Intelligent water quality monitoring method, system and computer equipment for collaborative sampling of bionic fish schools and drones

Abstract

The present invention relates to the field of environmental monitoring technology, and specifically to an intelligent water quality monitoring method, system, and computer equipment using collaborative sampling of bionic fish swarms and drones. The method includes: utilizing GIS data to divide monitoring grid cells, establishing a hierarchical communication architecture, dynamically deploying bionic fish swarms for real-time water quality monitoring, and optimizing resource allocation through fuzzy rule scheduling; integrating historical pollution event data and real-time sensor data to construct a water quality evaluation index system, employing a Nash equilibrium strategy to determine optimal weights and generate a dynamic pollution situation cloud map; constructing a poor-state fingerprint library to rapidly identify pollution events and trigger a two-level response mechanism; and finally, combining bionic fish sampling data with drone pre-diffusion coordinates to dynamically simulate the effects of remediation measures. This method achieves efficient integrated air-water monitoring, improves the accuracy and timeliness of water quality anomaly detection, and provides comprehensive technical support for aquatic environmental protection.

Description

Technical Field

[0001] The present invention relates to the field of environmental monitoring technology, and in particular to an intelligent water quality monitoring method, system, and computer equipment for collaborative sampling of bionic fish schools and drones. Background Art

[0002] With the increasing prevalence of environmental pollution, water quality monitoring has become a crucial tool for protecting the aquatic ecosystem. The continuous development of water quality monitoring technology has exposed a series of significant shortcomings in many past and current solutions. For example, patent document "CN115792165A - A Method and System for Intelligent Monitoring of Environmental Water Quality" demonstrates that this technology primarily utilizes a monitoring module, a communication module, and a display module to achieve water quality monitoring. The monitoring module comprises a water quality detection unit, a processor, and an edge computing module. The water quality detection unit uses a variety of sensors, such as temperature, turbidity, and pH, to collect water quality information. The processor then transmits this information to the edge computing module for screening and processing, ultimately computing it on a cloud computing platform and displaying it on a display terminal. However, this solution has significant shortcomings. While it improves some of the issues associated with manual monitoring, it is limited by a fixed sensor deployment model. Fixed sensors struggle to provide flexible and comprehensive coverage in vast, complex, and ever-changing waters, such as large river basins and offshore areas. They are unable to capture dynamic changes in water quality in a timely manner. Especially in areas with turbulent currents and complex terrain, monitoring data is prone to omissions or deviations, significantly compromising the integrity and accuracy of monitoring.

[0003] In recent years, with the advancement of artificial intelligence, the Internet of Things, and robotics, water quality monitoring using intelligent devices such as drones and bionic fish has become a research hotspot. However, existing technologies lack a robust synergy between drones and bionic fish, failing to fully leverage the strengths of both for efficient and accurate water quality monitoring. For example, challenges remain, such as dynamically allocating monitoring resources based on the geographic information system and hydrological characteristics of target waters, optimizing sampling strategies based on historical pollution incident data, and rapidly identifying and responding to sudden pollution incidents. Summary of the Invention

[0004] To address these issues, this paper proposes an intelligent water quality monitoring method and computer equipment that uses collaborative sampling between bionic fish swarms and drones. Through layered collaboration between drones and bionic fish swarms, this method optimizes the matching of monitoring resources with hydrological characteristics, improving the real-time, accurate, and intelligent nature of water quality monitoring. This enhances the ability to respond to sudden pollution incidents, enabling rapid response, dynamic resource scheduling, and three-dimensional situation visualization.

[0005] The technical solutions of the present invention are as follows:

[0006] One of the technical solutions of the present invention is to provide an intelligent water quality monitoring method using bionic fish schools and drones for collaborative sampling, comprising:

[0007] Based on the geographic information system data of the target waters, monitoring grid units are divided and a hierarchical communication architecture is established between drones and bionic fish schools. Based on the environmental parameters of the monitoring grid units, bionic fish schools are dynamically deployed. UAV remote sensing and bionic fish school sensors monitor water quality anomalies in real time. Fuzzy rules are used to schedule bionic fish schools to achieve initial adaptation of monitoring resources and hydrological characteristics.

[0008] The drone collects historical pollution event data and real-time fish sensor data to construct a water quality evaluation index system. Using a Nash equilibrium strategy, based on the optimal comprehensive weights of pollutant diffusion rate, concentration gradient change rate, and ecotoxicity index in the game theory evaluation index system, a dynamic pollution situation cloud map is generated based on the weighted results, directing fish schools to prioritize patrolling high-weighted areas.

[0009] A bad fingerprint library is formed based on stored historical pollution event peak data. When fish sensors in high-weight areas detect that the real-time data matches the fingerprint library features beyond the limit, a two-level response is immediately triggered: 1) the bionic fish automatically increases the encryption sampling frequency; 2) the drone sends the pollution pre-diffusion coordinates to the control center.

[0010] By integrating the bionic fish sampling data and the pollution pre-diffusion coordinates obtained by drones, the water quality evolution trajectory after the treatment measures are taken is dynamically simulated to determine the feasibility of the treatment measures.

[0011] As a further option of the present method, the layered communication architecture includes a three-order coordination mechanism of an air communication layer, a surface relay layer, and an underwater communication layer;

[0012] Among them, the air communication layer builds a dynamic topology network through drones, the surface relay layer uses adaptive beamforming technology to achieve cross-media signal conversion, and the underwater communication layer executes underwater acoustic communication and energy self-coordination protocols based on a distributed sensing network of bionic fish schools.

[0013] As a further option of this method, the fuzzy rule scheduling mechanism adopts a multi-input and multi-output fuzzy inference system. The input variables include pollutant concentration gradient, water temperature mutation rate, dissolved oxygen vertical stratification coefficient and turbidity spatiotemporal variation coefficient, and the output variables are the three-dimensional swimming trajectory offset of the bionic fish school and the sampling frequency domain gain coefficient.

[0014] As a further option of this method, the game model constructed by the Nash equilibrium strategy includes three types of decision-making entities in non-cooperative competition relationships: pollutant migration dynamics entity, ecotoxicity response entity, and monitoring resource constraint entity;

[0015] The form of the game model is as follows:

[0016] ;

[0017] in, 、 、 Represent the weights of pollutant diffusion rate, concentration gradient change rate and ecotoxicity index respectively. and is the adjustment factor;

[0018] The strategy space of each agent is defined as the probability distribution of weight allocation. The utility function comprehensively considers the timeliness of pollution warning, biological toxicity exposure risk, and energy cost constraints. The Pareto optimal solution of the indicator weight is obtained by solving the mixed strategy Nash equilibrium.

[0019] The calculation rules are as follows:

[0020] ;

[0021] ;

[0022] in, is the particle velocity, is the position of the particle, i.e., the weight configuration, is the historical optimal solution of the particle, is the global optimal solution, is the inertia weight, 、 is the acceleration factor, 、 is a random number.

[0023] As a further option of this method, the method for generating the dynamic pollution situation cloud map includes:

[0024] The optimal weight results obtained by solving the Nash equilibrium strategy are mapped to each monitoring grid unit to generate pollution risk values. Based on the distribution of pollution risk values, the trapezoidal membership function is used to divide the pollution risk levels.

[0025] GIS technology is used to map the pollution risk level onto a two-dimensional map of the target waters, and a situation cloud map is used to visualize the pollution risk distribution of the dynamic pollution situation cloud map.

[0026] As a further option of this method, the construction and matching method of the inferior fingerprint library includes:

[0027] Extract pollutant features from historical pollution event data. After feature extraction is completed, encode the pollutant features and build a bad fingerprint library.

[0028] The matching degree is evaluated by calculating the similarity of feature vectors and identifying pollution events based on thresholds.

[0029] As a further option of this method, the simulation formula for the water quality evolution trajectory is:

[0030] ;

[0031] in, For coordinates In time The concentration of pollutants, is the initial pollutant concentration, is the water velocity field, is the diffusion coefficient tensor, is the source term of the control measures, is the parameter vector of the governance scheme.

[0032] As a further option of this method, the feasibility assessment of the control measures includes: real-time comparison of the pollutant concentration gradient change rate under different control schemes with the pollution risk threshold defined by the Nash equilibrium weight, and identification of the intervention effect of the control measures on high-weight areas; when the simulation trajectory shows that the core indicators converge to the safe membership interval within the preset time window, the control measures are judged to be effective.

[0033] The second technical solution of the present invention is to provide an intelligent water quality monitoring system that uses bionic fish schools and drones for collaborative sampling, including:

[0034] The grid monitoring configuration unit is configured to: access geographic information system data to divide the monitoring grid units; build a layered communication architecture including an air communication layer, a surface relay layer, and an underwater communication layer; and generate dynamic deployment instructions for bionic fish schools based on the water depth gradient and flow velocity field;

[0035] The fuzzy scheduling control unit is connected to the bionic fish sensor network and is configured to: receive real-time data on water quality parameters, execute the fuzzy rule base to output bionic fish aggregation density correction instructions; and send path adjustment signals to the bionic fish through a layered communication architecture;

[0036] The Nash equilibrium decision engine is configured to: integrate historical pollution event data with real-time sensor data to construct an evaluation index system for pollutant diffusion rate, concentration gradient change rate, and ecotoxicity index; solve Pareto optimal weights based on a game model; generate a dynamic pollution situation cloud map and mark high-weight patrol areas;

[0037] The bad response execution module includes: a fingerprint library memory, which stores historical pollution event peak data encoded as multi-dimensional feature vectors; a real-time matcher, which calculates the weighted cosine similarity between the high-weight area sensor data and the fingerprint library;

[0038] Two-level response trigger, when the matching degree exceeds the limit: (a) send encrypted sampling instructions to the bionic fish school; (b) send pollution pre-diffusion coordinate generation instructions to the drone;

[0039] The control simulation verification platform is configured to: access encrypted sampling data from bionic fish schools and pollution pre-diffusion coordinates; parameterize control measures and feed them into a water quality evolution model for dynamic simulation; and output a control feasibility signal based on the convergence of the pollutant concentration gradient change rate.

[0040] The layered communication hardware architecture includes: an aerial communication layer: drones equipped with remote sensing equipment establish 5G / satellite backhaul links with ground control centers; a surface relay layer: self-organizing networking buoys equipped with air-water dual-frequency communication repeaters; and an underwater communication layer: bionic fish schools with built-in underwater acoustic modems form a dynamic routing sensor network.

[0041] The control center visualization terminal is configured to: render a dynamic pollution situation cloud map in real time, use gradient colors to mark pollution risk levels; and display the governance simulation trajectory and feasibility assessment results.

[0042] A third technical solution of the present invention is to provide a computer device, comprising: a processor and a memory, wherein the memory stores at least one instruction, at least one program, a code set or an instruction set, and the at least one instruction, the at least one program, the code set or the instruction set is loaded and executed by the processor to implement the intelligent water quality monitoring method of collaborative sampling of bionic fish schools and drones as described in one of the technical solutions.

[0043] The beneficial effects brought about by the technical solutions provided in the embodiments of the present application include at least the following beneficial effects:

[0044] This solution builds a multi-dimensional dynamic monitoring network by deeply integrating geographic information system (GIS) data, drone remote sensing information, and real-time biomimetic fish swarm sensor data. Its core advantage lies in its layered communication architecture: the air communication layer enables global situational awareness, the surface relay layer overcomes the bottleneck of cross-media signal conversion, and the underwater communication layer leverages the biomimetic fish swarm distributed sensor network to form an integrated air-space-ground collaborative monitoring system. This significantly improves data collection efficiency and coverage in complex water environments.

[0045] A Nash equilibrium strategy based on game theory innovatively addresses the challenge of dynamically allocating weights to water quality assessment indicators. By modeling non-cooperative games based on pollutant diffusion rates, concentration gradient rates, and ecotoxicity indices, Pareto-optimal decision-making is achieved under multi-objective conflicts. Combined with a fuzzy rule-based scheduling mechanism, this system can drive bionic fish schools to adaptively adjust sampling density and swimming paths in real time, precisely matching monitoring resources to hydrological characteristics and forming an intelligent closed-loop system of "risk prediction, dynamic response, and continuous optimization."

[0046] By building a database of historical pollution event fingerprints, achieving sublinear matching efficiency, and triggering a two-tiered response mechanism, the system simultaneously initiates high-frequency, intensified sampling by biomimetic fish schools and coordinate prediction of pollution spread from drones. Combined with water quality evolution simulations, this system not only dynamically predicts pollutant migration trajectories after the implementation of control measures, but also, by incorporating parameterized models of control measures (such as nonlinear terms for adsorption materials and correction terms for physical barrier boundaries), provides a high-precision digital twin verification platform for pollution control, enabling closed-loop control of the entire "monitoring-early warning-simulation-decision-making" process. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 This is a schematic diagram of the overall process of the intelligent water quality monitoring method with collaborative sampling of bionic fish schools and drones;

[0048] Figure 2 A detailed flow chart of the steps of the intelligent water quality monitoring method S100 for collaborative sampling of bionic fish schools and drones;

[0049] Figure 3 A detailed flow chart of the steps S200 of the intelligent water quality monitoring method for collaborative sampling of bionic fish schools and drones;

[0050] Figure 4 A detailed flow chart of the steps S300 of the intelligent water quality monitoring method for collaborative sampling of bionic fish schools and drones;

[0051] Figure 5 Detailed flow chart of step S400 of the intelligent water quality monitoring method for collaborative sampling of bionic fish schools and drones. DETAILED DESCRIPTION

[0052] The following will be combined with the accompanying drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making any creative efforts are within the scope of protection of this application.

[0053] Among existing water quality monitoring technologies, traditional manual sampling methods have problems such as limited monitoring range, poor real-time performance, and high labor costs. Fixed-point sensor monitoring networks are unable to flexibly adapt to the dynamic changes in complex water environments and are insufficiently responsive to sudden pollution incidents. Although single drone monitoring has a certain degree of maneuverability, it is limited by its endurance and monitoring depth, and cannot comprehensively obtain water quality data at different levels of the water body. In addition, existing technologies lack intelligent collaborative mechanisms for the dynamic allocation of monitoring resources, pollution risk assessment, and early warning, making it difficult to achieve accurate and efficient monitoring of water quality conditions. To address the above issues, please refer to Figure 1, which illustrates an intelligent water quality monitoring method for collaborative sampling of a bionic fish school and a drone, provided by an embodiment of the present invention, comprising:

[0054] S100: Based on the geographic information system data of the target water area, the monitoring grid units are divided and a hierarchical communication architecture between drones and bionic fish schools is established. The bionic fish schools are dynamically deployed according to the environmental parameters of the monitoring grid units. The drone remote sensing and bionic fish school sensors monitor water quality anomalies in real time. The bionic fish schools are dispatched through fuzzy rules to achieve initial adaptation of monitoring resources and hydrological characteristics.

[0055] S200: The drone collects historical pollution event data and real-time fish sensor data to construct a water quality evaluation index system. Using a Nash equilibrium strategy, based on the optimal comprehensive weights of pollutant diffusion rate, concentration gradient change rate, and ecotoxicity index in the game theory evaluation index system, it generates a dynamic pollution situation cloud map based on the weighted results and directs fish schools to prioritize patrolling high-weighted areas.

[0056] S300: A bad fingerprint library is formed based on the stored peak data of historical pollution events. When the fish school sensor in the high-weight area detects that the real-time data matches the fingerprint library features beyond the limit, a two-level response is immediately triggered: 1) The bionic fish school automatically increases the encryption sampling frequency; 2) The drone sends the pollution pre-diffusion coordinates to the control center.

[0057] S400: Connects bionic fish sampling data and pollution pre-diffusion coordinates obtained by drones to dynamically simulate the water quality evolution trajectory after treatment measures are taken to determine the feasibility of the treatment measures.

[0058] S100-S400 proposes a collaborative monitoring technology solution that combines bionic fish schools and drones. Through GIS data grid division, fuzzy rule scheduling, Nash equilibrium weight evaluation and dynamic pollution simulation, it achieves efficient and accurate water quality monitoring and treatment optimization. It has the significant advantages of intelligent resource allocation, rapid pollution response and scientific decision-making.

[0059] The specific plan is as follows:

[0060] In the intelligent water quality monitoring method using bionic fish schools and drones for collaborative sampling, S100 integrates geographic information system (GIS) data, a hierarchical communication architecture, and a dynamic resource allocation strategy to build a basic framework for collaborative water quality monitoring. S100 includes: GIS-based pollution risk and hydrological characteristic analysis to achieve adaptive division and dynamic optimization of the monitoring grid. A hierarchical communication architecture design addresses the challenges of stability and real-time data transmission across media. A bionic fish-like depth-velocity collaborative allocation mechanism uses multi-parameter sensors to perceive water quality in real time and dynamically adjust sampling density. UAV remote sensing and fuzzy rule scheduling, combined with remote sensing anomaly identification and bionic fish-like aggregation strategies, achieve initial adaptation of monitoring resources to hydrological characteristics.

[0061] Please refer to Figure 2 , which shows a flow chart of an exemplary water quality intelligent monitoring method S100 of the present application for collaborative sampling of bionic fish schools and drones, including:

[0062] S110: Obtain geographic information system (GIS) data of the target water area and divide the monitoring grid units based on the GIS data.

[0063] The acquisition of GIS data mainly relies on remote sensing images, topographic maps, hydrological monitoring data and historical pollution event databases.

[0064] Specifically, remote sensing images, acquired through satellite or drone aerial photography, capture high-resolution information on the distribution of land and water bodies, which can be used to identify water boundaries, river network structures, and surrounding land use types. Topographic maps provide information on elevation changes in target waters, enabling analysis of flow direction, water depth distribution, and the impact of topography on pollutant diffusion. Hydrological monitoring data, including flow velocity, water depth, and temperature, typically originate from ground-based monitoring stations or drone-mounted LiDAR scanning systems. Historical pollution event databases record the locations, pollutant types, peak concentrations, and diffusion paths of past pollution incidents, providing a reference for monitoring grid division and risk prediction.

[0065] After completing GIS data acquisition and preprocessing, the monitoring grid unit division phase begins. This phase divides the target water area into several monitoring grid units and dynamically adjusts the grid layout based on hydrological characteristics and pollution risks to optimize the allocation of monitoring resources.

[0066] In one possible implementation, a spatial partitioning method is used to divide the monitoring grid cells, and hydrological characteristics are used to optimize the grid layout. The specific process is as follows:

[0067] a. Convert remote sensing images and topographic map data into vector data to extract water body boundaries, river network distribution, and key monitoring points.

[0068] b. Use a spatial partitioning method to divide the target water area into a number of initial grid cells. Exemplarily, the spatial partitioning method uses Thiessen polygons or a rasterization method. Thiessen polygons are suitable for irregular water areas, while rasterization is suitable for regularly distributed monitoring areas.

[0069] c. Dynamically adjust the initial grid based on the hydrological model. For example, in areas with higher pollution risk, the grid cell area can be reduced to 50m×50m, while in areas with lower pollution risk, the grid cell area can be expanded to 200m×200m.

[0070] S120: Establish a hierarchical communication architecture between drones and bionic fish swarms.

[0071] The layered communication architecture between the drone and the bionic fish school consists of an aerial communication layer, a surface relay layer, and an underwater communication layer, ensuring efficient data interaction between the bionic fish school and the drone.

[0072] Specifically, the aerial communication layer consists of drones, responsible for high-altitude remote sensing data collection, global monitoring grid generation, and communication with the ground control center. The surface relay layer, comprised of surface base stations and buoy-type communication nodes, serves as a bridge between aerial and underwater communications. The underwater communication layer, comprised of bionic fish swarms and their onboard sensor networks, is responsible for underwater environment data collection and local communications.

[0073] S130: Dynamically deploying the bionic fish school based on the monitored grid unit environmental parameters.

[0074] After completing the monitoring grid division and establishing the communication architecture, the dynamic allocation phase of the bionic fish swarm begins. Based on environmental parameters, the bionic fish swarm is rationally deployed to optimize data collection coverage and monitoring accuracy.

[0075] In one possible implementation, the deployment of bionic fish schools is based on water depth, an environmental parameter. Water depth directly affects the vertical distribution of pollutants, so the deployment of bionic fish schools needs to be combined with water depth data. Specific deployment strategies include:

[0076] Using GIS data from topographic maps and hydrological monitoring data, the water depth distribution of each grid cell is calculated and the deployment density of the bionic fish school is adjusted accordingly. In deep water areas, the bionic fish school is deployed at a relatively low density to reduce energy consumption and avoid oversampling; in shallow water areas, the bionic fish school is deployed at a higher density to improve the spatiotemporal resolution of monitoring.

[0077] In another possible implementation, the deployment of bionic fish schools is based on water velocity, an environmental parameter. Water velocity directly affects the diffusion path and spatial distribution of pollutants, so the distribution of bionic fish schools needs to be dynamically adjusted based on flow velocity data. Specific deployment strategies include:

[0078] In river areas with faster flow rates, the deployment density of bionic fish schools is higher to improve the ability to track the migration of pollutants; in lake or reservoir areas with slower flow rates, the deployment density of bionic fish schools can be appropriately reduced to extend the endurance of the equipment.

[0079] S140: Scheduling bionic fish schools via fuzzy rules.

[0080] The input variables for fuzzy rule scheduling come from sensors carried by the bionic fish and remote sensing data from drones. Specifically, they include pollutant concentration, water temperature, dissolved oxygen (DO) content, pH value, conductivity, and turbidity.

[0081] In a possible implementation, the rule base of the fuzzy rules is determined based on the environmental parameters and is composed of a series of "IF-THEN" rules for describing the mapping relationship between the environmental parameters and the bionic fish school scheduling instructions.

[0082] Example:

[0083] IF the pollutant concentration is high AND the diffusion rate is fast THEN the density of bionic fish gathering increases significantly;

[0084] IF the pollutant concentration is medium AND the water toxicity index is high THEN the bionic fish aggregation density increases moderately;

[0085] IF the pollutant concentration is low AND the diffusion rate is slow THEN the bionic fish aggregation density remains unchanged.

[0086] UAV remote sensing and bionic fish sensors monitor water quality anomalies in real time. Based on fuzzy logic scheduling rules and combined with multi-source environmental parameters, the distribution density and swimming path of bionic fish schools are dynamically adjusted to achieve initial adaptation of monitoring resources and hydrological characteristics.

[0087] In the intelligent water quality monitoring method that uses collaborative sampling of bionic fish schools and drones, the S200 focuses on the construction and optimization of a water quality evaluation index system. First, the drone and bionic fish schools collaborate to collect historical pollution event data and real-time water quality sensor data, and then build an evaluation system based on the three core indicators of pollutant diffusion rate, concentration gradient change rate, and ecotoxicity index. Based on the Nash equilibrium strategy of game theory, the optimal comprehensive weight of each indicator is dynamically solved, and the weight is adaptively adjusted through the particle swarm optimization algorithm to ensure the accuracy of the evaluation under different pollution scenarios. Subsequently, the pollution risk level is divided into different levels using weight mapping and trapezoidal membership functions, and a dynamic pollution situation cloud map is generated in combination with Kriging interpolation to guide the bionic fish schools to prioritize patrolling high-risk areas.

[0088] Please refer to Figure 3 , which shows a flow chart of an exemplary water quality intelligent monitoring method S200 of the present application for collaborative sampling of bionic fish schools and drones, including:

[0089] S210: The drone collects historical pollution event data and real-time fish school sensor data.

[0090] Drones collect historical pollution event data from target waters and real-time sensor data from biomimetic fish schools to construct a comprehensive water quality assessment dataset. This historical pollution event database records the locations, pollutant types, peak concentrations, and diffusion paths of past pollution events, providing a reference for monitoring grid division and risk prediction.

[0091] At the same time, the sensor network carried by the bionic fish collects real-time water quality parameters, including pollutant concentration, water temperature, dissolved oxygen (DO) content, pH value, conductivity, and turbidity. These water quality parameters are transmitted via the underwater communication layer to the surface relay node, which is then received and integrated by the drone.

[0092] In one possible implementation, after integrating historical pollution event data with real-time fish sensor data, the two data types are fused. This process includes data standardization, feature extraction, and spatial interpolation to eliminate dimensional differences between the different data sources and enhance spatial continuity. Ultimately, this integrated dataset provides the foundation for the subsequent development of a water quality assessment indicator system.

[0093] S220: Construct a water quality evaluation index system.

[0094] After integrating historical pollution incident data with real-time bionic fish sensor data, a scientific and quantifiable evaluation standard was established based on this multi-source data to accurately reflect the pollution status of the target waters. The water quality evaluation index system mainly includes three key indicators: pollutant diffusion rate, concentration gradient change rate, and ecotoxicity index. These indicators are used to measure the migration capacity of pollutants, spatial distribution trends, and potential harm to the ecological environment.

[0095] Specifically, the diffusion rate of pollutants is an important parameter to measure the ability of pollutants to migrate in water bodies, and it directly affects the expansion speed of the pollution range.

[0096] In one possible implementation, the pollutant diffusion rate is calculated based on the pollutant diffusion paths in the historical pollution event database, combined with the water flow rate, water temperature, and water depth data collected in real time by the bionic fish school. The calculation formula is as follows:

[0097] ;

[0098] in, is the pollutant diffusion rate, For pollutants in time intervals This formula can be used to quantify the diffusion capacity of pollutants in different areas and to judge the urgency of pollution incidents.

[0099] Specifically, the concentration gradient change rate is used to describe the spatial variation trend of pollutant concentration, reflecting the diffusion intensity of pollution sources and the evolution of pollution range.

[0100] In one possible implementation, the concentration gradient change rate is calculated based on the pollutant concentration data collected by the bionic fish school and combined with the concentration peak information in the historical pollution event database. The calculation formula is as follows:

[0101] ;

[0102] in, is the concentration gradient change rate, is the pollutant concentration difference between adjacent grid cells, is the distance between grid cells. This formula can be used to identify areas where pollutant concentrations rise or fall rapidly, thereby optimizing the sampling strategy of the bionic fish school.

[0103] Specifically, the ecotoxicity index is used to assess the potential harm of pollutants to aquatic ecosystems, taking into account the toxicity and bioaccumulation capacity of pollutants.

[0104] In one possible implementation, the ecotoxicity index is calculated based on the pollutant type, concentration, and bioconcentration factor, combined with ecological impact records in the historical pollution event database. The calculation formula is as follows:

[0105] ;

[0106] in, is the ecotoxicity index, is the weight of the pollutant type, is the pollutant concentration, is the bioconcentration factor of the pollutant. This formula is used to quantify the impact of pollutants on aquatic organisms and adjust the sampling priority of the bionic fish school accordingly.

[0107] S230: Use the Nash equilibrium strategy to solve the optimal comprehensive weight of the water quality evaluation index system.

[0108] Because pollutant diffusion rate, concentration gradient change rate, and ecotoxicity index vary in importance across pollution events, a single fixed weight is difficult to adapt to complex and changing hydrological environments. Therefore, a Nash equilibrium strategy is adopted, optimizing the combined weights of each indicator based on game theory to ensure that the water quality assessment system maintains optimal decision-making capabilities under different pollution scenarios.

[0109] In one possible implementation, the implementation of the Nash equilibrium strategy mainly includes the following steps:

[0110] a. Taking pollutant diffusion rate, concentration gradient change rate, and ecotoxicity index as game participants, we set their respective strategy spaces, i.e., weight adjustment ranges, and establish a game model to measure the impact of different weight combinations on water quality assessment. The form of the game model is as follows:

[0111] ;

[0112] in, 、 、 Represent the weights of pollutant diffusion rate, concentration gradient change rate and ecotoxicity index respectively. and is the adjustment coefficient used to balance the weight of accuracy and adaptability.

[0113] b. In the game model, the process of finding the Nash equilibrium solution involves iterative calculations to find the optimal weight combination. The iterative calculation rules are as follows:

[0114] ;

[0115] ;

[0116] in, is the particle velocity, is the position of the particle, i.e., the weight configuration, is the historical optimal solution of the particle, is the global optimal solution, is the inertia weight, 、 is the acceleration factor, 、 is a random number. Through this algorithm, it is possible to quickly converge to the Nash equilibrium solution in the weight space.

[0117] S240: Generate a dynamic pollution situation cloud map based on the weighted results.

[0118] After optimizing the weights for the water quality assessment index system, a dynamic pollution situation cloud map is generated based on the weighted results. This map visually displays the distribution of pollution risks in the target waters and guides the bionic fish to prioritize high-weighted areas, thereby improving monitoring efficiency and data collection accuracy.

[0119] In one possible implementation, the process of generating a dynamic pollution situation cloud map includes:

[0120] a. Map the optimal weight results obtained from solving the Nash equilibrium strategy to each monitoring grid unit to generate a pollution risk value. Based on the distribution of pollution risk values, a trapezoidal membership function is used to classify pollution risk levels.

[0121] In one possible implementation, since the weights of pollutant diffusion velocity, concentration gradient change rate, and ecotoxicity index reflect the importance of different pollution characteristics, the pollution risk value of each grid cell is calculated by weighted summation.

[0122] In one possible implementation, the trapezoidal membership function is defined as:

[0123] ;

[0124] in, 、 、 、 The pollution risk thresholds correspond to the four levels of "safe," "warning," "dangerous," and "highly toxic." Using the trapezoidal membership function, we can convert pollution risk values ​​into probabilities at different levels, forming pollution risk levels that provide a basis for the subsequent visualization of dynamic pollution situation cloud maps.

[0125] b. Use GIS technology to map the pollution risk level onto a two-dimensional map of the target waters, and use a situation cloud map visualization method to display the pollution risk distribution of the dynamic pollution situation cloud map.

[0126] In one possible implementation, the dynamic pollution situation cloud map uses a color gradient, with blue representing "safe" areas, yellow "warning" areas, orange "dangerous" areas, and red "highly toxic" areas. This visualization method allows monitoring personnel to intuitively identify areas with higher pollution risks and adjust the bionic fish's patrol strategy accordingly.

[0127] After generating a dynamic pollution situation cloud map, a path planning strategy for the bionic fish school is developed based on the pollution risk level. This dynamic pollution situation cloud map effectively guides the bionic fish school to prioritize patrolling high-pollution risk areas, improving the relevance of water quality monitoring and the accuracy of data collection, providing a scientific basis for subsequent pollution control decisions.

[0128] The S300 intelligent water quality monitoring method, which uses bionic fish swarms and drones for collaborative sampling, includes the construction of a poor-quality fingerprint library and a pollution event response mechanism. First, pollutant signatures are extracted based on historical pollution event data to construct a poor-quality fingerprint library. Water quality data collected by the bionic fish swarm sensors is then compared with the poor-quality fingerprint library in real time. When the match exceeds a threshold, a two-level response is triggered. This ultimately forms a closed-loop management system of "real-time monitoring - event identification - response and control," ensuring rapid response to pollution events and efficient remediation.

[0129] Please refer to Figure 4 , which shows a flow chart of an exemplary water quality intelligent monitoring method S300 of the present application for collaborative sampling of bionic fish schools and drones, including:

[0130] S310: Build a bad fingerprint library.

[0131] The characteristics of pollutants are extracted through historical pollution event data. After the feature extraction is completed, the pollutant characteristics are encoded to build a bad fingerprint library.

[0132] In one possible implementation, the goal of feature extraction is to extract key pollutant features from historical pollution event data, including pollutant type, peak concentration, diffusion rate, toxicity index, and spatial distribution pattern.

[0133] In one possible implementation, pollutant signatures are encoded using multidimensional feature vectors, with each pollutant signature corresponding to a feature dimension, forming a multidimensional feature space. For example, pollutant types are encoded using one-hot encoding, converting different pollutant categories into binary vectors. Ultimately, each historical pollution event is represented as a high-dimensional feature vector with the following format:

[0134] ;

[0135] in, Code the pollutant type, is the diffusion rate, is the toxicity index, is the spatial distribution pattern of the pollutant. This feature vector constitutes the basic unit of the bad fingerprint library and is stored in the database for subsequent real-time data comparison.

[0136] As an optional step, in one possible implementation, to improve the efficiency of the bad fingerprint library, an inverted index is used to optimize the indexing of pollutant features. The inverted index establishes an index table based on key features such as pollutant type, peak concentration, diffusion rate, and toxicity index, enabling rapid location of similar pollution events.

[0137] S320: Matching analysis between real-time data and inferior fingerprint database.

[0138] After the construction of the poor fingerprint library is completed, the real-time water quality data collected by drone remote sensing and bionic fish sensors are compared with the characteristics of historical pollution events in the poor fingerprint library to identify potential pollution events and evaluate their matching degree.

[0139] In a possible implementation, the evaluation of the matching degree mainly relies on the calculation of feature vector similarity and the identification of pollution events based on a threshold.

[0140] Specifically, cosine similarity is used to calculate the matching degree between the real-time data feature vector and the historical pollution event feature vector in the bad fingerprint library. The calculation formula of cosine similarity is as follows:

[0141] ;

[0142] in, is the real-time data feature vector, is the feature vector in the inferior fingerprint library, represents the vector dot product, and The formula quantifies the similarity between real-time data and historical pollution events. The closer the similarity value is to 1, the higher the matching degree.

[0143] In an optional embodiment, weighted cosine similarity is used to further optimize the matching calculation to enhance the weight of key features.

[0144] Specifically, when the matching degree of real-time data exceeds a threshold, the current water quality is determined to be highly similar to a historical pollution incident, triggering the pollution incident identification process. This process includes pollutant type identification and pollution source location. Pollutant type identification is based on the highest-matching inferior fingerprint database record, determining the primary pollutant category of the current pollution incident. Pollution source location uses the real-time position of the bionic fish school to identify the most likely pollution source.

[0145] S330: If the matching degree exceeds the set threshold, the two-level response mechanism is immediately triggered.

[0146] After the real-time data is analyzed for a match against the bad fingerprint database, if the match exceeds a set threshold, a two-level response mechanism is immediately triggered to ensure timely handling of the pollution incident. This response mechanism includes: a first-level response, in which the bionic fish school automatically increases the sampling frequency, and a second-level response, in which the drone sends the pollution pre-diffusion coordinates to the control center.

[0147] First-level response: The bionic fish school automatically increases the sampling frequency.

[0148] Specifically, the bionic fish swarm's sampling frequency is dynamically adjusted based on the pollution risk level and the rate of pollutant diffusion. In areas with higher pollution risk, the bionic fish swarm's sampling frequency will be increased to 2-3 times the normal rate.

[0149] Second level response: The drone sends the pollution pre-diffusion coordinates to the control center.

[0150] While the bionic fish swarm increases its sampling frequency, the drone will perform a second-level response, sending pre-spreading coordinates of the pollution to ground control. This response mechanism aims to provide early warning for emergency response and support the prediction of pollution spread paths.

[0151] In one possible implementation, the drone predicts the diffusion path of pollutants based on real-time data collected by bionic fish schools and a database of historical pollution events, combined with a hydrological model.

[0152] In the intelligent water quality monitoring method that uses bionic fish schools and drones for collaborative sampling, the S400 system involves the development, simulation, and evaluation of pollution control measures. By developing and parameterizing control measures, simulating water quality evolution, and evaluating control effectiveness, it optimizes and verifies pollution control plans.

[0153] Please refer to Figure 5 , which shows a flow chart of an exemplary water quality intelligent monitoring method S400 of the present application for collaborative sampling of bionic fish schools and drones, including:

[0154] S410: Develop and parameterize governance measures.

[0155] The governance plan is based on real-time monitoring data from bionic fish schools and drones, as well as dynamic pollution situation cloud maps, and is formulated in combination with governance experience in the historical pollution event database.

[0156] For example, for heavy metal pollution incidents, the remediation solutions include: placement of adsorption materials, setting up physical barriers or artificial aeration and oxygenation.

[0157] To simulate the impact of treatment measures on water quality evolution, the treatment measures are parameterized as source terms or boundary conditions in the model. Based on the parameterized treatment measures, the impact of different treatment options on the water quality evolution trajectory is simulated, and the water quality status after treatment is predicted.

[0158] S420: Connect the bionic fish sampling data and the pollution pre-diffusion coordinates obtained by the drone to simulate the water quality evolution trajectory.

[0159] The simulation formula for the water quality evolution trajectory is as follows:

[0160] ;

[0161] in, For coordinates In time The concentration of pollutants, is the initial pollutant concentration, is the water velocity field, is the diffusion coefficient tensor, is the source term of the control measures, is the parameter vector of the governance scheme.

[0162] S430: Feasibility assessment of governance measures.

[0163] The real-time comparison of the pollutant concentration gradient change rate under different control schemes with the pollution risk threshold defined by the Nash equilibrium weights identifies the intervention effect of control measures on high-weighted areas. When the simulation trajectory shows that the core indicators converge to the safe membership range within the preset time window, the control measures are deemed effective.

[0164] S100-S400 include: using GIS data to divide monitoring grid units, establishing a hierarchical communication architecture, dynamically deploying bionic fish schools for real-time water quality monitoring, and optimizing resource allocation through fuzzy rule scheduling; integrating historical pollution event data and real-time sensor data to construct a water quality evaluation index system, using a Nash equilibrium strategy to determine the optimal weights to generate a dynamic pollution situation cloud map; constructing a bad fingerprint library to quickly identify pollution events and trigger a two-level response mechanism; and finally combining bionic fish sampling data with drone pre-diffusion coordinates to dynamically simulate the effects of treatment measures. This achieves efficient integrated air-water monitoring, improving the accuracy and timeliness of water quality anomaly detection; optimizing monitoring resource allocation through intelligent scheduling and data analysis; and leveraging game theory and dynamic simulation to enhance the scientific nature and feasibility of pollution control decisions, providing comprehensive technical support for aquatic environmental protection.

[0165] This application also provides an intelligent water quality monitoring system that uses bionic fish schools and drones for collaborative sampling, including:

[0166] The grid monitoring configuration unit is configured to: access geographic information system data to divide the monitoring grid units; build a layered communication architecture including an air communication layer, a surface relay layer, and an underwater communication layer; and generate dynamic deployment instructions for bionic fish schools based on the water depth gradient and flow velocity field;

[0167] The fuzzy scheduling control unit is connected to the bionic fish sensor network and is configured to: receive real-time data on water quality parameters, execute the fuzzy rule base to output bionic fish aggregation density correction instructions; and send path adjustment signals to the bionic fish through a layered communication architecture;

[0168] The Nash equilibrium decision engine is configured to: integrate historical pollution event data with real-time sensor data to construct an evaluation index system for pollutant diffusion rate, concentration gradient change rate, and ecotoxicity index; solve Pareto optimal weights based on a game model; generate a dynamic pollution situation cloud map and mark high-weight patrol areas;

[0169] The bad response execution module includes: a fingerprint library memory, which stores historical pollution event peak data encoded as multi-dimensional feature vectors; a real-time matcher, which calculates the weighted cosine similarity between the high-weight area sensor data and the fingerprint library;

[0170] Two-level response trigger, when the matching degree exceeds the limit: (a) send encrypted sampling instructions to the bionic fish school; (b) send pollution pre-diffusion coordinate generation instructions to the drone;

[0171] The control simulation verification platform is configured to: access encrypted sampling data from bionic fish schools and pollution pre-diffusion coordinates; parameterize control measures and feed them into a water quality evolution model for dynamic simulation; and output a control feasibility signal based on the convergence of the pollutant concentration gradient change rate.

[0172] The layered communication hardware architecture includes: an aerial communication layer: drones equipped with remote sensing equipment establish 5G / satellite backhaul links with ground control centers; a surface relay layer: self-organizing networking buoys equipped with air-water dual-frequency communication repeaters; and an underwater communication layer: bionic fish schools with built-in underwater acoustic modems form a dynamic routing sensor network.

[0173] The control center visualization terminal is configured to: render a dynamic pollution situation cloud map in real time, use gradient colors to mark pollution risk levels; and display the governance simulation trajectory and feasibility assessment results.

[0174] The present application also provides a computer device, which includes: a processor and a memory, wherein the memory stores at least one instruction, at least one program, a code set or an instruction set, and the at least one instruction, the at least one program, the code set or the instruction set is loaded and executed by the processor to realize an intelligent water quality monitoring method of collaborative sampling of bionic fish schools and drones.

[0175] The basic principles of the present application have been described above in conjunction with specific embodiments. However, it should be noted that the advantages, strengths, and effects mentioned in this application are merely illustrative and not restrictive, and it should not be assumed that these advantages, strengths, and effects are required of each embodiment of this application. In addition, the specific details disclosed above are merely illustrative and facilitating understanding, and are not restrictive. The above details do not limit this application to necessarily being implemented using the above specific details.

[0176] The block diagrams of the devices, apparatuses, equipment involved in this application are intended only as illustrative examples and are not intended to require or imply that they must be connected, arranged, or configured in the manner shown in the block diagrams. As will be appreciated by those skilled in the art, these devices, apparatuses, and equipment may be connected, arranged, or configured in any manner. Words such as "include," "comprise," "have," and the like are open-ended words, meaning "including but not limited to," and may be used interchangeably therewith. The words "or" and "and" used herein refer to the words "and / or" and may be used interchangeably therewith, unless the context clearly indicates otherwise. The word "such as" used herein refers to the phrase "such as but not limited to," and may be used interchangeably therewith.

[0177] It should also be noted that in the apparatus, device, and method of the present application, each component or each step can be decomposed and / or recombined, and such decomposition and / or recombination should be regarded as equivalent solutions of the present application.

[0178] The above description of the disclosed aspects is provided to enable any person skilled in the art to make or use the present application. Various modifications to these aspects will be readily apparent to those skilled in the art, and the general principles defined herein may be applied to other aspects without departing from the scope of the present application. Therefore, the present application is not intended to be limited to the aspects shown herein, but rather to be applied in the widest sense consistent with the principles and novel features of the present invention.

[0179] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent replacements, etc. made within the spirit and principles of the present application shall be included in the scope of protection of the present application.

Claims

1. An intelligent water quality monitoring method using collaborative sampling of bionic fish schools and drones, characterized by: include: Based on the geographic information system data of the target waters, monitoring grid units are divided and a hierarchical communication architecture is established between drones and bionic fish schools. Based on the environmental parameters of the monitoring grid units, bionic fish schools are dynamically deployed. UAV remote sensing and bionic fish school sensors monitor water quality anomalies in real time. Fuzzy rules are used to schedule bionic fish schools to achieve initial adaptation of monitoring resources and hydrological characteristics. The drone collects historical pollution event data and real-time fish sensor data to construct a water quality evaluation index system. Using a Nash equilibrium strategy, based on the optimal comprehensive weights of pollutant diffusion rate, concentration gradient change rate, and ecotoxicity index in the game theory evaluation index system, a dynamic pollution situation cloud map is generated based on the weighted results, directing fish schools to prioritize patrolling high-weighted areas. A bad fingerprint library is formed based on stored historical pollution event peak data. When fish sensors in high-weight areas detect that the real-time data matches the fingerprint library features beyond the limit, a two-level response is immediately triggered: 1) the bionic fish automatically increases the encryption sampling frequency; 2) the drone sends the pollution pre-diffusion coordinates to the control center. By integrating bionic fish sampling data with pollution pre-diffusion coordinates acquired by drones, the water quality evolution trajectory after treatment measures are implemented is dynamically simulated to determine the feasibility of the treatment measures. The layered communication architecture includes a three-level coordination mechanism of an air communication layer, a surface relay layer, and an underwater communication layer; The aerial communication layer uses drones to build a dynamic topology network, the surface relay layer uses adaptive beamforming technology to achieve cross-media signal conversion, and the underwater communication layer uses a distributed sensing network based on bionic fish schools to perform underwater acoustic communication and energy self-coordination protocols. The fuzzy rule scheduling mechanism adopts a multi-input and multi-output fuzzy inference system. The input variables include the pollutant concentration gradient, the water temperature mutation rate, the dissolved oxygen vertical stratification coefficient and the turbidity spatiotemporal variation coefficient. The output variables are the three-dimensional swimming trajectory offset of the bionic fish school and the sampling frequency domain gain coefficient. The game model constructed by the Nash equilibrium strategy includes three types of decision-making entities in non-cooperative competition relationships: pollutant migration dynamics entity, ecotoxicity response entity, and monitoring resource constraint entity; The form of the game model is as follows: ； in, Represent the weights of pollutant diffusion rate, concentration gradient change rate and ecotoxicity index respectively. is the adjustment coefficient, which is used to balance the weight of accuracy and adaptability; The strategy space of each agent is defined as the probability distribution of weight allocation. The utility function comprehensively considers the timeliness of pollution warning, biological toxicity exposure risk, and energy cost constraints. The Pareto optimal solution of the indicator weight is obtained by solving the mixed strategy Nash equilibrium. The calculation rules are as follows: ； ； in, is the particle velocity, is the position of the particle, i.e., the weight configuration, is the historical optimal solution of the particle, is the global optimal solution, is the inertia weight, is the acceleration factor, is a random number.

2. The intelligent water quality monitoring method of bionic fish school and drone collaborative sampling according to claim 1 is characterized in that: The method for generating the dynamic pollution situation cloud map includes: The optimal weight results obtained by solving the Nash equilibrium strategy are mapped to each monitoring grid unit to generate pollution risk values. Based on the distribution of pollution risk values, the trapezoidal membership function is used to divide the pollution risk levels. GIS technology is used to map the pollution risk level onto a two-dimensional map of the target waters, and a situation cloud map is used to visualize the pollution risk distribution of the dynamic pollution situation cloud map.

3. The intelligent water quality monitoring method of bionic fish school and drone collaborative sampling according to claim 1 is characterized in that: The construction and matching method of the inferior fingerprint library includes: Extract pollutant features from historical pollution event data. After feature extraction is completed, encode the pollutant features and build a bad fingerprint library. The matching degree is evaluated by calculating the similarity of feature vectors and identifying pollution events based on thresholds.

4. The intelligent water quality monitoring method of bionic fish school and drone collaborative sampling according to claim 1 is characterized in that: The simulation formula for the water quality evolution trajectory is: ； in, For coordinates In time The concentration of pollutants, is the initial pollutant concentration, is the water velocity field, is the diffusion coefficient tensor, is the source term of the control measures, is the parameter vector of the governance scheme.

5. The intelligent water quality monitoring method of bionic fish school and drone collaborative sampling according to claim 4 is characterized in that: The feasibility assessment of the control measures includes: real-time comparison of the pollutant concentration gradient change rate under different control schemes with the pollution risk threshold defined by the Nash equilibrium weight, and identification of the intervention effect of the control measures on high-weight areas; when the simulation trajectory shows that the core indicators converge to the safe membership interval within the preset time window, the control measures are judged to be effective.

6. A water quality intelligent monitoring system using a bionic fish school and drone collaborative sampling method, using the method according to any one of claims 1 to 5, characterized in that: include: The grid monitoring configuration unit is configured to: access geographic information system data to divide the monitoring grid units; build a layered communication architecture including an air communication layer, a surface relay layer, and an underwater communication layer; and generate dynamic deployment instructions for bionic fish schools based on the water depth gradient and flow velocity field; The fuzzy scheduling control unit is connected to the bionic fish sensor network and is configured to: receive real-time data on water quality parameters, execute the fuzzy rule base to output bionic fish aggregation density correction instructions; and send path adjustment signals to the bionic fish through a layered communication architecture; The Nash equilibrium decision engine is configured to: integrate historical pollution event data with real-time sensor data to construct an evaluation index system for pollutant diffusion rate, concentration gradient change rate, and ecotoxicity index; and solve for Pareto optimal weights based on a game model; Generate dynamic pollution situation cloud map and mark high-weight patrol areas; The bad response execution module includes: a fingerprint library memory, which stores historical pollution event peak data encoded as multi-dimensional feature vectors; a real-time matcher, which calculates the weighted cosine similarity between the high-weight area sensor data and the fingerprint library; Two-level response trigger, when the matching degree exceeds the limit: (a) send encrypted sampling instructions to the bionic fish school; (b) send pollution pre-diffusion coordinate generation instructions to the drone; The control simulation verification platform is configured to: access encrypted sampling data from bionic fish schools and pollution pre-diffusion coordinates; parameterize control measures and feed them into a water quality evolution model for dynamic simulation; and output a control feasibility signal based on the convergence of the pollutant concentration gradient change rate. The layered communication hardware architecture includes: an aerial communication layer: drones equipped with remote sensing equipment establish 5G / satellite backhaul links with ground control centers; a surface relay layer: self-organizing networking buoys equipped with air-water dual-frequency communication repeaters; and an underwater communication layer: bionic fish schools with built-in underwater acoustic modems form a dynamic routing sensor network. The control center visualization terminal is configured to: render a dynamic pollution situation cloud map in real time, use gradient colors to mark pollution risk levels; and display the governance simulation trajectory and feasibility assessment results.

7. A computer device, characterized in that: The computer device includes: a processor and a memory, wherein the memory stores at least one instruction, at least one program, a code set or an instruction set, and the at least one instruction, the at least one program, the code set or the instruction set is loaded and executed by the processor to implement the water quality intelligent monitoring method of collaborative sampling of bionic fish schools and drones as described in any one of claims 1 to 5.

Images