Comprehensive calculation method for river connectivity affected by small hydropower stations

Abstract

The present invention discloses a comprehensive calculation method for water system connectivity affected by small hydropower stations, including: dynamically identifying the connectivity state of tiny water bodies based on spatio-temporal multi-source data fusion, processing the water body presence frequency matrix through an adaptive threshold segmentation method, and combining hydrodynamic constraints to determine the water body connectivity state; identifying the implicit connectivity of the dewatered reach of diversion-type small hydropower stations; intelligently perceiving the operation state of small hydropower stations based on multi-modal data, and constructing a dynamic prediction model for diversion flow; establishing a spatio-temporal coupling model between the water system connectivity state and the operation of small hydropower stations to determine the connectivity critical thresholds for different ecological objectives; and constructing a feedback regulation mechanism between the connectivity state of tiny water bodies and the operation of small hydropower stations. The present invention realizes the accurate evaluation and optimal regulation of water system connectivity affected by small hydropower stations, and provides a scientific basis for the rectification plan of small hydropower stations.

Description

Technical Field

[0001] The present invention relates to the related technologies of hydropower stations, and in particular to a comprehensive calculation method for river connectivity affected by small hydropower stations. Background Art

[0002] As an important form of clean energy utilization, small hydropower development plays an important role in the global energy structure adjustment and carbon reduction process. However, the construction of small hydropower projects has a profound impact on river connectivity, changing the river hydrological regime, which may lead to river section drying up, fragmentation of aquatic biological habitats, and degradation of ecological functions. Scientifically evaluating the river connectivity affected by small hydropower is of great significance for balancing the relationship between energy development and ecological protection and realizing river health management.

[0003] Currently, the evaluation of river connectivity mainly adopts methods such as graph theory method, hydrological-hydraulic method, landscape method, and biological method. Traditional evaluation methods usually based on the static river network structure, and use indicators such as River Continuity Index (RCI), Dendritic Connectivity Index (DCI), and Channel Alteration Fragmentation Index (CAFI) to evaluate the impact of obstacles such as dams on river connectivity. These methods mainly focus on the impact of large-scale water conservancy projects, and the evaluation of connectivity at the basin scale and the main river channels is relatively mature. Remote sensing technology has been widely used in water body identification and river system extraction, and common methods include threshold method, supervised classification method, and water body index method, etc., but mainly for large water bodies.

[0004] However, the existing technologies still face the following key challenges in the evaluation of river connectivity affected by small hydropower: First, it is difficult to dynamically identify the connectivity state of small water bodies. Traditional water body segmentation methods have insufficient accuracy in identifying small or seasonally significantly changing water bodies, especially in the dewatered reaches of diversion-type small hydropower stations, and it is difficult to accurately reflect the spatio-temporal variation characteristics of the river connectivity state. Second, there is a lack of quantitative description methods for the dynamic coupling relationship between the operation state of small hydropower and river connectivity. Existing evaluation models are mostly based on static parameters and are difficult to reflect the dynamic changes of river connectivity under different operation modes. Especially in areas with intensive distribution of diversion-type small hydropower stations, the traditional Channel Alteration Fragmentation Index cannot accurately evaluate the hidden connectivity of the dewatered reaches, and there is also a lack of connectivity regulation mechanisms for different ecological objectives, resulting in the evaluation results being difficult to directly guide the operation management of small hydropower and the formulation of ecological restoration measures.

[0005] In response to these technical challenges, there is an urgent need to develop a comprehensive evaluation method that integrates the dynamic identification of small water bodies and the coupling of the operation state of small hydropower. Summary of the Invention

[0006] The object of the invention is to provide a comprehensive calculation method for river connectivity affected by small hydropower stations, in order to solve at least one technical problem existing in the prior art.

[0007] Technical solution: A comprehensive calculation method for river system connectivity affected by small hydropower stations, comprising the following steps:

[0008] Collect research data sets, including remote sensing image data, hydrological data, small hydropower station data, and water ecological environment data of the research area;

[0009] Using remote sensing image data, adopt deep learning methods to conduct refined segmentation of water bodies to generate river system network data;

[0010] Based on the river system network data and small hydropower station data, combined with hydrological data, calculate the basic indicators of river network connectivity;

[0011] According to the basic indicators of river network connectivity and the operation mode of small hydropower stations in the small hydropower station data, construct a spatio-temporal coupling model to generate dynamic connectivity indicators;

[0012] Based on the dynamic connectivity indicators and water ecological environment data, conduct a comprehensive evaluation of river system connectivity to form an evaluation result of the rectification priority of small hydropower stations.

[0013] Beneficial effects: Achieved precise evaluation of complex river system networks under the influence of small hydropower. Overcame the technical defects of traditional connectivity evaluation methods that cannot accurately identify tiny water bodies and cannot consider spatio-temporal variation characteristics. The relevant technical effects will be described in detail in combination with specific embodiments. Description of the drawings

[0014] Figure 1 is the flowchart of the present invention.

[0015] Figure 2 is the flowchart of the refined segmentation of the present invention.

[0016] Figure 3 is the flowchart of calculating the basic indicators of river network connectivity of the present invention.

[0017] Figure 4 is the flowchart of generating dynamic connectivity indicators of the present invention.

[0018] Figure 5 is the flowchart of conducting a comprehensive evaluation of river system connectivity of the present invention. Detailed implementation manners

[0019] As Figures 1 to 5 shown, the technical solution will be described in detail.

[0020] According to one aspect of the present application, the steps of using remote sensing image data and adopting deep learning methods to conduct refined segmentation of water bodies include:

[0021] Construct a spatio-temporal multi-source database, including multi-temporal optical remote sensing images, radar remote sensing data, and high-resolution images obtained by drones;

[0022] Based on the spatio-temporal multi-source database, calculate the water body change rate between different time intervals, and construct a water body presence frequency matrix;

[0023] Use the adaptive threshold segmentation method to process the water body presence frequency matrix and generate the segmentation results of small water bodies;

[0024] Based on the segmentation results of small water bodies and the digital elevation model, adopt the hydrodynamic constraint discrimination method to generate the dynamic connectivity state map of small water bodies;

[0025] Identify the intake and tailrace positions of diversion power stations in the location data of small hydropower projects, and combine with the dynamic connectivity state map of small water bodies to generate the implicit connectivity state of the dewatered river reaches;

[0026] Integrate the dynamic connectivity state map of small water bodies and the implicit connectivity state of the dewatered river reaches into the water system network data.

[0027] By constructing a spatio-temporal multi-source database and a series of water body segmentation algorithms, the technical problems that traditional remote sensing methods are difficult to accurately identify small water bodies and intermittent water bodies are solved. It is especially suitable for water body identification under complex terrain conditions in mountainous areas. By establishing a water body presence frequency matrix, the limitations of single-temporal remote sensing data affected by factors such as cloud cover and vegetation occlusion are effectively overcome. The application of the adaptive threshold segmentation method further improves the identification accuracy of small water bodies, with the accuracy rate increased by more than 25%. The introduction of the hydrodynamic constraint discrimination method makes the segmentation results conform to the actual hydrodynamic laws, effectively avoiding misclassification caused by terrain shadows, cloud cover, etc. By identifying the intake and tailrace positions of diversion power stations and combining with the dynamic connectivity state map of small water bodies, the quantitative expression of the implicit connectivity state of the dewatered river reaches is realized, providing a scientific basis for evaluating the actual impact of small hydropower on the water ecosystem.

[0028] According to one aspect of the present application, the steps of calculating the basic indicators of river network water system connectivity based on the water system network data, small hydropower station data, and combined with hydrological data include:

[0029] Based on the water system network data, combine the river network structure, river level, and catchment area information to calculate the dendritic connectivity index, river continuity index, and river fragmentation index;

[0030] Combine the development method and installed capacity parameters in the small hydropower station data to calculate the improved river fragmentation index for dam-type developed small hydropower;

[0031] Combine the small hydropower station data and hydrological data to calculate the improved river fragmentation index considering the impact of dewatering flow for diversion-type developed small hydropower stations;

[0032] Based on the spatial hierarchical relationship of the river network data, calculate the parameter values of the above indicators (dendritic connectivity index, river continuity index, river fragmentation index, improved river fragmentation index, improved river fragmentation index) at different river scales, regional scales, and basin scales respectively, and form the basic indicators of river network water system connectivity.

[0033] By integrating the dendritic connectivity index, river continuity index, and river fragmentation index, a multi-dimensional connectivity evaluation system is constructed. In view of the special impact of small hydropower, two different types of development, namely dam-type development and diversion-type development, are distinguished, and an improved calculation method for the river fragmentation index is designed respectively, which improves the accuracy and pertinence of the connectivity evaluation. In particular, for the improved river fragmentation index considering the impact of dewatering flow in diversion-type small hydropower development, it solves the technical problem that traditional indicators cannot quantitatively evaluate the impact of diversion-type hydropower on the reduction of water flow in river sections. By calculating the index parameter values at different river scales, regional scales, and basin scales respectively, a multi-scale connectivity evaluation framework is constructed, getting rid of the limitations of traditional single-scale evaluation, and enabling the evaluation results to meet different needs from micro-river section management to macro-basin planning. This method can quantitatively reflect the impact degree of small hydropower on the water system connectivity at different spatial scales, providing a strong technical support for scientifically formulating the optimized operation scheduling plan of small hydropower stations.

[0034] According to one aspect of the present application, the steps of constructing a spatio-temporal coupling model and generating dynamic connectivity indicators based on the basic indicators of river network water system connectivity and the operation mode data of small hydropower stations include:

[0035] According to the gate dam scheduling rules in the operation mode data of small hydropower stations and hydrological data, construct a time matrix of the connectivity situation of river sections, and divide each river section into three situations: continuous connectivity, periodic connectivity, and temporary connectivity;

[0036] Based on the operation mode data of small hydropower stations and hydrological data, construct a matrix of the change of the diversion flow of power stations and river flow;

[0037] According to the time matrix of the connectivity situation of river sections and the matrix of the change of the diversion flow of power stations and river flow, combined with the basic indicators of river network water system connectivity, calculate the dynamic change indicators of connectivity at different time scales;

[0038] Construct a spatio-temporal coupling analysis framework, integrate the dynamic connectivity state of tiny water bodies and the operation state data of small hydropower, and generate dynamic connectivity indicators.

[0039] By constructing a spatio-temporal coupling model, the evaluation has been realized from static connectivity evaluation to dynamic connectivity evaluation. Based on the time matrix of river reach connectivity and the matrix of the change in the diversion flow of power stations and river flow, this model accurately depicts the dynamic impact of small hydropower operation on river system connectivity, overcoming the limitation of traditional evaluation methods that cannot reflect the time-varying characteristics of connectivity. By classifying river reaches into three situations: continuous connectivity, periodic connectivity, and temporary connectivity, the model can accurately reflect the variation law of river connectivity status under different hydrological conditions, and the evaluation accuracy has been improved by more than 35% compared with traditional static evaluation. In particular, the constructed spatio-temporal coupling analysis framework realizes the organic integration of the dynamic connectivity status of small water bodies and the operation status data of small hydropower, making the evaluation results closer to the actual hydrological and ecological processes. The generation of dynamic connectivity indicators provides a scientific basis for formulating small hydropower operation and scheduling schemes that meet ecological requirements, which can minimize the adverse impact on the water ecosystem while ensuring power generation benefits, and realize the coordinated development of small hydropower and the ecological environment.

[0040] According to one aspect of the present application, the steps for comprehensively evaluating river system connectivity based on dynamic connectivity indicators and water ecological environment data include:

[0041] Based on water ecological environment data, determine the set of influencing indicators for connectivity evaluation, including the proportion of surface water, comprehensive water quality index, ecological flow, and biological indicators;

[0042] Perform dimensionless processing on the data of each indicator in the set of influencing indicators for connectivity evaluation, calculate the information entropy of each indicator, and determine the indicator weights;

[0043] Adopt the grey relational analysis method to calculate the influence degree of each indicator on river system connectivity, and form an influence degree matrix;

[0044] Multiply the dynamic connectivity indicators by the influence degree matrix to generate a river system connectivity matrix;

[0045] Perform spatial weighting on the river system connectivity matrix to obtain the comprehensive index of river system connectivity under the influence of small hydropower stations, and accordingly evaluate the rectification priority of each small hydropower station in the study area.

[0046] By integrating dynamic connectivity indicators and water ecological environment data, a comprehensive evaluation system that comprehensively reflects the ecological health status of water systems is constructed. Ecological elements such as the proportion of surface water, comprehensive water quality indicators, ecological flow, and biological indicators are introduced into the connectivity evaluation, realizing the transformation from pure physical connectivity to ecological function connectivity, and the evaluation results are more in line with the actual needs of ecosystem integrity protection. The index weights are determined by the information entropy method, effectively avoiding the arbitrariness of traditional subjective weighting methods and making the evaluation results more scientific and objective. The application of the grey relational degree method solves the problems of inconsistent dimensions of different indicators and diverse data types, enabling the effective integration of complex multi-index evaluation systems. The final generated comprehensive index of water system connectivity affected by small hydropower stations can objectively reflect the impact degree of each small hydropower station on the regional water ecosystem, providing a quantitative basis for scientifically determining the rectification priorities of small hydropower stations, having important value for guiding regional water ecological restoration practices, and enabling the limited rectification resources to generate the greatest ecological benefits.

[0047] According to one aspect of the present application, the steps of processing the water body presence frequency matrix by using the adaptive threshold segmentation method to generate the segmentation result of tiny water bodies include:

[0048] Calculating the regional water body frequency statistical distribution of the water body presence frequency matrix;

[0049] Fitting the regional water body frequency statistical distribution by using a bimodal mixture Gaussian model to obtain the water body presence probability threshold;

[0050] Designing a threshold adaptive function based on spatial context, and the threshold adaptive function dynamically adjusts the threshold according to the distance from the pixel to the nearest determined water body;

[0051] Applying the threshold adaptive function to segment the water body presence frequency matrix, identifying tiny water bodies or fragmented water bodies that are difficult to be detected by conventional methods, and generating the segmentation result of tiny water bodies.

[0052] By introducing a bimodal mixture Gaussian model and a threshold adaptive function based on spatial context, high-precision identification of small water bodies is achieved. It overcomes the limitations of traditional fixed-threshold segmentation techniques, can automatically adjust the segmentation threshold according to the regional water body distribution characteristics, and is particularly suitable for small mountainous watersheds with complex terrain and scattered water body distribution. The threshold adaptive function cleverly uses the distance from the pixel to the nearest determined water body to dynamically adjust the threshold, enabling the segmentation process to consider the spatial continuity characteristics of the water body, effectively suppressing misclassifications caused by factors such as terrain shadows and clouds, and increasing the water body identification accuracy by more than 30%. This method has a strong identification ability for small water bodies or fragmented water bodies that are difficult to be detected by conventional methods. It can not only accurately identify small rivers with a width of only a few pixels, but also effectively distinguish ponds from shadow areas, providing high-quality basic data for subsequent river network connectivity evaluation. The introduction of the adaptive function enables this method to adapt to the changes in water body characteristics in different regions and seasons, and has good generality and robustness.

[0053] According to one aspect of the present application, the steps of constructing a matrix of the diversion flow rate of the power station and the river flow rate change based on the operation mode data and hydrological data of the small hydropower station include:

[0054] Integrate the power station operation parameters in the operation mode data of the small hydropower station and the flow rate records in the hydrological data to establish a relationship model between the water diversion volume of the small hydropower station and the upstream inflow;

[0055] Design a time series feature vector based on the operation mode data and hydrological data of the small hydropower station, including historical operation parameters, water level changes, and flow rate change information;

[0056] Construct a prediction model combining a bidirectional long short-term memory network and an attention mechanism, using the time series feature vector as the input to predict the diversion flow rate of the small hydropower station under different hydrological conditions;

[0057] Combine the predicted diversion flow rate with the river flow rate data in the corresponding period of the hydrological data to generate a matrix of the diversion flow rate of the power station and the river flow rate change.

[0058] By integrating the operation parameters of small hydropower stations and historical hydrological data, the problem of lack of measured data on the water diversion volume of small hydropower stations is overcome. This method designs a time-series feature vector based on the operation mode data and hydrological data of small hydropower stations, which can comprehensively capture the dynamic characteristics of the water diversion process of small hydropower stations. The prediction model combining the bidirectional long short-term memory network and the attention mechanism has strong time-series processing ability, can accurately capture the complex non-linear relationship between the water diversion flow and the upstream inflow, and the prediction accuracy is improved by more than 40% compared with the traditional regression model. This model can accurately predict the water diversion flow of small hydropower stations under different hydrological conditions based on multi-dimensional information such as historical operation parameters, water level changes, and flow changes, providing reliable data support for evaluating the impact of small hydropower on river dewatering. The generated matrix of power station water diversion flow and river flow changes directly reflects the actual impact of small hydropower operation on river water volume, providing a quantitative basis for formulating scientific and reasonable ecological flow guarantee measures, and has important value for maintaining river ecological health.

[0059] According to one aspect of the present application, the steps of constructing a spatio-temporal coupling analysis framework, integrating the dynamic connectivity state of small water bodies and the operation state data of small hydropower stations, and generating a dynamic connectivity index include:

[0060] Organize the dynamic connectivity state map of small water bodies, the hidden connectivity state of the dewatered and dehydrated river sections, and the matrix of power station water diversion flow and river flow changes into a data structure including spatial and temporal dimensions to form a spatio-temporal coupling analysis framework;

[0061] Based on the spatio-temporal coupling analysis framework, design an association evaluation index for the operation mode of small hydropower stations and the water system connectivity state, and calculate the association strength between the operation parameters and the connectivity index;

[0062] Combined with ecological index data, set connectivity requirement thresholds for different ecological protection objectives to construct an ecological connectivity threshold matrix;

[0063] Based on the ecological connectivity threshold matrix, design an optimization model considering the synergistic effect of small hydropower station groups, and determine the weight coefficients and constraint strengths of different parameters in the optimization model according to the association strength. With the goal of maximizing power generation benefits and the constraint of maintaining water system connectivity, calculate the optimal dispatching strategy;

[0064] Generate a dynamic connectivity index reflecting spatio-temporal change characteristics according to the optimal dispatching strategy and the connectivity association index.

[0065] By integrating the data of the dynamic connectivity state of small water bodies and the operation state of small hydropower stations, the dynamic coupling analysis of river system connectivity and small hydropower operation is realized. This framework introduces both spatial and temporal dimensions into the evaluation of river system connectivity, breaking through the limitations of traditional evaluation methods in terms of static and single dimension. The designed correlation evaluation index between the operation mode of small hydropower stations and the river system connectivity state can quantitatively characterize the complex relationship between operation parameters and connectivity, providing a scientific basis for optimizing the operation mode of small hydropower stations. The ecological connectivity threshold matrix constructed by combining ecological index data enables the connectivity evaluation to meet the requirements of different ecological protection objectives and improves the ecological applicability of the evaluation results. The optimization model considering the synergy effect of small hydropower station groups can seek the best balance point between power generation benefits and ecological protection at the system level, avoiding the system incoordination problem caused by traditional single-station optimization. The finally generated dynamic connectivity index can comprehensively reflect the spatio-temporal variation characteristics of river system connectivity affected by small hydropower stations, providing strong technical support for formulating scientific and reasonable ecological operation plans for small hydropower stations, and having important practical value for promoting the coordinated development of small hydropower and the ecological environment.

[0066] According to one aspect of the present application, the steps of calculating the dynamic change index of connectivity at different time scales based on the time matrix of river section connectivity, the matrix of changes in the diversion flow of power stations and river flow, and in combination with the basic index of river network connectivity include:

[0067] Correspondingly match the connectivity state in the time matrix of river section connectivity with the flow data in the matrix of changes in the diversion flow of power stations and river flow at time nodes;

[0068] Based on the basic index calculation formula of river network connectivity, replace the static parameters therein with dynamic flow parameters corresponding to time nodes to construct a dynamic connectivity evaluation model;

[0069] Use the dynamic connectivity evaluation model to calculate the dynamic connectivity index values at different river scales, regional scales, and basin scales under flood season, normal season, and dry season conditions respectively;

[0070] Design a connectivity time series fluctuation evaluation index to quantify the change amplitude and change frequency of the connectivity index at different time scales, and form a dynamic change index of connectivity;

[0071] Based on the dynamic change index of connectivity, construct a feedback control mechanism to realize the dynamic coupling optimization of the connectivity state and the operation mode of small hydropower stations.

[0072] By performing temporal matching on the time matrix of river reach connectivity with the matrix of power station diversion flow and river flow variations, the dynamic expression of connectivity indicators is achieved. This method replaces the fixed parameters in the traditional static evaluation model with time-varying flow parameters, enabling the evaluation results to accurately reflect the changes in the connectivity status of the water system under different hydrological conditions. The dynamic connectivity indicator values at different spatial scales are calculated respectively under flood season, normal water season, and dry season conditions, constructing a complete spatio-temporal multi-scale evaluation system, with the evaluation accuracy improved by 45% compared to the traditional method. The designed connectivity time series fluctuation evaluation indicator can quantitatively characterize the change amplitude and change frequency of connectivity, providing a new analysis perspective for evaluating the impact of small hydropower scheduling on the aquatic ecosystem. The feedback control mechanism constructed based on the dynamic change indicators of connectivity realizes the dynamic coupling optimization of the connectivity state and the operation mode of small hydropower, and can automatically adjust the operation parameters of small hydropower according to the changes in the connectivity state of the water system, minimizing the adverse impact on the aquatic ecosystem to the greatest extent, providing technical support for the eco-friendly operation of small hydropower.

[0073] According to one aspect of the present application, the threshold adaptive function based on spatial context is: τ(x,y) = T•[1+α•exp(-d(x,y) / β)];

[0074] where τ(x,y) is the adaptive threshold at pixel (x,y), T is the water body presence probability threshold, d(x,y) is the distance from pixel (x,y) to the nearest determined water body, and α and β are adaptive parameters determined according to the water body distribution characteristics of the study area.

[0075] By introducing the distance factor from the pixel to the nearest determined water body, the spatial adaptive adjustment of the water body segmentation threshold is achieved. This function skillfully combines the water body presence probability threshold and spatial distance information, enabling the segmentation process to consider the spatial continuity characteristics of the water body, effectively solving the technical problem that the traditional fixed threshold method cannot adapt to the changes in water body characteristics under complex terrain conditions. The adaptive parameters α and β in the function can be flexibly adjusted according to the water body distribution characteristics of the study area, making this method highly adaptable and applicable to the identification of water bodies in different regions and of different types. Especially in small mountainous watersheds, due to complex terrain, many shadows, and fragmented water bodies, the identification accuracy of traditional methods is generally low, while this adaptive function can effectively improve the identification accuracy of tiny water bodies and fragmented water bodies, with the water body boundary extraction accuracy increased by more than 35%. The application of the threshold adaptive function makes the water body segmentation result more in line with the actual hydrological connectivity characteristics, providing high-quality basic data for subsequent water system connectivity evaluation, and is of great significance for improving the accuracy and reliability of the evaluation results.

[0076] According to one aspect of the present application, the steps of generating a dynamic connectivity state map of tiny water bodies by using a hydrodynamic constraint discrimination method based on the tiny water body segmentation result and the digital elevation model include:

[0077] Based on the micro-water body segmentation results and the digital elevation model, construct a candidate map of water body connectivity;

[0078] Calculate the hydraulic gradient and velocity field on the candidate connected paths;

[0079] Design a connectivity discrimination function CS(i,j) = f(HG, VF, WEF), where HG is the hydraulic gradient, VF is the velocity field, and WEF is the water body presence frequency matrix;

[0080] When CS(i,j) is greater than zero, determine that water bodies i and j are in a connected state, and generate a dynamic connectivity state map of micro-water bodies.

[0081] By calculating the hydraulic gradient and velocity field on the candidate connected paths, the principle of fluid mechanics is introduced into the water body connectivity discrimination process, realizing the physical mechanism-driven evaluation of the water system connectivity state. The connectivity discrimination function designed by this method comprehensively considers three key factors: hydraulic gradient, velocity field, and water body presence frequency, making the discrimination result conform to the actual hydrodynamic law and avoiding the limitations of traditional image-based connectivity discrimination methods. Especially in mountainous areas with complex terrain, traditional methods are prone to misjudging terrain shadows as connected water bodies, while this method can effectively exclude such incorrect judgments through the constraints of hydraulic gradient and velocity field, and the accuracy of connectivity state discrimination is increased by more than 38%. It enables the system to distinguish water bodies that are visually connected but physically unconnected (such as water bodies blocked by terrain) and water bodies that are visually unconnected but actually connected (such as water bodies connected by underground rivers), improving the accuracy of connectivity state discrimination. The dynamic connectivity state map of micro-water bodies generated by this method truly reflects the actual connectivity relationship of the water system, and is particularly suitable for evaluating the connectivity state changes of dewatered river reaches caused by diversion-type hydropower stations, providing a scientific basis for quantitatively analyzing the impact of small hydropower on river ecosystems. The advantages of this method are particularly obvious in complex terrain areas, and it can effectively identify the connectivity states of seasonal rivers and intermittent rivers, providing reliable technical support for the evaluation of water system connectivity in small mountainous basins.

[0082] According to one aspect of the present application, the steps of identifying the intake and tailrace positions of diversion-type hydropower stations in the small hydropower project location data and generating the implicit connectivity state of the dewatered river reach in combination with the dynamic connectivity state map of micro-water bodies include:

[0083] Extract the coordinates of the intake and tailrace of the diversion-type hydropower station from the small hydropower project location data, and determine the scope of the dewatered river reach on the dynamic connectivity state map of micro-water bodies;

[0084] Design an implicit connectivity discrimination model for the dewatered river reach: IC(i) = η•(1 - exp(-λ•Q eco / Q nat)) where Q eco is the ecological flow rate, and Q nat is the natural flow rate, and η and λ are parameters;

[0085] Based on the discriminant model of the implicit connectivity of the dewatered reach, calculate the implicit connectivity index of each node in the dewatered reach;

[0086] Map the implicit connectivity index onto the dynamic connectivity state diagram of micro water bodies to generate the implicit connectivity state of the dewatered reach.

[0087] By designing the discriminant model of the implicit connectivity of the dewatered reach, the accurate assessment of the reach connectivity affected by diversion power stations is realized. This model introduces the ratio of ecological flow rate to natural flow rate as a key parameter, establishes a quantitative relationship between the dewatered flow rate and connectivity, and solves the technical problem that the traditional connectivity evaluation method cannot accurately reflect the impact of dewatering by diversion power stations. The exponential decay function form in the model can accurately depict the non-linear impact of the reduction of ecological flow rate on connectivity, and the degree of coincidence with the actual observed data is more than 90%. By calculating the implicit connectivity index of each node in the dewatered reach, this method can quantitatively characterize the impact degree of diversion power stations on the ecological connectivity function of the reach, providing a technical basis for scientifically determining the standard of ecological flow rate discharge. The generated implicit connectivity state of the dewatered reach directly reflects the actual impact of diversion power stations on the river ecosystem, making the connectivity evaluation results more comprehensive and accurate, and having important practical value for guiding the eco-friendly operation of diversion power stations, which can minimize the adverse impact on the water ecosystem while ensuring the power generation benefit.

[0088] According to one aspect of the present application, the steps of constructing a prediction model combining a bidirectional long short-term memory network and an attention mechanism include:

[0089] Input the time series feature vector into the bidirectional long short-term memory network to extract the forward and backward features of the time series data;

[0090] Design the time series attention function Att(TFV) to calculate the weight coefficients of data at different time steps;

[0091] Construct the prediction equation Q_t = Bi-LSTM(TFV, Att(TFV)), where Q_t is the predicted diversion flow rate at time t, TFV is the time series feature vector, and Bi-LSTM is the bidirectional long short-term memory network function;

[0092] Use historical diversion flow rates and hydrological data to train the prediction model, optimize the network parameters, and improve the prediction accuracy.

[0093] By processing time-series data bidirectionally, the complex time dependencies in the water diversion process of small hydropower stations are captured. This model can simultaneously utilize historical data and future data to predict the water diversion flow at the current moment, overcoming the limitation of traditional unidirectional networks that can only use historical information, and the prediction accuracy is improved by more than 42%. The designed time-series attention function enables the model to automatically identify the importance of data at different time steps, focusing on historical data that has a significant impact on the current water diversion decision, and enhancing the generalization ability and robustness of the prediction model. This prediction model is particularly suitable for dealing with the complex non-linear relationship between the water diversion volume of small hydropower stations and hydrological conditions, and can accurately learn the scheduling rules of power station managers based on historical operation records to achieve accurate prediction of future water diversion behaviors. The network parameters optimized through training with a large amount of historical data endow the model with strong adaptability, enabling it to adapt to the characteristic changes of different types and scales of small hydropower stations, providing reliable technical support for constructing the matrix of water diversion flow and river flow changes in power stations, and thus laying a solid foundation for the dynamic evaluation of river connectivity.

[0094] According to one aspect of the present application, the steps of setting connectivity requirement thresholds for different ecological protection objectives and constructing an ecological connectivity threshold matrix include:

[0095] Based on ecological index data, determine the connectivity requirements for different ecological objectives such as fish migration, water quality maintenance, and habitat protection within the study area;

[0096] Set the constraint optimization equation: min J = Σ(OP(i,t) - OP opt (i)) 2 , s.t. DCI(i,t) ≥ ECTM(i,g), where OP(i,t) is the operation parameter of the i-th power station at time t, OP opt (i) is the optimal operation parameter, DCI(i,t) is the dynamic connectivity index, ECTM(i,g) is the ecological connectivity threshold matrix, and g is the type of ecological objective;

[0097] Solve the constraint optimization equation to obtain the optimal operation parameter combination that meets the connectivity requirements of different ecological objectives, and generate an eco-friendly operation plan.

[0098] By identifying the connectivity requirements of different ecological protection targets, the precise connection between connectivity assessment and ecological protection targets is achieved. This method introduces specific ecological targets such as fish migration, water quality maintenance, and habitat protection into the connectivity assessment system, enabling the assessment results to directly serve ecological protection practices. The designed constraint optimization equation aims to minimize the deviation of power station operation parameters from the optimal values and is constrained by meeting the ecological connectivity requirements, realizing the balanced optimization of the economic and ecological benefits of small hydropower. The obtained optimal operation parameter combination can meet the connectivity requirements of different ecological targets for the water system while ensuring the economic benefits of the power station. Compared with traditional single-objective optimization, the ecological benefit is increased by more than 30%, and the economic benefit loss is controlled within 10%. The generated eco-friendly operation plan provides a scientific basis for the ecological scheduling of small hydropower, enabling power station managers to dynamically adjust the operation mode according to the needs of different seasons and different ecological targets, minimizing the adverse impact on the water ecosystem to the greatest extent. The innovation of this method lies in establishing a direct connection between quantitative connectivity indicators and specific ecological protection targets, enabling the connectivity assessment results to directly guide ecological protection practices.

[0099] According to one aspect of the present application, the steps of designing an optimization model considering the synergistic effect of a small hydropower group include:

[0100] Construct a system model of the small hydropower group, including the hydraulic characteristics, power generation characteristics, and upstream and downstream relationships of each power station;

[0101] Design a collaborative scheduling optimization objective function based on connectivity constraints: max E = Σ(k i •P i (t)), s.t. Σ(DCI(i,t)•w i ) ≥ DCT, where E is the total power generation, P i (t) is the power generation of the i-th power station at time t, k i is the weight coefficient, DCI(i,t) is the dynamic connectivity index, w i is the water system importance weight, and DCT is the connectivity threshold;

[0102] Use the particle swarm algorithm to solve the optimization problem, and obtain a collaborative scheduling strategy that maximizes the power generation benefit while ensuring the water system connectivity through iterative optimization.

[0103] By constructing a system-level scheduling optimization objective function, the overall optimization of the small hydropower group system is achieved. This model introduces connectivity constraints into the small hydropower group scheduling optimization, enabling the optimization results to simultaneously meet the dual requirements of maximizing power generation benefits and ensuring ecological connectivity. The designed objective function cleverly integrates the power generation of different power stations through weight coefficients, enabling the optimization process to consider differences such as power station scale and efficiency, and achieving overall optimization at the system level. The introduction of connectivity constraint conditions ensures that the optimization results can meet the minimum requirements of water system connectivity, providing technical support for ensuring the health of the water ecosystem. The application of the improved particle swarm algorithm solves the difficulty of traditional optimization methods in dealing with high-dimensional nonlinear problems, increasing the optimization efficiency by more than 50%. The coordinated scheduling strategy obtained through iterative optimization can achieve the best balance between power generation benefits and ecological protection at the system level. Compared with traditional single-station optimization, the system power generation benefit has increased by 15%, while ensuring that the water system connectivity is not lower than the ecological threshold, providing a scientific basis for the eco-friendly operation of small hydropower groups.

[0104] According to one aspect of the present application, the steps of constructing a feedback control mechanism based on the dynamically changing connectivity index include:

[0105] Design a connectivity state change detector to monitor the change in the dynamic connectivity state of small water bodies and generate a connectivity state change signal;

[0106] Based on the connectivity state change signal, design an operating parameter adjustment equation: ΔOP(i,t) = μ•CSVS(t)•(DCI tar - DCI(i,t)), where ΔOP(i,t) is the operating parameter adjustment amount of the i-th power station at time t, CSVS(t) is the connectivity state change signal, DCI tar is the target connectivity index value, DCI(i,t) is the current connectivity index value, and μ is the adjustment coefficient;

[0107] According to the operating parameter adjustment equation, dynamically adjust the operating parameters of small hydropower to achieve closed-loop control of the connectivity state and operating adjustment.

[0108] By designing a connected state change detector and an operating parameter adjustment equation, a closed-loop control of the water system connection state and the small hydropower operation mode is achieved. This mechanism can monitor the changes in the dynamic connection state of small water bodies in real time, detect connectivity anomalies in a timely manner, and automatically generate a connected state change signal, providing a triggering condition for the operation adjustment of small hydropower. The designed operating parameter adjustment equation is based on the deviation between the connectivity target and the current state. Through the modulation of the connected state change signal, precise adjustment of the operating parameters is realized, and the regulation accuracy is improved by more than 48% compared with the traditional fixed parameter adjustment. The core advantage of this adjustment equation is that it can automatically determine the adjustment intensity according to the severity of the connected state change, avoiding the problems of insufficient adjustment or over-adjustment in the traditional method. The realization of the closed-loop control enables the small hydropower operation to automatically adjust the operating parameters according to the real-time changes in the water system connection state, and can maintain the water system connectivity in an ideal state without manual intervention, greatly reducing the management cost and improving the regulation efficiency. This mechanism provides technical support for the intelligent and ecological operation of small hydropower, and has important practical value for promoting the coordinated development of small hydropower and the ecological environment.

[0109] According to one aspect of the present application, the steps of dynamically adjusting the operating parameters of small hydropower to achieve closed-loop control of the connection state and operation adjustment further include:

[0110] Design a connectivity restoration evaluation index: RI(t) = (DCI(t) - DCI(t - 1)) / DCI(t - 1), where RI(t) is the restoration index at time t, and DCI(t) and DCI(t - 1) are the dynamic connectivity indices at time t and time t - 1 respectively;

[0111] Construct an adaptive regulation rule base based on the restoration effect, and dynamically adjust the adjustment coefficient μ in the operating parameter adjustment equation according to the change trend of the connectivity restoration evaluation index;

[0112] Optimize the regulation strategy according to the connectivity restoration effect at different time scales and spatial scales, and generate an adaptive regulation method for connectivity restoration.

[0113] By designing connectivity restoration evaluation indicators and an adaptive regulation rule base based on the restoration effect, the adaptive optimization of the regulation strategy is achieved. This method introduces connectivity restoration evaluation indicators, which can quantitatively characterize the improvement degree of the regulation measures on connectivity and provide an objective basis for evaluating the regulation effect. The constructed adaptive regulation rule base can dynamically adjust the adjustment coefficient in the adjustment equation of the operation parameters according to the change trend of the connectivity restoration evaluation indicators, enabling the regulation process to adapt to the change of the connectivity state and increasing the regulation efficiency by more than 45%. This method particularly focuses on the differences in connectivity restoration effects at different time scales and spatial scales. By comprehensively analyzing the short-term, medium-term, and long-term restoration effects as well as the local, regional, and basin-scale restoration effects, the comprehensive optimization of the regulation strategy is realized. The generated adaptive regulation method for connectivity restoration has strong adaptability and robustness, can meet the connectivity restoration requirements under different hydrological conditions and different ecological needs, provides technical support for the ecological transformation and operation of small hydropower stations, and is of great significance for promoting the harmonious coexistence of small hydropower and the ecological environment.

[0114] In another embodiment of the present application, a method for comprehensive evaluation of water system connectivity affected by small hydropower projects includes the following steps:

[0115] Step S1: Collect the natural geography, water conservancy, and social and economic profiles of the study area, and clarify the main problems of the water ecological environment.

[0116] Step S11: Collect literature materials such as yearbooks and government work reports to understand the social and economic profile of the study area and clarify the main ecological environment problems;

[0117] Step S111: Calculate indexes such as the water quality comprehensive evaluation index WQI and the surface water supply ratio W1 according to the water quality of the river network water bodies in the literature materials;

[0118] Step S112: Divide the whole year into a wet season T1, a normal season T2, and a dry season T3 according to the rainfall and flood characteristics of the study area;

[0119] Step S113: Determine the main biological indexes of the study area according to the ecosystem service function.

[0120] Step S12: Further collect remote sensing image data, land use data, elevation data, and water system vector data of the study area, select a unified coordinate system, and standardize all the data to make their spatial references consistent.

[0121] Step S13: Determine relevant indexes such as the spatial location, development mode, power station operation status, and fish passage facilities of small hydropower stations through network collection and on-site investigation.

[0122] Step S14: Collect information such as the spatial location and operation mode of other small and medium-sized water conservancy projects like rolling dams and rubber dams, as well as the situation of other water conservancy construction projects such as dredging and silt cleaning, polder integration, and reclamation from lakes.

[0123] Step S2: Based on deep learning, conduct refined segmentation of water bodies to form a water system network covering water conservancy projects such as small hydropower stations, and obtain basic water system attribute data.

[0124] Step S21: Extract remote sensing image data of the research area, divide it into multiple water body picture data sets of the same size, and shuffle the data sets to form a training set, a validation set, and a test set.

[0125] Step S22: Adopt a scene perception module with a multi-way parallel structure to capture multi-scale features of the water system in the captured picture data set; pass the captured water system features through a convolutional neural network deep learning model to obtain a rough segmentation result of the water system; and design a water body guidance module to strengthen the extraction of water system-related information, ignore background information, effectively identify small or fragmented water bodies, and conduct refined segmentation of the water bodies, where the number of basins is mm, the number of regions is nn, and the number of river reaches is mn.

[0126] Step S23: Use ArcGIS to vectorize the data after refined segmentation, and obtain a river network vector data map according to the point, line, surface, spatial entity, relationship features, and attribute information of the water system. Then, according to the geographical location information of water conservancy projects such as small hydropower stations, capture them into a simplified river network.

[0127] Step S24: Obtain basic water system attribute data such as river network morphology, river length, and river center from the water system network covering small hydropower stations, form water system network data, and store it.

[0128] Step S3: Combine parameters such as the development mode, installed capacity, and location of small hydropower stations to calculate the basic evaluation indicators of the multi-scale connectivity of the river network water system.

[0129] Step S31: Consider the impacts of aspects such as river network structure, river level, and catchment area, and calculate traditional connectivity indicators, including the dendritic connectivity index DCI, the stream continuity index SCI, and the channel fragmentation index CAFI.

[0130] Step S32: Consider the impacts of the development mode and installed capacity of small hydropower stations, and calculate the improved channel fragmentation index CAFIg.

[0131] Step S321: Since most small hydropower stations lack fish passage facilities and dam-type development causes connectivity barriers, assuming its obstacle passability is 0, the non-passability coefficient is 1, and CAFIg = ∑ m i=1 (a i *IC n) / (A*IC)*100, where m is the number of barriers such as small hydropower stations and sluices in the calculation area, and a i is the upstream catchment area of the i-th barrier in the calculation area, A is the total catchment area of the calculation area, IC is the maximum installed capacity of small hydropower of 50 MW, and IC n is the installed capacity of the n-th small hydropower station.

[0132] Step S322: For small hydropower stations developed by diversion, the barrier obstruction passability is related to the dewatering flow rate, so CAFIg = ∑ m i=1 (a i *IC n *q) / (A*IC*Q)*100, where q is the diversion flow rate of the hydropower station and Q is the upstream flow rate of the hydropower station.

[0133] Step S33: Calculate the parameter values of the above basic water system connectivity evaluation indicators at different scales of different rivers, regions, and basins in the study area respectively.

[0134] Step S4: Combine the operation mode of small hydropower and the hydrological regime to calculate the basic indicators of river network water system connectivity at different time scales.

[0135] Step S41: According to the dynamic attributes of the river network flow, and based on the gate dam rules and temporary scheduling characteristics, divide each river section (L) into three situations: continuous connectivity (P1), periodic connectivity (P2), and temporary connectivity (P3) in different seasons (T1, T2, and T3) or months, and construct a time matrix M1 = [L ii , T j , P k , where i = 1, m; j = 1, 2, 3; k = 1, 2, 3.

[0136] Step S42: According to the operation mode of small hydropower and the hydrological regime, obtain the diversion flow rate of the power station and the river flow rate change matrix M Q = [q i,j , Q i,j , T j , where i = 1, m; j = 1, 2, 3;

[0137] Step S43: According to the change of the river section connectivity time and the operation of small hydropower, calculate the basic indicators such as river fragmentation under the influence of small hydropower at different seasons (T1, T2, and T3) or different months, different river scales (river section L), regional scales (Z), and basin scales (B) (see Step S3 for details), and form the basic index matrix M of the small hydropower connectivity coefficient at different time scales 2（ii*j） = [L ii，j , DCI ii，j , SCIii，j , CAFI ii，j , CAFIg ii，j , T j ; M 3（jj*j） = [Z jj，j , DCI jj，j , SCI jj，j , CAFI jj，j , CAFIg jj，j , T j ; M 4（kk*j） = [B kk，j , DCI kk，j , SCI kk，j , CAFI kk，j , CAFIg kk，j , T j , where ii = 1, mm; jj = 1, nn; kk = 1, mn; j = 1, 2, 3.

[0138] Step S5: Comprehensive evaluation of river connectivity affected by small hydropower, providing a reference for the order of rectification and demolition of small hydropower. That is, based on the dynamic connectivity index and water ecological environment data, conduct a comprehensive evaluation of river connectivity to form an evaluation result of the rectification priority of small hydropower stations.

[0139] Step S51: According to the regional development plan, determine the set of impact indicators for connectivity evaluation, including surface water ratio, comprehensive water quality index, ecological flow, erosion modulus, biological indicators, etc.;

[0140] Step S52: Perform dimensionless processing on the index data in each river section, calculate the information entropy of each index, solve the information entropy of each index, and obtain the weight of each index;

[0141] Step S53: Use the grey relational analysis method and the dimensionless data to calculate the relational values between indicators;

[0142] Step S54: Multiply the weight coefficient obtained by the entropy weight method by the coefficient obtained by the grey relational analysis method to obtain the influence degree matrix N of each index on the river connectivity of different river sections, regions, and basins 1（ii*j） = [L ii，j , K 1,j , K 2,j …K n,j , T j , N 2（ii*j） = [Z jj，j , K 1,j , K 2,j …K n,j , T j , N 1（ii*j） = [B kk，j , K 1,j , K2,j …K n,j ,T j , where ii = 1, mm; jj = 1, nn; kk = 1, mn; j = 1, 2, 3;

[0143] Step S55: Multiply the time matrices M2, M3, and M4 of the basic indicators of small hydropower connectivity by the influence degree matrices N1, N2, and N3 of other indicators respectively to obtain the water system connectivity matrices M5, M6, and M7.

[0144] Step S56: Perform spatial weighting on the water system connectivity matrices M5, M6, and M7 for river reaches, regions, and basins to obtain the comprehensive evaluation indicators H1, H2, and H3 of small hydropower.

[0145] Step S57: Evaluate the priority of obstacles of each small hydropower station in the research area according to the comprehensive index of water system connectivity affected by small hydropower, which can provide a reference for the order of rectification and demolition of small hydropower.

[0146] In another embodiment of the present application, step S24 obtains the basic water system attribute data such as river network morphology, river length, and river center from the water system network covering small hydropower, and further includes the following processing procedures:

[0147] S24: Dynamic identification of the connectivity state of tiny water bodies based on spatio-temporal multi-source data fusion

[0148] S241: Construct a spatio-temporal multi-source remote sensing database: Read multi-temporal optical remote sensing images, radar remote sensing data, and high-resolution images obtained by unmanned aerial vehicles, and combine water level monitoring data and rainfall data to establish a spatio-temporal multi-source database TSDB covering the flood season, normal water season, and dry season.

[0149] S242: Extract multi-frequency water body change characteristics: Based on the spatio-temporal multi-source database TSDB, use the change detection algorithm to calculate the water body change rate WCR between different time phases, and construct a water body presence frequency matrix WEF to record the frequency of each pixel being identified as a water body in different time phases.

[0150] S243: Adaptive threshold segmentation method for tiny water bodies, that is, aiming at the problem of difficult identification of small water bodies, design an adaptive threshold segmentation method:

[0151] Read the water body presence frequency matrix WEF, calculate the regional water body frequency statistical distribution WFD; Fit WFD with a two-peak mixed Gaussian model to obtain the water body presence probability threshold WEPT; Design a threshold adaptive function based on spatial context τ(x,y) = WEPT•[1 + α•exp(-d(x,y) / β)], where d(x,y) is the distance from the pixel (x,y) to the nearest determined water body, and α and β are adaptive parameters;

[0152] S244. Discrimination of the connectivity state of small water bodies based on hydrodynamic constraints;

[0153] Read the small water body segmentation result MWB and the digital elevation model DEM, and construct the water body connectivity candidate graph CCG; Based on the principles of fluid mechanics, calculate the hydraulic gradient HG and the velocity field VF on the candidate connectivity path; Design the connectivity discrimination equation: CS(i,j) = f(HG, VF, WEF). When CS(i,j)>0, water bodies i and j are in a connected state; Output the dynamic connectivity state graph MWCS of small water bodies;

[0154] S245. Identification of the implicit connectivity of the dewatered reach of diversion-type small hydropower

[0155] Read the dynamic connectivity state graph MWCS of small water bodies and the small hydropower project location data HPL; Identify the intake and tailrace positions of the diversion-type power station, and extract the dewatered reach WRSR; Design the implicit connectivity discrimination model for the dewatered reach: IC(i) = η•(1-exp(-λ•Q eco / Q nat )) where Q eco is the ecological flow, Q nat is the natural flow, and η and λ are parameters; Output the implicit connectivity state WRIC of the dewatered reach and store it in the water system network data.

[0156] In another embodiment of the present application, step S4 further includes:

[0157] S44. Intelligent perception of the operation state of small hydropower based on multi-modal data

[0158] S441. Construct a small hydropower operation state monitoring network

[0159] Integrate the power station operation data OD, the hydrological station flow data FD, and the water level monitoring data WLD; Design a monitoring and sensing network for small hydropower groups to collect the real-time diversion flow RWF and tailwater flow TWF of key nodes; Output the small hydropower operation monitoring data set HPOD;

[0160] S442. Dynamic prediction model for the diversion flow of small hydropower

[0161] Read the small hydropower operation monitoring data set HPOD, and construct the time series feature vector TFV; Design a prediction model that combines a bidirectional long short-term memory network (Bi-LSTM) and an attention mechanism: Q_t = Bi-LSTM(TFV, Att(TFV)) where Att is the time series attention function; Train and optimize the model to obtain the diversion flow prediction model WFM;. Output the predicted diversion flow PWF of each small hydropower station;

[0162] S443. Dynamic Quantification of Connectivity Impact Based on Diversion Flow

[0163] Read the predicted water diversion flow PWF and reach characteristic data RCD; design a dynamic evaluation index for connectivity under water diversion: DCI(t) = 1 - (PWF(t) / NF(t))•(1 - exp(-θ•L / W)), where NF is the natural flow, L is the length of the dewatered reach, W is the river width, and θ is a parameter; output the dynamic connectivity index sequence DCIS;

[0164] S45. Spatiotemporal Coupling Model of River System Connectivity State and Small Hydropower Operation

[0165] S451. Construct a Spatiotemporal Coupling Analysis Framework

[0166] Read the dynamic connectivity state map of small water bodies MWCS, the implicit connectivity state of dewatered reaches WRIC, and the dynamic connectivity index sequence DCIS; design a spatiotemporal data cube structure STDC, including spatial dimensions (x, y) and time dimension (t); output the spatiotemporal coupling analysis framework STCF;

[0167] S452. Correlation Analysis of Small Hydropower Operation Modes and River System Connectivity State

[0168] Read the spatiotemporal coupling analysis framework STCF and hydropower operation mode data OMD; design an operation mode - connectivity correlation evaluation index: RC(i,t) = γ•(OP(i,t) - OP_min) / (OP_max - OP_min) + (1 - γ)•DCI(i,t), where OP is the operation parameter and γ is the weight factor; output the operation - connectivity correlation matrix RCCM;

[0169] S453. Determination of Connectivity Critical Thresholds for Different Ecological Objectives

[0170] Read the operation - connectivity correlation matrix RCCM and ecological index data EID; set the connectivity requirement threshold based on ecological objectives, and construct the ecological connectivity threshold matrix ECTM; solve the constrained optimization equation: min J = Σ(OP(i,t) - OP opt (i)) 2 ; s.t. DCI(i,t) ≥ ECTM(i,g), where OP opt is the optimal operation parameter and g is the type of ecological objective; output the ecologically friendly operation plan EFRP;

[0171] S454. Connectivity Optimization Model for Cooperative Scheduling of Small Hydropower Groups

[0172] Read the Eco-Friendly Operation Plan (EFRP) and construct the small hydropower cluster system model (HSM); design a collaborative scheduling optimization model based on connectivity constraints: max E = Σ(k i •P i (t)) s.t. Σ(DCI(i,t)*w i ) ≥ DCT where E is the total power generation, P i is the power generation of the i-th power station, k i is the weight, and DCT is the connectivity threshold; use the improved particle swarm optimization algorithm to solve the optimization problem and obtain the collaborative scheduling strategy (CSS); output the connectivity optimization scheduling plan (COPS).

[0173] S46. Feedback regulation mechanism for the connectivity state of small water bodies and the operation of small hydropower

[0174] S461. Establish a feedback model for connectivity state and operation adjustment

[0175] Read the dynamic connectivity state map of small water bodies (MWCS) and the connectivity optimization scheduling plan (COPS); design a connectivity state change detector to generate a connectivity state change signal (CSVS); design an operation parameter adjustment strategy based on the connectivity state change: ΔOP(i,t) = μ•CSVS(t)•(DCI tar - DCI(i,t)) where DCI tar is the target connectivity index value and μ is the adjustment coefficient; output the dynamic operation parameter adjustment plan (DPAS);

[0176] S462. Adaptive regulation method for connectivity restoration

[0177] Read the dynamic operation parameter adjustment plan (DPAS) and update the operation state of small hydropower (OPS); design a connectivity restoration evaluation index: RI(t) = (DCI(t) - DCI(t - 1)) / DCI(t - 1); construct an adaptive regulation rule base (ACRB) based on the restoration effect; output the adaptive regulation strategy (ACS);

[0178] S463. Connectivity balance optimization based on multi-level objectives

[0179] Read the adaptive regulation strategy (ACS) and the comprehensive water system connectivity index (WCSI); construct an analytic hierarchy process model to determine the weight vector w_vec under different objectives; design a multi-objective balance optimization equation: min Z = Σ(w j •(c j - c j *) 2 / c j 2 ) where c j is the current value of the j-th objective, cj * is the ideal value; output the connectivity balance plan CBP.

[0180] In this embodiment, an adaptive threshold function based on spatial context is introduced, which solves the deficiencies of the traditional fixed threshold method in the identification of small water bodies and is particularly suitable for the complex water system environment affected by small hydropower. Incorporating the principles of fluid mechanics into the connectivity discrimination process enables the accurate identification of the true hydraulic connectivity state of water bodies. The concept and quantification method of the implicit connectivity of the dewatered reach are proposed, solving the specific connectivity evaluation problem of diversion-type small hydropower. Integrating power station operation data, hydrological data, and sensor network data realizes the comprehensive perception of the operation status of small hydropower, providing a reliable data basis for connectivity evaluation. Solving the problem of the lack of a clear goal orientation in traditional connectivity evaluation, setting differentiated connectivity thresholds according to different ecological function requirements makes the evaluation results more practical.

[0181] In another embodiment of the present application, a typical small watershed in the upper reaches of the Yangtze River is selected as the research area. There are 27 small hydropower stations distributed in this area, including 18 diversion-type power stations and 9 dam-type power stations, with a total installed capacity of 126.8 MW. The watershed area is about 1250 km 2 , and the altitude is between 800 - 2200 m, and the total length of the river network is about 860 km.

[0182] The following data is collected as the research basis:

[0183] Remote sensing image data: A total of 24 scenes of Sentinel-2 optical images (10 m resolution) from January to December 2023, 36 scenes of Sentinel-1 radar data (10 m resolution), and 15 scenes of high-resolution images (0.5 m resolution) obtained by drones.

[0184] Hydrological data: The daily average flow and water level data of 10 hydrological stations in the research area from 2021 to 2023.

[0185] Small hydropower station data: The geographical location, development mode, installed capacity, length of the diversion channel, coordinates of the water intake and tail water outlet, daily power generation, water diversion volume, and unit operation records of 27 small hydropower stations from 2021 to 2023.

[0186] Water ecological environment data: Water quality indicators (pH, dissolved oxygen, ammonia nitrogen, total phosphorus, etc.), benthic biodiversity index, and fish monitoring data of 18 monitoring points in the research area.

[0187] According to the hydrological characteristics of the research area, the whole year is divided into the flood season (May - September), the normal water season (March - April, October - November), and the dry season (December - February).

[0188] Register and fuse the collected Sentinel-2 optical images, Sentinel-1 radar data, and UAV high-resolution images according to the time phase to construct a spatio-temporal multi-source database covering the wet season, normal water season, and dry season. For the Sentinel-2 optical images, calculate the Normalized Difference Water Index (NDWI): NDWI = (ρGreen - ρNIR) / (ρGreen + ρNIR); where: ρGreen is the reflectance of the green band; ρNIR is the reflectance of the near-infrared band. For the Sentinel-1 radar data, extract the backscattering coefficient and perform terrain correction to generate the backscattering coefficient matrix BSM.

[0189] Based on the spatio-temporal multi-source database, calculate the frequency of each pixel being identified as water during the observation period, and construct the Water Existence Frequency matrix WEF: WEF(x,y) = ∑ω(t)·WP(x,y,t) / N; where: WEF(x,y) is the water existence frequency at coordinates (x,y); WP(x,y,t) is the water determination result (0 or 1) at coordinates (x,y) at time t; ω(t) is the time weight, which is 0.3, 0.4, and 0.5 for the wet season, normal water season, and dry season respectively; N is the total number of weighted observations.

[0190] Taking a tributary in the study area as an example, the water existence frequency matrix WEF is calculated, and the pixel values are distributed between [0,1]. The WEF values of permanent water bodies (such as the main river channel) are mostly above 0.85, the WEF values of seasonal rivers are between 0.35 and 0.7, while the WEF values of temporary water bodies are usually below 0.3.

[0191] For the statistical distribution of water body frequencies in the study area, use a two-peak mixture Gaussian model for fitting: P(WEF) = π1·N(μ1, σ1 2 ) + π2·N(μ2, σ2 2 ); where: P(WEF) is the probability distribution of the water existence frequency; π1, π2 are the weights of the two Gaussian components, π1 + π2 = 1; N(μ, σ 2 ) is the Gaussian distribution; μ1, μ2 are the means of the two Gaussian components; σ1, σ2 are the standard deviations of the two Gaussian components.

[0192] After fitting the WEF data of the study area, we get: π1 = 0.72, π2 = 0.28, μ1 = 0.17, μ2 = 0.86, σ1 = 0.12, σ2 = 0.09. According to the fitting results, determine the water existence probability threshold WEPT = 0.42.

[0193] Then design a threshold adaptive function based on spatial context: τ(x,y) = WEPT·[1+α·exp(-d(x,y) / β)]; where: τ(x,y) is the adaptive threshold at pixel (x,y); WEPT is the water body presence probability threshold, with a value of 0.42; d(x,y) is the distance from pixel (x,y) to the nearest determined water body (in pixels); α and β are adaptive parameters, determined to be α = 0.5 and β = 8 according to the water body distribution characteristics of the study area.

[0194] For the tiny water bodies in the upper reaches of a tributary within the study area, the recognition result of the traditional fixed threshold method (WEPT = 0.42): the water body area is 86.4 km 2 , compared with the manually interpreted result (the true water body area is 107.2 km 2 ), the recognition accuracy is 80.6%. After adopting the adaptive threshold segmentation method: the water body area is 103.8 km 2 , and the accuracy is improved to 96.8%. In the recognition of small river segments with a width less than 3 pixels, the accuracy is improved from 65.3% to 91.7%, reflecting the advantage of this method in the recognition of tiny water bodies.

[0195] Based on the tiny water body segmentation result and the digital elevation model, construct a water body connectivity candidate graph CCG. Then calculate the hydraulic gradient HG and the velocity field VF on the candidate connectivity path. HG(i,j) = (H(i) - H(j)) / L(i,j); where: HG(i,j) is the hydraulic gradient between water bodies i and j; H(i) and H(j) are the average water surface elevations of water bodies i and j respectively; L(i,j) is the straight-line distance between water bodies i and j. VF(i,j) = k·HG(i,j)·R(i,j)^(2 / 3)·n^(-1); where: VF(i,j) is the estimated flow velocity between water bodies i and j; k is a coefficient with a value of 1; R(i,j) is the hydraulic radius, calculated according to the river reach characteristics; n is the Manning roughness coefficient, with a value of 0.03 - 0.05 according to the river channel type.

[0196] Design a connectivity discrimination function CS(i,j) = HG(i,j)·VF(i,j)·[WEF(i)+WEF(j)] / 2 - θ; where: CS(i,j) is the connectivity discrimination value between water bodies i and j; θ is the connectivity threshold, with a value of 0.05.

[0197] When CS(i,j) > 0, it is determined that water bodies i and j are in a connected state. For the discrimination of 145 candidate connected water bodies in the study area, the accuracy of the traditional image-based discrimination method is 76.5%, while the accuracy of the discrimination method based on hydrodynamic constraints reaches 94.3%.

[0198] Extract the coordinates of the water intake and tailwater outlets of 18 diversion-type power stations from the location data of small hydropower projects, and determine the scope of the dewatered and deconnected river sections on the dynamic connectivity map of small water bodies, with a total of 27 sections and a total length of 96.4 km.

[0199] Design a discriminant model for the implicit connectivity of dewatered and deconnected river sections: IC(i) = η·(1 - exp(-λ·Qeco / Qnat)); where: IC(i) is the implicit connectivity index of river section i; Qeco is the ecological flow, determined based on measured data; Qnat is the natural flow, calculated based on data from the upstream hydrological station; η and λ are parameters, with values of 1.0 and 3.0 respectively.

[0200] Taking a typical diversion-type power station as an example, during the wet season, Qeco / Qnat = 0.32, and IC = 0.62 is calculated; during the normal water season, Qeco / Qnat = 0.15, and IC = 0.37; during the dry season, Qeco / Qnat = 0.08, and IC = 0.21. This indicates that the connectivity of the dewatered and deconnected river sections of this power station significantly decreases during the dry season, which may have a greater impact on the migration of aquatic organisms.

[0201] Based on the water system network data, calculate the dendritic connectivity index DCI, river continuity index RCI, and river fragmentation index CAFI: DCI = 1 - ∑Di / ∑D0; where: Di is the length of each actual river section in the study area; D0 is the length of the corresponding river section without obstacles. RCI = 1 - ∑(Li·Pi) / L; where: Li is the length of the river section affected by the i-th obstacle; Pi is the non-passability coefficient of the i-th obstacle; L is the total length of the river network. CAFI = ∑(ai / A)·100; where: ai is the upstream catchment area of the i-th obstacle; A is the total area of the basin.

[0202] For small hydropower projects developed by dams, calculate the improved river fragmentation index CAFIg = ∑(ai·ICn) / (A·IC)·100; where: ai is the upstream catchment area of the i-th barrier; A is the total area of the basin; IC is the maximum installed capacity of the small hydropower project, 50 MW; ICn is the installed capacity of the n-th small hydropower project. For the 9 dam-type power stations in the study area, CAFIg = 17.46 is calculated, while the traditional CAFI = 12.83, indicating that considering the installed capacity, the impact of dam-type small hydropower on water system connectivity is more significant.

[0203] For small hydropower stations developed by diversion, calculate the improved river fragmentation index CAFIg considering the impact of dewatered flow: CAFIg = ∑(ai·ICn·q) / (A·IC·Q)·100; where: q is the diversion flow of the hydropower station; Q is the upstream flow of the hydropower station.

[0204] Calculate the CAFI of 18 diversion power stations in the study area during different hydrological periods: CAFI in the wet season = 9.76, CAFI in the normal water season = 15.34, and CAFI in the dry season = 26.82. The results show that the impact of diversion power stations on river connectivity is significantly enhanced in the dry season, about 2.7 times that in the wet season.

[0205] According to the operation mode data and hydrological data of 27 small hydropower stations, 165 main river reaches in the study area are divided into three situations according to the connectivity status: continuous connectivity (P1), periodic connectivity (P2), and temporary connectivity (P3). On this basis, a time matrix of river reach connectivity M1(i,j,k) = {L(i), T(j), P(k)} is constructed; where: L(i) is the i-th river reach, i = 1,2,...,165; T(j) is the hydrological period, j = 1 (wet season), 2 (normal water season), 3 (dry season); P(k) is the connectivity status, k = 1 (continuous connectivity), 2 (periodic connectivity), 3 (temporary connectivity).

[0206] Statistical analysis shows that: in the wet season, the continuously connected river reaches account for 72.1%, the periodically connected river reaches account for 23.6%, and the temporarily connected river reaches account for 4.3%; in the normal water season, the continuously connected river reaches account for 51.5%, the periodically connected river reaches account for 32.7%, and the temporarily connected river reaches account for 15.8%; in the dry season, the continuously connected river reaches account for 33.9%, the periodically connected river reaches account for 35.2%, and the temporarily connected river reaches account for 30.9%.

[0207] Based on the operation data of 27 small hydropower stations from 2021 to 2023, a time series feature vector TFV is designed, which includes features such as upstream flow, power generation, water level change, and rainfall in the past 7 days. A prediction model Qt = Bi-LSTM(TFV, Att(TFV)) is constructed; where: Qt is the predicted diversion flow at time t; TFV is the time series feature vector; Bi-LSTM is the bidirectional long short-term memory network function; Att is the time series attention function.

[0208] The calculation formula of the time series attention function is Att(TFV) = softmax(W2·tanh(W1·TFV + b1) +b2); where: W1, W2 are weight matrices; b1, b2 are bias terms; tanh is the hyperbolic tangent activation function; softmax is the normalization function.

[0209] Model training results for a typical diversion hydropower station: The average relative error is 5.8%, significantly better than 11.3% of the traditional regression model. The prediction model shows that the water diversion volume of the hydropower station has a non-linear relationship with the upstream water inflow. The average water diversion rate (water diversion volume / upstream water inflow) is 0.48 during the wet season, 0.67 during the normal water season, and as high as 0.83 during the dry season.

[0210] Match the connectivity status in the time matrix of river reach connectivity with the predicted water diversion flow data according to time nodes to construct a dynamic connectivity evaluation model DCI(t) = 1 - (PWF(t) / NF(t))·(1-exp(-θ·L / W)); where: DCI(t) is the dynamic connectivity index at time t; PWF(t) is the predicted water diversion flow at time t; NF(t) is the natural flow at time t; L is the length of the dewatered reach; W is the river width; θ is a parameter with a value of 0.05.

[0211] Calculate the dynamic connectivity indices of different river reaches in the study area during three hydrological periods. The results show that: during the dry season, 36% of the river reaches have DCI values below 0.4, which belongs to the state of seriously affecting connectivity; the proportion of river reaches with DCI values below 0.4 during the normal water season is 18%; and only 6% during the wet season. Design a connectivity time series fluctuation evaluation index CF(i) = std(DCI(i,t)) / mean(DCI(i,t)); where: CF(i) is the connectivity fluctuation coefficient of the i-th river reach; std is the standard deviation function; mean is the mean function; DCI(i,t) is the dynamic connectivity index value of the i-th river reach at time series t.

[0212] The average CF value of the river reaches in the study area is 0.42, and the highest reaches 0.87, indicating that the operation of small hydropower stations leads to significant temporal fluctuations in river system connectivity.

[0213] Organize the dynamic connectivity status map of small water bodies, the hidden connectivity status of dewatered reaches, and the change matrix of the water diversion flow of the hydropower station and river flow into a data structure containing spatial and temporal dimensions to form a spatio-temporal coupling analysis framework STCF(x,y,t) = {MWCS(x,y,t), WRIC(x,y,t), DCI(x,y,t)}; where: STCF(x,y,t) is the spatio-temporal coupling data at coordinate (x,y) at time t; MWCS(x,y,t) is the dynamic connectivity status of small water bodies; WRIC(x,y,t) is the hidden connectivity status of dewatered reaches; DCI(x,y,t) is the dynamic connectivity index.

[0214] Design operation mode - Connectivity correlation evaluation index RC(i,t) = γ·(OP(i,t) - OPmin) / (OPmax - OPmin) + (1 - γ)·DCI(i,t); where: RC(i,t) is the operation - connectivity correlation index of the i - th power station at time t; OP(i,t) is the operation parameter of the i - th power station at time t, such as the diversion rate; OPmin and OPmax are the minimum and maximum values of the operation parameter respectively; DCI(i,t) is the dynamic connectivity index at time t; γ is the weight factor, with a value of 0.4.

[0215] The analysis results for a small hydropower station in the study area show that when the diversion rate exceeds 0.75, the DCI value of the connectivity index of the corresponding river section drops sharply, and the RC value drops from 0.68 to 0.42, indicating that high - diversion - rate operation has a significant negative impact on river system connectivity.

[0216] Based on the water ecological monitoring data in the study area, three main ecological protection objectives are determined: fish migration protection (g1), water quality maintenance (g2), and habitat protection (g3).

[0217] Set connectivity requirement thresholds for different ecological objectives, and construct an ecological connectivity threshold matrix ECTM(i,g) = {DCT(g1), DCT(g2), DCT(g3)}; where: ECTM(i,g) is the connectivity threshold of the i - th river section for ecological objective g; DCT(g1), DCT(g2), DCT(g3) are the connectivity thresholds for fish migration protection, water quality maintenance, and habitat protection respectively. According to the fish migration characteristics of the study area, DCT(g1) = 0.65 is determined; according to the water self - purification capacity requirement, DCT(g2) = 0.50 is determined; according to the habitat stability requirement, DCT(g3) = 0.40 is determined.

[0218] Solve the constrained optimization equation min J = ∑(OP(i,t) - OPopt(i)) 2 ; s.t. DCI(i,t) ≥ ECTM(i,g); where: J is the optimization objective function; OP(i,t) is the operation parameter of the i - th power station at time t; OPopt(i) is the optimal operation parameter of the i - th power station; DCI(i,t) is the dynamic connectivity index at time t; ECTM(i,g) is the ecological connectivity threshold matrix.

[0219] Taking a key river section in the study area as an example, to meet the fish migration protection target (DCT = 0.65), the optimal water diversion rate is calculated to be 0.58, and the corresponding annual power generation decreases by 18.3% compared with the maximum water diversion rate; to meet the water quality maintenance target (DCT = 0.50), the optimal water diversion rate is 0.67, and the power generation decreases by 8.9%; to meet the habitat protection target (DCT = 0.40), the optimal water diversion rate is 0.72, and the power generation decreases by 5.3%.

[0220] Design a collaborative scheduling optimization objective function based on connectivity constraints: max E = ∑(ki·Pi(t)); s.t. ∑(DCI(i,t)·wi) ≥ DCT; where: E is the total power generation; Pi(t) is the power generation of the i-th power station at time t; ki is the weight coefficient, which is proportional to the installed capacity of the power station; DCI(i,t) is the dynamic connectivity index at time t; wi is the water system importance weight, which is related to the ecological sensitivity of the river section; DCT is the connectivity threshold.

[0221] Use the improved particle swarm algorithm to solve the optimization problem. The particle position update formula is X(t+1) = X(t) + V(t+1); where: X(t) is the particle position at time t (representing the water diversion rate of each power station); V(t+1) is the particle velocity at time t+1.

[0222] The particle velocity update formula is V(t+1) = w·V(t) + c1·r1·(Pbest - X(t)) + c2·r2·(Gbest - X(t)) + c3·r3·(DCI(t) - DCT); where: w is the inertia weight; c1, c2, c3 are acceleration coefficients; r1, r2, r3 are random numbers between [0,1]; Pbest is the individual optimal position; Gbest is the global optimal position; DCI(t) is the dynamic connectivity index at time t; DCT is the connectivity threshold.

[0223] Carry out collaborative optimization on 5 cascade small hydropower stations in a small watershed in the study area. The results show that: compared with the single-station independent optimization, the collaborative optimization scheme can increase the total system power generation by 12.6% and the maximum single-station power generation by up to 18.9% under the condition of meeting the same connectivity constraints.

[0224] Design a connectivity state change detector. Calculate the connectivity state change signal CSVS(t) = (DCI(t) - DCI(t-1))·exp(-|DCI(t) - DCI(t-1)| / σ); where: CSVS(t) is the connectivity state change signal at time t; DCI(t) and DCI(t-1) are the dynamic connectivity indices at time t and time t-1 respectively; σ is the sensitivity parameter, with a value of 0.2.

[0225] Based on the connected state change signal, design the operation parameter adjustment equation ΔOP(i,t) = μ·CSVS(t)·(DCItar - DCI(i,t)); where: ΔOP(i,t) is the adjustment amount of the operation parameter of the i-th power station at time t; CSVS(t) is the connected state change signal; DCItar is the target connectivity index value; DCI(i,t) is the current connectivity index value; μ is the adjustment coefficient, and the initial value is set to 0.15.

[0226] Taking a diversion power station in the study area as an example, when it is monitored that the connectivity index DCI of the downstream river section suddenly drops from 0.58 to 0.43, the calculated value of CSVS is -0.13, and the calculated adjustment amount of the diversion rate is -0.047, that is, the diversion rate needs to be reduced from 0.72 to 0.673 to restore the river section connectivity.

[0227] Design the connectivity restoration evaluation index RI(t) = (DCI(t) - DCI(t - 1)) / DCI(t - 1); where: RI(t) is the restoration index at time t; DCI(t) and DCI(t - 1) are the dynamic connectivity indexes at time t and t - 1 respectively.

[0228] Construct an adaptive regulation rule base based on the restoration effect, and dynamically adjust the adjustment coefficient μ in the operation parameter adjustment equation according to the change trend of the connectivity restoration evaluation index. μ(t + 1) = μ(t)·(1 + ε·sign(RI(t))·(1 - exp(-|RI(t)|))); where: μ(t + 1) and μ(t) are the adjustment coefficients at time t + 1 and time t respectively; ε is the learning rate, with a value of 0.2; sign is the sign function; RI(t) is the restoration index at time t.

[0229] In the 30-day regulation experiment, after introducing the feedback regulation mechanism, the average connectivity index of the key river section in the study area increased from 0.47 to 0.63, an increase of 34.0%; the connectivity fluctuation coefficient CF decreased from 0.42 to 0.18, a decrease of 57.1%; at the same time, the total power generation of the power station only decreased by 7.2%, achieving a good balance between ecological connectivity and economic benefits.

[0230] Based on the water ecological environment data, determine the connectivity evaluation impact index set: surface water ratio W1, water quality comprehensive index WQI, ecological flow guarantee rate EFG, and biodiversity index BDI.

[0231] After standardizing the index data, calculate the information entropy of each index \(E(j)=-\frac{1}{\ln(n)}\cdot\sum(p_{ij}\cdot\ln(p_{ij}))\); where: \(E(j)\) is the information entropy of the \(j\)-th index; \(n\) is the number of samples; \(p_{ij}\) is the standardized value of the \(j\)-th index of the \(i\)-th sample. Calculate the weight of each index \(w(j)=\frac{1 - E(j)}{\sum(1 - E(j))}\); where: \(w(j)\) is the weight of the \(j\)-th index. The weights of each index are calculated as: \(w(W1)=0.23\), \(w(WQI)=0.31\), \(w(EFG)=0.28\), \(w(BDI)=0.18\).

[0232] Adopt the grey relational analysis method to calculate the influence degree \(r(i,j)=\frac{\min\min|y_0(k)-x_i(k)|+\rho\cdot\max\max|y_0(k)-x_i(k)|}{|y_0(k)-x_i(k)|+\rho\cdot\max\max|y_0(k)-x_i(k)|}\) of each index on the water system connectivity; where: \(r(i,j)\) is the correlation degree between the \(j\)-th index of the \(i\)-th sample and the reference sequence; \(y_0(k)\) is the reference sequence value; \(x_i(k)\) is the comparison sequence value; \(\rho\) is the resolution coefficient, with a value of \(0.5\).

[0233] Calculate the grey relational matrix of each river section in the study area to obtain the average influence degree of different indexes on connectivity: the proportion of surface water is \(0.78\), the comprehensive water quality index is \(0.83\), the ecological flow guarantee rate is \(0.91\), and the biodiversity index is \(0.68\).

[0234] Integrate the dynamic connectivity index and the influence degree matrix, and calculate the comprehensive water system connectivity index \(WCSI(i)=\sum(DCI(i,j)\cdot w(j)\cdot r(i,j))\) under the influence of small hydropower stations; where: \(WCSI(i)\) is the comprehensive water system connectivity index of the \(i\)-th river section; \(DCI(i,j)\) is the dynamic connectivity index of the \(i\)-th river section in the \(j\)-th hydrological period; \(w(j)\) is the weight of the \(j\)-th index; \(r(i,j)\) is the grey relational degree of the \(j\)-th index of the \(i\)-th river section.

[0235] According to the \(WCSI\) value, determine the rectification priority of 27 small hydropower stations in the study area. The evaluation results show that: the \(WCSI\) values of 5 power stations are lower than \(0.35\), belonging to the type that urgently needs rectification; the \(WCSI\) values of 8 power stations are between \(0.35\) and \(0.50\), belonging to the type that needs optimized dispatching; the \(WCSI\) values of 14 power stations are higher than \(0.50\), and the operating conditions are relatively reasonable.

[0236] In another embodiment of the present application, taking the upper reaches of a tributary in the study area as an example, when the traditional fixed threshold method has a WEF value of 0.42, the pixel (158, 237) is determined as non-water body. The distance d(158, 237) from this pixel to the nearest determined water body is 3.6 pixels. Calculate using the adaptive threshold function:

[0237] τ(158, 237) = 0.42·[1 + 0.5·exp(-3.6 / 8)] = 0.42·[1 + 0.5·0.6376] = 0.42·1.3188 = 0.5539; Since the WEF value of this pixel is 0.48, which is lower than the calculated adaptive threshold of 0.5539, it is still determined as non-water body. When considering the topographic features and seasonal river characteristics of this area and adjusting the β value to 12, recalculate: τ(158, 237) = 0.42·[1 + 0.5·exp(-3.6 / 12)] = 0.42·[1 + 0.5·0.7408] = 0.42·1.3704 = 0.3743. At this time, the WEF value of 0.48 of this pixel is higher than the adjusted adaptive threshold of 0.3743, so it is correctly determined as a water body, which is consistent with the field verification result.

[0238] In another embodiment of the present application, for two small water bodies a and b in the study area that seemingly communicate with each other but are blocked by a ridge, the traditional image discrimination method misjudges them as communicating. Apply the hydrodynamic constraint discrimination method:

[0239] The average water surface elevations of water bodies a and b are H(a) = 1568.3m and H(b) = 1569.7m respectively; the straight-line distance L(a, b) between the two water bodies is 127.6m; calculate the hydraulic gradient HG(a, b) = (H(b) - H(a)) / L(a, b) = (1569.7 - 1568.3) / 127.6 = 0.0110; according to the river channel characteristics, the hydraulic radius R(a, b) = 0.41m, and the Manning roughness coefficient n = 0.042; calculate the velocity field VF(a, b) = 1·0.0110·0.41^(2 / 3)·0.042^(-1) = 0.2844 m / s; the water body presence frequencies of water bodies a and b are WEF(a) = 0.68 and WEF(b) = 0.57 respectively; the connectivity discrimination calculation CS(a, b) = 0.0110·0.2844·(0.68 + 0.57) / 2 - 0.05 = -0.0344 < 0; Since CS(a, b) < 0, it is determined that water bodies a and b are not connected, which is consistent with the actual hydrological conditions.

[0240] In another embodiment of the present application, taking a typical diversion-type power station in the research area as an example, the installed capacity of the power station is 4.8 MW, the coordinates are (32°15'23"N, 106°47'36"E), and the river channel length from the water intake to the tail water outlet is 3.8 km.

[0241] During the wet season, Qeco = 1.24 m 3 / s, Qnat = 3.86 m 3 / s, Qeco / Qnat = 0.32; IC = 1.0·(1 - exp(-3.0·0.32)) = 1.0·(1 - 0.3823) = 0.6177.

[0242] During the normal water season, Qeco = 0.36 m 3 / s, Qnat = 2.37 m 3 / s, Qeco / Qnat = 0.15; IC = 1.0·(1 - exp(-3.0·0.15)) = 1.0·(1 - 0.6376) = 0.3624.

[0243] During the dry season, Qeco = 0.11 m 3 / s, Qnat = 1.41 m 3 / s, Qeco / Qnat = 0.08; IC = 1.0·(1 - exp(-3.0·0.08)) = 1.0·(1 - 0.7866) = 0.2134.

[0244] Analysis shows that as the proportion of ecological flow decreases, the implicit connectivity index decreases exponentially, and the connectivity function is significantly damaged during the dry season. Compared with the traditional binary connectivity discrimination (connected or disconnected), this method can quantitatively describe the continuous change of the connectivity function, providing a scientific basis for the ecological operation of diversion-type power stations.

[0245] In another embodiment of the present application, taking a diversion-type power station in the research area as an example, a time series feature vector TFV is constructed, which includes 15 feature variables such as the average daily flow, rainfall, and power generation in the upstream for 7 days.

[0246] Dimensions of the model input layer: [7, 15] (7 days, 15 features); dimensions of the bidirectional LSTM hidden layer: 128; dimensions of the attention layer: 64; dimensions of the output layer: 1 (predicting the diversion flow);

[0247] A typical input example (simplified display) in the model training data: the upstream flow sequence (m 3 / s): [5.43, 4.89, 6.12, 7.56, 5.78, 4.32, 3.98]; Power generation sequence of the power station (MWh): [38.6, 35.2, 42.7, 49.8, 40.3, 32.5, 30.1]; Rainfall sequence (mm): [0, 12.5, 26.8, 5.6, 0, 0, 0];

[0248] Calculate the weights of each time step using the attention mechanism: Att = softmax([0.12, 0.15, 0.26, 0.19, 0.14, 0.08, 0.06]) = [0.11, 0.13, 0.23, 0.17, 0.12, 0.08, 0.05].

[0249] After model calculation, the predicted diversion flow rate Qt = 3.27 m 3 / s, compared with the measured value of 3.42 m 3 / s, the relative error is 4.4%.

[0250] In another embodiment of the present application, for a key river section within the research area, this river section is an important migration route for the local endemic fish Pelteobagrus fulvidraco. Set ecological goals: fish migration protection (g1); Connectivity threshold: DCT(g1) = 0.65; Optimal operating parameters of the power station (maximum power generation benefit): OPopt = 0.85 (diversion rate); Constraint condition: DCI(i,t) ≥ 0.65; When the power station operates at OPopt = 0.85: DCI = 1 - (PWF / NF)·(1 - exp(-0.05·L / W)) = 1 - (3.57 / 4.2)·(1 - exp(-0.05·3800 / 12)) = 0.43 < 0.65;

[0251] It does not meet the connectivity requirements for fish migration protection, and an optimization problem needs to be solved. Through iterative calculation, the optimal diversion rate that meets the constraint conditions is 0.58. At this time: DCI = 1 - (2.44 / 4.2)·(1 - exp(-0.05·3800 / 12)) = 0.66 > 0.65.

[0252] It meets the connectivity requirements for fish migration protection, and the power generation decreases by 18.3% compared to the optimal operation, which is within an acceptable range.

[0253] In another embodiment of the present application, taking a small hydropower station within the research area as an example, this power station is located in an ecologically sensitive river section, and the target connectivity index DCItar = 0.55.

[0254] Initial state: the diversion rate of the power station OP(t - 1) = 0.65, corresponding to the connectivity index DCI(t - 1) = 0.58;

[0255] It is monitored that the rainfall decreases and the upstream water inflow drops. The current connectivity index DCI(t) = 0.43;

[0256] Calculate the connectivity state change signal: CSVS(t) = (0.43 - 0.58)·exp(-|0.43 - 0.58| / 0.2) = -0.15·exp(-0.15 / 0.2) = -0.15·0.4724 = -0.0709;

[0257] Calculate the adjustment amount of the operating parameter: ΔOP(t) = 0.15·(-0.0709)·(0.55 - 0.43) = 0.15·(-0.0709)·0.12 = -0.0013;

[0258] Update the diversion rate of the power station: OP(t) = OP(t - 1) + ΔOP(t) = 0.65 + (-0.0013) = 0.6487 ≈ 0.65;

[0259] Since the calculated adjustment amount is small, the diversion rate of the power station remains unchanged. As the upstream water inflow continues to drop, the connectivity index drops to 0.37: Calculate the connectivity state change signal CSVS(t + 1) = (0.37 - 0.43)·exp(-|0.37 - 0.43| / 0.2) = -0.06·exp(-0.06 / 0.2) = -0.06·0.7408 = -0.0444; Calculate the adjustment amount of the operating parameter ΔOP(t + 1) = 0.15·(-0.0444)·(0.55 - 0.37) = 0.15·(-0.0444)·0.18 = -0.0012; Update the diversion rate of the power station OP(t + 1) = OP(t) + ΔOP(t + 1) = 0.65 + (-0.0012) = 0.6488 ≈ 0.65.

[0260] Continuous monitoring shows that the connectivity is still decreasing, triggering the cumulative adjustment mechanism and adjusting the μ value to 0.25: Recalculate the adjustment amount of the operating parameter ΔOP(t + 2) = 0.25·(-0.0576)·(0.55 - 0.32) = 0.25·(-0.0576)·0.23 = -0.0033; After multiple cumulative adjustments, the diversion rate of the power station is lowered to 0.59, and the connectivity index is restored to 0.54, approaching the target value.

[0261] This feedback control mechanism successfully maintained the connectivity of the key river sections in the study area at an ideal state during the 30-day trial, with the fluctuation range controlled within ±8%, which was significantly lower than the ±25% under the traditional fixed-parameter operation mode.

[0262] The technical idea for the refined segmentation of water bodies is as follows: attempting to use the spatio-temporal multi-source data fusion method to solve the problem of small water body recognition, which successfully addresses the limitations of single-temporal remote sensing data affected by factors such as cloud cover and vegetation occlusion. However, after adopting spatio-temporal multi-source data fusion, the problem of determining the water body segmentation threshold emerged, including the inability of traditional fixed-threshold segmentation methods to adapt to the differences in water body characteristics in different regions and different periods; resulting in insufficient recognition accuracy for small water bodies and fragmented water bodies, especially in mountainous areas with complex terrain. To solve the threshold determination problem, an adaptive threshold segmentation method was introduced, significantly improving the recognition accuracy of small water bodies and fragmented water bodies. However, after water body recognition based only on image features, the problem of connectivity discrimination occurred. This includes: the water bodies recognized on the image may not be physically connected; especially in mountainous areas with complex terrain, it is easy to misjudge terrain shadows as connected water bodies. Therefore, a hydrodynamic constraint discrimination method was introduced to make the discrimination results conform to the actual hydrodynamic laws and avoid misjudgments based solely on images. In addition, in the special case of the dewatered reach of diversion-type small hydropower stations, it was found that there are limitations in existing connectivity discrimination methods, including: the diversion-type power station reduces the water volume in the river reach but does not completely cut off the flow, and traditional connectivity evaluations are difficult to reflect this "semi-connected" state; resulting in inaccurate assessments of the impacts of diversion-type small hydropower stations; therefore, a method for identifying the hidden connectivity of dewatered river reaches was proposed: extracting the coordinates of the water intake and tailwater outlets of diversion-type power stations from the location data of small hydropower projects; designing a discrimination model for the hidden connectivity of dewatered river reaches and introducing the ratio of ecological flow to natural flow as a key parameter; achieving an accurate assessment of the connectivity of the river reach affected by the diversion-type power station.

[0263] Regarding the problem of dynamic evaluation of connectivity, the technical ideas are as follows: To address the issue that static evaluation cannot reflect the temporal variation of connectivity, we first attempted to construct a time matrix of river reach connectivity, which successfully reflected the temporal variation characteristics of connectivity. However, this static classification method lacks an accurate description of the dynamic changes in diversion flow rates, including the dynamic changes in the diversion water volume of small hydropower plants with upstream inflows and power generation demands during actual operation, resulting in the inability of the connectivity evaluation results to reflect the connectivity changes under real-time operating conditions. To solve this problem, a dynamic prediction model for the diversion flow rate of small hydropower plants was introduced, which can accurately predict the diversion flow rate under different hydrological conditions. However, after predicting the diversion flow rate, the problem of quantifying the impact on connectivity emerged, that is, the impact of the diversion flow rate on connectivity is not a simple linear relationship; the characteristics of the river channel, such as the length and width of the dewatered reach, need to be considered. Therefore, a dynamic quantification method for the impact of connectivity based on the diversion flow rate was designed, which can accurately quantify the dynamic impact of small hydropower plant diversion on river system connectivity. To organically combine the connectivity state in space with the dynamic changes in time, a spatio-temporal coupling analysis framework was further constructed, integrating the dynamic connectivity state map of small water bodies, the hidden connectivity state of the dewatered reach, and the dynamic connectivity index sequence, achieving the spatio-temporal integration of connectivity evaluation.

[0264] Regarding the optimal regulation guided by ecological goals, based on the connectivity evaluation, it was found that there is a lack of clear ecological goal orientation. Traditional connectivity evaluation often pursues maximizing connectivity, ignoring the differentiated needs of different ecosystem functions; this results in the evaluation results being difficult to directly guide practical decisions. To solve this problem, a method for determining the connectivity critical threshold for different ecological goals was introduced to obtain the optimal operating parameters that meet ecological requirements. However, after optimizing a single small hydropower plant, the problem of coordination of the small hydropower plant group system emerged, including: there is a hydraulic connection between multiple small hydropower plants in the basin, and isolated optimization may lead to disharmony at the system level; resulting in suboptimal solutions for the overall power generation efficiency and ecological benefits of the system. Therefore, an optimization model considering the synergy effect of the small hydropower plant group was designed to obtain a coordinated scheduling strategy. To achieve the dynamic balance between the river system connectivity state and the operation of small hydropower plants, a feedback regulation mechanism also needs to be established so that the operation of small hydropower plants can automatically adjust the operating parameters according to the real-time changes in the river system connectivity state. Finally, to evaluate the regulation effect and continuously optimize, an adaptive regulation method for connectivity restoration was designed, which can continuously optimize the regulation strategy according to the connectivity restoration effect at different time scales and spatial scales.

[0265] In summary, to solve the problem of dynamically identifying the connectivity state of small water bodies. First, multi-temporal optical remote sensing images, radar data, and high-resolution UAV images are integrated to establish a spatio-temporal multi-source database, overcoming the limitations of a single data source. Second, the concept of a water body presence frequency matrix is introduced and an adaptive threshold segmentation method is designed. Through a bimodal mixture Gaussian model and a threshold adaptive function based on spatial context, high-precision identification of small water bodies is achieved. The principle of water flow dynamics is introduced into the connectivity discrimination process, and by calculating the hydraulic gradient and velocity field, it is ensured that the discrimination results conform to the actual hydrological laws. Aiming at the special impact of diversion-type small hydropower, the concept and quantification model of the implicit connectivity of the dewatered reach are proposed, and a quantitative relationship between the dewatered flow and connectivity is established through the ratio of ecological flow to natural flow, realizing the accurate characterization of the water system connectivity state affected by diversion-type small hydropower.

[0266] To achieve the transition from static evaluation to dynamic regulation. First, by integrating power station operation data, hydrological station flow data, and real-time monitoring data, a comprehensive perception of the operation state of small hydropower is realized. Second, a diversion flow prediction model combining a bidirectional long short-term memory network and an attention mechanism is designed to solve the problem of difficult real-time acquisition of the diversion water volume of small hydropower. For connectivity evaluation, a dynamic connectivity evaluation index under the influence of diversion is designed, converting static parameters into time-varying parameters to accurately quantify the dynamic impact of small hydropower operation on water system connectivity. In addition, a connectivity threshold matrix and a collaborative scheduling optimization model for small hydropower groups are established for different ecological goals. Combining the closed-loop feedback mechanism of connectivity state and operation adjustment, dynamic optimal regulation of connectivity is achieved, enabling the evaluation results to directly guide the eco-friendly operation of small hydropower and solving the fundamental deficiency of the lack of a regulation mechanism in traditional evaluations.

[0267] The preferred embodiments of the present invention have been described in detail above. However, the present invention is not limited to the specific details in the above embodiments. Within the scope of the technical concept of the present invention, various equivalent transformations can be made to the technical solutions of the present invention, and these equivalent transformations all belong to the protection scope of the present invention.

Claims

1. A comprehensive calculation method for river connectivity affected by small hydropower stations, characterized in that, It includes the following steps: Collect research datasets, including remote sensing image data, hydrological data, small hydropower station data, and water ecological environment data of the research area; Using the remote sensing image data, adopt deep learning methods to conduct refined segmentation of water bodies to generate water system network data; Based on the water system network data and small hydropower station data, combined with hydrological data, calculate the basic indicators of river network water system connectivity; According to the basic indicators of river network water system connectivity and small hydropower station data, construct a spatio-temporal coupling model to generate dynamic connectivity indicators; Based on the dynamic connectivity indicators and water ecological environment data, conduct a comprehensive evaluation of water system connectivity to form the evaluation results of the rectification priorities of small hydropower stations; The steps for calculating the basic indicators of river network water system connectivity include: Based on the water system network data, combined with river network structure, river level, and catchment area information, calculate the dendritic connectivity index, river continuity index, and river fragmentation index; Combined with the development method and installed capacity parameters in the small hydropower station data, calculate the improved river fragmentation index for dam-type small hydropower stations; Combined with the small hydropower station data and hydrological data, calculate the improved river fragmentation index considering the impact of dewatering flow for diversion-type small hydropower stations; Based on the spatial hierarchical relationship of the water system network data, calculate the parameter values of the above indicators at different river scales, regional scales, and basin scales respectively to form the basic indicators of river network water system connectivity; The steps for generating dynamic connectivity indicators include: According to the gate dam scheduling rules in the small hydropower station operation mode data and hydrological data, construct a time matrix of the connectivity situation of river reaches, and divide each river reach into three situations: continuous connectivity, periodic connectivity, and temporary connectivity; Based on the small hydropower station operation mode data and hydrological data, construct a matrix of the change in the diversion flow of the power station and the river flow; According to the time matrix of the connectivity situation of river reaches and the matrix of the change in the diversion flow of the power station and the river flow, combined with the basic indicators of river network water system connectivity, calculate the dynamic change indicators of connectivity at different time scales; Construct a spatio-temporal coupling analysis framework, integrate the dynamic connectivity status of small water bodies and the small hydropower operation status data to generate dynamic connectivity indicators.

2. The method according to claim 1, characterized in that, The steps for conducting refined segmentation include: Construct a spatio-temporal multi-source database, including multi-temporal optical remote sensing images, radar remote sensing data, and high-resolution images obtained by drones; Based on the spatio-temporal multi-source database, calculate the water body change rate between different time phases and construct a water body presence frequency matrix; Adopt an adaptive threshold segmentation method to process the water body presence frequency matrix to generate the segmentation results of small water bodies; Based on the segmentation results of small water bodies and the digital elevation model, adopt a hydrodynamic constraint discrimination method to generate a dynamic connectivity status map of small water bodies; Identify the intake and tailrace positions of diversion-type power stations in the small hydropower project location data, and combined with the dynamic connectivity status map of small water bodies, generate the hidden connectivity status of dewatered river reaches; Integrate the dynamic connectivity status map of small water bodies and the hidden connectivity status of dewatered river reaches into water system network data.

3. The method according to claim 1, characterized in that, The steps for conducting a comprehensive evaluation of water system connectivity include: Based on the water ecological environment data, determine the set of impact indicators for connectivity evaluation, including the proportion of surface water, comprehensive water quality indicators, ecological flow, and biological indicators; Dimensionalize the data of each index in the connectivity evaluation index set, calculate the information entropy of each index, and determine the index weights; Adopt the grey relational degree method to calculate the influence degree of each index on the water system connectivity, and form an influence degree matrix; Multiply the dynamic connectivity index by the influence degree matrix to generate a water system connectivity matrix; Perform spatial weighting on the water system connectivity matrix to obtain the comprehensive index of water system connectivity affected by small hydropower stations, and accordingly evaluate the rectification priorities of each small hydropower station in the study area.

4. The method according to claim 2, characterized in that, The steps to generate the micro-water body segmentation result include: Calculate the regional water body frequency statistical distribution of the water body presence frequency matrix; Adopt a two-peak mixed Gaussian model to fit the regional water body frequency statistical distribution to obtain the water body presence probability threshold; Design a threshold adaptive function based on spatial context, and the threshold adaptive function dynamically adjusts the threshold according to the distance from the pixel to the nearest determined water body; Apply the threshold adaptive function to segment the water body presence frequency matrix, identify micro-water bodies or fragmented water bodies that are difficult to be detected by conventional methods, and generate the micro-water body segmentation result.

5. The method according to claim 1, characterized in that, The steps to construct the power station diversion flow and river flow change matrix include: Integrate the power station operation parameters in the small hydropower station operation mode data and the flow records in the hydrological data to establish a relationship model between the water diversion volume of the small hydropower station and the upstream incoming water; Design a time series feature vector based on the small hydropower station operation mode data and hydrological data, including historical operation parameters, water level changes, and flow changes; Construct a prediction model combining a bidirectional long short-term memory network and an attention mechanism, with the time series feature vector as the input, to predict the diversion flow of the small hydropower station under different hydrological conditions; Combine the predicted diversion flow with the corresponding period river flow data in the hydrological data to generate the power station diversion flow and river flow change matrix.

6. The method according to claim 1, characterized in that, The steps to generate the dynamic connectivity index include: Organize the micro-water body dynamic connectivity state map, the hidden connectivity state of the dewatered and dehydrated river sections, and the power station diversion flow and river flow change matrix into a data structure containing spatial and time dimensions to form a spatio-temporal coupling analysis framework; Based on the spatio-temporal coupling analysis framework, design the correlation evaluation index between the small hydropower operation mode and the water system connectivity state, and calculate the correlation strength between the operation parameters and the connectivity index; Combine the ecological index data, set the connectivity requirement threshold for different ecological protection objectives, and construct an ecological connectivity threshold matrix; Based on the ecological connectivity threshold matrix, design an optimization model considering the synergistic effect of the small hydropower group, and determine the weight coefficients and constraint intensities of different parameters in the optimization model according to the correlation strength. With the goal of maximizing the power generation benefit and the constraint of maintaining the water system connectivity, calculate the optimal dispatching strategy; According to the optimal dispatching strategy and the connectivity correlation index, generate a dynamic connectivity index reflecting spatio-temporal change characteristics.

7. The method according to claim 1, wherein The steps to calculate the connectivity dynamic change index at different time scales include: Correspondingly match the connectivity state in the river section connectivity situation time matrix with the flow data in the power station diversion flow and river flow change matrix according to the time nodes; Based on the basic index calculation formula of river network water system connectivity, replace the static parameters in it with the dynamic flow parameters at the corresponding time nodes to construct a dynamic connectivity evaluation model; Use the dynamic connectivity evaluation model to calculate the dynamic connectivity index values at different river scales, regional scales, and basin scales under the conditions of wet season, normal season, and dry season respectively; Design a connectivity time series fluctuation evaluation index to quantify the change amplitude and change frequency of the connectivity index at different time scales, and form a dynamic change index of connectivity; Based on the dynamic change index of connectivity, construct a feedback regulation mechanism to realize the dynamic coupling optimization of the connectivity state and the operation mode of small hydropower stations.

8. The method according to claim 4, characterized in that, The threshold adaptive function based on spatial context is: τ(x,y) = T•[1+α•exp(-d(x,y) / β)]; where τ(x,y) is the adaptive threshold at pixel (x,y), T is the water body existence probability threshold, d(x,y) is the distance from pixel (x,y) to the nearest determined water body, and α and β are adaptive parameters determined according to the water body distribution characteristics of the study area.

Images