A Collaborative Intelligent Water Quality Monitoring Method Based on a Dual Platform of Unmanned Aerial Vehicles and Bionic Fish

Abstract

This invention discloses a collaborative intelligent water quality monitoring method using a dual-platform approach combining a drone and a biomimetic fish. It includes the following steps: The drone subsystem transmits the location and extent information of the initially screened contaminated area, along with spectral data, back to an intelligent remote monitoring platform in real time; a biomimetic fish precision detection task is generated, and the coordinates of the initially screened contaminated area and a preset monitoring path are sent to the biomimetic fish subsystem; the biomimetic fish subsystem collects measured data from the initially screened contaminated area and transmits the collected data back to the intelligent remote monitoring platform; based on the spectral data from the drone and the measured data from the biomimetic fish, data fusion is performed using spatiotemporal registration to obtain the fused contamination concentration inside and outside the contaminated area; an early warning level is automatically triggered according to the water quality anomaly level, and the coordinates of the contaminated area are pushed to the operation and maintenance terminal, completing the early warning response closed loop. This invention not only enables rapid response to pollution events and improves source tracing efficiency but also shortens the monitoring cycle.

Description

Technical Field

[0001] This invention belongs to the field of water quality monitoring technology, specifically referring to a collaborative intelligent water quality monitoring method using a dual-platform system of drones and biomimetic fish. Background Technology

[0002] Bionic fish dual-platform collaborative intelligent water quality monitoring is an innovative environmental monitoring method that achieves efficient and accurate monitoring of the aquatic environment through the collaborative work of two or more bionic fish platforms. A typical bionic fish dual-platform collaborative intelligent water quality monitoring system consists of multiple bionic fish platforms, a ground control base station, and an ecological big data platform. Each bionic fish platform adopts a biomimetic design (such as tail fin actuation), is equipped with a multi-channel sampling chamber and scalable online water quality sensors (such as ammonia nitrogen, COD, turbidity, dissolved oxygen, etc.), and supports 24-hour continuous operation.

[0003] However, existing biomimetic fish platforms have the following technical problems: While single bionic fish or unmanned surface vessel monitoring can obtain accurate data, it suffers from insufficient energy supply and endurance, limited mobility, and high deployment costs, making it impossible to meet the requirements of "wide coverage, high efficiency, and high precision". Second, the existing monitoring system lacks a multi-device collaboration mechanism, resulting in fragmented data and difficulty in forming a closed-loop monitoring process of "initial screening, location, precise detection and source tracing", leading to delayed pollution response and low source tracing efficiency. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention proposes a collaborative intelligent water quality monitoring method using a dual-platform drone and biomimetic fish, forming a closed-loop monitoring process. This method not only enables rapid response to pollution incidents and improves source tracing efficiency, but also shortens the monitoring cycle and avoids problems related to insufficient energy supply and endurance compared to a single biomimetic fish.

[0005] To achieve the above objectives, this invention designs a collaborative intelligent water quality monitoring method using a dual-platform system of unmanned aerial vehicles (UAV) and biomimetic fish, characterized by the following steps: S1) Build an unmanned aerial vehicle (UAV) subsystem, a bionic fish roe system, an intelligent remote monitoring platform, and an early warning and response system; S2) The UAV subsystem is used to conduct preliminary screening of water pollution areas, perform spectral inversion on the pre-screened pollution areas to obtain spectral data, and transmit the location and range information of the pre-screened pollution areas and spectral data back to the intelligent remote monitoring platform in real time, while receiving the platform's cruise path instructions. After receiving the information transmitted back from the UAV subsystem, the intelligent remote monitoring platform generates a bionic fish precision detection task and sends the coordinates of the initial screening pollution area and the preset monitoring path instruction to the bionic fish subsystem. S3) The bionic fish roe system receives the coordinates of the initial screening contamination area from the intelligent remote monitoring platform, arrives at the initial screening contamination area according to the preset monitoring path, collects actual measurement data, and transmits the collected actual measurement data back to the intelligent remote monitoring platform. S4) The intelligent remote monitoring platform is based on the spectral data of the UAV and the measured data of the bionic fish. It uses spatiotemporal registration to fuse the data, build a pollution model in the polluted area, and obtain the fused accurate pollution concentration in the polluted area. Then, with the pollution source as the center, a diffusion time is set to build a pollution diffusion model outside the polluted area, and obtain the fused pollution concentration at a certain moment when the pollution has diffused outside the polluted area. S5) The early warning response system generates a complete monitoring report based on the fusion analysis results. The monitoring report includes details of the polluted area, the level of water quality anomaly, and the trend of pollution spread. The system automatically triggers an early warning level based on the level of water quality anomaly and pushes the coordinates of the polluted area to the operation and maintenance terminal to complete the early warning response closed loop.

[0006] Furthermore, in S1), the UAV subsystem is equipped with a hyperspectral camera, a positioning module, and a communication module; the biomimetic fish roe system is equipped with a multi-parameter water quality sensor, an IMU inertial navigation unit, a GPS positioning unit, an infrared obstacle avoidance sensor, and a watertight sampling chamber.

[0007] Furthermore, in S3), the biomimetic fish roe system uses a hierarchical data feature extraction method to collect measured data. The specific steps are as follows: S31) The bionic fish completes underwater mobile monitoring according to the preset monitoring path, and simultaneously collects raw data from the water quality sensing layer and raw data from the biological sensing layer. The two types of data are bound to spatiotemporal labels according to the collection time and underwater monitoring point to form a raw dataset of water quality sensing and a raw dataset of biological sensing. S32) Remove outliers and fill in missing values ​​in the raw data of the water quality sensing layer to eliminate equipment monitoring errors; perform targeted processing on the raw data of the biological sensing layer, complete noise reduction and contour extraction according to the target contour image, complete the filtering of abnormal frequencies of aquatic organism swimming frequency and time series normalization, and obtain standardized water quality sensing dataset and standardized biological sensing dataset respectively. S33) Extract core water quality features such as turbidity change trend, pH fluctuation range and real-time index values ​​from the standardized water quality sensing dataset to form the initial water quality sensing feature set; extract core biological features such as target contour morphology features, contour area ratio, average swimming frequency of aquatic organisms and frequency mutation features from the standardized biological sensing dataset to form the initial biological sensing feature set. S34) Differentiated weights are assigned to each feature parameter in the initial feature set of water quality perception and the initial feature set of biological perception. The water quality perception layer focuses on the pollution characterization of basic water quality indicators, and core indicators are given high weights. The biological perception layer focuses on the response characteristics of biological behavior to pollution, and core features are given high weights. Non-core features are given low weights according to their characterization relevance, thus completing the construction of the weight system.

[0008] Furthermore, in S4), the specific steps for fusion using spatiotemporal registration include: S41) The bionic fish simultaneously collects underwater water quality and biological stratification monitoring data and binds its own underwater relative coordinates and collection timestamps. The drone collects water surface and airspace pollution-related data and binds GPS absolute coordinates and collection timestamps to form a fish-drone heterogeneous dataset with original spatiotemporal labels. At the same time, the basic parameters of the two devices' collection frequency and positioning accuracy are recorded. S42) Establish a fish-machine collaborative spatiotemporal reference system, determine a unified spatial reference coordinate system and time synchronization reference, extract the absolute GPS coordinates of the UAV as the spatial reference and the high-precision time synchronization of the UAV as the time reference, and establish a unified reference for subsequent coordinate transformation and time compensation. S43) Introduce a spatial coordinate transformation coefficient to establish a mapping relationship between the absolute GPS coordinates of the UAV and the underwater relative coordinates of the bionic fish. Convert the underwater relative coordinates of the bionic fish to the GPS spatial reference coordinate system of the UAV through the transformation coefficient, eliminate the spatial positioning dimension difference between the fish and the UAV across media, and achieve unified matching of the spatial positions of the monitoring data of the two. S44) Introducing a time synchronization compensation factor, based on the data collection timestamp deviation, data transmission delay, and device response time difference of the fish-machine equipment, to compensate and correct the data collection time of the bionic fish and the drone, eliminate the time asynchrony problem across devices and media, and achieve accurate alignment of the monitoring data of the two in the time dimension. S45) For the fish-machine heterogeneous data that have completed spatial coordinate transformation and time synchronization compensation, data matching is performed according to the same monitoring area and time period to verify the consistency of spatiotemporal registration results. Abnormal data with spatial misalignment and time deviation caused by coordinate transformation and time compensation are eliminated to form an initial registration heterogeneous dataset.

[0009] Furthermore, in S4), the pollution model is expressed by the following formula: In the formula, This indicates the precise concentration of contaminants within the contaminated area after fusion. This indicates the spectral pollution concentration within the polluted area obtained from the spectral inversion by the drone. This indicates the measured pollution concentration within the contaminated area collected by the bionic fish sensor. Indicates the fusion weight. Indicates the fusion weight. For environmental correction items, The standard deviation of the pollution concentration obtained from the UAV spectral inversion reflects the degree of fluctuation and uncertainty of this set of spectral inversion data. The standard deviation of the pollution concentration measured by the bionic fish sensor reflects the degree of fluctuation and uncertainty of the measured data from this set of sensors.

[0010] Furthermore, in S4), the pollution diffusion model is expressed by the following equation: In the formula, This indicates the coordinates outside the contaminated area at time t. The concentration of contaminants after fusion at ( ) Indicates the coordinates of the pollution source ( , Precise pollution concentration at the location after fusion. Indicates the water diffusion coefficient. Indicates diffusion time. ( , The coordinates of the pollution source are indicated by drones for initial screening and bionic fish for fixed-point detection.

[0011] Furthermore, in S5), the water quality anomaly level is represented by the following formula. In the formula, L Indicates the level of water quality abnormality. C Indicates the concentration of pollutants after fusion. Indicates the standard threshold for pollution concentration. z represents the set parameter. p indicates the setting parameter. q indicates the setting parameter.

[0012] Furthermore, in S5), the warning levels include Level 1, Level 2, and Level 3 warnings; Level 1 warning includes popping up a green warning pop-up window, displaying details of the polluted area and abnormal parameters, and automatically pushing prompt information to the operation and maintenance terminal; Level 2 warning includes triggering an audible and visual alarm, pushing the coordinates of the polluted area and preliminary handling suggestions, and recording the warning log; Level 3 warning includes activating a red emergency alarm, simultaneously pushing information to the operation and maintenance terminal, the management personnel's mobile APP, and the relevant regulatory department's system, and automatically generating an emergency response plan.

[0013] Furthermore, in S5, a time limit for handling early warning levels is set. If no handling result is received within the time limit, the early warning level will be automatically upgraded and the scope of notification will be expanded.

[0014] The advantages of this invention are: 1. This invention breaks through the limitations of single equipment. Through aerial and underwater collaborative monitoring and data fusion, it achieves broad-coverage initial screening and high-precision monitoring. Ultimately, it obtains the fused precise pollution concentration within the polluted area and the fused pollution concentration outside the polluted area. Compared with a single drone, the monitoring effect and efficiency are significantly improved; compared with a single bionic fish, the monitoring cycle for small lakes is shortened. 2. This invention integrates the entire process of monitoring, sampling, tracing, and early warning, enabling rapid response and precise control of pollution risks. It avoids resource waste caused by excessive early warnings, prevents the omission of major pollution hazards, and safeguards the safety of the aquatic ecological environment. The present invention provides a collaborative intelligent water quality monitoring method using a dual-platform drone and biomimetic fish, forming a closed-loop monitoring process. This method not only enables rapid response to pollution incidents and improves source tracing efficiency, but also shortens the monitoring cycle and avoids problems related to insufficient energy supply and endurance compared to a single biomimetic fish. Attached Figure Description

[0015] Figure 1 This is a flowchart of the UAV-bionic fish dual-platform collaborative intelligent water quality monitoring method of the present invention. Detailed Implementation

[0016] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0017] In the description of this invention, it should be understood that the terms "length", "width", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on the invention.

[0018] like Figure 1 As shown, a collaborative intelligent water quality monitoring method using a drone-bionic fish dual-platform includes the following steps.

[0019] S1) Build an unmanned aerial vehicle subsystem, a bionic fish roe system, an intelligent remote monitoring platform, and an early warning and response system.

[0020] The unmanned aerial vehicle (UAV) subsystem is a rotary-wing UAV equipped with a hyperspectral camera, a positioning module, and a communication module.

[0021] The biomimetic fish roe system is based on the trevally family fish and features a three-jointed, submersible structure equipped with a multi-parameter water quality sensor, an IMU inertial navigation unit, a GPS positioning unit, an infrared obstacle avoidance sensor, and a watertight sampling chamber.

[0022] The intelligent remote monitoring platform is based on MATLAB to build a visual interface and integrates the InfluxDB time series database with the SDN edge-cloud collaborative architecture.

[0023] S2) The UAV subsystem is used to conduct preliminary screening of water pollution areas, perform spectral inversion on the pre-screened pollution areas to obtain spectral data, and transmit the location and range information of the pre-screened pollution areas and spectral data back to the intelligent remote monitoring platform in real time, while receiving the platform's cruise path instructions. After receiving the information from the UAV subsystem, the intelligent remote monitoring platform generates a bionic fish precision detection task and sends the coordinates of the initial pollution area and the preset monitoring path instruction to the bionic fish subsystem.

[0024] The UAV subsystem conducts aerial patrols, using high-definition imaging and multi-sensor data collection to complete a full-area scan of the polluted water area, initially screening out suspected anomaly areas and marking location information. The hyperspectral camera in the UAV subsystem covers the visible to near-infrared band (400-1000nm), achieving initial screening of polluted water areas through hyperspectral scanning. In this embodiment, the hyperspectral camera acquires water reflectance spectral data at a spectral resolution of 2nm, generating 30 frames per second of images containing 256 spectral channels. It simultaneously records the GPS and BeiDou dual-mode positioning coordinates (accuracy ±5cm) and shooting timestamp for each frame. The UAV's built-in edge computing module preprocesses the raw spectral data in real time, eliminating the influence of illumination changes through radiometric correction, correcting flight attitude deviations through geometric correction, and removing salt-and-pepper noise through median filtering. Subsequently, the spectral anomaly index of each grid cell is extracted.

[0025] Based on the initial high-altitude screening results, the intelligent remote monitoring platform automatically plans the cruise path and coordinates the operation sequence of the drone and the bionic fish. The platform controls the drone's cruise path, the bionic fish's navigation and operation through command issuance, while simultaneously monitoring the equipment's operating status in real time.

[0026] S3) The bionic fish roe system receives the coordinates of the initial screening pollution area from the intelligent remote monitoring platform, achieves underwater ascent and descent through center of gravity lifting technology, arrives at the initial screening pollution area through autonomous navigation according to the preset monitoring path, collects actual measurement data using multi-parameter water quality sensors, retains water samples through a watertight sampling chamber, and transmits the collected actual measurement data back to the intelligent remote monitoring platform.

[0027] The biomimetic fish roe system uses a hierarchical data feature extraction method to collect measured data. The specific steps are as follows: S31) Layered raw data collection.

[0028] The bionic fish completes underwater mobile monitoring according to the preset monitoring path, and simultaneously collects raw data from the water quality sensing layer and the biological sensing layer. The two types of data are then bound with spatiotemporal tags according to the collection time and the underwater monitoring point, forming a raw dataset of water quality sensing and a raw dataset of biological sensing.

[0029] S32) Independent preprocessing of hierarchical data.

[0030] Outlier removal and missing value completion were performed on the raw data of the water quality sensing layer to eliminate equipment monitoring errors. The raw data of the biological sensing layer were processed in a targeted manner, including noise reduction and contour extraction based on the target contour image, filtering of abnormal frequencies of aquatic organism swimming frequency and time series normalization, to obtain standardized water quality sensing datasets and standardized biological sensing datasets.

[0031] S33) Initial extraction of hierarchical features.

[0032] Core water quality features, such as turbidity change trends, pH fluctuations, and real-time values ​​of indicators, are extracted from the standardized water quality sensing dataset to form an initial water quality sensing feature set. Core biological features, such as target contour morphology, contour area ratio, average swimming frequency of aquatic organisms, and frequency mutation characteristics, are extracted from the standardized biological sensing dataset to form an initial biological sensing feature set.

[0033] S34) Assigning weights to differentiated parameters.

[0034] Differential weights were assigned to each feature parameter in the initial feature set of water quality perception and the initial feature set of biological perception. The water quality perception layer focuses on the pollution characterization of basic water quality indicators, and core indicators are given high weights. The biological perception layer focuses on the response characteristics of biological behavior to pollution, and core features are given high weights. Non-core features are given low weights according to their characterization relevance, thus completing the construction of the weight system.

[0035] S4) The intelligent remote monitoring platform is based on the spectral data of the UAV and the measured data of the bionic fish. It uses spatiotemporal registration to fuse the data and build a pollution model in the polluted area to obtain the fused accurate pollution concentration in the polluted area. Then, with the pollution source as the center, the diffusion time is set to build a pollution diffusion model outside the polluted area to obtain the fused pollution concentration at a certain moment when the pollution has diffused outside the polluted area.

[0036] The specific steps for fusion using spatiotemporal registration include: S41) Heterogeneous raw data acquisition and spatiotemporal labeling.

[0037] The biomimetic fish simultaneously collects underwater water quality and biological stratification monitoring data and binds its own underwater relative coordinates and collection timestamps. The drone collects water surface and airspace pollution-related data and binds them to GPS absolute coordinates and collection timestamps, forming a fish-drone heterogeneous dataset with original spatiotemporal labels. At the same time, the basic parameters of the two devices' collection frequency and positioning accuracy are recorded.

[0038] S42) Unified initialization of the spacetime reference.

[0039] A fish-drone collaborative spatiotemporal reference system was established, a unified spatial reference coordinate system and time synchronization reference were determined, and the absolute GPS coordinates of the drone were extracted as the spatial reference and the high-precision time synchronization of the drone as the time reference, so as to establish a unified reference for subsequent coordinate transformation and time compensation.

[0040] S43) Spatial coordinate transformation adaptation.

[0041] By introducing a spatial coordinate transformation coefficient, a mapping relationship is established between the absolute GPS coordinates of the UAV and the underwater relative coordinates of the bionic fish. The underwater relative coordinates of the bionic fish are converted to the GPS spatial reference coordinate system of the UAV through the transformation coefficient, eliminating the spatial positioning dimension difference between the fish and the UAV across media, and realizing the unified matching of the spatial positions of the monitoring data of the two.

[0042] S44) Time synchronization compensation calibration.

[0043] By introducing a time synchronization compensation factor, the collection time of the bionic fish and the drone is compensated and corrected based on the data collection timestamp deviation, data transmission delay, and device response time difference between the fish and the drone. This eliminates the time asynchrony problem across devices and media, and achieves accurate alignment of the monitoring data of the two in the time dimension.

[0044] S45) Initial verification of spatiotemporal registration.

[0045] For the fish-machine heterogeneous data that have completed spatial coordinate transformation and time synchronization compensation, data matching is performed according to the same monitoring area and time period to verify the consistency of spatiotemporal registration results. Abnormal data with spatial misalignment and time deviation caused by coordinate transformation and time compensation are eliminated to form an initial registered heterogeneous dataset.

[0046] The pollution model is expressed by the following formula. In the formula, This indicates the precise concentration of contaminants within the contaminated area after fusion. This indicates the spectral pollution concentration within the polluted area obtained from the spectral inversion by the drone. This indicates the measured pollution concentration within the contaminated area collected by the bionic fish sensor. Indicates the fusion weight. Indicates the fusion weight. For environmental correction items, The standard deviation of the pollution concentration obtained from the UAV spectral inversion reflects the degree of fluctuation and uncertainty of this set of spectral inversion data. The standard deviation of the pollution concentration measured by the bionic fish sensor reflects the degree of fluctuation and uncertainty of the measured data from this set of sensors.

[0047] The pollution diffusion model is expressed by the following equation. In the formula, This indicates the coordinates outside the contaminated area at time t. The concentration of contaminants after fusion at ( ) Indicates the coordinates of the pollution source ( , Precise pollution concentration at the location after fusion. Indicates the water diffusion coefficient. Indicates diffusion time. ( , The coordinates of the pollution source are indicated by drones for initial screening and bionic fish for fixed-point detection.

[0048] S5) The early warning response system generates a complete monitoring report based on the fusion analysis results. The monitoring report includes details of the polluted area, the level of water quality anomaly, and the trend of pollution spread. The system automatically triggers an early warning level based on the level of water quality anomaly and pushes the coordinates of the polluted area to the operation and maintenance terminal to complete the early warning response closed loop.

[0049] The water quality anomaly level is expressed by the following formula. In the formula, L Indicates the level of water quality abnormality. C Indicates the concentration of pollutants after fusion. Indicates the standard threshold for pollution concentration. z represents the set parameter. p indicates the setting parameter. q indicates the setting parameter.

[0050] The warning levels include Level 1 (mild anomaly), Level 2 (moderate anomaly), and Level 3 (severe anomaly). For Level 1 warnings, the warning response system displays a green warning pop-up window, showing details of the contaminated area and abnormal parameters, and automatically pushes a notification to the operation and maintenance terminal. For Level 2 warnings, the warning response system triggers an audible and visual alarm, pushes the coordinates of the contaminated area and preliminary handling suggestions, and records the warning log. For Level 3 warnings, the warning response system activates a red emergency alarm, simultaneously pushes information to the operation and maintenance terminal, the administrator's mobile app, and relevant regulatory department systems, and automatically generates an emergency response plan.

[0051] Once an alert is triggered, the alert response system tracks the progress of operation and maintenance, receives feedback on the handling results, and updates the alert status (not handled, handling in progress, completed), forming a closed-loop response process of "alert, push, handling and feedback".

[0052] In addition, a time limit for handling early warning levels is set (24 hours for Level 1 warning, 12 hours for Level 2 warning, and 2 hours for Level 3 warning). If no response is received within the time limit, the system will automatically upgrade the warning level (e.g., if a Level 2 warning is upgraded to a Level 3 warning after 1 hour) and expand the scope of notification (e.g., notify the superior leader).

[0053] After the maintenance personnel complete the treatment, they need to upload on-site photos and water quality retest data (which can be collected by portable sensors and synchronized to the intelligent remote monitoring platform in real time). The intelligent remote monitoring platform compares the changes in pollution concentration before and after the treatment. If the pollution concentration returns to the normal range, the treatment is deemed effective, and the closed-loop status is updated to "completed". If the pollution concentration does not improve significantly, the maintenance personnel are prompted to adjust the treatment plan and a secondary monitoring task is triggered (instructing the drone and bionic fish to re-test).

[0054] Once the closed loop is complete, the system automatically integrates the early warning trigger time, push records, handling process, and result verification data into an early warning response file, which is then stored in association with the corresponding monitoring report to support subsequent auditing and accountability. This invention achieves rapid response and precise control of pollution risks through tiered early warning and closed-loop management, which avoids the waste of resources caused by excessive early warning, prevents the omission of major pollution hazards, and ensures the safety of the aquatic ecological environment.

[0055] The present invention provides a collaborative intelligent water quality monitoring method using a dual-platform drone and biomimetic fish, forming a closed-loop monitoring process. This method not only enables rapid response to pollution incidents and improves source tracing efficiency, but also shortens the monitoring cycle and avoids problems related to insufficient energy supply and endurance compared to a single biomimetic fish.

[0056] The above embodiments are preferred embodiments of the present invention, but the embodiments of the present invention are not limited to the above embodiments. Any changes, modifications, substitutions, combinations, or simplifications made without departing from the spirit and principle of the present invention shall be considered equivalent substitutions and shall be included within the protection scope of the present invention.

Claims

1. A collaborative intelligent water quality monitoring method using a dual-platform system of unmanned aerial vehicles (UAV) and biomimetic fish, characterized in that... Includes the following steps: S1) Build an unmanned aerial vehicle (UAV) subsystem, a bionic fish roe system, an intelligent remote monitoring platform, and an early warning and response system; S2) The UAV subsystem is used to conduct preliminary screening of water pollution areas, perform spectral inversion on the pre-screened pollution areas to obtain spectral data, and transmit the location and range information of the pre-screened pollution areas and spectral data back to the intelligent remote monitoring platform in real time, while receiving the platform's cruise path instructions. After receiving the information transmitted back from the UAV subsystem, the intelligent remote monitoring platform generates a bionic fish precision detection task and sends the coordinates of the initial screening pollution area and the preset monitoring path instruction to the bionic fish subsystem. S3) The bionic fish roe system receives the coordinates of the initial screening contamination area from the intelligent remote monitoring platform, arrives at the initial screening contamination area according to the preset monitoring path, collects actual measurement data, and transmits the collected actual measurement data back to the intelligent remote monitoring platform. S4) The intelligent remote monitoring platform is based on the spectral data of the UAV and the measured data of the bionic fish. It uses spatiotemporal registration to fuse the data, build a pollution model in the polluted area, and obtain the fused accurate pollution concentration in the polluted area. Then, with the pollution source as the center, a diffusion time is set to build a pollution diffusion model outside the polluted area, and obtain the fused pollution concentration at a certain moment when the pollution has diffused outside the polluted area. S5) The early warning response system generates a complete monitoring report based on the fusion analysis results. The monitoring report includes details of the polluted area, the level of water quality anomaly, and the trend of pollution spread. The system automatically triggers an early warning level based on the level of water quality anomaly, and pushes the coordinates of the polluted area to the operation and maintenance terminal to complete the early warning response closed loop.

2. The UAV-bionic fish dual-platform collaborative intelligent water quality monitoring method according to claim 1, characterized in that: In S1), the UAV subsystem is equipped with a hyperspectral camera, a positioning module and a communication module; the biomimetic fish roe system is equipped with a multi-parameter water quality sensor, an IMU inertial navigation unit, a GPS positioning unit, an infrared obstacle avoidance sensor and a watertight sampling chamber.

3. The UAV-bionic fish dual-platform collaborative intelligent water quality monitoring method according to claim 2, characterized in that: In S3), the biomimetic fish roe system uses a hierarchical data feature extraction method to collect measured data. The specific steps are as follows: S31) The bionic fish completes underwater mobile monitoring according to the preset monitoring path, and simultaneously collects raw data from the water quality sensing layer and raw data from the biological sensing layer. The two types of data are bound to spatiotemporal labels according to the collection time and underwater monitoring point to form a raw dataset of water quality sensing and a raw dataset of biological sensing. S32) Remove outliers and fill in missing values ​​in the raw data of the water quality sensing layer to eliminate equipment monitoring errors; perform targeted processing on the raw data of the biological sensing layer, complete noise reduction and contour extraction according to the target contour image, complete the filtering of abnormal frequencies of aquatic organism swimming frequency and time series normalization, and obtain standardized water quality sensing dataset and standardized biological sensing dataset respectively. S33) Extract core water quality features such as turbidity change trend, pH fluctuation range and real-time index values ​​from the standardized water quality sensing dataset to form the initial water quality sensing feature set; extract core biological features such as target contour morphology features, contour area ratio, average swimming frequency of aquatic organisms and frequency mutation features from the standardized biological sensing dataset to form the initial biological sensing feature set. S34) Differentiated weights are assigned to each feature parameter in the initial feature set extracted from water quality perception and the initial feature set extracted from biological perception. The water quality perception layer focuses on the pollution characterization of basic water quality indicators, with core indicators assigned high weights. The biological perception layer focuses on the response characteristics of biological behavior to pollution, with core features assigned low weights. High-weighted features are assigned low weights based on their relevance to other features, thus completing the weighting system.

4. The UAV-bionic fish dual-platform collaborative intelligent water quality monitoring method according to claim 1, characterized in that: In S4), the specific steps for fusion using spatiotemporal registration include: S41) The bionic fish simultaneously collects underwater water quality and biological stratification monitoring data and binds its own underwater relative coordinates and collection timestamps. The drone collects water surface and airspace pollution-related data and binds GPS absolute coordinates and collection timestamps to form a fish-drone heterogeneous dataset with original spatiotemporal labels. At the same time, the basic parameters of the two devices' collection frequency and positioning accuracy are recorded. S42) Establish a fish-machine collaborative spatiotemporal reference system, determine a unified spatial reference coordinate system and time synchronization reference, extract the absolute GPS coordinates of the UAV as the spatial reference and the high-precision time synchronization of the UAV as the time reference, and establish a unified reference for subsequent coordinate transformation and time compensation. S43) Introduce a spatial coordinate transformation coefficient to establish a mapping relationship between the absolute GPS coordinates of the UAV and the underwater relative coordinates of the bionic fish. Convert the underwater relative coordinates of the bionic fish to the GPS spatial reference coordinate system of the UAV through the transformation coefficient, eliminate the spatial positioning dimension difference between the fish and the UAV across media, and achieve unified matching of the spatial positions of the monitoring data of the two. S44) Introducing a time synchronization compensation factor, based on the data collection timestamp deviation, data transmission delay, and device response time difference of the fish-machine equipment, to compensate and correct the data collection time of the bionic fish and the drone, eliminate the time asynchrony problem across devices and media, and achieve accurate alignment of the monitoring data of the two in the time dimension. S45) For the fish-machine heterogeneous data that have completed spatial coordinate transformation and time synchronization compensation, data matching is performed according to the same monitoring area and time period to verify the consistency of spatiotemporal registration results. Abnormal data with spatial misalignment and time deviation caused by coordinate transformation and time compensation are eliminated to form an initial registration heterogeneous dataset.

5. The UAV-bionic fish dual-platform collaborative intelligent water quality monitoring method according to claim 4, characterized in that: In S4), the pollution model is expressed by the following formula: In the formula, This indicates the precise concentration of contaminants within the contaminated area after fusion. This indicates the spectral pollution concentration within the polluted area obtained from the spectral inversion by the drone. This indicates the measured pollution concentration within the contaminated area collected by the bionic fish sensor. Indicates the fusion weight. Indicates the fusion weight. For environmental correction items, This represents the standard deviation of the pollution concentration obtained from the spectral inversion of the UAV. This represents the standard deviation of the pollution concentration measured by the bionic fish sensor.

6. The UAV-bionic fish dual-platform collaborative intelligent water quality monitoring method according to claim 5, characterized in that: In S4), the pollution diffusion model is expressed by the following formula: In the formula, This indicates the coordinates outside the contaminated area at time t. The concentration of contaminants after fusion at ( ) Indicates the coordinates of the pollution source ( , Precise pollution concentration at the location after fusion. Indicates the water diffusion coefficient. Indicates diffusion time. ( , The coordinates of the pollution source are indicated by drones for initial screening and bionic fish for fixed-point detection.

7. The UAV-bionic fish dual-platform collaborative intelligent water quality monitoring method according to claim 1, characterized in that: In S5), the water quality anomaly level is expressed by the following formula. In the formula, L indicates the level of water quality abnormality. C represents the contamination concentration after fusion. Indicates the standard threshold for pollution concentration. z represents the set parameter. p indicates the setting parameter. q indicates the setting parameter.

8. The UAV-bionic fish dual-platform collaborative intelligent water quality monitoring method according to claim 7, characterized in that: In S5), the warning levels include Level 1 warning, Level 2 warning, and Level 3 warning; Level 1 warning includes popping up a green warning pop-up window, displaying details of the polluted area and abnormal parameters, and automatically pushing prompt information to the operation and maintenance terminal; The Level 2 warning includes triggering an audible and visual alarm, pushing the coordinates of the polluted area and preliminary treatment suggestions, and recording the warning log; the Level 3 warning includes activating a red emergency alarm, simultaneously pushing information to the operation and maintenance terminal, the mobile APP of the management personnel and the relevant regulatory department system, and automatically generating an emergency response plan.

9. The UAV-bionic fish dual-platform collaborative intelligent water quality monitoring method according to claim 8, characterized in that: In S5, a time limit for handling early warning levels is set. If no handling result is reported within the time limit, the early warning level will be automatically upgraded and the scope of push notification will be expanded.

Images