A river and lake health intelligent evaluation and treatment method and device, a terminal and a medium

Abstract

The application discloses a river and lake health intelligent evaluation and treatment method and device, a terminal and a medium. The scheme provided by the application firstly obtains multi-dimensional monitoring index data of a target river and lake, obtains subjective and objective weights by analytic hierarchy process and entropy weight method respectively, and combines a river section landform type to introduce a correction coefficient for scene calibration. Then, a convex optimization model is constructed based on game theory, and a global optimal combination weight is obtained by solving, so that balanced fusion of subjective and objective weighting is realized. Subsequently, a health comprehensive score is calculated and graded, the degradation contribution proportion of each index is quantified by combining an XGBoost model and a SHAP interpretability framework, and precise tracing is realized. Finally, by means of a feature mapping matching algorithm, the degradation factor and the treatment countermeasure knowledge base are matched in similarity in a unified feature space, and a river and lake grading treatment scheme is output. The application breaks through the "data input-health evaluation-degradation diagnosis-countermeasure generation" full-intelligent closed loop, greatly improves the scientificity of river and lake health evaluation and the adaptability of treatment schemes, and can be widely applied to the health evaluation and ecological treatment decision-making scene of river and lake basins.

Description

Technical Field

[0001] This application relates to the field of river and lake ecological governance technology, and in particular to a method, device, terminal and medium for intelligent evaluation and governance of river and lake health. Background Technology

[0002] River and lake health assessment is a core foundational task for watershed ecological protection and governance. Scientific and accurate assessment results are crucial for formulating governance plans and carrying out ecological restoration. With the full implementation of my country's river chief system, the demand for river and lake health assessment continues to grow, placing higher demands on the scientific rigor, accuracy, and efficiency of the assessment.

[0003] Existing river and lake health assessment technologies have the following three main shortcomings: First, the methods for assigning weights to indicators have significant limitations. Traditional weighting methods fall into two categories: subjective and objective. Subjective weighting, such as the Analytic Hierarchy Process (AHP), relies on expert experience, is highly arbitrary, and is easily influenced by personal judgment biases. Objective weighting, such as the entropy weighting method, is entirely data-driven, easily deviating from the common sense of river and lake management, resulting in weight allocations that do not conform to hydrological logic. Most existing combined weighting schemes adopt a linear weighting method with manually specified proportions, such as 60% subjective weight and 40% objective weight. Essentially, this still relies on human decision-making, lacks mathematical optimality support, and cannot achieve a balanced integration of subjective and objective information. Furthermore, most existing schemes use uniform weights across the entire region, failing to adapt to the differences in health-dominant factors across different geomorphological and functional river sections, resulting in a "one-size-fits-all" evaluation bias.

[0004] Second, intelligent evaluation models generally suffer from the black box problem. In recent years, machine learning models have been gradually applied to river and lake health assessments, but most of them only remain at the level of "input indicators, output scores," failing to explain the reasons for the evaluation results or quantify the contribution of each indicator to river degradation. Management departments can only arrive at the conclusion of "unhealthy," but cannot pinpoint the core causes of degradation, making it difficult to support precise governance decisions and significantly reducing the practical value of the evaluation results.

[0005] Third, there is a serious disconnect between the evaluation and governance stages. Most existing river and lake health assessment technologies end at the health level determination stage, while the development of governance plans still relies on manual analysis of problems and matching of measures. This is not only inefficient but also highly dependent on the experience of the personnel involved, which can easily lead to insufficient adaptability of measures to degradation problems and a lack of specificity. It is impossible to form an intelligent closed loop of "evaluation-diagnosis-governance" and cannot meet the needs of large-scale, routine watershed evaluation and governance.

[0006] Therefore, there is an urgent need in this field to develop a river and lake health assessment technology that is scientifically objective, highly interpretable, and capable of intelligently matching governance solutions, in order to overcome the aforementioned shortcomings of existing technologies. Summary of the Invention

[0007] In view of this, this application provides a method, device, terminal and medium for intelligent evaluation and management of river and lake health, which is used to solve the technical problems of strong subjectivity in weighting, black box evaluation and disconnect between evaluation and management in the prior art.

[0008] To address the aforementioned technical problems, the first aspect of this application provides a method for intelligent assessment and management of river and lake health, comprising: Acquire multidimensional monitoring index data for the target river section, complete data preprocessing, and obtain a standardized index dataset; The subjective weight vector of the indicator system is calculated using the analytic hierarchy process (AHP), and the objective weight vector is calculated using the entropy weight method. The geomorphological correction coefficient vector is obtained based on the geomorphological type of the target river channel, and the subjective and objective weights are corrected and normalized. A game optimization model is constructed based on relative entropy, and the globally optimal combined weight vector is obtained by solving the model. The overall health score is calculated and the health level is divided based on the globally optimal combined weight vector. At the same time, the XGBoost regression model is constructed and trained. The degradation contribution ratio of each monitoring indicator was calculated based on the SHAP interpretability framework, and the core degradation driving factors were obtained by ranking them. A pre-built knowledge base of structured governance strategies is constructed, which maps the core degradation drivers and governance strategies to a unified feature space. The comprehensive matching degree is calculated and sorted and filtered to obtain a set of suitable governance strategies and generate a structured health governance plan.

[0009] Preferably, the step of constructing a game optimization model based on relative entropy and solving for the globally optimal combined weight vector specifically includes: Treating the modified subjective weight vector and the modified objective weight vector as two game participants, and using relative entropy to measure the information deviation between the combined weight and the two types of weights, a game optimization objective function is constructed with the goal of minimizing the total information deviation: In the formula, For demand dimensional combined weight vector, The total number of evaluation indicators; The corrected subjective weight vector The Each indicator weight, The corrected objective weight vector The Weight of each indicator; It is the natural logarithm; Introduce weight constraints: , This constitutes a convex optimization problem; Constructing the Lagrangian function and solving for the analytical solution, we finally obtain the globally optimal combined weights: , Preferably, the step of obtaining the geomorphic correction coefficient vector based on the geomorphic type of the target river section, and correcting and normalizing the subjective and objective weights, specifically includes: The landform types include four categories: mountain canyon type, hill transition type, plain meandering type, and urban waterfront type. Each landform type corresponds to a set of correction coefficient vectors. The correction coefficient of the dominant health factor is greater than 1, the correction coefficient of the non-dominant factor is less than 1, and the correction coefficient of the general index is equal to 1. Optionally, based on the landform correction, a water season (high water season, normal water season, low water season) correction coefficient is superimposed. The total correction coefficient is the product of the corresponding elements of the correction coefficients of each dimension. After correction, it is normalized again.

[0010] Preferably, the step of calculating the degradation contribution ratio of each monitoring indicator based on the SHAP interpretability framework and ranking them to obtain the core degradation driving factors specifically includes: The TreeExplainer tool, based on the SHAP framework, calculates the SHAP value of each indicator for each sample in the trained XGBoost tree model, representing the marginal contribution of the indicator to the health score of the sample. Calculate the average absolute SHAP value of all indicators and the contribution percentage of each indicator to the degradation of the river section's health status. Sort by contribution percentage from highest to lowest, and select the top N indicators as the core degradation driving factors, where N is a preset positive integer.

[0011] Preferably, it further includes: Based on the measured improvement data of historical governance projects, the basic weight coefficients of each governance strategy are reverse-calibrated, and the governance strategy knowledge base is iteratively updated in combination with the adaptation effect of river section scenarios. Based on the set of governance countermeasures obtained through matching, the extent of improvement in health scores after the implementation of governance is pre-assessed, and a prediction result of governance effectiveness is generated.

[0012] Meanwhile, the second aspect of this application provides a smart assessment and management device for river and lake health, comprising: The basic data preprocessing unit is used to acquire multi-dimensional monitoring indicator data of the river section to be evaluated, complete data preprocessing, and obtain a standardized indicator dataset. The weight calculation and evaluation unit is used to calculate subjective and objective weight vectors, calibrate them in combination with the topography correction coefficient, solve the global optimal combination weight vector through a game theory model, calculate the comprehensive health score and classify health levels, and train the XGBoost regression model. The degradation source analysis unit is used to calculate the degradation contribution ratio of each indicator based on the SHAP interpretability framework, and output the core degradation driving factors in a sorted manner. The countermeasure matching and solution generation unit has a built-in structured governance countermeasure knowledge base, which is used to output suitable hierarchical governance countermeasures through feature mapping matching algorithms and generate structured health governance solutions.

[0013] Preferably, the weight calculation and evaluation unit is specifically used for: The modified subjective weight vector and the modified objective weight vector are regarded as two game participants. The information deviation between the combined weight and the two types of weight is measured by relative entropy. A game optimization objective function with the goal of minimizing the total information deviation is constructed. By introducing constraints that the sum of the weights is 1 and the weights are non-negative, a Lagrangian function is constructed to solve for the analytical solution, thus obtaining the globally optimal combined weight vector.

[0014] Preferably, it further includes: a knowledge base iterative optimization unit, used for: Based on the measured improvement data of historical governance projects, the basic weight coefficients of each governance strategy are reverse-calibrated, and the governance strategy knowledge base is iteratively updated in combination with the adaptation effect of river section scenarios. Based on the set of governance countermeasures obtained through matching, the extent of improvement in health scores after the implementation of governance is pre-assessed, and a prediction result of governance effectiveness is generated.

[0015] The third aspect of this application provides a smart evaluation and governance terminal for river and lake health based on game theory empowerment and interpretable traceability, including: a memory and a processor; The memory is used to store program code, which corresponds to the intelligent evaluation and governance method for river and lake health based on game theory weighting and interpretable traceability provided in the first aspect of this application; The processor is used to read and execute the program code.

[0016] The fourth aspect of this application provides a computer-readable storage medium storing program code, which is read and executed by a processor to implement a method for intelligent evaluation and governance of river and lake health based on game theory weighting and interpretable traceability, as provided in the first aspect of this application.

[0017] As can be seen from the above technical solutions, this application has the following beneficial effects: The scientific rigor of weighting is significantly enhanced, and its adaptability to various scenarios is strengthened. This application constructs a game theory convex optimization model based on relative entropy to solve for the globally optimal combination of weights. This eliminates the need for manually specifying the combination coefficients of subjective and objective weights, ensuring a balanced integration of subjective and objective information from a mathematical perspective and avoiding the subjective bias of human decision-making. At the same time, a geomorphological zoning correction coefficient is introduced, which can dynamically calibrate the weights according to the geomorphological type of the river section, solving the problem of "uniformity across all rivers" caused by traditional uniform weights. The evaluation results are more in line with the actual situation of the river channel.

[0018] The evaluation results are interpretable, enabling precise tracing of degradation sources. This application introduces the SHAP interpretability framework and combines it with the XGBoost model to quantify the marginal contribution of each indicator to the health status of the river section, clearly identifying the core degradation drivers for ranking. This breaks through the black box problem of traditional intelligent evaluation models, allowing the evaluation results to not only provide conclusions but also traceable causes, directly supporting precise governance decisions.

[0019] This application establishes a closed-loop evaluation and governance system to achieve intelligent solution generation. It constructs a structured knowledge base of governance strategies and uses a feature mapping matching algorithm to calculate the similarity between degradation factors and governance strategies within a unified feature space. Combined with river segment scenario coefficients, it achieves precise matching and automatically generates tiered "one river, one policy" governance solutions. This directly transforms evaluation results into implementable governance measures, significantly improving the efficiency and adaptability of governance solution development. It fully realizes a complete intelligent closed loop of "data input - health assessment - degradation diagnosis - strategy generation."

[0020] The solution is highly scalable and widely applicable. The geomorphic correction coefficient in this application can be expanded to include multi-dimensional corrections during different water seasons, and the knowledge base can be continuously updated with new governance measures without altering the core algorithm logic. It can adapt to the health assessment and governance needs of different watersheds and types of rivers and lakes, and has strong engineering application value. Attached Figure Description

[0021] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0022] Figure 1 This is a flowchart illustrating an embodiment of an intelligent evaluation method for river and lake health provided in this application.

[0023] Figure 2 This application provides a schematic diagram of the segmented distribution of landforms along a river channel for intelligent river and lake health assessment.

[0024] Figure 3 This is a flowchart illustrating the process of assigning weights to the globally optimal combination based on relative entropy in this application.

[0025] Figure 4 A schematic diagram of the feature mapping intelligent strategy matching process provided in this application.

[0026] Figure 5 This is a schematic diagram of an embodiment of a smart evaluation device for river and lake health provided in this application.

[0027] Figure 6 This is a structural schematic diagram of an embodiment of a smart assessment and management terminal for river and lake health provided in this application. Detailed Implementation

[0028] This application provides a method, device, terminal, and medium for intelligent evaluation and management of river and lake health, which addresses the technical problems of strong subjectivity in weighting, black-box evaluation, and disconnect between evaluation and management in the prior art.

[0029] To make the inventive objectives, features, and advantages of this application more apparent and understandable, the technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the embodiments described below are only some embodiments of this application, and not all embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0030] First, a detailed description of an embodiment of the intelligent assessment and management method for river and lake health provided in this application is as follows: Please refer to Figure 1. An embodiment of the intelligent assessment method for river and lake health provided in this application includes: Step 101: Obtain multi-dimensional monitoring indicator data for the river section to be evaluated, complete data preprocessing, and obtain a standardized indicator dataset.

[0031] It should be noted that, firstly, for the target river section requiring health assessment, data on five major categories of monitoring indicators should be obtained: hydrology and water resources, physical morphology, water environment, aquatic organisms, and social services. These indicators cover a range of specific metrics, including ecological flow satisfaction, flow velocity, shoreline hardening rate, beach encroachment ratio, COD concentration, ammonia nitrogen concentration, total phosphorus concentration, dissolved oxygen concentration, benthic biodiversity, vegetation coverage, and landscape comfort. Because different indicators have different dimensions, value ranges, and degrees of favorability, standardized preprocessing is necessary to ensure the scientific validity of subsequent weight calculations and comprehensive evaluations.

[0032] More specifically, the data preprocessing process includes: removing outliers exceeding reasonable ranges and supplementing missing data using linear interpolation; normalizing positive indicators (higher values ​​indicate higher health, such as dissolved oxygen and vegetation coverage) and negative indicators (higher values ​​indicate lower health, such as pollutant concentration and shoreline hardening rate) respectively, mapping all indicator values ​​to the [0,1] interval to obtain a standardized indicator dataset. The required river dataset is shown in Table 1.

[0033] Step 102: Calculate the subjective and objective weight vectors respectively, perform calibration in conjunction with terrain correction, and solve for the global optimal combined weight vector.

[0034] It should be noted that this step first obtains the initial subjective and objective weights using the analytic hierarchy process (AHP) and the entropy weight method, respectively.

[0035] like Figure 2 As shown, scenario-based modifications are made in conjunction with the river section's landform type, and the global optimal combination weights are finally solved using a game theory model based on relative entropy.

[0036] As shown in Figure 3, the specific execution process of this step is as follows: (1) Calculation of initial subjective and objective weights Subjective weight calculation: Based on industry standard documents, an indicator judgment matrix is ​​constructed. After passing the consistency test, the subjective weight vector is calculated using the Analytic Hierarchy Process (AHP). This represents the subjective experience's judgment on the importance of the indicator.

[0037] Objective weight calculation: Based on historical monitoring data of the watershed, the information entropy of each indicator is calculated using the entropy weight method. The smaller the information entropy, the more information the indicator provides, and the higher its weight. The final objective weight vector is obtained. This characterizes the importance of the indicators reflected by the measured data.

[0038] (2) Adjustment of geomorphic zoning weights Based on the topography and landforms of the target river section and the intensity of human activities, its landform type is determined, including four categories: mountain canyon type, hilly transition type, plain meandering type, and urban waterfront type. Each landform type corresponds to a set of preset correction coefficient vectors. The range of the correction factor is controlled within... For this type of river section, the correction coefficient of the dominant health factor corresponding to the indicator is greater than 1, and the weight ratio is increased; the correction coefficient of the non-dominant factor corresponding to the indicator is less than 1, and the weight ratio is decreased; the correction coefficient of the general importance indicator is equal to 1, and the weight remains unchanged.

[0039] The subjective and objective weights are adjusted using element-weighted methods, and then re-normalized to ensure that the sum of the weights is 1. , in This is the Hadamard product, which is the multiplication of elements at corresponding positions.

[0040] Optionally, based on the geomorphological correction, a water season correction coefficient (high water season, normal water season, and low water season) can be superimposed. The total correction coefficient is multiplied by the corresponding elements of the correction coefficients for each dimension, and then normalized again to achieve multi-dimensional scene adaptation. The river geomorphological correction coefficients can be seen in Table 2.

[0041] Note: The total correction factor is the product of the corresponding elements of the correction factors of each dimension. After correction, it needs to be renormalized to ensure that the weight sum is 1. The wet / normal / dry season correction factors can be superimposed on this basis.

[0042] (3) Solving for the global optimal combination weight using game theory The revised subjective weight With objective weight We consider two game participants, representing two independent information sources: objective empirical information and measured data information. The weights are essentially probability distributions of indicator importance. Relative entropy is used to measure the information deviation between the two probability distributions, which better reflects the physical meaning of the weights than Euclidean distance. We construct a game optimization objective function that minimizes the total information deviation: Introduce weight constraints: all weights are non-negative, and the sum of the weights is 1, i.e. , Since the relative entropy function is a strictly convex function, the objective function is also a strictly convex function, and the constrained feasible region is a convex set. Therefore, this problem has a unique global optimal solution.

[0043] Introducing Lagrange daily numbers Construct the Lagrange function: Finally, the analytical solution for the globally optimal combined weights is obtained: The result is a nonlinear combination of the objective and objective weights after geometric mean normalization. It is derived entirely from the game optimization model without the need for manual specification of the combination ratio. It is an equilibrium fusion of the two types of information under the premise of minimizing information loss, corresponding to the Nash equilibrium state in game theory.

[0044] Step 103: Calculate the overall health score and classify the health levels based on the globally optimal combined weights, and at the same time construct and train the XGBoost regression model.

[0045] It should be noted that the overall health score of the river segment is obtained by weighting and summing the standardized indicator data with the globally optimal combined weights. in For the first The standardized values ​​of each indicator.

[0046] Simultaneously, long-term monitoring data and corresponding health scores of the river section were collected to construct a sample dataset. This dataset was then divided into training and test sets according to a certain ratio (generally 6:4). An XGBoost regression model was trained using standardized indicator data as input and the comprehensive health score as output. After training, the model's fitting accuracy was verified using the test set to ensure its usability for subsequent degradation source analysis.

[0047] Step 104: Calculate the degradation contribution ratio of each monitoring indicator based on the SHAP interpretability framework, and sort them to obtain the core degradation driving factors.

[0048] It should be noted that this step is based on the trained XGBoost model and uses the SHAP interpretability framework to conduct degradation tracing, quantify the marginal contribution of each indicator to the health status of the river section, and break the black box of evaluation.

[0049] The specific calculation process is as follows: (1) The TreeExplainer tool with the SHAP framework was used to calculate the SHAP value of each indicator in the target river segment sample for the tree-structured XGBoost model. The SHAP value represents the marginal contribution of the indicator value to the current sample health score. A positive value indicates an increase in the health score, and a negative value indicates a decrease in the health score. The larger the absolute value, the greater the contribution.

[0050] (2) Calculate the average absolute SHAP value of each indicator and the contribution percentage of each indicator to river degradation: (3) Rank the indicators from highest to lowest according to their contribution ratio, select the top N indicators as the core degradation driving factors, and clarify the main causes of the river degradation.

[0051] Step 105: Pre-build a structured governance strategy knowledge base, map the core degradation driving factors and governance strategies to a unified feature space, calculate the comprehensive matching degree and sort and filter to obtain a set of suitable governance strategies, and generate a structured health governance plan.

[0052] like Figure 4 As shown, this step uses a feature mapping matching algorithm to achieve intelligent matching of degradation factors to governance strategies. The specific process is as follows: (1) Construct a knowledge base for governance strategies The knowledge base uses a three-level tree structure for storage, including three levels: degradation categories, driving factor tags, and governance strategy entries. Each governance strategy entry includes fields such as strategy ID, applicable degradation feature tags, applicable river type, graded measures, measure content, expected effects, regulatory basis, and basic weight. The measures are graded into two categories: short-term emergency measures and medium- and long-term ecological restoration measures.

[0053] (2) Construct a unified feature space A unified feature space with fixed dimensions is constructed based on a pre-defined degenerate feature label dictionary. The dictionary dimension is consistent with the total number of degenerate driving factors, and each index position corresponds to a degenerate factor.

[0054] (3) Feature vectorization Degradation feature vector: The contribution ratio of the core degradation driving factor is filled into the corresponding index position, and the rest are 0, to generate a weighted degradation feature vector, so that the dominant degradation factor has a higher matching weight.

[0055] Countermeasure feature vector: For each countermeasure in the knowledge base, the position corresponding to the solvable degradation factor is assigned a value of 1, and the rest are 0, thus generating a countermeasure feature vector.

[0056] (4) Calculate the overall matching degree First, calculate the cosine similarity between the degenerate feature vector and the feature vector of each countermeasure to measure the degree of feature overlap between the two: in For degenerate feature vectors, This represents the feature vector of the strategy.

[0057] Combining river type adaptation coefficient Basic weights of countermeasures Calculate the overall matching degree: Among them, the river section type adaptation coefficient is 1.2 when the applicable river section matches the current river section, 1.0 for general river sections, and 0 for inapplicable river sections and is directly filtered; the basic weight of the countermeasure is set by domain experts in combination with policy priority, technology maturity and certainty of governance effectiveness, with a benchmark value of 1.0 and a value range of [0.7, 1.3].

[0058] Set a comprehensive matching degree threshold, filter out invalid countermeasures below the threshold, sort them from high to low comprehensive matching degree, and select the top N countermeasures as the governance countermeasure set.

[0059] (5) Generate structured governance solutions Following the "one river, one policy" standard format in the water conservancy industry, a structured river health management plan document is generated by integrating the river section's health overview, core degradation drivers, tiered management measures, and long-term control recommendations.

[0060] Furthermore, this embodiment also includes knowledge base iteration and performance prediction steps: Based on measured indicator improvement data from historical governance projects, the basic weight coefficients of each governance strategy are reverse-calibrated. Combined with the adaptation effect of river section scenarios in practical applications, the governance strategy knowledge base is continuously iterated and updated. Simultaneously, based on the matched set of governance strategies and the expected improvement range of each strategy, the health score improvement after governance implementation is pre-assessed, generating a governance effectiveness prediction result to provide a quantitative reference for governance decisions. The river strategy knowledge base is shown in Table 3.

[0061] The above is a detailed description of an embodiment of a smart assessment and management method for river and lake health provided in this application. The following is a detailed description of an embodiment of a smart assessment and management device for river and lake health provided in this application.

[0062] Please refer to Figure 5. An embodiment of a smart river and lake health assessment and management device based on game theory weighting and interpretable traceability provided in this application includes: The basic data preprocessing unit 201 is used to acquire multi-dimensional monitoring indicator data of the river section to be evaluated, complete data preprocessing, and obtain a standardized indicator dataset. The weight calculation and evaluation unit 202 is used to calculate the subjective and objective weight vectors and calibrate them in combination with the landform correction coefficient. The Health Assessment and Modeling Module 203 uses a game theory model to solve for the globally optimal combination of weight vectors, calculates the comprehensive health score and classifies health levels, and trains the XGBoost regression model. The degradation source analysis unit 204 is used to calculate the degradation contribution ratio of each indicator based on the SHAP interpretability framework and output the core degradation driving factors in a sorted manner. The countermeasure matching and solution generation unit 205 has a built-in structured governance countermeasure knowledge base, which is used to output suitable hierarchical governance countermeasures through feature mapping matching algorithms and generate structured health governance solutions.

[0063] Furthermore, the weight calculation and evaluation units 202 and 203 are specifically used for: The modified subjective weight vector and the modified objective weight vector are regarded as two game participants. The information deviation between the combined weight and the two types of weight is measured by relative entropy. A game optimization objective function with the goal of minimizing the total information deviation is constructed. By introducing constraints that the sum of the weights is 1 and the weights are non-negative, a Lagrangian function is constructed to solve for the analytical solution, thus obtaining the globally optimal combined weight vector.

[0064] Furthermore, it also includes: a knowledge base iterative optimization unit 206, used for: Based on the measured improvement data of historical governance projects, the basic weight coefficients of each governance strategy are reverse-calibrated, and the governance strategy knowledge base is iteratively updated in combination with the adaptation effect of river section scenarios. Based on the set of governance countermeasures obtained through matching, the extent of improvement in health scores after the implementation of governance is pre-assessed, and a prediction result of governance effectiveness is generated.

[0065] like Figure 6 As shown, the third aspect of this application provides a smart evaluation terminal for river and lake health. The types of terminals include, but are not limited to, personal computers, servers, and embedded smart terminals. The main components of the terminal include a memory 33 and a processor 31, which can be connected via a communication bus 34. The memory 33 is used to store program code, which corresponds to a method for intelligent evaluation of river and lake health as provided in the above embodiments; Processor 31 is used to read and execute program code.

[0066] The fourth aspect of this application provides a computer-readable storage medium storing program code, which is read and executed by a processor to implement a river and lake health intelligent evaluation method as provided in the above embodiments.

[0067] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the terminals, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0068] In the several embodiments provided in this application, it should be understood that the disclosed terminals, devices, and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection between devices or units through some interfaces, and may be electrical, mechanical, or other forms.

[0069] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a particular order or sequence. It should be understood that such data can be interchanged where appropriate so that embodiments of the application described herein can be implemented, for example, in orders other than those illustrated or described herein. Furthermore, the terms “comprising” and “having,” and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0070] It should be understood that in this application, "at least one (item)" means one or more, and "more than" means two or more. "And / or" is used to describe the relationship between related objects, indicating that three relationships can exist. For example, "A and / or B" can represent three cases: only A exists, only B exists, and both A and B exist simultaneously, where A and B can be singular or plural. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. "At least one (item) of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one (item) of a, b, or c can represent: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, and c can be single or multiple.

[0071] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0072] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0073] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0074] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A method for intelligent assessment and management of river and lake health, characterized in that, include: S1. Obtain multi-dimensional monitoring index data of the target river channel, complete data preprocessing, and obtain a standardized index dataset; S2. Calculate the subjective weight vector of the indicator system using the Analytic Hierarchy Process (AHP). The objective weight vector of the indicator system is calculated using the entropy weight method. ); S3. Based on the topographic features of the target river channel, obtain the corresponding topographic correction coefficient vector ( The subjective weight and objective weight vectors are weighted and corrected respectively, and then normalized to obtain the corrected subjective weight vector. ) and the corrected objective weight vector ( ); S4. Using the modified subjective weight vector and objective weight vector as game participants, construct an optimization model with the goal of minimizing the total weight deviation, and obtain the globally optimal combined weight vector by solving the Lagrange operator. S5. Based on the global optimal combined weight vector, calculate the comprehensive health score of the target river channel and classify the river channel health level according to the grading threshold; at the same time, construct an XGBoost regression model, using standardized index data as input and the comprehensive health score of the river channel as output to complete the model training. S6. Based on the SHAP interpretable framework, calculate the SHAP value of each monitoring indicator in the trained XGBoost regression model, quantify the contribution of each indicator to river degradation, and rank them to obtain the core degradation driving factors. S7. Construct a structured governance strategy knowledge base, map the core degradation driving factors and each governance strategy scheme in the knowledge base to a unified feature space, and generate degradation feature vectors and strategy feature vectors respectively. S8. Calculate the cosine similarity between the degradation feature vector and the feature vector of each countermeasure, and combine it with the type fit coefficient of the evaluated river channel to obtain the comprehensive matching degree of each countermeasure; sort and filter according to the comprehensive matching degree to obtain the set of governance countermeasures suitable for the river to be evaluated. S9. Based on the river health level, core degradation driving factors and governance countermeasures, generate a structured health governance plan for the target river.

2. The method for river health assessment and management according to claim 1, characterized in that, Step S4 specifically includes: S41. Treat the modified subjective weight vector and the modified objective weight vector as two game participants, use relative entropy to measure the information deviation between the combined weight and the two types of weights, and construct a game optimization objective function with the goal of minimizing the total information deviation: In the formula, For those in demand dimensional combined weight vector, The total number of evaluation indicators; The corrected subjective weight vector The Each indicator weight, The corrected objective weight vector The Weight of each indicator; It is the natural logarithm; Introduce weight constraints: ， S42. Constructing Lagrange daily numbers Construct the Lagrange function: S43. Substituting the normalization constraints, we obtain the globally optimal solution, and finally, the globally optimal combination weights in game theory: ， S44. Optimal solution constraint verification: After the solution is obtained, two verifications are performed: normalization verification and non-negativity verification, to ensure that the result meets the physical constraints.

3. The method for assessing the health of rivers and lakes according to claim 1, characterized in that, In step S3, the landform types include four categories: mountain canyon type, hill transition type, plain meandering type, and urban waterfront type. Each type of landform corresponds to a set of correction coefficient vectors. Optionally, based on the landform correction, a water season correction coefficient is superimposed, and the total correction coefficient is the product of the corresponding elements of the correction coefficients of each dimension, and then normalized again after correction.

4. The intelligent assessment and management method for river and lake health according to claim 1, characterized in that, Step S6 specifically includes: S61. The TreeExplainer tool based on the SHAP framework calculates the SHAP value of each indicator for each sample in the trained XGBoost tree model, representing the marginal contribution of the indicator to the health score of the sample. S62. Calculate the average absolute SHAP value of all indicators and calculate the contribution ratio of each indicator to the degradation of the river health status. S63. Sort by contribution percentage from highest to lowest, and select the top N indicators as the core degradation driving factors, where N is a preset positive integer.

5. The intelligent evaluation method for river and lake health according to claim 1, characterized in that, In step S7, the governance strategy knowledge base is stored in a three-level tree structure, including three levels: degradation category, driving factor label, and governance strategy entry. Each governance strategy entry includes strategy ID, applicable degradation feature label, applicable river type, measure level, measure content, expected effect, standard basis, and basic weight field. The unified feature space is constructed based on a preset degradation feature label dictionary. The degradation feature vector uses the contribution ratio of each degradation factor as the element value, and the strategy feature vector is assigned a value of 1 for the degradation factor that the strategy can solve, and 0 for the rest.

6. The intelligent evaluation method for river and lake health according to claim 5, characterized in that, In step S7, the formula for calculating the overall matching degree is: in, The cosine similarity between the degenerate feature vector and the game feature vector is given. The river segment type fit coefficient is determined by the matching degree between the applicable river segment and the river segment to be evaluated. The basic weights are pre-set for the countermeasures.

7. The intelligent assessment and management method for river and lake health according to claim 1, characterized in that, In step S7, the governance strategies are classified into two categories according to the measures: short-term emergency measures and medium- and long-term ecological restoration measures. The structured health governance plan generated in step S8 is in the "one river, one policy" format that conforms to the water conservancy industry standards. It includes five parts: river health overview, main degradation problems and driving factors, a list of short-term emergency remediation measures, a list of medium- and long-term ecological restoration measures, and long-term management and control recommendations.

8. A smart assessment and management device for river and lake health, characterized in that, include: The data preprocessing module is used to acquire multidimensional monitoring index data of the river to be evaluated and output a standardized index dataset. The weight calculation module is used to calculate subjective and objective weights, and to perform calibration by combining the terrain correction coefficient. It solves the global optimal combination weight vector through a game theory model. The health assessment module is used to classify health levels based on the combined weighted comprehensive health score and to train the XGBoost regression model. The degradation tracing module is used to calculate the degradation contribution ratio of each indicator based on the SHAP interpretability framework and output the core degradation driving factors. The countermeasure matching module has a built-in structured governance countermeasure knowledge base, which is used to output appropriate hierarchical governance countermeasures through feature mapping matching algorithms. The scheme generation module is used to generate and output structured river section health management schemes.

9. A terminal for an intelligent evaluation system for river and lake health, characterized in that, include: Memory and processor; The memory is used to store program code, which corresponds to the intelligent evaluation method for river and lake health as described in any one of claims 1 to 7; The processor is used to read and execute the program code.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium contains program code that is read and executed by a processor to implement a method for intelligent evaluation and management of river and lake health as described in any one of claims 1 to 7.

Images