An artificial intelligence driven water ecological sensitive area coupling analysis and evaluation method

Abstract

This invention relates to the technical field of eco-hydrological analysis and discloses an artificial intelligence-driven method for coupled analysis and assessment of water ecologically sensitive areas. The method includes: collecting standardized water quality data from different sampling points in a water body using a distributed sensor network; dividing the water body into multiple water regions with different spatial heterogeneity using a spatial heterogeneity identification model and calculating spatial heterogeneity indices for different water regions; dynamically adjusting the sampling frequency and sampling point deployment density of the water regions based on the spatial heterogeneity indices; synchronously collecting multi-source coupled hydrological data of the water regions; constructing a water health status assessment model to assess the health status of the water regions; and identifying ecologically vulnerable water regions based on the health status and location information of the water regions. This invention achieves intelligent acquisition and dynamic scheduling of multi-source coupled hydrological data and accurately identifies ecologically vulnerable areas based on spatial heterogeneity and health status.

Description

Technical Field

[0001] This invention relates to the field of big data processing technology, and more particularly to the field of eco-hydrological analysis, specifically an artificial intelligence-driven method for coupled analysis and evaluation of water ecologically sensitive areas. Background Technology

[0002] With the continuous increase in watershed pollution load and the deepening of ecosystem degradation, problems such as eutrophication, decline in benthic biodiversity, and weakening of aquatic environmental functions are becoming increasingly prominent. The watershed's aquatic ecosystem exhibits significant spatiotemporal heterogeneity and complex multi-factor coupling characteristics, making it difficult for traditional monitoring and assessment methods to comprehensively and dynamically reflect changes in ecological status. Establishing a water ecological monitoring and sensitive area identification system with high spatiotemporal resolution, intelligent sensing, and adaptive analysis capabilities has become a key foundation for comprehensive watershed management and ecological protection.

[0003] Currently, existing monitoring systems mostly rely on manual fixed-point sampling or fixed-location online monitoring. These methods have significant limitations in terms of sampling time, spatial distribution, and data update frequency, and cannot fully reflect the dynamic changes of water bodies under different seasons, meteorological conditions, and human activities. At the same time, observation methods based on a single data source are difficult to reveal the nonlinear coupling relationships between multiple factors such as water quality, aquatic ecology, hydrology, and meteorology, which limits the accurate identification of ecological anomalies and potential risks.

[0004] Existing research and patents have attempted to apply artificial intelligence and sensor networks to water quality and aquatic ecosystem monitoring in large river basins. A typical example is the authorized patent CN118822084B, which proposes an AI-based method and system for water quality and aquatic ecosystem monitoring. This patent proposes acquiring water and soil chemical composition data in the Yangtze River basin, deploying a distributed sensor network based on regional environmental analysis, and collecting meteorological data to calculate the hydrological gradient index, thereby achieving basin-wide water quality and aquatic ecosystem monitoring and analysis. This patent demonstrates the feasibility of combining multi-source environmental data with sensor coverage strategies for basin-wide ecological analysis, which helps to improve the dimensionality of monitoring and the depth of data mining. However, this patented approach also faces some technical challenges: first, the cost and energy consumption control issues associated with large-scale sensor coverage and long-term operation; second, the lack of unified indicators for quantifying spatial heterogeneity of water quality and interpretable classification targets, making it difficult to support adaptive sampling or on-demand deployment; and third, the failure to integrate ecological data for more comprehensive ecological analysis and identification of ecologically vulnerable areas.

[0005] To address this issue, this invention proposes an AI-driven coupled analysis and assessment method for water ecologically sensitive areas. This method utilizes a combination of distributed sensor networks and machine learning algorithms to collect water data in real time, capture minute changes in water bodies, and delineate water areas. It overcomes the shortcomings of traditional methods in terms of spatial resolution, data acquisition frequency, and environmental adaptability, enabling more accurate identification of spatial heterogeneity within water ecologically sensitive areas and timely intervention and protection. Summary of the Invention

[0006] This invention provides an AI-driven method for coupled analysis and assessment of water ecological sensitive areas. S1 utilizes a distributed sensor network to achieve parallel data acquisition by multiple sensing units, effectively addressing the shortcomings of traditional manual sampling methods in terms of spatial resolution and real-time performance. Furthermore, denoising and standardization processes reduce sensor noise and dimensional differences, improving data consistency and comparability. S2 introduces a spatial heterogeneity identification model, using KL divergence and water quality distribution differences to construct an objective function. This divides the locations of multiple sampling points into water areas with significant spatial variations in water quality, achieving intelligent classification of water area spatial structure differences and overcoming the problems of strong subjectivity and insufficient identification of complex watershed structures in manual experience-based classification. S3 utilizes a spatial heterogeneity identification model... The adaptive sampling frequency and sampling point deployment density adjustment mechanism for heterogeneity indicators dynamically adjust the sampling strategy of node sensing units, concentrating sampling resources in areas with large water quality fluctuations. This ensures monitoring accuracy in highly heterogeneous areas while reducing overall energy consumption and data redundancy. S4 integrates multi-source water quality, ecological, hydrological, and meteorological data of water areas, uses support vector machines to construct a water health status assessment model to achieve nonlinear feature mapping and ecological health scoring, and calculates ecological vulnerability indicators based on spatial location, thereby identifying potential ecological risk areas. This achieves closed-loop optimization of the entire process from data collection to feature recognition, dynamic scheduling, and intelligent assessment, significantly improving the identification accuracy and management intelligence level of water ecologically sensitive areas.

[0007] To achieve the above objectives, this invention provides an artificial intelligence-driven method for coupled analysis and evaluation of water ecological sensitive areas, comprising the following steps:

[0008] S1: Use a distributed sensor network to collect water quality data at different sampling points in the water area, and perform noise reduction and standardization on the collected water quality data to obtain standardized water quality data at different sampling points.

[0009] S2: Based on standardized water quality data at different sampling point deployment locations, the water area is divided into multiple water areas with different spatial heterogeneity using a spatial heterogeneity identification model, and spatial heterogeneity indices for different water areas are calculated.

[0010] S3: Based on spatial heterogeneity indicators, dynamically adjust the sampling frequency and sampling point deployment density of the water area, use a distributed sensor network to synchronously collect regional water quality data, regional ecological data, regional hydrological data and regional meteorological data of the water area, and perform noise reduction and standardization processing on the collected data to form multi-source coupled hydrological data of the water area.

[0011] S4: Construct a water health status assessment model. Receive multi-source coupled hydrological data, assess the health status of water areas, and identify ecologically vulnerable water areas based on the health status and location information of the water areas.

[0012] As a further improvement of the present invention:

[0013] Further, step S1 includes:

[0014] The distributed sensor network consists of multiple node sensing units and a regional aggregation unit. The node sensing units are deployed at sampling points in the water area to collect water quality data at different sampling points and send the collected water quality data to the regional aggregation unit. The regional aggregation unit is used to aggregate the data collected by all node sensing units and control and adjust the number of node sensing units and the sampling frequency of the node sensing units.

[0015] The water quality data is in the form of sequence data of multiple water quality parameters, wherein the length of the sequence data is N, and the water quality data includes pH data sequence, conductivity data sequence, dissolved oxygen data sequence, turbidity data sequence, ammonia nitrogen concentration data sequence and phosphorus concentration data sequence, which correspond to the water quality parameters respectively: pH, conductivity, dissolved oxygen, turbidity, ammonia nitrogen concentration and phosphorus concentration.

[0016] The water quality data is denoised and standardized. The denoising process includes outlier repair and sequence smoothing. The standardization process involves performing extreme value normalization on the sequence data of multiple water quality parameters in sequence. The outlier repair includes outlier detection and interpolation repair of the detected outliers.

[0017] Furthermore, the spatial heterogeneity identification model in step S2 includes a water area division module and a spatial heterogeneity index calculation module;

[0018] The water area division module takes the standardized water quality data at the sampling point deployment location and the sampling point deployment location as input. By dividing the sampling point deployment location into K different water areas, it obtains the water quality distribution difference and KL divergence of the standardized water quality data at all sampling point deployment locations in each water area. Based on the KL divergence and water quality distribution difference, a spatial heterogeneity detection and identification objective function is constructed, where K represents the target number of water area divisions.

[0019] The spatial heterogeneity index calculation module calculates the spatial heterogeneity index of the water area based on standardized water quality data at all sampling point deployment locations within the water area.

[0020] The spatial heterogeneity detection and identification objective function aims to maximize the spatial heterogeneity between different water areas. It adjusts the sampling point deployment location to the water area division result. Spatial heterogeneity includes geographical distribution differences and water quality differences. Water quality differences include water quality distribution differences and KL divergence.

[0021] Furthermore, the expression for the spatial heterogeneity detection and identification objective function is:

[0022] ;

[0023] Where C represents the set of all sampling point deployment locations. They represent the division up to the th, respectively. The set of sampling point deployment locations for the j-th and j-th water areas Indicates the first Location distance information between the set of sampling points deployed in the j-th and j-th water areas. Indicates the first Geographical distribution differences between the j-th and j-th water areas Indicates the first Differences in water quality distribution between the j-th and j-th water areas Indicates the first KL divergence of the j-th and j-th water regions These represent the proportional coefficients for geographical distribution differences, water quality distribution differences, and KL divergence, respectively.

[0024] By solving the spatial heterogeneity detection and identification objective function, the division results of the sampling point deployment locations into K different water areas are obtained;

[0025] The spatial heterogeneity index of the water area is calculated as follows:

[0026] ;

[0027] in, Indicates the first Spatial heterogeneity indicators for each water area , This indicates the solution obtained for the first... A set of sampling point deployment locations for each water area. Represents the set of sampling point deployment locations The number of sampling point deployment locations in the data. Represents the set of sampling point deployment locations Any sampling point deployment location in the process, Indicates the location of sampling points Standard deviation of standardized water quality data Represents the set of sampling point deployment locations The mean of the standard deviations of standardized water quality data at all sampling point locations. Indicates the location of sampling points With the The Euclidean distance between the centers of the ... The center location of each water area is The mean value of all sampling point deployment locations. Indicates the distance control coefficient. This represents an exponential function with the natural constant as its base.

[0028] Furthermore, step S3 dynamically adjusts the sampling frequency and sampling point deployment density of the water area based on spatial heterogeneity indicators, including:

[0029] The dynamic adjustment formula for the sampling frequency and sampling point deployment density in the water area is as follows:

[0030] ;

[0031] ;

[0032] ;

[0033] in, Indicates the first Sampling point deployment density in each water area This indicates the default sampling point deployment density, which represents the number of node sensing units deployed in the water area. Represents the logarithmic function. Indicates the first Sampling frequency for each water area This indicates the default sampling frequency, which is the sampling time interval for data collected by the node sensing unit. Indicates selection The maximum value between.

[0034] Furthermore, step S3, which utilizes a distributed sensor network to synchronously collect regional water quality data, regional ecological data, regional hydrological data, and regional meteorological data of the water area, also includes:

[0035] The node sensing units in the water area are redeployed according to the sampling frequency and sampling point deployment density of the water area, and the sampling frequency of the node sensing units is adjusted. The distributed sensor network is updated, and the updated distributed sensor network is used to synchronously collect regional water quality data, regional ecological data, regional hydrological data, and regional meteorological data of the water area.

[0036] Furthermore, step S4 involves constructing a water area health status assessment model by receiving multi-source coupled hydrological data and assessing the health status of the water area, including:

[0037] The water area health status assessment model includes an input layer, an evaluation layer, and an output layer. The input layer is used to receive multi-source coupled hydrological data of the water area. The evaluation layer adopts a support vector machine model structure to perform high-dimensional nonlinear mapping on the multi-source coupled hydrological data to obtain the health status score of the water area. The output layer is used to output the health status of the water area. The health status score ranges from 0 to 100. The higher the health status score, the higher the health level of the water area.

[0038] Furthermore, step S4, based on the health status and location information of the water area, identifies ecologically vulnerable water areas, and also includes:

[0039] S41: Obtain the health status and location information of each water area, where the location information is the center location of the water area. The center location of the water area is obtained by calculating the average location of all sampling points deployed in the water area.

[0040] S42: Calculate the distance between the location information of any different water areas. For any water area, select the D water areas that are closest to the water area to form the D nearest neighbor-water area set of the water area, where D represents the number of water areas in the D nearest neighbor-water area set.

[0041] S43: For any water area, calculate the rate of change in the health status of the water area relative to the D nearest neighbor-water area set of the water area;

[0042] S44: The ecological vulnerability index of a water area is calculated by combining the health status and the rate of change of health status. If the ecological vulnerability index of a water area is lower than the vulnerability index threshold, the water area is marked as an ecologically vulnerable water area.

[0043] Furthermore, the formula for calculating the ecological vulnerability index is as follows:

[0044] , ;

[0045] in, Indicates the first Ecological vulnerability indicators for each aquatic area Indicates the first The health status of each water area Indicates the first For each water region in the set of D nearest neighbor water regions, there are water regions. The rate of change in health status Indicates the first The d-th water region in the set of D nearest neighbor water regions of a given water region. , Indicates the selection of a set The maximum value in, This represents the indicator weighting coefficient.

[0046] Compared with existing technologies, this invention proposes an artificial intelligence-driven coupled analysis and evaluation method for water ecological sensitive areas, which has the following beneficial effects:

[0047] First, this invention defines the sampling point deployment density to automatically increase the number of node sensing units deployed in water areas with large water quality fluctuations and high spatial heterogeneity, thereby enhancing local monitoring accuracy; and the sampling frequency adjustment formula... This enables adaptive shortening of the sampling period, allowing for continuous observation of highly spatially heterogeneous regions with higher temporal resolution. This dynamic adjustment mechanism balances monitoring coverage and data representativeness under limited sampling resources. Through dual scheduling of density enhancement and frequency self-adjustment, it effectively captures the dynamic processes of water quality changes and responds rapidly, improving the ability to respond to abnormal events such as sudden pollution and ecological disturbances. This achieves a transformation from static sampling to an intelligent and dynamic sampling system, significantly improving the spatiotemporal perception efficiency of the water quality monitoring system.

[0048] Meanwhile, by integrating the health status of a water area with the health change rate of its spatial neighbors, this invention can comprehensively reflect the ecological vulnerability index of an water area, reflecting both its internal ecological stability and sensitivity to external disturbances. The health status item... The rate of change reflects the overall condition of the water quality and ecosystem of the water body. This reveals the gradient differences in ecological health between the water area and the surrounding water areas. The combination of the two can accurately assess the ecological pressure of the water area in space, and thus effectively identify water areas with low health status that are in the high health gradient area. This provides a data-driven basis for key governance, monitoring resource allocation and ecological restoration planning of water ecologically sensitive areas, and realizes intelligent and precise water ecological risk assessment. Attached Figure Description

[0049] Figure 1This is a flowchart illustrating an artificial intelligence-driven coupling analysis and evaluation method for water ecologically sensitive areas, provided as an embodiment of the present invention.

[0050] Figure 2 This is a flowchart of water quality data denoising processing provided in an embodiment of the present invention.

[0051] Figure 3 This is a flowchart of the objective function solution provided in an embodiment of the present invention.

[0052] Figure 4 This is a comparison chart of water quality indicators in different water areas provided in an embodiment of the present invention. Detailed Implementation

[0053] The realization of the objectives, functional characteristics, and advantages of this invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0054] This invention provides an artificial intelligence-driven method for coupled analysis and assessment of water ecologically sensitive areas. The executing entity of this method includes, but is not limited to, at least one of the following electronic devices that can be configured to execute the method provided in this invention: a server, a terminal, etc. In other words, the method can be executed by software or hardware installed on a terminal device or a server device, and the software may be a blockchain platform. The server includes, but is not limited to, a single server, a server cluster, a cloud server, or a cloud server cluster.

[0055] Reference Figure 1 , Figure 2 as well as Figure 3 Embodiment 1 of the present invention is as follows:

[0056] An AI-driven method for coupled analysis and assessment of water ecologically sensitive areas, the method comprising:

[0057] S1: Use a distributed sensor network to collect water quality data at different sampling points in the water area, and perform noise reduction and standardization on the collected water quality data to obtain standardized water quality data at different sampling points.

[0058] The distributed sensor network consists of multiple node sensing units and a regional aggregation unit. The node sensing units are deployed at sampling points in the water area to collect water quality data at different sampling point locations, and the collected water quality data is sent to the regional aggregation unit.

[0059] It should be noted that the water area is a two-dimensional planar region;

[0060] The node sensing unit integrates a water quality sensing module, an ecological sensing module, a hydrological sensing module, and a meteorological sensing module. It sequentially collects water quality data, ecological data, hydrological data, and meteorological data at the sampling point deployment locations, respectively. The sampling point deployment locations in the water area are the deployment locations of the node sensing unit, and the distance between adjacent sampling point deployment locations is set to be greater than 500 meters.

[0061] The regional aggregation unit is used to aggregate water quality data, ecological data, hydrological data, and meteorological data collected by all node sensing units, and to control and adjust the number of node sensing units and the sampling frequency of the node sensing units.

[0062] The node sensing units and regional aggregation units in the distributed sensor network achieve collaborative communication through wireless self-organizing networks (such as LoRa, NB-IoT or 5G IoT communication).

[0063] As an embodiment of the present invention, various sensing modules are composed of a multi-parameter composite sensing probe, a signal acquisition circuit, and a microcontroller unit (MCU). The water quality sensing module adopts the electrochemical and optical measurement principles and simultaneously monitors multiple water quality indicators through a multi-parameter integrated probe. The multi-parameter composite sensing probe includes a pH sensor (measuring the acidity and alkalinity of water using the glass electrode method), a conductivity sensor (measuring the change in ion concentration in liquid based on four-electrode AC measurement technology), a dissolved oxygen sensor (measuring dissolved oxygen using the fluorescence quenching method), a turbidity sensor (detecting the concentration of suspended particles based on the 90° scattered light detection principle), an ammonia nitrogen sensor (detecting the activity of NH4⁺ ions in liquid using an ion-selective electrode (ISE), and a phosphorus sensor (detecting the orthophosphate content in liquid using ultraviolet absorption spectroscopy, and obtaining the phosphorus concentration through chemical conversion and photometric comparison). The multi-parameter composite sensing probe sequentially collects the sequence data of water quality parameters pH, conductivity, dissolved oxygen, turbidity, ammonia nitrogen concentration, and phosphorus concentration as water quality data.

[0064] The multi-parameter sensing probes in the ecological sensing module include a chlorophyll sensor (based on the principle of fluorescence spectroscopy detection, which excites chlorophyll a molecules in the water body through a specific excitation wavelength (about 470 nm) and detects the intensity of its emitted light to reflect algal biomass) and an algal density sensor (which uses the multispectral absorption ratio method to measure the difference in light absorption rate in different bands to distinguish algal types such as cyanobacteria and green algae, and calculates the relative density), and sequentially collect the sequence data of ecological parameters (algal biomass and algal density) as ecological data;

[0065] The multi-parameter composite sensing probe in the hydrological sensing module includes a flow velocity sensor (which uses ultrasonic Doppler (ADCP) to measure water flow velocity), a flow rate sensor (which combines a cross-sectional flow velocity integration algorithm to calculate the volumetric flow rate per unit time through the cross-section at the sampling point deployment location by measuring flow velocity at multiple points), and a level gauge (which uses a pressure-type or radar-type level detection principle to measure water level height). It sequentially collects the sequence data of hydrological parameters (water flow velocity, flow rate, and water level height) as hydrological data.

[0066] The multi-parameter composite sensing probe in the meteorological sensing module includes a temperature and humidity sensor and a wind speed sensor (using an ultrasonic anemometer), which sequentially collects the sequence data of meteorological parameters (temperature, humidity and wind speed) as meteorological data.

[0067] The water quality data is in the form of sequence data of multiple water quality parameters, wherein the length of the sequence data is N, and the water quality data includes pH data sequence, conductivity data sequence, dissolved oxygen data sequence, turbidity data sequence, ammonia nitrogen concentration data sequence and phosphorus concentration data sequence, which correspond to the water quality parameters respectively: pH, conductivity, dissolved oxygen, turbidity, ammonia nitrogen concentration and phosphorus concentration.

[0068] It should be noted that in step S1, water quality data of equal length are collected simultaneously at each sampling point deployment location to divide the water area into multiple water areas with different spatial heterogeneity, thereby adjusting the sampling frequency and sampling point deployment density of different water areas separately to improve the analysis and detection frequency of abnormal water areas.

[0069] The water quality data is denoised and standardized. The denoising process includes outlier repair and sequence smoothing. The standardization process involves performing extreme value normalization on the sequence data of multiple water quality parameters in sequence. The outlier repair includes outlier detection and interpolation repair of the detected outliers.

[0070] For reference Figure 2 The flowchart shown below illustrates the denoising process for water quality data. The denoising process for the water quality data is as follows:

[0071] S101: Sequentially extract the sequence data of any water quality parameter from the water quality data, and use the 3 sigma criterion to detect outliers in the sequence data; specifically, by calculating the mean and standard deviation of the sequence data, if the absolute value of the difference between the data value and the mean in the sequence data is higher than 3 times the standard deviation, then the data value is regarded as an outlier. The 3 sigma criterion assumes that the sequence data approximately follows a normal distribution, and the probability of a data value exceeding 3 times the standard deviation is less than 0.3%, which can be regarded as a non-random anomaly. In order to avoid misjudgment due to sudden environmental changes, a sliding time window strategy can be introduced in practical applications, that is, the mean and standard deviation are calculated only within the time window, and outlier detection is performed on the data values ​​within the time window, making the detection results more timely and locally sensitive.

[0072] S102: Treat outliers as missing data and use an autoregressive model to interpolate and complete the missing data, thus achieving outlier repair. Optionally, compared to the traditional autoregressive model, attenuation weights for different data values ​​in the sequence data can be introduced to attenuate the data values, thereby improving the completion accuracy in non-stationary environments. The formula for calculating the attenuation weights is as follows: t represents the order of the data value in the sequence data. The higher the order, the later the data value was collected, and the lower the decay weight. Indicates the attenuation control coefficient, set It is 0.6. Represents an exponential function with the natural constant as its base;

[0073] S103: The sequence data after outlier repair is smoothed using a sequence smoothing method. Optionally, the sequence smoothing can be performed using moving average filtering or wavelet denoising filtering.

[0074] S2: Based on standardized water quality data at different sampling point deployment locations, the water area is divided into multiple water areas with different spatial heterogeneity using a spatial heterogeneity identification model, and spatial heterogeneity indices for different water areas are calculated.

[0075] In step S2, the spatial heterogeneity identification model includes a water area division module and a spatial heterogeneity index calculation module;

[0076] The water area division module takes the standardized water quality data at the sampling point deployment location and the sampling point deployment location as input. By dividing the sampling point deployment location into K different water areas, it obtains the water quality distribution differences and KL divergence of the standardized water quality data at all sampling point deployment locations in each water area. Based on the KL divergence and water quality distribution differences, a spatial heterogeneity detection and identification objective function is constructed; optionally, K is set to 10.

[0077] The spatial heterogeneity detection and identification objective function aims to maximize the spatial heterogeneity between different water areas. It adjusts the sampling point deployment location to the water area division result. Spatial heterogeneity includes geographical distribution differences and water quality differences. Water quality differences include water quality distribution differences and KL divergence.

[0078] The expression for the spatial heterogeneity detection and identification objective function is:

[0079] ;

[0080] ;

[0081] ;

[0082] ;

[0083] ;

[0084] ;

[0085] Where C represents the set of all sampling point deployment locations. They represent the division up to the th, respectively. The set of sampling point deployment locations for the j-th and j-th water areas Indicates the first Location distance information between the set of sampling points deployed in the j-th and j-th water areas. Indicates the first Geographical distribution differences between the j-th and j-th water areas Indicates the first Differences in water quality distribution between the j-th and j-th water areas Indicates the first KL divergence of the j-th and j-th water regions The proportional coefficients representing geographical distribution differences, water quality distribution differences, and KL divergence are respectively set. The values ​​are 0.3, 0.3, and 0.4 respectively. The larger the KL divergence, the greater the water quality difference between the two water areas. This indicates that the two water areas are in a weak coupling relationship or a state of over-sensitivity in the aquatic ecosystem, and one of the water areas may be in an unhealthy state. They represent The number of sampling point deployment locations in the data. , , These represent the sets of sampling point deployment locations, respectively. Any sampling point deployment location in the process, The sampling point deployment locations are represented in sequence. Standard deviation of standardized water quality data Indicates the location of sampling points Deployment location distance coefficient, ,in Indicates location distance information The deployment location of the extracted sampling points Euclidean distance, Indicates the distance control coefficient, set It is 2;

[0086] Specifically, the standard deviation of the standardized water quality data at the sampling point deployment location is calculated as follows:

[0087] Extract standardized water quality data from the sampling point deployment locations, and calculate the standard deviation of the sequence data corresponding to the water quality parameters pH, conductivity, dissolved oxygen, turbidity, ammonia nitrogen concentration and phosphorus concentration in the standardized water quality data. The calculated standard deviations are formed into a sequence and used as the standard deviation of the standardized water quality data at the sampling point deployment locations.

[0088] In turn, respectively The mean sequence of standardized water quality data at the locations of sampling points within the j-th and j-th water areas, where T represents the transpose; specifically, the mean sequence It is a sequence of length 6, the mean sequence. The sequence values ​​in the sequence are respectively the first... The average values ​​of standardized water quality parameters pH, conductivity, dissolved oxygen, turbidity, ammonia nitrogen concentration, and phosphorus concentration at the sampling points deployed in the water area.

[0089] Represents the mean sequence The covariance matrix, Represents the mean sequence The covariance matrix, This represents the computation of the trace of a matrix. This represents the numerical value of the determinant corresponding to the calculated matrix (the numerical representation of the determinant is calculated using the algebraic cofactor method). Describing the L2 norm, Describing the L1 norm, as well as All represent variables in the calculation process. Indicates the number of water quality parameters. It is 6. Represents the natural logarithm function;

[0090] It should be noted that the spatial heterogeneity detection and identification objective function proposed in this invention achieves a comprehensive characterization of water quality data differences at sampling point deployment locations in multiple water areas by jointly measuring geographical distribution differences, water quality distribution differences, and KL divergence. Specifically, the geographical distribution difference term reflects the aggregation degree of sampling point deployment locations. If the standard deviation distance of standardized water quality data in relatively close locations in different water areas is large, it indicates that there is a local abrupt change in one of the water areas, ensuring that the model can identify local locations with prominent spatial features. The water quality distribution difference term is introduced to characterize the difference in the mean of standardized water quality data in each water area, which can accurately capture the overall water quality differences in the water area. At the same time, combined with the KL divergence term based on the multidimensional Gaussian assumption, the inconsistency of water quality feature distribution in different areas is characterized from the perspective of probability distribution, improving the sensitivity and robustness of heterogeneity identification. By constructing this objective function, the division of water areas at the sampling point deployment locations can be dynamically adjusted, highlighting areas with drastic water quality changes and high ecological sensitivity. This spatial heterogeneity detection and identification objective function integrates geospatial and water quality characteristics, significantly improving the accuracy of spatial heterogeneity identification in water areas compared to traditional single-index analysis, and providing a scientific basis for ecological monitoring and water quality regulation.

[0091] By solving the spatial heterogeneity detection and identification objective function, the division results of the sampling point deployment locations into K different water areas are obtained;

[0092] As an embodiment of the present invention, refer to as follows Figure 3 The flowchart shown illustrates the process of solving the objective function for spatial heterogeneity detection and identification.

[0093] S201: Obtain the Euclidean distance between different sampling point deployment locations, and use a clustering algorithm to cluster the sampling point deployment locations into K clusters. All sampling point deployment locations in the clusters constitute a set of sampling point deployment locations within the same water area; optionally, the clustering algorithm is the K-means clustering algorithm.

[0094] S202: Perform a local perturbation operation on the set of sampling point deployment locations in each water area to generate multiple sets of water area division schemes; specifically, the local perturbation operation involves randomly selecting several sampling point deployment locations and attempting to exchange them between different sets of sampling point deployment locations;

[0095] S203: Calculate the function value of each water area division scheme in the spatial heterogeneity detection and identification objective function;

[0096] S204: Compare the function values ​​of all water area division schemes and select the water area division scheme with the highest function value as the solution result;

[0097] Through the above steps, adaptive optimization of the sampling point deployment locations is achieved, which can maximize the difference between regions while ensuring the continuity of geographic space. Furthermore, by adopting a clustering algorithm combined with a local perturbation strategy, it can break out of local optima based on the initial clustering results, thereby improving the global search capability for spatial heterogeneity identification.

[0098] The spatial heterogeneity index calculation module calculates the spatial heterogeneity index of the water area based on standardized water quality data at all sampling point deployment locations within the water area.

[0099] It should be noted that the spatial heterogeneity index of the aforementioned water area is calculated as follows:

[0100] ;

[0101] in, Indicates the first Spatial heterogeneity indicators for each water area , This indicates the solution obtained for the first... A set of sampling point deployment locations for each water area. Represents the set of sampling point deployment locations The number of sampling point deployment locations in the data. Represents the set of sampling point deployment locations Any sampling point deployment location in the process, Indicates the location of sampling points Standard deviation of standardized water quality data Represents the set of sampling point deployment locations The mean of the standard deviations of standardized water quality data at all sampling point locations. Indicates the location of sampling points With the The Euclidean distance between the centers of the ... The center location of each water area is The average value of all sampling point deployment locations.

[0102] It should be noted that the calculation method for this spatial heterogeneity index comprehensively considers the differences in water quality between the sampling point deployment location and the regional center location, as well as the spatial distance attenuation effect, by introducing a Gaussian weighting term. This enables spatial correlation modeling where the closer the sampling point is to the center, the greater its contribution to regional heterogeneity, using the standard deviation of water quality data. To measure the degree of local water quality fluctuation, the overall regional standard deviation is used. It characterizes the intensity of water quality fluctuations within a region and, combined with the normalization of the number of sampling points, avoids calculation biases caused by uneven deployment of sampling points in different water areas. Thus, it can quantify the spatial dispersion of water quality in different water areas at a unified scale and depict the differences in water quality characteristics within water areas.

[0103] S3: Based on spatial heterogeneity indicators, dynamically adjust the sampling frequency and sampling point deployment density of the water area, and use a distributed sensor network to synchronously collect regional water quality data, regional ecological data, regional hydrological data and regional meteorological data of the water area. Then, perform noise reduction and standardization processing on the collected data to form multi-source coupled hydrological data of the water area.

[0104] Dynamically adjust the sampling frequency and sampling point deployment density of the water area based on spatial heterogeneity indicators, including:

[0105] The dynamic adjustment formula for the sampling frequency and sampling point deployment density in the water area is as follows:

[0106] ;

[0107] ;

[0108] ;

[0109] in, Indicates the first Sampling point deployment density in each water area This indicates the default sampling point deployment density (e.g., set to 10). Indicates the first The spatial heterogeneity index for a water area, wherein the sampling point deployment density represents the number of node sensing units deployed in the water area. Represents the logarithmic function. Indicates the first Sampling frequency for each water area This indicates the default sampling frequency (e.g., set to 60 seconds), where the sampling frequency is the sampling time interval for data collected by the node sensing unit. Indicates selection The maximum value between.

[0110] The node sensing units in the water area are redeployed according to the sampling frequency and sampling point deployment density of the water area, and the sampling frequency of the node sensing units is adjusted. The distributed sensor network is updated, and the updated distributed sensor network is used to synchronously collect regional water quality data, regional ecological data, regional hydrological data, and regional meteorological data of the water area.

[0111] It should be noted that the node sensing unit is used to collect water quality data, ecological data, hydrological data, and meteorological data at the sampling point deployment locations. After noise reduction and standardization, the mean values ​​of water quality data, ecological data, hydrological data, and meteorological data at all sampling point deployment locations in the water area are calculated sequentially and used as the regional water quality data, regional ecological data, regional hydrological data, and regional meteorological data for the water area. The regional water quality data, regional ecological data, regional hydrological data, and regional meteorological data are all in the form of sequential data and have the same data length as the water quality data, ecological data, hydrological data, and meteorological data at the sampling point deployment locations.

[0112] Specifically, the denoising and standardization methods for the ecological data, hydrological data, and meteorological data are consistent with the denoising and standardization methods for the water quality data in step S1.

[0113] S4: Construct a water health status assessment model. Receive multi-source coupled hydrological data, assess the health status of water areas, and identify ecologically vulnerable water areas based on the health status and location information of the water areas.

[0114] A water health status assessment model is constructed by receiving multi-source coupled hydrological data to assess the health status of water areas, including:

[0115] The water area health status assessment model includes an input layer, an evaluation layer, and an output layer. The input layer is used to receive multi-source coupled hydrological data of the water area. The evaluation layer adopts a support vector machine model structure to perform high-dimensional nonlinear mapping on the multi-source coupled hydrological data to obtain the health status score of the water area. The output layer is used to output the health status of the water area. The health status score ranges from 0 to 100. The higher the health status score, the higher the health level of the water area.

[0116] In this embodiment of the invention, the water health status assessment model uses a support vector machine model to model and classify the complex nonlinear relationships between different parameters in a high-dimensional feature space. The health status output by the assessment layer appears in the form of a score, with the score ranging from 0 to 100. A score of 80 to 100 indicates that the water area is in a good or excellent state, 50 to 80 indicates that the water quality is stable but there is a slight risk, and a score below 50 indicates that the health level of the water area is poor and key treatment needs to be implemented.

[0117] Specifically, the true health status scores of multiple sets of multi-source coupled hydrological data are obtained as the training set. The health status of each set of multi-source coupled hydrological data output by the water health status assessment model is used as the predicted value. The mean square error between the true health status score and the predicted value is constructed. The training objective function of the parameters to be trained in the assessment layer is constructed with the goal of minimizing the mean square error and the correction term. The correction term is the square of the L1 norm of the hyperplane normal vector of the support vector machine in the parameters to be trained. The parameters to be trained include the hyperplane normal vector of the support vector machine and the bias term. The true health status score is obtained by expert judgment. Optionally, the gradient descent algorithm is used to solve the training objective function.

[0118] Based on the health status and location information of aquatic areas, ecologically vulnerable aquatic areas are identified, including:

[0119] S41: Obtain the health status and location information of each water area, where the location information is the center location of the water area. The center location of the water area is obtained by calculating the average location of all sampling points deployed in the water area.

[0120] S42: Calculate the distance between the location information of any different water areas. For any water area, select the D water areas that are closest to the water area to form the D nearest neighbor-water area set of the water area, where D represents the number of water areas in the D nearest neighbor-water area set, and D is set to 10; where the distance is the Euclidean distance between the location information.

[0121] S43: For any water area, calculate the rate of change in health status of that water area relative to the D nearest neighbor water area set:

[0122] ;

[0123] in, Indicates the first For each water region in the set of D nearest neighbor water regions, there are water regions. The rate of change in health status Indicates the first The d-th water region in the set of D nearest neighbor water regions of a given water region. , They represent the number respectively. individual water areas and water areas health status, Indicates the first individual water areas and water areas The distance between them;

[0124] S44: The ecological vulnerability index of a water area is calculated by combining the health status and the rate of change of health status. If the ecological vulnerability index of a water area is lower than the vulnerability index threshold, the water area is marked as an ecologically vulnerable water area.

[0125] Optionally, the mean and standard deviation of the ecological vulnerability index for all water areas can be calculated, and the sum of the two can be used as the vulnerability index threshold.

[0126] The formula for calculating the ecological vulnerability index is as follows:

[0127] , ;

[0128] in, Indicates the first Ecological vulnerability indicators for each aquatic area Indicates the first The health status of each water area Indicates the first For each water region in the set of D nearest neighbor water regions, there are water regions. The rate of change in health status Indicates the first The d-th water region in the set of D nearest neighbor water regions of a given water region. , Indicates the selection of a set The maximum value in, Indicates the indicator weight coefficient, set It is 0.6.

[0129] Example 2: As another embodiment of the present invention, after dividing the water area into multiple water areas with different spatial heterogeneity using the spatial heterogeneity identification model in step S2, four water areas are selected for water quality parameter comparison to verify the effectiveness of the water area division.

[0130] For reference Figure 4 The comparison chart of water quality indicators in different water areas shows that: Water area 1 has stable overall water quality, with moderate pH and dissolved oxygen levels, low pollutant concentrations, and a good aquatic environment; Water area 2 is significantly affected by external factors, with higher turbidity, ammonia nitrogen, and phosphorus concentrations, higher conductivity, lower dissolved oxygen, and weaker self-purification capacity; Water area 3 has the best water quality, with slightly acidic pH, high dissolved oxygen content, low turbidity, and an excellent ecological environment; Water area 4 has moderate water quality, with indicators between those of water areas 1 and 2, slightly higher conductivity and pollutant concentrations, posing a certain risk of pollution input and requiring strengthened management.

[0131] It should be understood that the embodiments described are for illustrative purposes only and are not limited to this structure in the scope of the patent application.

[0132] It should be noted that the sequence numbers of the above embodiments of the present invention are merely for descriptive purposes and do not represent the superiority or inferiority of the embodiments. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, apparatus, article, or method that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, apparatus, article, or method. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, apparatus, article, or method that includes that element.

[0133] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) as described above, and includes several instructions to cause a terminal device (which may be a mobile phone, computer, server, or network device, etc.) to execute the methods described in the various embodiments of the present invention.

[0134] The above are merely preferred embodiments of the present invention and do not limit the scope of the patent. Any equivalent structural or procedural transformations made based on the description and drawings of the present invention, or direct or indirect applications in other related technical fields, are similarly included within the scope of patent protection of the present invention.

Claims

1. An artificial intelligence driven water ecological sensitive area coupling analysis and evaluation method, characterized in that, The method includes: S1: Use a distributed sensor network to collect water quality data at different sampling points in the water area, and perform noise reduction and standardization on the collected water quality data to obtain standardized water quality data at different sampling points. S2: Based on standardized water quality data at different sampling point deployment locations, the water area is divided into multiple water areas with different spatial heterogeneity using a spatial heterogeneity identification model, and spatial heterogeneity indices for different water areas are calculated. S3: Based on spatial heterogeneity indicators, dynamically adjust the sampling frequency and sampling point deployment density of the water area, use a distributed sensor network to synchronously collect regional water quality data, regional ecological data, regional hydrological data and regional meteorological data of the water area, and perform noise reduction and standardization processing on the collected data to form multi-source coupled hydrological data of the water area. S4: Construct a water health status assessment model. Receive multi-source coupled hydrological data, assess the health status of water areas, and identify ecologically vulnerable water areas based on the health status and location information of the water areas. The spatial heterogeneity identification model described in step S2 includes a water area division module and a spatial heterogeneity index calculation module; The water area division module takes the standardized water quality data at the sampling point deployment location and the sampling point deployment location as input. By dividing the sampling point deployment location into K different water areas, it obtains the water quality distribution difference and KL divergence of the standardized water quality data at all sampling point deployment locations in each water area. Based on the KL divergence and water quality distribution difference, a spatial heterogeneity detection and identification objective function is constructed, where K represents the target number of water area divisions. The spatial heterogeneity index calculation module calculates the spatial heterogeneity index of the water area based on standardized water quality data at all sampling point deployment locations within the water area. The spatial heterogeneity detection and identification objective function aims to maximize the spatial heterogeneity between different water areas, and adjusts the sampling point deployment location to the water area division result. Spatial heterogeneity includes geographical distribution differences and water quality differences. Water quality differences include water quality distribution differences and KL divergence. The expression for the spatial heterogeneity detection and identification objective function is: ； Where C represents the set of all sampling point deployment locations. They represent the division up to the th, respectively. The set of sampling point deployment locations for the j-th and j-th water areas Indicates the first Location distance information between the set of sampling points deployed in the j-th and j-th water areas. Indicates the first Geographical distribution differences between the j-th and j-th water areas Indicates the first Differences in water quality distribution between the j-th and j-th water areas Indicates the first KL divergence of the j-th and j-th water regions These represent the proportional coefficients for geographical distribution differences, water quality distribution differences, and KL divergence, respectively. By solving the spatial heterogeneity detection and identification objective function, the result of dividing the sampling point deployment locations into K different water areas is obtained.

2. The AI-driven coupled analysis and evaluation method for water ecologically sensitive areas as described in claim 1, characterized in that, Step S1 includes: The distributed sensor network consists of multiple node sensing units and a regional aggregation unit. The node sensing units are deployed at sampling points in the water area to collect water quality data at different sampling points and send the collected water quality data to the regional aggregation unit. The regional aggregation unit is used to aggregate the data collected by all node sensing units and control and adjust the number of node sensing units and the sampling frequency of the node sensing units. The water quality data is in the form of sequence data of multiple water quality parameters, wherein the length of the sequence data is N, and the water quality data includes pH data sequence, conductivity data sequence, dissolved oxygen data sequence, turbidity data sequence, ammonia nitrogen concentration data sequence and phosphorus concentration data sequence, which correspond to the water quality parameters respectively: pH, conductivity, dissolved oxygen, turbidity, ammonia nitrogen concentration and phosphorus concentration. The water quality data is denoised and standardized. The denoising process includes outlier repair and sequence smoothing. The standardization process involves performing extreme value normalization on the sequence data of multiple water quality parameters in sequence. The outlier repair includes outlier detection and interpolation repair of the detected outliers.

3. The artificial intelligence driven water ecological sensitive area coupling analysis and evaluation method according to claim 1, characterized in that, Step S3 dynamically adjusts the sampling frequency and sampling point deployment density of the water area based on spatial heterogeneity indicators, including: The dynamic adjustment formula for the sampling frequency and sampling point deployment density in the water area is as follows: ； ； ； in, Indicates the first Sampling point deployment density in each water area Indicates the first Spatial heterogeneity indicators for each water area This indicates the default sampling point deployment density, which represents the number of node sensing units deployed in the water area. Represents the logarithmic function. Indicates the first Sampling frequency for each water area This indicates the default sampling frequency, which is the sampling time interval for data collected by the node sensing unit. Indicates selection The maximum value between.

4. The artificial intelligence driven water ecological sensitive area coupling analysis and evaluation method according to claim 1, characterized in that, Step S3, which utilizes a distributed sensor network to synchronously collect regional water quality data, regional ecological data, regional hydrological data, and regional meteorological data for the water area, also includes: The node sensing units in the water area are redeployed according to the sampling frequency and sampling point deployment density of the water area, and the sampling frequency of the node sensing units is adjusted. The distributed sensor network is updated, and the updated distributed sensor network is used to synchronously collect regional water quality data, regional ecological data, regional hydrological data, and regional meteorological data of the water area.

5. The artificial intelligence driven water ecological sensitive area coupling analysis and evaluation method according to claim 1, characterized in that, Step S4 involves constructing a water area health status assessment model by receiving multi-source coupled hydrological data and assessing the health status of the water area, including: The water area health status assessment model includes an input layer, an evaluation layer, and an output layer. The input layer is used to receive multi-source coupled hydrological data of the water area. The evaluation layer adopts a support vector machine model structure to perform high-dimensional nonlinear mapping on the multi-source coupled hydrological data to obtain the health status score of the water area. The output layer is used to output the health status of the water area. The health status score ranges from 0 to 100. The higher the health status score, the higher the health level of the water area.

6. The artificial intelligence driven water ecological sensitive area coupling analysis and evaluation method according to claim 5, characterized in that, Step S4, based on the health status and location information of the water area, identifies ecologically vulnerable water areas and further includes: S41: Obtain the health status and location information of each water area, where the location information is the center location of the water area. The center location of the water area is obtained by calculating the average location of all sampling points deployed in the water area. S42: Calculate the distance between the location information of any different water areas. For any water area, select the D water areas that are closest to the water area to form the D nearest neighbor-water area set of the water area. S43: For any water area, calculate the rate of change of the health status of that water area relative to the D nearest neighbor-water area set of that water area; S44: The ecological vulnerability index of a water area is calculated by combining the health status and the rate of change of health status. If the ecological vulnerability index of a water area is lower than the vulnerability index threshold, the water area is marked as an ecologically vulnerable water area.

7. The AI-driven coupled analysis and assessment method for water ecologically sensitive areas as described in claim 6, characterized in that, The formula for calculating the ecological vulnerability index is as follows: ， ； in, Indicates the first Ecological vulnerability indicators for each aquatic area Indicates the first The health status of each water area Indicates the first For each water region in the set of D nearest neighbor water regions, there are water regions. The rate of change in health status Indicates the first The d-th water region in the set of D nearest neighbor water regions of a given water region. , Indicates the selection of a set The maximum value in, This represents the indicator weighting coefficient.

Images