Evaluation method based on water ecological environment monitoring

Abstract

According to the evaluation method based on water ecological environment monitoring, a three-level monitoring system is constructed, a drainage basin is divided into a first-level monitoring area, a second-level sub-area and a third-level micro-unit, a multi-level monitoring layout from a macroscopic level to a microscopic level is achieved, monitoring comprehensiveness is improved, monitoring blind areas are effectively avoided, and water ecological environment data are comprehensively captured; meanwhile, according to the cross-unit compensation method based on the space-time correlation network and the graph attention mechanism, when the data missing rate exceeds a preset threshold value, the space-time correlation network is constructed through node feature vectors, the hydraulic distance, the flow velocity and a time attenuation factor, compensation data are generated through the graph attention mechanism and are subjected to credibility weighted fusion with actually measured data, and the data missing rate is calculated. The mechanism can reduce the influence of data missing on the evaluation result.

Description

Technical Field

[0001] The present invention relates to the technical field of water ecological environment monitoring, and in particular to an evaluation method based on water ecological environment monitoring. Background Art

[0002] With the acceleration of industrialization and urbanization, water ecological and environmental problems are becoming increasingly severe. Accurate and efficient water ecological and environmental monitoring and assessment have become key to pollution prevention and control and ecological protection. However, existing water ecological and environmental monitoring and assessment technologies have many shortcomings and cannot meet actual needs.

[0003] In terms of monitoring layout, traditional water ecological environment monitoring often uses a single or limited distribution approach, lacking a systematic hierarchical division. Data processing also presents challenges. During the monitoring process, data loss is frequent due to factors such as equipment failure, communication interruptions, and environmental interference. Existing data restoration technologies often rely on simple interpolation or empirical formula compensation, failing to fully consider the spatiotemporal correlations of data within the water ecological environment. This approach is unable to accurately fill in missing data, leading to biased assessment results and seriously impacting their reliability and accuracy. Summary of the Invention

[0004] In order to solve the above technical problems, the present invention provides an evaluation method based on water ecological environment monitoring.

[0005] The technical solution adopted in the present invention is as follows:

[0006] An assessment method based on water ecological environment monitoring includes the following steps:

[0007] Construct a three-level monitoring system, dividing the target watershed into N primary monitoring areas, each of which is divided into M secondary sub-areas, each of which is divided into K tertiary micro-units, and each of which is equipped with monitoring equipment to collect data in real time;

[0008] Acquire a monitoring data set collected in real time by monitoring equipment of each of the three-level micro-units in the three-level monitoring system; if the data missing rate of the target three-level micro-unit monitoring data set exceeds a preset threshold, perform cross-unit compensation on the monitoring data set to obtain a repaired data set; wherein the monitoring data set includes water quality parameters, flow velocity vectors, and biological activity indicators;

[0009] Based on the restoration data set, a pollution assessment index for each third-level micro-unit is generated through a spatiotemporal feature extraction model, and the coefficient of variation of the biological activity index is simultaneously extracted; based on the hydraulic topology weight, the pollution assessment index of all third-level micro-units within the corresponding second-level sub-area is aggregated to generate a sub-basin comprehensive assessment value; the assessment results of adjacent first-level monitoring areas are corrected for lag effects through a time series recursive model to generate a basin-level comprehensive assessment index;

[0010] The basin-level comprehensive assessment index is integrated with the output data of the HEC-RAS hydrodynamic model to generate a dynamic pollution diffusion heat map; the pollutant migration path is annotated using the Lagrangian particle tracking algorithm; and a three-dimensional interactive map containing the ecological risk level, diffusion range, and remediation priority is output.

[0011] Furthermore, the method for constructing a three-level monitoring system is as follows: obtaining a digital elevation model data set, and calculating the river curvature parameters and the water flow direction matrix based on the digital elevation model data set; generating a dynamic monitoring grid according to the river curvature parameters, and generating an encrypted monitoring grid when the river curvature parameters are greater than a preset threshold, otherwise generating a standard monitoring grid; constructing a three-level monitoring system based on the encrypted monitoring grid and the standard monitoring grid, dividing the target watershed into N first-level monitoring areas, each first-level area is divided into M second-level sub-areas, and each second-level sub-area is divided into K third-level micro-units.

[0012] Furthermore, the cross-unit compensation method is as follows: construct a spatiotemporal correlation network, in which the nodes of the spatiotemporal correlation network are adjacent three-level micro-units, and the edge weights are dynamically calculated based on the hydraulic distance and flow velocity between the three-level micro-units; based on the spatiotemporal correlation network, compensation data is generated through a graph attention mechanism; the compensation data is credibility-weightedly fused with the effective measured data of the target three-level micro-unit to obtain a repaired data set.

[0013] Furthermore, the method for constructing the spatiotemporal correlation network includes: defining node feature vectors, including water quality parameters, biological activity indicators and historical pollution trends; introducing a time decay factor when calculating edge weights, wherein the time decay factor is associated with the half-life period of pollutants; and normalizing the correlation strength of adjacent nodes through the Softmax function to generate a spatiotemporal correlation matrix.

[0014] Furthermore, when the compensation data is fused with the effective measured data of the target tertiary micro-unit in a credibility-weighted manner, a credibility coefficient of the compensation data is set, and the credibility coefficient is inversely proportional to the hydraulic distance between the target tertiary micro-unit and the adjacent tertiary micro-unit; when the credibility coefficient is lower than the set credibility coefficient threshold, the manual verification process is triggered and the automatic evaluation is suspended.

[0015] Furthermore, if the mutation rate of the basin-level comprehensive evaluation index exceeds the historical benchmark value, an incremental training set is intercepted based on a sliding time window, and a meta-learning optimizer is used to perform online parameter updates on the spatiotemporal feature extraction model and the temporal recursive model.

[0016] Beneficial effects of the present invention:

[0017] (1) This invention constructs a three-level monitoring system, dividing the watershed into primary monitoring areas, secondary sub-areas, and tertiary micro-units, achieving a multi-level monitoring layout from macro to micro. In areas with large river curvature and complex water flows, an encrypted monitoring grid is automatically generated to ensure data collection density in key areas. Compared with traditional monitoring methods, this improves monitoring comprehensiveness, effectively avoids monitoring blind spots, and comprehensively captures water ecological environment data.

[0018] (2) The present invention proposes a cross-unit compensation method based on a spatiotemporal association network and a graph attention mechanism. When the data missing rate exceeds a preset threshold, a spatiotemporal association network is constructed using node feature vectors, hydraulic distance, flow velocity, and time attenuation factor. Compensation data is generated through a graph attention mechanism and weightedly fused with the credibility of the measured data. This mechanism can reduce the impact of data missing on the evaluation results.

[0019] (3) The present invention combines the spatiotemporal feature extraction model, the hydraulic topology weight aggregation algorithm and the time series recursive model to conduct step-by-step assessment from the third-level micro-unit to the watershed level; the spatiotemporal feature extraction model integrates CNN and LSTM to accurately extract the spatiotemporal features of the data to generate a pollution assessment index; the sub-regional assessment values ​​are aggregated based on the hydraulic topology weights to highlight the impact of key areas; the time series recursive model corrects the lag effect to generate a more realistic watershed-level comprehensive assessment index, thereby enhancing the assessment accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 This is a flow chart of an evaluation method based on water ecological environment monitoring according to an embodiment of the present invention. DETAILED DESCRIPTION

[0021] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0022] like Figure 1 As shown, an evaluation method based on water ecological environment monitoring according to an embodiment of the present invention includes the following steps:

[0023] S1: Build a three-level monitoring system, dividing the target watershed into N primary monitoring areas, each of which is divided into M secondary sub-areas, each of which is divided into K tertiary micro-cells. Monitoring equipment is deployed in each of these micro-cells to collect real-time data.

[0024] In one embodiment of the present invention, the three-level monitoring system is constructed as follows:

[0025] S11: Acquire a digital elevation model dataset, and calculate river curvature parameters and a water flow direction matrix based on the digital elevation model dataset.

[0026] In the embodiment of the present invention, the digital elevation model dataset contains the height information of the basin terrain. The river curvature parameter calculation formula is:

[0027]

[0028] Where, Indicates the actual meandering length of the river; It represents the straight-line distance between the starting and ending points of a river.

[0029] The river curvature parameter directly reflects the degree of river curvature. For example, in mountainous rivers, the river is winding and twisting. The value is larger; while in the straight river channel on the plain, The value is close to 1.

[0030] The flow direction matrix is ​​determined based on elevation differences, clearly defining the direction of water flow and providing a directional reference for subsequent monitoring grid division. It should be noted that the flow direction matrix is ​​a two-dimensional matrix whose size corresponds to the number of grids in the watershed model. Each element in the matrix represents the flow direction of the corresponding grid. For example, if the element in the i-th row and j-th column of the matrix is ​​1, it indicates that the water is flowing from the grid in the i-th row and j-th column to an adjacent grid (the specific flow direction is determined based on actual calculations).

[0031] The flow direction matrix can be calculated using the D8 algorithm. Specifically, in the D8 algorithm, the flow direction of each grid is defined as pointing toward the grid with the lowest elevation among its eight adjacent grids. If multiple adjacent grids have the same and lowest elevation, the flow direction can be determined based on specific rules (such as random selection or prioritizing a particular direction). The specific calculation scheme is as follows: First, for each grid, calculate the elevation difference between it and its eight adjacent grids; then find the adjacent grid with the largest elevation difference and use the direction of that grid as the flow direction for the current grid; finally, record the flow direction in the flow direction matrix.

[0032] S12: Generate a dynamic monitoring grid according to the river curvature parameter. When the river curvature parameter is greater than a preset threshold, generate an encrypted monitoring grid; otherwise, generate a standard monitoring grid.

[0033] Specifically, when (preset threshold), the river flow is complex, and it is easy to have water stagnation and pollutant accumulation. At this time, an encrypted monitoring grid is generated to improve monitoring accuracy; otherwise, a standard monitoring grid is generated. For example, Take 1.2, When the curvature is 1.2, an encrypted monitoring grid is generated with a resolution of 0.5km×0.5km, covering river sections with a curvature exceeding 1.2; otherwise, a standard monitoring grid is generated with a resolution of 1km×1km, covering straight sections of the river and slow-flowing areas.

[0034] S13: Construct a three-level monitoring system based on the encrypted monitoring grid and the standard monitoring grid, divide the target watershed into N first-level monitoring areas, each first-level area into M second-level sub-areas, and each second-level sub-area into K third-level micro-units.

[0035] The present invention disassembles the watershed from macro to micro through three-level division, ensuring comprehensive and targeted collection of data such as water quality parameters, flow velocity vectors, and biological activity indicators.

[0036] S2: Obtain the monitoring data set collected in real time by the monitoring equipment of each of the three-level micro-units in the three-level monitoring system. If the data missing rate of the target three-level micro-unit monitoring data set exceeds a preset threshold, perform cross-unit compensation on the monitoring data set to obtain a repaired data set, wherein the monitoring data set includes water quality parameters, flow velocity vectors and biological activity indicators.

[0037] In the actual monitoring process, due to various reasons such as equipment failure and communication interruption, the monitoring data set of the three-level micro-unit may be missing data. The present invention quantifies the integrity of the monitoring data by calculating the data missing rate, and the calculation formula is:

[0038]

[0039] Where, represents the data missing rate; Indicates the number of missing data, that is, the number of values ​​not obtained in the monitoring data; Indicates the total amount of data that should be collected, that is, the total amount of monitoring data that needs to be collected in theory.

[0040] For example, in a three-level micro-unit, the original plan was to collect 100 data points, but only 70 were successfully obtained. is 30, is 100. Then the data missing rate of the three-level micro unit is =0.3. The preset threshold is defined as ,like , exemplary A value of 0.2 indicates serious data missing. If these data are used directly for evaluation, the results will be highly biased. In this case, cross-unit compensation of the monitoring dataset is required.

[0041] In one embodiment of the present invention, the cross-unit compensation method is:

[0042] S21: constructing a spatiotemporal association network, wherein the nodes of the spatiotemporal association network are adjacent three-level micro-units, and the edge weights are dynamically calculated based on the hydraulic distance and flow velocity between the three-level micro-units.

[0043] The method for constructing the spatiotemporal correlation network is as follows:

[0044] Define node feature vectors, including water quality parameters, biological activity indicators, and historical pollution trends;

[0045] When calculating edge weights, a time decay factor is introduced, and the time decay factor is associated with the half-life period of the pollutant;

[0046] The association strength of adjacent nodes is normalized by the Softmax function to generate a spatiotemporal association matrix.

[0047] In the embodiment of the present invention, in order to fully describe the state of each three-level micro unit, the node feature vector is defined as .in, Indicates water quality parameters, such as chemical oxygen demand (COD), ammonia nitrogen content, etc. These parameters directly reflect the pollution status of the water body in the current micro unit; Represents biological activity indicators, such as microbial ATP concentration, which can reflect the metabolic activity of microorganisms and indirectly reflect the health of the ecosystem; For historical pollution trends, by analyzing the changes in pollutant concentrations over a period of time, we can predict future pollution trends. These eigenvectors provide basic information for subsequent calculation of edge weights and generation of compensation data.

[0048] Edge weight is used to measure the strength of the connection between nodes, and its calculation formula is:

[0049]

[0050] Where, Representation node and nodes The edge weight between them is used to measure the strength of the association between two nodes; represents the time decay factor, which is related to the pollutant half-life period; Representation node and nodes The hydraulic distance between two nodes is used to measure the water flow distance between them. It takes into account factors such as the length of the water flow path and the terrain slope. The closer the hydraulic distance is, the closer the water flow connection between the two third-level micro-units is, and the higher the data similarity may be. Representation node and nodes The average flow velocity between nodes is used to measure the speed of water flow between two nodes. Faster flow means faster pollutant spread, and the data correlation between two third-level micro-cells is stronger. Therefore, adjacent third-level micro-cells with close distances, fast flow rates, and short time intervals have a greater impact on the target third-level micro-cell.

[0051] Among them, for the time decay factor , and its calculation formula is:

[0052] Where, Represents a time interval, that is, the time difference from the current moment to a certain moment in the past; It represents the pollutant half-life period, that is, the time required for the pollutant concentration to drop to half of the initial concentration.

[0053] For example, a pollutant =10 days, if =5 days, then Approximately equal to 0.607.

[0054] The edge weights are normalized by the Softmax function to obtain the spatiotemporal correlation matrix . Among them, the Softmax function expression is , which converts edge weights into probability distribution form, making the association strength between all nodes comparable, which is convenient for subsequent graph attention mechanism to calculate.

[0055] For example, for node With three adjacent nodes 、 、 The calculated edge weights are 、 、 1. After being processed by the Softmax function, the normalized weights are obtained, thus forming part of the spatiotemporal correlation matrix.

[0056] S22: Based on the spatiotemporal association network, compensatory data is generated through a graph attention mechanism.

[0057] The graph attention mechanism is based on the node feature vector and the spatiotemporal correlation matrix , calculate the attention weight of each adjacent node to the target node, highlighting the influence of key adjacent units on the target unit. For example, for the target node , the graph attention mechanism will be based on the spatiotemporal correlation matrix Center and target node The relevant edge weights and the feature vectors of the adjacent nodes are used to calculate the relationship between each adjacent node and the target node. The contribution of the adjacent units is then used to generate compensation data. This method can fully utilize the data information of the adjacent units to generate reasonable compensation data. .

[0058] Specifically, for the target three-level microcell node , its own node feature vector is ; At the same time, obtain the feature vector set of all its adjacent nodes, assuming that the adjacent nodes are , and the corresponding eigenvectors are ,These feature vectors contain rich information such as water quality parameters, biological activity indices, and historical pollution trends.

[0059] In order to better capture the relationship between node features, the feature vector of each node is first linearly transformed. Through a learnable weight matrix W, the node feature vector is mapped to a new feature space, that is, the node The eigenvector of Transform to get , for adjacent nodes Transform to get ,in, .

[0060] Next, calculate the target node With adjacent nodes The attention coefficient between , which reflects the adjacent nodes A preliminary estimate of the importance of 𝑖. Specifically, The calculation formula is:

[0061]

[0062] Where, Represents the vector splicing operation, which splices the transformed target node feature vector with the adjacent node feature vector; is a learnable attention weight vector used to calculate the degree of association between nodes; It is an activation function used to introduce nonlinear factors to avoid the gradient disappearance problem. Its expression is:

[0063]

[0064] In order to make the attention coefficients of different adjacent nodes comparable, the Softmax function is used to normalize the attention coefficients. Its adjacent nodes Normalized attention weights The calculation formula is:

[0065]

[0066] After normalization, The value of is between 0 and 1, and the sum of the attention weights of all adjacent nodes is 1, which clearly reflects the attention weight of each adjacent node on the target node when generating compensation data. relative contribution.

[0067] After getting the normalized attention weight Finally, the feature information of adjacent nodes is aggregated by weighted summation to generate the target three-level micro-unit node Compensation data , and its calculation formula is:

[0068]

[0069] Therefore, it is important to emphasize that the compensation data is the linearly transformed feature vectors of adjacent nodes, weighted summed according to their respective attention weights. In other words, adjacent nodes that are more closely associated with the target node and have higher attention weights have their feature information accounted for a larger proportion of the compensation data. This allows the generated compensation data to fully utilize the data information of adjacent units with high correlation with the target node, more reasonably fill in the missing data of the target node, and provide strong support for subsequent data fusion and accurate evaluation.

[0070] S23: performing credibility-weighted fusion on the compensated data and the effective measured data of the target tertiary micro-unit to obtain a repaired data set.

[0071] Among them, when the compensation data is fused with the effective measured data of the target tertiary micro-unit in a credibility weighted manner, a credibility coefficient of the compensation data is set, and the credibility coefficient is inversely proportional to the hydraulic distance between the target tertiary micro-unit and the adjacent tertiary micro-unit; when the credibility coefficient is lower than the set credibility coefficient threshold, the manual verification process is triggered and the automatic evaluation is suspended.

[0072] In this embodiment of the present invention, since the compensated data is derived from adjacent units using a spatiotemporal correlation network and graph attention mechanism, its credibility is related to the closeness of the association between the adjacent units and the target unit. In contrast, the measured data is directly collected by monitoring equipment and has higher reliability, but may contain missing data. Weighted fusion can balance the advantages of these two methods and improve data quality.

[0073] Specifically, in order to quantify the credibility of compensation data, the credibility coefficient of compensation data is introduced , where the credibility coefficient expression is:

[0074]

[0075] Where, Represents the target three-level micro unit With adjacent three-level micro-units The hydraulic distance between them.

[0076] The hydraulic distance comprehensively considers factors such as the length of the water flow path, topography, and water flow resistance, and reflects the closeness of the water flow connection between two micro-units. For example, in areas where the river channel is relatively straight and the water flow is smooth, the hydraulic distance between adjacent micro-units is short. The larger the value, the higher the credibility of the compensation data. In areas with complex terrain and tortuous flow paths, the hydraulic distance is long. The smaller the value, the lower the credibility of the compensation data. The setting of this coefficient conforms to the physical laws of pollutant propagation and data association in the aquatic ecological environment. The closer the micro-units are, the more similar their data characteristics and environmental conditions are, and the more reliable the compensation data will be.

[0077] The present invention is based on the determined credibility coefficient , the credibility weighted fusion of the compensation data and the effective measured data of the target three-level micro-unit is performed, and the fusion expression is shown as follows:

[0078]

[0079] Where, Compensatory data generated by the graph attention mechanism is used to fill the missing data part of the target micro-unit; It is the effective measured data of the target three-level micro-unit and is the reliable data obtained by actual monitoring.

[0080] Among them, when When it is close to 1, that is, the hydraulic distance between the target micro-unit and the adjacent micro-unit is very close, the compensation data accounts for a large proportion in the fusion, indicating that the compensation data derived from the adjacent units is highly credible and can better supplement the missing values; when When it is close to 0, it means that the hydraulic distance is far and the credibility of the compensation data is low. The fused data is based on the measured data. Mainly, to avoid unreliable compensation data from interfering with the overall data.

[0081] To ensure data quality, set a credibility coefficient threshold ,when , it means that the credibility of the compensation data is seriously insufficient, at which point the manual verification process is triggered and the automatic evaluation is suspended. Among them, the manual verification link can be intervened by professionals, and the compensation data and measured data can be carefully checked and analyzed based on the actual situation of the basin, the status of the monitoring equipment, historical data and other information. For example, check whether there is a monitoring equipment failure that causes data anomalies, whether there are special pollution sources in adjacent units that affect data association, etc. If necessary, re-collect data or adopt other more reliable compensation methods to ensure that the repair data set entering the subsequent evaluation process is true and accurate, avoid deviations in the evaluation results due to unreliable data, and provide guarantees for the scientificity and effectiveness of water ecological environment monitoring and evaluation.

[0082] S3: Based on the restoration data set, the pollution assessment index of each third-level micro-unit is generated through the spatiotemporal feature extraction model, and the coefficient of variation of the biological activity index is extracted simultaneously; based on the hydraulic topology weight aggregation, the pollution assessment index of all third-level micro-units in the second-level sub-area is generated to generate a sub-basin comprehensive assessment value; the assessment results of the adjacent first-level monitoring areas are corrected for the lag effect through the time series recursive model to generate a basin-level comprehensive assessment index.

[0083] Specifically, based on the repair dataset, the spatiotemporal feature extraction model is used to generate the pollution assessment index of each third-level micro-unit. The spatiotemporal feature extraction model uses a combination of convolutional neural networks (CNNs) and long short-term memory networks (LSTMs). The spatial feature vectors extracted by the CNNs are fused with the temporal feature vectors extracted by the LSTMs and input into the fully connected layer for training. This learns the mapping relationship between the spatiotemporal features of the data and the pollution assessment index. Ultimately, the model outputs the pollution assessment index corresponding to each three-level micro-unit. ,This index quantifies the pollution degree of the micro-unit, and the larger the value, the more serious the pollution.

[0084] In an embodiment of the present invention, the stability of an ecosystem is assessed by extracting the coefficient of variation of a biological activity indicator. The method for calculating the coefficient of variation of the biological activity indicator includes:

[0085] Microbial ATP concentration B1 and fish movement trajectory dispersion B2 were selected as core biological indicators. They are extremely sensitive to environmental changes and can intuitively reflect the activity of the ecosystem.

[0086] The calculation formula of the coefficient of variation of the biological activity index is:

[0087]

[0088] Where, and Represent the mean and standard deviation of a group of biological activity indicators, respectively.

[0089] When the coefficient of variation of the bioactivity index When the value exceeds a set threshold, it is marked as an ecological anomaly. For example, if the coefficient of variation of microbial ATP concentration in a tertiary microunit exceeds the set threshold of 0.3, it means that the microbial metabolic activity in that area is abnormal, and the stability of the ecosystem is threatened. Further investigation is needed to determine the cause, such as whether pollutants are poisoning the microorganisms.

[0090] In the embodiment of the present invention, the pollution assessment index of all the third-level micro-units in the second-level sub-area is aggregated based on the hydraulic topology weight to generate a sub-basin comprehensive assessment value. , where its aggregation formula is:

[0091]

[0092] Where, Three-level microcell Pollution assessment index; is the total number of third-level microunits in the second-level sub-area; Represents the hydraulic topology weight; reflects the three-level micro unit The importance of a microcell in the sub-basin's flow topology takes into account factors such as its location, flow path, and connections with other microcells. For example, microcells located at key flow nodes and having a greater impact on pollutant dispersion are assigned a higher hydraulic topology weight.

[0093] In the present invention, the comprehensive assessment value of the sub-basin is the result of weighted summation of the pollution assessment index of each third-level micro-unit according to its hydraulic topology weight, highlighting the impact of key micro-units on the pollution status of the entire sub-basin, and can more comprehensively reflect the water ecological environment quality of the sub-basin.

[0094] The present invention uses a time series recursive model (such as LSTM) to correct the hysteresis effect of the evaluation results of adjacent first-level monitoring areas and generate a basin-level comprehensive evaluation index. In the water ecological environment, the spread and impact of pollutants have a lag. For example, pollutants discharged from upstream areas may take some time to affect the water quality and ecological conditions of downstream areas. The time series recursive model can learn the lag relationship in this time series data, and predict the impact of pollutants on adjacent areas at the current moment by analyzing historical evaluation results. The model will correct the current evaluation results based on the input evaluation data sequence of the adjacent first-level monitoring area and the learned lag law, eliminate the evaluation bias caused by the lag effect, and finally output a basin-level comprehensive evaluation index that accurately reflects the overall pollution status of the basin.

[0095] S4: The basin-level comprehensive assessment index is integrated with the output data of the HEC-RAS hydrodynamic model to generate a dynamic pollution diffusion heat map; the pollutant migration path is marked using the Lagrangian particle tracking algorithm; and a three-dimensional interactive map containing the ecological risk level, diffusion range, and remediation priority is output.

[0096] It should be noted that the HEC-RAS (Hydrologic Engineering Center-River Analysis System) hydrodynamic model is a professional software used to simulate river hydrological and hydrodynamic processes. It can output the distribution and changes of hydrodynamic parameters such as water velocity, water level, and flow in the basin. The HEC-RAS hydrodynamic model output reflects the overall pollution status of the watershed, while the HEC-RAS hydrodynamic model output data reflects the dynamic characteristics of the water flow. When the two are integrated, the pollution level of each region is linked to its hydrodynamic conditions, using spatial location as a link. For example, a high pollution assessment index in a region with fast water flow indicates that pollutants are spreading rapidly and the impact area may expand rapidly. In areas with slow water flow, even if the pollution assessment index is not high, pollutants may accumulate due to long-term retention, causing serious pollution.

[0097] By integrating this data and utilizing visualization technology, a dynamic pollution diffusion heat map is generated. This heat map uses visual elements such as color depth and brightness to intuitively display the pollution levels and diffusion trends in different areas within the watershed. The redder and brighter the area, the more severe the pollution. The dynamic color changes provide a real-time reflection of the pollution's spread over time, enabling managers and decision makers to quickly grasp the overall pollution situation in the watershed.

[0098] In an embodiment of the present invention, the expression of the Lagrangian particle tracking algorithm is:

[0099]

[0100] Where, Indicates that the particles The location at the moment, represents the initial position of the particle, For location exist The flow rate of time.

[0101] In a specific embodiment of the present invention, the pollutants are assumed to be a large number of particles, and the initial emission location of the pollutants is known (i.e. ), according to the flow rate at each moment and each location , calculate the position of the particle at each subsequent moment by integration For example, in a river, pollutant particles released from a factory's sewage outlet (initial location) will be in different positions at different times as the water flows. This algorithm can accurately calculate the movement trajectories of these particles, thereby marking the migration paths of pollutants within the watershed. The marked migration paths are presented as lines or animations on the dynamic pollution diffusion heat map, clearly showing the complete process of pollutants from the source to the diffusion area, helping decision makers analyze the spread of pollutants and identify affected areas and potential risks.

[0102] Based on dynamic pollution diffusion heat maps and marked pollutant migration paths, combined with ecological risk assessment models, the present invention ultimately outputs a three-dimensional interactive map containing ecological risk levels, diffusion ranges, and restoration priorities.

[0103] Ecological risk levels are determined using pre-set risk assessment rules based on factors such as the degree of pollution in different areas of the basin and the coefficient of variation of bioactivity indicators (e.g., a CV > 0.3 indicates an abnormal ecological state). For example, ecological risk is categorized into low, medium, and high levels. Areas with severe pollution and abnormal bioactivity indicators are classified as high risk, indicating that the ecosystem in the area is facing a serious threat and requires immediate action.

[0104] The diffusion range of pollutants within the watershed can be determined based on dynamic pollution diffusion heat maps and pollutant migration paths. Furthermore, through visualization technology, the diffusion range can be displayed as a three-dimensional area on a three-dimensional map, visualizing the spatial distribution of pollution.

[0105] This method determines the restoration priority for each region by comprehensively considering factors such as ecological risk level, spread range, and control costs. For high-risk areas with rapid spread, restoration priority is highest, and resources are allocated first. For low-risk areas with slow spread, restoration priority is relatively low.

[0106] In one embodiment of the present invention, if the mutation rate of the basin-level comprehensive evaluation index exceeds the historical benchmark value, an incremental training set is intercepted based on a sliding time window, and a meta-learning optimizer is used to perform online parameter updates on the spatiotemporal feature extraction model and the temporal recursive model.

[0107] In the embodiment of the present invention, online model updating is a key link to ensure the accuracy and timeliness of the assessment. It can timely optimize the assessment model when the watershed environment changes significantly, so that it can continuously adapt to the new environmental conditions.

[0108] Specifically, the model is updated online with the mutation rate of the basin-level comprehensive assessment index Exceeding historical benchmarks The comparison results are based on the Indicates the time interval In terms of comprehensive evaluation index at the river basin level The rate of change is obtained by calculating the ratio of the difference between the evaluation indexes of two adjacent time points to the time interval, that is, , where and They are and The basin-level comprehensive assessment index at the moment.

[0109] The historical benchmark value It is a reference threshold value obtained through statistical analysis based on the historical monitoring data and assessment results of the basin, which reflects the general level of the assessment index change rate of the basin under normal circumstances. When the water ecological environment in the basin changes dramatically in a short period of time, this change may be caused by sudden pollution incidents (such as factory wastewater leakage), extreme climate impacts (heavy rain causing pollutants to be washed away and spread), or other sudden changes in environmental factors. At this time, the original assessment model may not be able to accurately reflect the current environmental conditions and needs to be updated.

[0110] When an update is needed, the system of the present invention extracts incremental training sets based on a sliding time window. It should be noted that the size of the sliding time window is a key parameter and can be flexibly set based on actual needs. For example, in watersheds with frequent environmental changes, a smaller time window (e.g., the last week) can be selected to quickly capture details of environmental changes. In watersheds with relatively stable environments, a larger time window (e.g., one month) can be used to reduce the amount of data processing while ensuring that the training set contains sufficient valid information.

[0111] The intercepted incremental training set contains new data after environmental changes. This data covers monitoring information for each of the three-level micro-units in the three-level monitoring system, such as water quality parameters, flow velocity vectors, and biological activity indicators. Before updating the model, the data in the incremental training set undergoes preprocessing, including data cleaning (removing outliers and duplicates) and data normalization (unifying data of different magnitudes to the same scale), to improve data quality and model training efficiency.

[0112] The present invention adopts a meta-learning optimizer to perform online parameter update on a spatiotemporal feature extraction model and a temporal recursive model, wherein the meta-learning optimizer may adopt MAML (Model-Agnostic Meta-Learning).

[0113] During the update process, the incremental training set is first divided into a support set and a query set. The support set is used to calculate the model's gradient on new data to update model parameters; the query set is used to evaluate the performance of the updated model. For spatiotemporal feature extraction models (such as those combining CNN and LSTM), MAML calculates the gradients of parameters in various layers, such as convolutional and fully connected layers, based on the data in the support set. It then adjusts these parameters through backpropagation to enable the model to better extract spatiotemporal features from new data. For time series recursive models (such as LSTM), MAML optimizes parameters such as memory cells and gating structures to enhance the model's ability to capture lag effects in new time series data.

[0114] After each update, the model is verified using the query set. If the model performance (such as the accuracy of the prediction evaluation index, mean square error, and other indicators) meets expectations, the updated parameters are retained; otherwise, the update strategy is adjusted and the model is continuously optimized until the model can accurately adapt to the new environmental data.

[0115] By updating the model online, this method provides dynamic adaptability, enabling timely responses to sudden environmental changes and avoiding distortions in assessment results caused by model lags. Furthermore, by continuously optimizing the model, it helps improve the accuracy of long-term monitoring and assessment.

[0116] In the description of the present invention, the terms "first" and "second" are used for descriptive purposes only and should not be understood to indicate or imply relative importance or implicitly specify the number of the technical features indicated. Therefore, a feature specified as "first" or "second" may explicitly or implicitly include one or more of such features. "Multiple" means two or more, unless otherwise specifically defined.

[0117] In the present invention, unless otherwise expressly specified or limited, the terms "mounted," "connected," "connect," "fixed," etc. should be understood broadly. For example, they may refer to fixed connection, detachable connection, or integration; mechanical connection or electrical connection; direct connection or indirect connection through an intermediate medium; internal communication between two components or interaction between two components. Those skilled in the art will understand the specific meanings of the above terms in the present invention based on specific circumstances.

[0118] In the present invention, unless otherwise expressly specified or limited, when a first feature is "above" or "below" a second feature, it may mean that the first and second features are in direct contact, or that the first and second features are in indirect contact through an intermediary. Furthermore, when a first feature is "above," "above," or "above" a second feature, it may mean that the first feature is directly above or diagonally above the second feature, or simply means that the first feature is at a higher level than the second feature. When a first feature is "below," "below," or "below" a second feature, it may mean that the first feature is directly below or diagonally below the second feature, or simply means that the first feature is at a lower level than the second feature.

[0119] In the description of this specification, the reference terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" mean that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art can combine and combine different embodiments or examples described in this specification and features of different embodiments or examples without contradiction.

[0120] Any process or method description in a flowchart or otherwise described herein may be understood to represent a module, segment or portion of code comprising one or more executable instructions for implementing the steps of a specific logical function or process, and the scope of the preferred embodiments of the present invention includes alternative implementations in which functions may be performed out of the order shown or discussed, including performing functions in a substantially simultaneous manner or in the reverse order depending on the functions involved, which should be understood by those skilled in the art to which the embodiments of the present invention pertain.

[0121] The logic and / or steps represented in a flowchart or otherwise described herein, for example, can be considered a sequenced list of executable instructions for implementing the logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (e.g., a computer-based system, a system including a processor, or other system that can fetch and execute instructions from an instruction execution system, apparatus, or device). For purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transport a program for use by, or in conjunction with, an instruction execution system, apparatus, or device. More specific examples (not exhaustive) of computer-readable media include: an electrical connection with one or more wires (electronic devices), a portable computer disk cartridge (magnetic device), random access memory (RAM), read-only memory (ROM), erasable and programmable read-only memory (EPROM or flash memory), fiber optic devices, and a portable compact disc read-only memory (CDROM). Furthermore, the computer-readable medium may even be paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium and then editing, interpreting or processing it in another suitable manner if necessary, and then storing it in a computer memory.

[0122] It should be understood that various components of the present invention may be implemented using hardware, software, firmware, or a combination thereof. In the above-described embodiments, multiple steps or methods may be implemented using software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented using hardware, as in another embodiment, any one of the following technologies known in the art or a combination thereof may be used: a discrete logic circuit having logic gate circuits for implementing logic functions on data signals, an application-specific integrated circuit having suitable combinational logic gate circuits, a programmable gate array (PGA), a field-programmable gate array (FPGA), etc.

[0123] Those skilled in the art will understand that all or part of the steps in the method of the above embodiment can be completed by instructing related hardware through a program, and the program can be stored in a computer-readable storage medium. When the program is executed, it includes one or a combination of the steps of the method embodiment.

[0124] In addition, the functional units in the various embodiments of the present invention may be integrated into a single processing module, or each unit may exist physically separately, or two or more units may be integrated into a single module. The aforementioned integrated modules may be implemented in the form of hardware or in the form of software functional modules. If the integrated modules are implemented in the form of software functional modules and sold or used as independent products, they may also be stored in a computer-readable storage medium.

[0125] Although the embodiments of the present invention have been shown and described above, it will be understood that the above embodiments are illustrative and are not to be construed as limitations on the present invention. A person skilled in the art may change, modify, replace and modify the above embodiments within the scope of the present invention.

Claims

1. An assessment method based on water ecological environment monitoring, characterized in that: The following steps are involved: Construct a three-level monitoring system, dividing the target watershed into N primary monitoring areas, each of which is divided into M secondary sub-areas, each of which is divided into K tertiary micro-units, and each of which is equipped with monitoring equipment to collect data in real time; Acquire a monitoring data set collected in real time by monitoring equipment of each of the three-level micro-units in the three-level monitoring system; if the data missing rate of the target three-level micro-unit monitoring data set exceeds a preset threshold, perform cross-unit compensation on the monitoring data set to obtain a repaired data set; wherein the monitoring data set includes water quality parameters, flow velocity vectors, and biological activity indicators; Based on the restoration data set, a pollution assessment index for each third-level micro-unit is generated through a spatiotemporal feature extraction model, and the coefficient of variation of the biological activity index is simultaneously extracted; based on the hydraulic topology weight, the pollution assessment index of all third-level micro-units within the corresponding second-level sub-area is aggregated to generate a sub-basin comprehensive assessment value; the assessment results of adjacent first-level monitoring areas are corrected for lag effects through a time series recursive model to generate a basin-level comprehensive assessment index; The basin-level comprehensive assessment index is integrated with the output data of the HEC-RAS hydrodynamic model to generate a dynamic pollution diffusion heat map; the pollutant migration path is annotated using the Lagrangian particle tracking algorithm; and a three-dimensional interactive map containing the ecological risk level, diffusion range, and remediation priority is output.

2. The evaluation method based on water ecological environment monitoring according to claim 1 is characterized in that: The three-level monitoring system is constructed as follows: Acquiring a digital elevation model dataset, and calculating river channel curvature parameters and a water flow direction matrix based on the digital elevation model dataset; Generating a dynamic monitoring grid according to the river curvature parameter, generating an encrypted monitoring grid when the river curvature parameter is greater than a preset threshold, otherwise generating a standard monitoring grid; A three-level monitoring system is constructed based on the encrypted monitoring grid and the standard monitoring grid, and the target basin is divided into N first-level monitoring areas, each first-level area is divided into M second-level sub-areas, and each second-level sub-area is divided into K third-level micro-units.

3. The evaluation method based on water ecological environment monitoring according to claim 2 is characterized in that: The cross-unit compensation method is: Constructing a spatiotemporal association network, wherein the nodes of the spatiotemporal association network are adjacent three-level micro-units, and the edge weights are dynamically calculated based on the hydraulic distance and flow velocity between the three-level micro-units; Based on the spatiotemporal correlation network, compensatory data is generated through a graph attention mechanism; The compensation data is fused with the effective measured data of the target tertiary micro-unit in a credibility-weighted manner to obtain a repair data set.

4. The evaluation method based on water ecological environment monitoring according to claim 3 is characterized in that: The method for constructing the spatiotemporal correlation network includes: Define node feature vectors, including water quality parameters, biological activity indicators, and historical pollution trends; When calculating edge weights, a time decay factor is introduced, and the time decay factor is associated with the half-life period of the pollutant; The association strength of adjacent nodes is normalized by the Softmax function to generate a spatiotemporal association matrix.

5. The evaluation method based on water ecological environment monitoring according to claim 4 is characterized in that: When the compensation data is fused with the effective measured data of the target tertiary micro-unit in a credibility-weighted manner, a credibility coefficient of the compensation data is set, and the credibility coefficient is inversely proportional to the hydraulic distance between the target tertiary micro-unit and the adjacent tertiary micro-unit; when the credibility coefficient is lower than the set credibility coefficient threshold, the manual verification process is triggered and the automatic evaluation is suspended.

6. The evaluation method based on water ecological environment monitoring according to claim 5 is characterized in that: If the mutation rate of the basin-level comprehensive evaluation index exceeds the historical benchmark value, an incremental training set is intercepted based on a sliding time window, and a meta-learning optimizer is used to perform online parameter updates on the spatiotemporal feature extraction model and the temporal recursive model.

Images