the Institute of Hydraulic Research: ,in: The gate opening is expressed in meters (m). The depth of the water in front of the sluice gate, in meters, is obtained by comparing the laser range measured by the UAV at the hovering point on the water surface in front of the sluice gate in step S1 with the elevation of the sluice gate bottom plate. When the ratio is in the range of 0.05 to 0.60, the above empirical formula has good fitting accuracy. Submergence coefficient of gate outflow. The value is determined based on the outflow state of the gate: when the gate is in a free outflow state... Take 1.0; when the gate is in a submerged outflow state, Based on the ratio of downstream water depth to upstream water depth as well as The ratio is determined by interpolation using a pre-stored submergence coefficient curve, with a value typically ranging from 0.65 to 0.95. The criterion for distinguishing between submerged outflow and free outflow is: when the downstream water level is lower than the bottom edge elevation of the gate outlet plus the gate opening degree... When the water depth is reduced, it is determined to be free outflow; otherwise, it is determined to be submerged outflow.

[0037] Scenario 3: Flow model with flat ejection from the nozzle.

[0038] When step S2 determines that the water flow type is horizontal projectile flow from the pipe outlet, a hydraulic calculation model based on the laws of projectile motion is invoked. This scenario is applicable to situations where the water flow from the horizontal outlet of the pipe follows a parabolic trajectory. Its core lies in using a drone to measure the spatial parameters of the water flow trajectory to deduce the outlet velocity. In one embodiment of the invention, the flow rate calculation for horizontal projectile flow from the pipe outlet is divided into three steps.

[0039] The first step is to calculate the landing time. This involves using a drone to measure the vertical distance from the water outlet to the landing point. The landing time can be deduced by working backward from the laws of free fall. : ,in: The time it takes for water to travel from the nozzle to the point of impact, measured in seconds. The vertical distance from the pipe outlet to the water landing point is measured in meters. It is obtained by measuring the spatial line drawing function of the UAV, that is, after recording the spatial coordinates at the pipe outlet position and the water landing point position respectively, the vertical coordinate difference is calculated. The acceleration due to gravity is 9.81 m / s².

[0040] The second step is to calculate the horizontal velocity at the outlet. This involves using a drone to measure the horizontal distance from the water outlet to the point of impact. Combined with landing time Calculate the horizontal velocity at the exit : ,in: The horizontal velocity of the water flow at the pipe outlet is expressed in m / s. The distance from the pipe outlet to the point of impact with water is measured in meters (m), and is also obtained through the spatial line drawing function of the UAV.

[0041] Step 3: Calculation of ecological release flow: ,in: The ecological discharge flow rate is expressed as m³ / s when the pipe is horizontally ejected from the outlet. This refers to the cross-sectional area of ​​the pipe outlet, in m². The outlet horizontal velocity is calculated in the previous step. In a verification case of this invention, during the horizontal outflow measurement at the outlet of a hydropower station, a drone measured the horizontal distance. It is 1.5m, vertical distance The pipe diameter is 1.0m. The distance is 0.1m, and the landing time is calculated using the above model. Approximately 0.452s, exit horizontal velocity Approximately 3.322 m / s, pipe cross-sectional area Approximately 0.00785 m², corresponding to an ecological release flow rate. It is approximately 0.026 m³ / s.

[0042] Step S3 outputs the calculation results of the above three scenario models as ecological release flow values. The data is then passed to step S5 for compliance assessment. Simultaneously, the intermediate flow calculation parameters output from step S3 (including flow coefficient, effective head, cross-sectional area, etc.) are recorded for subsequent data traceability and report generation.

[0043] Step S4: Dynamic assessment of the water surface in the reduced-water section. In one embodiment of the present invention, step S4 utilizes the multispectral or thermal infrared images obtained in step S1 to identify and assess the water surface coverage status of the reduced-water section downstream of the discharge outlet. The reduced-water section refers to a river channel section within a certain distance downstream from the hydropower station discharge outlet where the riverbed may be exposed and the water surface may shrink due to reduced water volume. The water surface coverage status of this section directly reflects the actual ecological effect of ecological flow release. In a preferred embodiment of the present invention, the assessment range boundary of the reduced-water section is determined as follows: starting from the discharge outlet location, extending downstream to the next tributary confluence or the upper boundary of the downstream hydropower station backwater area, the closer of the two is taken as the downstream boundary of the assessment range; the lateral boundary of the assessment range is based on the dikes or natural shorelines on both sides of the river.

[0044] First, the multispectral imagery is preprocessed. Since multispectral images acquired by the UAV may exhibit geometric distortions due to changes in flight attitude, orthorectification is performed using the flight attitude parameters recorded in step S1. Subsequently, the digital quantization values ​​of each band are converted into surface reflectance values ​​using radiometric calibration coefficients. The green band reflectance is then extracted from the preprocessed multispectral imagery. (Corresponding to a center wavelength of approximately 560nm) and near-infrared reflectivity (Corresponding to a center wavelength of approximately 840 nm), the Normalized Difference Water Index (NDWI) is constructed using both. In one embodiment of the present invention, the calculation formula for NDWI is as follows: ,in: The normalized differential water index has a value range of -1 to 1. The reflectance in the green light band is dimensionless and ranges from 0 to 1. NDWI represents near-infrared reflectance, a dimensionless value ranging from 0 to 1. Water has high reflectance in the green light band and strong absorption in the near-infrared band; therefore, the NDWI value of water surface pixels is typically greater than 0, while the NDWI value of vegetation and bare land is typically less than 0. A preset water body threshold is set to an NDWI value greater than the specified threshold. The pixels are initially marked as candidate pixels for the water surface, preferably, The value ranges from 0.1 to 0.3, and in one embodiment of the present invention, it is 0.2.

[0045] Furthermore, cross-validation using thermal infrared imagery was employed to eliminate false positives. Due to the high thermal inertia of water bodies, the daytime temperature of the water surface area is typically lower than that of the surrounding bare land and vegetated areas. Surface temperature values ​​for each pixel were extracted from the thermal infrared imagery. Calculate the average temperature of the region To satisfy the temperature The pixels that meet the conditions are labeled as candidate pixels for the thermal infrared water surface, where To preset the temperature difference threshold, a value between 2K and 5K is preferred. The final determination of water surface pixels is based on the intersection of the NDWI and thermal infrared identification results, i.e., simultaneously satisfying... and The pixels that are confirmed as water surface pixels are identified. This dual verification mechanism effectively eliminates interference from shadowed areas and wet bare ground, improving recognition accuracy. In a preferred embodiment of the invention, the water surface recognition results after dual verification are further subjected to morphological post-processing operations, including filling isolated empty pixels within the water surface area using a 3×3 pixel dilation operation and eliminating scattered noise pixels at the edges using a 3×3 pixel erosion operation, thereby obtaining a water surface area mask with clear boundaries and good connectivity. Furthermore, isolated water surface patches with an area smaller than a preset minimum connected component threshold (usually 50 pixels) are removed to avoid including non-river water surfaces such as road surface water and small puddles in the statistical scope.

[0046] Subsequently, the water surface coverage rate was calculated within a pre-defined assessment area of ​​the dewatering section. This pre-defined assessment area was set according to the ecological flow monitoring requirements of the hydropower station, typically encompassing the river channel area within 0.5 km to 5 km downstream of the discharge outlet. Water surface coverage rate Calculate using the following formula: ,in: This represents the current water surface coverage rate, expressed in % %. The number of water surface pixels within the assessment area that have been double-verified; To assess the total number of pixels within the assessment area. The current water surface coverage. Compared with historical baseline water surface coverage By comparing the results, an assessment of water surface changes can be obtained. The average water surface coverage is obtained by performing the same identification and processing on multispectral images collected during the high-water season (usually June to September) within the same assessment area. It represents the baseline level of water surface coverage under normal hydrological conditions for that river section. Water surface change rate. The calculation is as follows: ,in: The rate of change of water surface is expressed as %; when This indicates that the water surface area in the dehydrated river section has shrunk significantly, requiring close attention. The output of step S4 includes the current water surface coverage rate. The results of the water surface change assessment are then passed to step S5 for comprehensive assessment of ecological flow compliance.

[0047] Step S5: Ecological flow compliance assessment and closed-loop feedback of collected parameters. In one embodiment of the present invention, step S5 receives the ecological release flow value output in step S3. and the water surface coverage rate output in step S4 The ecological flow status is comprehensively assessed. It should be noted that this comprehensive assessment employs a dual-condition strategy of assessing both flow rate compliance and water surface coverage compliance. The technical consideration behind this design is that relying solely on discharge outlet flow data may not fully reflect the actual aquatic ecological conditions of the downstream river channel. Therefore, introducing water surface coverage as an auxiliary assessment condition allows for a more comprehensive evaluation of the ecological flow release effect. The logic of the comprehensive assessment is as follows: when the ecological flow release value... Greater than or equal to the preset ecological flow threshold And water surface coverage Greater than or equal to the preset lower limit of water surface coverage When the ecological discharge flow value is [value missing], it is judged to be in a state of compliance; when the ecological discharge flow value is [value missing], it is judged to be in a state of compliance. Less than or water surface coverage Less than When this occurs, it is determined to be a non-compliant state. Preferably, the ecological flow threshold is... The lower limit of water surface coverage is set according to the minimum ecological discharge standard approved by each hydropower station. The design is based on the ecological protection requirements of the river section in question.

[0048] In a preferred embodiment of the present invention, the compliance status can be further subdivided into three levels. The first level is complete compliance, i.e. and The second level is the critical level, that is... or The third level is substandard, that is... or Different levels of compliance correspond to different warning levels and feedback strategies.

[0049] Furthermore, step S5 generates cruise parameter adjustment instructions based on the compliance status and feeds them back to the cruise control link in step S1, forming a closed-loop optimization. Specifically, for discharge outlets determined to be non-compliant, their sampling frequency is increased to twice the initial sampling frequency to obtain more comprehensive monitoring data in subsequent cruises. For discharge outlets determined to be fully compliant twice consecutively, their sampling frequency can be appropriately reduced to 0.5 times the initial sampling frequency to optimize cruise time and battery endurance allocation. This closed-loop feedback mechanism enables the entire monitoring system to dynamically adjust resource allocation based on actual monitoring results, prioritizing the allocation of limited UAV cruise time to discharge facilities with abnormal ecological flow conditions, thereby maximizing monitoring efficiency while ensuring monitoring coverage.

[0050] The final output of step S5 is a standardized monitoring report, which includes the flow type, flow calculation results, compliance status, water surface coverage and its changing trend for each discharge outlet, as well as suggestions for adjusting cruise parameters. In one embodiment of the invention, the monitoring report also includes data traceability information, including the timestamp of data collected at each discharge outlet, the location coordinates of the UAV, sensor parameter settings, and the complete intermediate parameter chain for flow calculation (including the original distance measurement value, head calculation process, flow coefficient calculation process, and final flow value), to facilitate subsequent review and quality control. The monitoring report supports storage, query, and historical traceability functions. Monitoring data from each cruise is archived and stored in a time series, supporting compliance trend analysis and statistical report generation across time periods, providing data support and decision-making basis for water administrative departments in ecological flow supervision.

[0051] In one verification embodiment of the present invention, 10 discharge outlets from 8 small hydropower stations in a certain river basin were selected as test objects, including 4 full-pipe discharge outlets, 3 gate outlet outlets, and 3 horizontal discharge outlets. The test area is located in a mountainous river basin with an altitude ranging from 200m to 800m, and the average distance between the hydropower stations is approximately 2km to 7km. The test drone is a quadcopter equipped with an RTK positioning module, with a maximum flight time of approximately 45 minutes, a flight altitude set at 30m to 50m, and terrain-following flight. The inspection time for a single station is approximately 15min to 25min, which is significantly shorter than the 1h to 2h of traditional manual on-site measurement, improving work efficiency by approximately 6 to 8 times.

[0052] Regarding flow calculation accuracy, comparative verification was conducted using manually measured flow rates on-site as a benchmark. For the scenario of full-pipe outflow from a pipe, 16 sets of data were collected from 4 vents, with an average relative error of 12.8%, a maximum relative error of 18.5%, and a minimum relative error of 5.2%. For the scenario of orifice outflow, 12 sets of data were collected from 3 vents, with an average relative error of 10.5%, a maximum relative error of 15.8%, and a minimum relative error of 4.7%. For the scenario of horizontal outflow from a pipe outlet, 12 sets of data were collected from 3 vents, with an average relative error of 14.2%, a maximum relative error of 19.6%, and a minimum relative error of 6.8%. The overall measurement accuracy reached over 85%, meeting the engineering accuracy requirements for ecological flow calibration. The main sources of error include the inherent error of laser ranging from 0.1m to 0.2m, the error in extracting geometric parameters from the image, and the applicability deviation of the empirical coefficients of the hydraulic model. Among these, the laser ranging error has the most significant impact on the full pipe outflow model and the gate outflow model, because the flow rate of these two models is proportional to the square root of the head, and the head measurement error will be amplified into the flow rate result through the square root calculation.

[0053] Regarding water surface coverage identification, the results were validated in five assessment areas (total area of ​​approximately 12.5 km²) of dehydrated river sections, based on visual interpretation. The water surface identification method based on the normalized difference water index and thermal infrared cross-validation achieved an overall accuracy of 91.3%, with a user accuracy of 89.6% and a producer accuracy of 93.1%. In contrast, when using only the normalized difference water index for identification, the overall accuracy was 83.7%, with the main sources of misclassification being shaded areas and wet bare land incorrectly marked as water surfaces. By introducing a thermal infrared temperature difference cross-validation mechanism, this type of misclassification rate was reduced from 16.3% to 8.7%, improving the water surface identification accuracy by 7.6 percentage points.

[0054] Regarding the closed-loop feedback mechanism, after three rounds of patrol iterations, the data collection completeness of non-compliant discharge outlets improved from 72% in the first round to 96%, indicating that the adaptive collection frequency adjustment strategy can effectively improve the monitoring coverage of abnormal discharge outlets. Simultaneously, the average inspection time for fully compliant discharge outlets was shortened by 35%, and the total patrol power consumption decreased by approximately 22%, verifying the effectiveness of resource optimization driven by closed-loop feedback. In summary, the UAV-based intelligent ecological flow monitoring method proposed in this invention meets the practical engineering supervision needs in terms of measurement accuracy, operational efficiency, and monitoring coverage, demonstrating good practical value and promising prospects for wider application.

[0055] Reference Figure 2As shown, this embodiment of the invention also provides an intelligent monitoring system for ecological traffic based on unmanned aerial vehicles (UAVs). This system corresponds one-to-one with each step in the above method embodiment and includes the following five functional modules. The modules transmit data to each other through a standardized data interface, forming a complete closed-loop collaborative monitoring system from data collection to target assessment and then to the feedback optimization of collected parameters.

[0056] The multi-sensor collaborative data acquisition and adaptive cruise control module is configured to perform all the functions of step S1 in the above method embodiment. This module includes a route planning subunit and a sensor collaborative acquisition subunit. The route planning subunit automatically generates a cruise route covering all discharge outlets based on the geographical coordinates and distribution information of the discharge facilities, and sets the corresponding acquisition frequency and hovering duration according to the acquisition priority of each discharge outlet. Preferably, the route planning subunit has a built-in shortest path planning algorithm, which can minimize the total cruise distance while meeting the coverage requirements of all discharge outlets, thereby extending the effective operating time of the UAV. The sensor collaborative acquisition subunit controls a high-definition visible light camera, a laser ranging module, and a multispectral imager and / or a thermal infrared imager to synchronously perform data acquisition operations at each preset acquisition point, and records spatial position coordinates and flight attitude parameters. In one embodiment of the present invention, the sensor collaborative acquisition subunit ensures that the acquisition times of each sensor are aligned through a time synchronization signal, with a time synchronization accuracy of not less than 10ms, thereby ensuring that the multi-source data acquired at the same acquisition point are strictly consistent in time and space. This module also receives cruise parameter adjustment instructions from the compliance assessment and closed-loop feedback module, dynamically updates the acquisition configuration parameters of each discharge port, and realizes adaptive optimization configuration of monitoring resources.

[0057] The intelligent positioning and flow morphology recognition module for the vent is configured to perform all the functions of step S2 in the above method embodiment. This module receives visible light image sequences output by the multi-sensor collaborative data acquisition and adaptive cruise control module. It achieves vent positioning through color space conversion, edge detection, and geometric fitting, and classifies and recognizes flow morphology through the extraction of gray-level co-occurrence matrix texture features and optical flow field motion vectors, using a random forest classifier. In one embodiment of the invention, the built-in random forest classifier supports online incremental learning. When misclassification occurs after manual verification of the actual classification results, misclassified samples can be added to the training set for model updates, thereby continuously improving classification accuracy. The module's output includes the structural geometric parameters of the vent and the flow type classification results, which are then passed to the multi-scenario adaptive hydraulic model coupling calculation module.

[0058] The multi-scenario adaptive hydraulic model coupling calculation module is configured to execute all the functions of step S3 in the above method embodiment. This module incorporates three hydraulic calculation models: full-pipe outflow model, gate outflow model, and horizontal outflow model. It adaptively calls the matching calculation model based on the flow type. The calculation formulas, coefficient determination methods, and parameter input requirements for each model are consistent with the detailed description of step S3 in the method embodiment. In one embodiment of the present invention, this module adopts a strategy pattern software architecture design, encapsulating the three hydraulic calculation models into independent strategy objects, and dynamically indexing and calling them through flow type tags. This design allows for the addition of other types of hydraulic calculation models only by adding the corresponding strategy objects without modifying the main control logic. This module outputs the ecological discharge flow value and intermediate calculation parameters, which are then transmitted to the compliance assessment and closed-loop feedback module.

[0059] The dehydrated river section water surface dynamic assessment module is configured to perform all the functions of step S4 in the above method embodiment. This module receives multispectral or thermal infrared images output by the multi-sensor collaborative data acquisition and adaptive cruise control module. It identifies water surface areas through normalized difference water index calculation or thermal infrared temperature difference feature analysis, then calculates the water surface coverage rate and compares it with historical benchmarks to generate a water surface change assessment result. In one embodiment of the invention, this module has a built-in historical benchmark database that stores benchmark water surface coverage data acquired in different seasons for each monitored river section, supporting time-series retrieval and trend analysis. The calculation process and parameter settings of this module are consistent with the description of step S4 in the method embodiment.

[0060] The compliance assessment and closed-loop feedback module is configured to execute all the functions of step S5 in the above method embodiment. This module receives the ecological discharge flow value output by the multi-scenario adaptive hydraulic model coupled calculation module and the water surface coverage rate output by the dewatering river section water surface dynamic assessment module. It comprehensively determines the ecological flow compliance status according to a preset compliance judgment logic and generates a standardized monitoring report. Simultaneously, this module generates cruise parameter adjustment instructions based on the compliance status and feeds them back to the multi-sensor collaborative data acquisition and adaptive cruise control module, forming a closed-loop collaborative optimization. In one embodiment of the invention, the standardized monitoring report includes basic data acquisition information (acquisition time, weather conditions, UAV model and flight parameters), flow calculation details for each discharge outlet (flow type, input parameters, calculation results, compliance status), and water surface assessment information for the dewatering river section (water surface coverage rate, change trend, warning level). The report format supports exporting to a common electronic document format for archiving and sharing. The above five functional modules can be deployed and upgraded independently, possessing good scalability and maintainability, and can adapt to ecological flow monitoring scenarios of different scales and management needs.

[0061] The embodiments of the present invention are not limited to the specific embodiments described above. Those skilled in the art can make various equivalent changes or substitutions based on the technical solutions of the present invention, and all such changes or substitutions should be included within the protection scope of the present invention.

Claims

1. A method for intelligent monitoring of ecological flow based on unmanned aerial vehicles (UAVs), characterized in that, Includes the following steps: Step S1, UAV multi-sensor collaborative data acquisition and adaptive cruise control: Based on the geographical coordinates of the water conservancy hub to be monitored and the distribution information of the discharge facilities, a UAV cruise route is generated. The UAV equipped with a high-definition visible light camera, a laser ranging module, and a multispectral imager and / or thermal infrared imager is controlled to fly along the cruise route. The UAV performs hovering acquisition operations at the preset acquisition points of each discharge facility to acquire visible light image sequences, laser ranging data, and multispectral images and / or thermal infrared images of the discharge outlet and its surrounding area. The UAV's spatial position coordinates and flight attitude parameters at each acquisition point are recorded simultaneously. Step S2, intelligent positioning and water flow pattern recognition of the discharge port: water flow region detection and discharge port positioning processing are performed on the visible light image sequence, the structural geometric parameters of the discharge port are extracted, and the water flow pattern is classified and identified based on water flow texture features and spatial motion pattern information to determine the water flow type of the current discharge port. The water flow type includes full pipe outflow, gate outlet outflow and horizontal discharge from pipe opening. Step S3, Ecological flow calculation of multi-scenario adaptive hydraulic model coupling: According to the water flow type, a hydraulic calculation model matching the current water flow type is called from the preset multi-scenario hydraulic model library. The structural geometric parameters and the head parameters obtained by the laser ranging data are substituted into the hydraulic calculation model to calculate the ecological discharge flow value of the current discharge outlet. Step S4, Dynamic assessment of the water surface in the dewatering section: The water surface coverage of the dewatering section downstream of the outlet is identified using the multispectral image and / or the thermal infrared image. The water surface area is extracted by constructing a normalized differential water body index or by combining thermal infrared temperature difference features. The water surface coverage rate is calculated and compared with the historical benchmark water surface coverage rate to obtain the water surface change assessment result.

2. The method according to claim 1, characterized in that, In step S1, the adaptive cruise control further includes: determining the initial acquisition priority of each release port based on the importance level of the release facility and historical compliance records; setting a first acquisition frequency for release ports with high acquisition priority and a second acquisition frequency for release ports with low acquisition priority, wherein the first acquisition frequency is greater than the second acquisition frequency; the laser ranging module keeps the lens vertically downward for distance measurement when hovering for acquisition, and obtains the elevation difference between the upstream and downstream water surfaces by hovering and measuring distances on the upstream and downstream water surfaces respectively.

3. The method according to claim 1, characterized in that, In step S2, the classification and recognition of water flow patterns specifically includes: extracting the gray-level co-occurrence matrix texture features and optical flow field motion vectors of the water flow region in the visible light image to construct a water flow pattern feature vector; inputting the water flow pattern feature vector into a pre-trained random forest classifier to output the water flow type classification result; the structural geometric parameters include at least one of the following: pipe inner diameter, pipe length, gate width, and gate opening, which are obtained through actual measurement, UAV ranging, or image pixel conversion.

4. The method according to claim 1, characterized in that, In step S3, the calculation of the head parameter includes: calculating the upstream and downstream water surface elevations based on the spatial coordinates of the UAV hovering point on the upstream water surface and the downstream outlet point, and the corresponding laser ranging values, thereby obtaining the head difference between the upstream and downstream; when the flow type is submerged outflow, the head parameter is the head difference between the upstream and downstream; when the flow type is full pipe outflow, the head parameter is the head from the water surface in front of the dam to the center elevation of the pipe; when the flow type is gate outflow, the head parameter is the water depth in front of the gate.

5. The method according to claim 1, characterized in that, In step S3, the multi-scenario hydraulic model library includes: The full-pipe outflow model calculates the flow rate based on the pressurized pipe flow formula, where the pipe flow coefficient is determined by the sum of the pipe calculation section length, pipe inner diameter, friction head loss coefficient, and local head loss coefficient. The gate outlet flow model calculates the flow rate based on the gate discharge formula. The gate flow coefficient is determined by the ratio of the gate opening to the water depth in front of the gate, and the submersion coefficient is determined by the gate outlet flow state, which is either free flow or submerged flow. The horizontal projectile flow model is based on the laws of horizontal projectile motion. It uses a drone to measure the horizontal and vertical distances from the water outlet to the landing point, and then calculates the flow rate by multiplying the outlet water velocity by the pipe cross-sectional area.

6. The method according to claim 5, characterized in that, In the full-pipe outflow model, the pipe flow coefficient is obtained by a composite calculation of the friction head loss coefficient, the ratio of the calculated pipe section length to the pipe inner diameter, and the sum of the local head loss coefficient; the friction head loss coefficient is determined based on the Darcy coefficient, which is related to the pipe roughness coefficient and the hydraulic radius according to the Chezy formula.

7. The method according to claim 1, characterized in that, In step S4, the construction of the normalized differential water index includes: extracting the green band reflectance and near-infrared band reflectance from the multispectral image, calculating the ratio of the difference between the green band reflectance and the near-infrared band reflectance to the sum of the values ​​as the water index; marking pixels with the water index greater than a preset water threshold as water surface pixels, or performing cross-validation by combining the thermal infrared image with areas where the temperature is lower than a preset temperature difference threshold to eliminate misjudgments of shadows and wet bare land.

8. The method according to claim 1, characterized in that, In step S4, the calculation of the water surface coverage rate includes: within the preset assessment range of the dehydrated river section, the ratio of the number of water surface pixels determined after water surface identification to the total number of pixels within the assessment range is used as the current water surface coverage rate; the historical benchmark water surface coverage rate is the average water surface coverage rate obtained by the same identification processing of multispectral images collected during the high-water season within the same assessment range.

9. The method according to claim 1, characterized in that, The process also includes step S5, ecological flow compliance assessment and closed-loop feedback of collected parameters: comparing the ecological discharge flow value with a preset ecological flow threshold, and simultaneously comparing the water surface coverage rate with a preset lower limit of water surface coverage rate, comprehensively determining the ecological flow compliance status, generating a monitoring report, and generating a cruise parameter adjustment instruction based on the compliance status and feeding it back to step S1, which is used to adjust the collection frequency of the corresponding discharge outlet in the subsequent cruise route. The comprehensive determination of the ecological flow compliance status includes: when the ecological discharge flow value is greater than or equal to the ecological flow threshold and the water surface coverage rate is greater than or equal to the lower limit of water surface coverage rate, it is determined to be compliant; when the ecological discharge flow value is less than the ecological flow threshold or the water surface coverage rate is less than the lower limit of water surface coverage rate, it is determined to be non-compliant; the cruise parameter adjustment instruction includes increasing the collection frequency of the non-compliant discharge outlet to a preset multiple of the initial collection frequency.

10. An intelligent ecological flow monitoring system based on unmanned aerial vehicles (UAVs), used to implement the method of claim 9, characterized in that, include: The multi-sensor collaborative data acquisition and adaptive cruise control module is configured to generate a cruise route based on the geographical coordinates of the water conservancy hub to be monitored and the distribution information of the discharge facilities. It controls a UAV equipped with a high-definition visible light camera, a laser ranging module, and a multispectral imager and / or a thermal infrared imager to fly along the cruise route and perform hovering acquisition operations at preset acquisition points of each discharge facility to acquire visible light image sequences, laser ranging data, multispectral images or thermal infrared images, and simultaneously record spatial position coordinates and flight attitude parameters. The intelligent positioning and water flow pattern recognition module for the vent is configured to receive the visible light image sequence, detect the water flow area and locate the vent, extract structural geometric parameters, and classify and recognize the water flow pattern based on water flow texture features and spatial motion pattern information to determine the water flow type. The multi-scenario adaptive hydraulic model coupling calculation module is configured to call a matching hydraulic calculation model from a preset multi-scenario hydraulic model library according to the water flow type, substitute the structural geometric parameters and the head parameters obtained by the laser ranging data into the hydraulic calculation model, and calculate the ecological release flow value. The dewatering section water surface dynamic assessment module is configured to use the multispectral image or the thermal infrared image to identify the water surface coverage of the dewatering section downstream of the outlet, extract the water surface area by normalized differential water body index or combined with thermal infrared temperature difference features, calculate the water surface coverage rate and compare it with historical benchmarks to obtain the water surface change assessment results. The compliance assessment and closed-loop feedback module is configured to compare the ecological discharge flow rate and the water surface coverage rate with preset thresholds respectively, comprehensively determine the compliance status of the ecological flow, generate a monitoring report, and generate cruise parameter adjustment instructions based on the compliance status and feed them back to the multi-sensor collaborative data acquisition and adaptive cruise control module.

Images