A river and lake recovery water quantity elastic allocation method coupled with ecological water demand response

Abstract

The present application belongs to the technical field of water conservancy projects and water resources allocation, and discloses a river and lake recovery water elastic allocation method and system coupled with ecological water demand response, aiming to solve the problem of rigid allocation and resource mismatch caused by treating ecological water demand as a static exogenous constant in the prior art. The method comprises the following steps: constructing a distributed sensing network to collect multi-source ecological hydrological elements; inputting an analysis model to output a dynamic ecological water demand index through fusion and mapping; inputting the index and available water into a double-layer elastic allocation decision model, the upper layer generates an elastic space of water demand satisfaction strategy, and the lower layer generates discharge flow and gate opening degree instructions in the elastic space with minimum cost; executing the instructions and monitoring the changes of ecological hydrological elements, feeding back to the analysis model to trigger index iteration correction, forming a closed-loop dynamic coupling. The present application realizes two-way mutual feedback of allocation decision and ecological response, and improves the river and lake recovery effect and water resources utilization efficiency.

Description

Technical Field

[0001] This invention belongs to the field of water conservancy engineering and water resource allocation technology, specifically relating to a flexible allocation method for river and lake recovery water volume coupled with ecological water demand response. Background Technology

[0002] Water conservancy projects and water resource allocation are fundamental technical fields for maintaining the ecological health of rivers and lakes and ensuring the carrying capacity of regional water environments. One of their core tasks is to restore the ecological functions of degraded rivers and lakes through scientific water resource allocation methods to compensate for water volume and reconstruct flow patterns. In this process, the effectiveness of restoration projects is directly determined by how accurately the ecological water demand of rivers and lakes is determined and dynamically met under the constraints of water resource supply. Therefore, developing allocation technologies that can couple the real-time response characteristics of ecological water demand is of fundamental supporting significance for achieving the dual goals of efficient water resource utilization and ecological restoration of rivers and lakes.

[0003] In the practical application of water allocation for river and lake restoration, a commonly used approach for determining and meeting ecological water demand—a core aspect—is an open-loop matching technique based on static pre-setting. Its operational mechanism is as follows: First, based on statistical analysis of historical hydrological data and combined with pre-set specific ecological protection targets, a minimum ecological flow threshold or target water demand process line that remains constant throughout the entire scheduling cycle is determined through offline calculation. Then, this static water demand value is used as a rigid boundary condition and unidirectionally input into the water allocation model. This model, based on upstream inflow forecasting and water balance principles, mechanically controls the downstream water release by adjusting reservoir storage and gate openings to precisely match the pre-set water demand. Some schemes further introduce segmented linear scheduling rules with phased and graded classifications, but essentially, they do not change the fundamental logic that ecological water demand is an independent exogenous constant and unidirectionally coupled with the scheduling process.

[0004] However, the aforementioned technical approach treats ecological water demand as an exogenous constant independent of scheduling behavior and unaffected by water flow conditions, fundamentally severing the real-time dynamic feedback mechanism of river and lake ecosystems to hydrological processes, thus leading to deep-seated allocation failures. The true ecological water demand of rivers and lakes is not a predetermined value, but a dynamic field variable determined by various environmental factors such as water temperature, water depth, flow velocity, nutrient transport cycles, and aquatic phenological rhythms. When the allocated water is released, the hydraulic conditions, material transport state, and thermodynamic state of the river and lake water body immediately change. This change in physical state directly triggers a shift in the immediate water demand threshold of the river and lake ecosystem. Because the ecological water demand value in existing technologies is completely independent of these physical state changes caused by scheduling actions, the water allocation model always relies on outdated water demand values ​​before the scheduling action, which no longer reflect the current true state of the system. It completely fails to capture the crucial physical feedback loop of how the released water flow changes the river and lake state, thereby altering water demand. This disconnect from the physical feedback path directly leads to a systemic misalignment of supply and demand: during periods of high water flow or when river and lake hydraulic conditions improve due to water replenishment, the actual water demand gap of the ecosystem has significantly narrowed due to the improved physical conditions. However, the scheduling model based on static water demand values ​​still forces the release of water at the original high standard, resulting in the ineffective disposal of water resources. Conversely, during periods of continuous low water flow that cause the hydrodynamics of rivers and lakes to shrink, the ecosystem's tolerance threshold is drastically compressed, and its water demand gap abruptly amplifies due to the deterioration of physical conditions. However, the static water demand setting remains unchanged because it is completely detached from changes in physical conditions, resulting in a severe lag in the scheduling response and further exacerbating ecological damage. This rigidity in allocation and resource misallocation caused by the complete lack of dynamic coupling between ecological water demand response and water allocation processes is the core technical challenge currently hindering the achievement of fundamental breakthroughs in river and lake recovery. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of the prior art and provide a method for elastically allocating river and lake recovery water volume that couples ecological water demand response.

[0006] To achieve the above objectives, the present invention provides the following technical solution: This application provides a method for elastically allocating river and lake recovery water volume coupled with ecological water demand response, including: S1. Construct a distributed sensing network in the target river sections of rivers and lakes awaiting restoration to collect multi-source eco-hydrological elements that characterize the response of river and lake ecosystems to hydrological processes in real time.

[0007] S2. Input the multi-source eco-hydrological elements into the pre-constructed ecological water demand response analytical model, and output the dynamic ecological water demand index through multi-element fusion and nonlinear mapping; the dynamic ecological water demand index represents the real-time water demand fluctuation range required by rivers and lakes to maintain or approach the target ecological function under the current hydraulic, water quality and biological conditions.

[0008] S3. Using the dynamic ecological water demand index and the real-time available water quantity index characterizing the watershed's water resource supply capacity as inputs, a two-layer flexible allocation decision model is used to generate an instruction set for downstream flow and gate opening; wherein, the upper layer of the two-layer flexible allocation decision model aims to maximize the comprehensive ecological benefits within the scheduling cycle and generates an elastic space for water demand satisfaction strategies; the lower layer, within the elastic space, aims to minimize water allocation costs and generates the instruction set.

[0009] S4. Execute the instruction set to release the allocated water volume to the target river section, and continuously monitor the changes in eco-hydrological elements after execution through the distributed sensing network. Feed back the monitoring results to the ecological water demand response analysis model to trigger the iterative correction of the dynamic ecological water demand index, forming a closed-loop dynamic coupling between water allocation and ecological response.

[0010] Preferably, the multi-source eco-hydrological elements include water body physicochemical indicators, hydrological and hydrodynamic indicators, and aquatic biological behavioral indicators.

[0011] Preferably, the construction of the distributed sensing network described in S1 includes: setting up measuring points in deep water, shallow water and near shore areas at multiple cross sections along the target river section, and integrating water temperature sensor, dissolved oxygen sensor, conductivity sensor, chlorophyll α fluorescence probe, flow profiler and hydrophone to simultaneously collect water temperature, dissolved oxygen, conductivity, chlorophyll α concentration, three-dimensional flow velocity of the flow field and underwater sound scene signals.

[0012] Preferably, the aquatic biological behavioral indicators include fish behavior telemetry data and aquatic vegetation remote sensing image data; wherein, the fish behavior telemetry data is acquired by implanting acoustic transmitters into target fish and using a hydrophone array deployed along the route to quantify the fish's behavioral response to flow velocity, water depth, and temperature fields; the aquatic vegetation remote sensing image data is acquired by deployed hyperspectral imaging terminals to quantify the response of aquatic vegetation cover and growth to hydrological changes.

[0013] Preferably, the ecological water demand response analytical model includes multiple parallel feature extraction branches, which are used to extract water quality evolution features, hydrodynamic change features, and biological response features from the time series of water quality elements, the time series of hydrodynamic elements, and the time series of biological response elements, respectively; the outputs of the multiple feature extraction branches are interactively fused by the feature fusion layer and then input into the regression layer to output the dynamic ecological water demand index.

[0014] Preferably, the feature fusion layer employs a multi-head attention mechanism, performing cross-attention operations on the outputs of each feature extraction branch and then fusing them.

[0015] Preferably, the dynamic ecological water demand index is composed of a baseflow maintenance index, a pulse replenishment index, and a habitat abundance index; the baseflow maintenance index represents the minimum flow requirement to maintain the current survival and basic metabolism of aquatic organisms; the pulse replenishment index represents the requirement for short-term high flow pulses to simulate natural hydrological rhythms, trigger fish spawning, or inhibit excessive algal growth; and the habitat abundance index represents the requirement for specific flow processes to submerge floodplains or restore aquatic vegetation through water level changes.

[0016] Preferably, in S3: The upper layer of the two-layer flexible allocation decision model is a global decision-maker based on Monte Carlo tree search. This global decision-maker takes the dynamic ecological water demand index and its components, reservoir storage capacity, and medium-term runoff forecast statistical characteristics as its state, and takes the floating value of the ecological water demand satisfaction rate as its action. During the search process, a dynamic exploration factor based on the uncertainty of medium-term runoff forecast is introduced, and a surrogate model trained based on historical data is used to simulate and extrapolate the state changes of multiple future steps under the water demand satisfaction strategy. The return is calculated based on the instantaneous benefit function based on the baseflow maintenance degree, pulse timing accuracy, and habitat inundation area, and the elastic space of the water demand satisfaction strategy is generated. The lower layer of the two-layer elastic dispatch decision model is a real-time scheduler based on mixed integer linear programming. It uses reservoir outflow, gate opening and unit load as decision variables, and the weighted sum of pump station energy consumption, water abandonment and head loss as the cost function. It solves to generate the instruction set under hydraulic balance constraints, equipment operation constraints and the water demand satisfaction rate interval constraints corresponding to the elastic space.

[0017] Preferably, triggering the iterative correction of the dynamic ecological water demand index in S4 includes: collecting snapshots of ecological and hydrological elements at multiple preset time points after the release of allocated water and generating a new dynamic ecological water demand index, and calculating the ecological water demand gap; using a fuzzy logic regulator, taking the rate of change and acceleration of change of the ecological water demand gap within a sliding window as input, and outputting the adjustment range of the satisfaction rate through fuzzy inference; when multiple consecutive snapshots show that the ecological water demand gap is continuously narrowing, sending a control message to the two-layer elastic allocation decision model to lower the water demand satisfaction rate, and sending a control message to raise the water demand satisfaction rate when the ecological water demand gap expands.

[0018] A flexible water allocation system for river and lake recovery coupled with ecological water demand response includes: A distributed eco-hydrological sensing subsystem is used to collect multi-source eco-hydrological elements that characterize the response of river and lake ecosystems to hydrological processes in real time. An ecological water demand response analysis engine is used to input the multi-source ecological hydrological elements into a pre-constructed ecological water demand response analysis model and output a dynamic ecological water demand index. A two-layer flexible allocation decision platform is used to generate an elastic space for water demand satisfaction strategy in the upper layer, and generate a set of instructions for discharge flow and gate opening in the lower layer within the elastic space, taking the dynamic ecological water demand index and real-time available water quantity index as input. The water volume execution and ecological feedback closed-loop control module is used to execute the instruction set, monitor changes in eco-hydrological elements, and feed them back to the ecological water demand response analysis engine to trigger the iterative correction of the dynamic ecological water demand index.

[0019] Compared with the prior art, this application has the following beneficial effects: This invention, through the construction of a technical architecture of distributed sensing, dynamic analysis, flexible decision-making, and closed-loop feedback, achieves for the first time a two-way dynamic coupling between allocation decisions and real-time ecosystem responses in river and lake recovery water allocation. Specifically, the distributed sensing network constructed in S1 collects multi-source eco-hydrological elements in real time, providing the system with a comprehensive and real-time data foundation for characterizing the ecological state of rivers and lakes. The ecological water demand response analytical model pre-constructed in S2 performs multi-element fusion and nonlinear mapping on multi-source heterogeneous ecological elements, outputting a dynamic ecological water demand index that accurately reflects the real-time fluctuation range of water demand under current hydraulic, water quality, and biological conditions. This fundamentally changes the traditional approach of treating ecological water demand as a static exogenous constant, enabling the water demand standard to respond to the physical state triggered by allocation actions. The system adapts and migrates in real time. Through a two-layer flexible allocation decision model in S3, the upper layer uses the dynamic ecological water demand index as the core boundary condition. Under the uncertainty of runoff forecasting, it intelligently generates a flexible space for water demand satisfaction strategies with floating capacity. The lower layer, within this space, balances operating costs and hydraulic constraints, achieving an optimal balance between ecological goals and resource economy. This allows the system to avoid ineffective water release during the wet season and strengthen ecological protection during the dry season. Through the continuous monitoring and feedback iterative correction mechanism in S4, the changes in eco-hydrological elements after execution are re-injected into the analytical model, forming a complete physical feedback loop where downstream water flow changes the state of rivers and lakes, state changes reshape water demand, and demand adjustments guide allocation decisions. This completely solves the technical problem of systematic misalignment of supply and demand in traditional open-loop methods. The organic combination of these technologies ensures that each water release precisely targets the period of most urgent ecosystem recovery needs, significantly improving the spatiotemporal accuracy, ecological effectiveness, and water resource utilization efficiency of river and lake recovery. Attached Figure Description

[0020] Figure 1 This is a schematic diagram of the overall technical architecture of the river and lake recovery water volume elastic allocation method coupled with ecological water demand response proposed in this invention.

[0021] Figure 2 This is a schematic diagram of the core principle framework of the closed-loop dynamic coupling of ecological water demand response and water allocation in this invention.

[0022] Figure 3 This is a flowchart illustrating the logical flow of the ecological water demand response analytical model in this invention.

[0023] Figure 4 This is a logical flow diagram of the two-layer flexible allocation decision model in this invention.

[0024] Figure 5 This is a schematic diagram of the module interaction relationship and data flow of the river and lake recovery water volume elastic allocation system coupled with ecological water demand response in this invention. Detailed Implementation

[0025] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0026] Furthermore, in this invention, an element referred to as fixed to or disposed on another element may be directly disposed on the other element, or there may be an intermediate element. When an element is considered to be connected to another element, it may be directly connected to the other element, or there may be an intermediate element present simultaneously. The terms vertical, horizontal, left, right, and similar expressions used herein are for illustrative purposes only and do not represent the only possible implementation.

[0027] See Figures 1-5 This application provides a method for elastically allocating river and lake recovery water volume coupled with ecological water demand response, including steps S1-S4: S1. Construct a distributed sensing network in the target river sections of rivers and lakes awaiting restoration to collect multi-source eco-hydrological elements that characterize the response of river and lake ecosystems to hydrological processes in real time.

[0028] Specifically, in the target river sections of the rivers and lakes to be revitalized, a distributed sensing network needs to be deployed. This network is essentially a three-dimensional sensing system composed of multiple monitoring sections along the river and various types of measuring points. This sensing network collects in real-time multi-source eco-hydrological elements that characterize the response of the river and lake ecosystem to hydrological processes. Characterizing the response of the river and lake ecosystem to hydrological processes means that these elements are not isolated environmental parameters, but rather indicative variables that reflect how changes in water flow conditions cause changes in the ecosystem's state. These multi-source elements cover multiple dimensions, from the physicochemical properties of water bodies and hydrodynamic characteristics to the behavioral responses of aquatic organisms, collectively forming a comprehensive depiction of the river and lake's ecological state. Through this sensing network, the system obtains the real-time data foundation for implementing subsequent intelligent allocation decisions.

[0029] S2. Input the multi-source eco-hydrological elements into the pre-constructed ecological water demand response analytical model, and output the dynamic ecological water demand index through multi-element fusion and nonlinear mapping; the dynamic ecological water demand index represents the real-time water demand fluctuation range required by rivers and lakes to maintain or approach the target ecological function under the current hydraulic, water quality and biological conditions.

[0030] Specifically: The multi-source eco-hydrological elements collected in S1 are input into a pre-constructed ecological water demand response analytical model. The core functions of this model are multi-element fusion and nonlinear mapping. Multi-element fusion means that the model can collaboratively process various eco-hydrological elements from different monitoring devices, with different physical meanings and spatiotemporal scales, rather than simply listing or using them independently. Nonlinear mapping means that the model uses complex nonlinear function transformations to map high-dimensional, heterogeneous sensing data onto a scalar or low-dimensional index that can comprehensively characterize the ecological water demand state of rivers and lakes. The output of the model is the dynamic ecological water demand index. This index is a dynamically changing quantity, not a fixed constant. It represents the real-time water volume fluctuation range required by the river and lake ecosystem to maintain its current state or approach the preset target ecological function level under the combined effects of specific hydraulic, water quality, and biological conditions at the current moment. This means that the index not only reflects how much water is needed, but more importantly, it reflects that this demand fluctuates in real time with changes in the state of the rivers and lakes themselves.

[0031] S3. Using the dynamic ecological water demand index and the real-time available water quantity index characterizing the watershed's water resource supply capacity as inputs, a two-layer flexible allocation decision model is used to generate an instruction set for downstream flow and gate opening; wherein, the upper layer of the two-layer flexible allocation decision model aims to maximize the comprehensive ecological benefits within the scheduling cycle and generates an elastic space for water demand satisfaction strategies; the lower layer, within the elastic space, aims to minimize water allocation costs and generates the instruction set.

[0032] Specifically, this step uses the dynamic ecological water demand index output by S2 as the core boundary condition, and simultaneously introduces a real-time available water quantity index characterizing the watershed's water resource supply capacity. Both are input into a flexible allocation decision model with a two-layer structure. The upper layer of this model takes a global perspective, aiming to maximize the comprehensive ecological benefits throughout the entire scheduling cycle. Based on the uncertainty of the dynamic ecological water demand index and inflow forecast, it generates a flexible space for water demand satisfaction strategies. This flexible space is not a rigid value, but a range that allows the ecological water demand satisfaction rate to fluctuate within a reasonable range, thus giving the system the ability to flexibly adjust to cope with uncertainties. The lower layer of the model, within the fluctuation range set by the aforementioned flexible space, aims to minimize the economic cost of water allocation. It comprehensively considers economic factors such as pump station energy consumption and water wastage losses, and generates a specific and executable set of discharge flow and gate opening instructions. This two-layer architecture realizes the hierarchical optimization and synergy of ecological and economic benefits.

[0033] S4. Execute the instruction set to release the allocated water volume to the target river section, and continuously monitor the changes in eco-hydrological elements after execution through the distributed sensing network. Feed back the monitoring results to the ecological water demand response analysis model to trigger the iterative correction of the dynamic ecological water demand index, forming a closed-loop dynamic coupling between water allocation and ecological response.

[0034] First, the specific instruction set generated by S3 is executed, and the allocated water volume is actually released to the target river section through control gates and other equipment. After execution, it does not end there, but immediately initiates closed-loop monitoring: through the distributed sensing network deployed in S1, the actual response of the river and lake ecosystem after the allocated water volume is released is continuously monitored, that is, how the eco-hydrological elements have changed. These monitored changes are then fed back into the ecological water demand response analytical model in S2 as feedback information, triggering the model to perform a new round of iterative correction of the dynamic ecological water demand index. The corrected index will then affect the allocation decision in the next round of S3. This process repeats, forming a complete closed-loop dynamic coupling process of allocation execution, state response, water demand reassessment, and decision adjustment, enabling the system to have adaptive and self-regulating capabilities.

[0035] As a preferred embodiment, the multi-source eco-hydrological elements include water body physicochemical indicators, hydrological and hydrodynamic indicators, and aquatic biological behavioral indicators.

[0036] In this embodiment, these three types of indicators constitute a complete description of the ecological state of rivers and lakes from three dimensions: water chemical environment, water flow physical processes, and biological response.

[0037] Physicochemical indicators of water bodies are fundamental parameters reflecting the chemical properties of river water and the quality of the aquatic environment. They mainly include water temperature, dissolved oxygen, conductivity, and chlorophyll α concentration. Water temperature directly affects the metabolic rate of aquatic organisms and the saturation of dissolved oxygen. Dissolved oxygen is a key factor for maintaining the respiration and survival of aquatic animals. Conductivity reflects the total amount of ions in the water and can indicate pollution or saline intrusion. Chlorophyll α concentration is an important indicator for measuring phytoplankton biomass and primary productivity, and is also a core parameter for early warning of algal bloom risks.

[0038] Hydrological and hydrodynamic indicators are parameters that describe the characteristics of river flow. The most important one is the three-dimensional velocity of the flow field, which is the velocity component of the water flow in three dimensions: along the river channel, laterally, and vertically. The three-dimensional velocity determines the mixing and diffusion capacity of the water body, the rate of material transport, and the conditions for the initiation and settling of bottom sediments, and has a decisive impact on the hydraulic suitability of aquatic organism habitats.

[0039] Aquatic organism behavioral indicators are used to infer the response of aquatic organisms to changes in the aquatic environment by monitoring their behavioral responses. They serve as a direct bridge connecting physical hydrological processes and ecological effects. Typical indicators include remote sensing data of fish behavior and remote sensing image data of aquatic vegetation, and their acquisition methods will be described in detail below.

[0040] As a preferred implementation, the construction of the distributed sensing network described in S1 includes: setting up measuring points in deep water, shallow water and near-shore areas at multiple cross-sections along the target river section, and integrating water temperature sensors, dissolved oxygen sensors, conductivity sensors, chlorophyll α fluorescence probes, flow profilers and hydrophones to simultaneously collect water temperature, dissolved oxygen, conductivity, chlorophyll α concentration, three-dimensional flow velocity of the flow field and underwater sound signals.

[0041] In this embodiment, different types of measuring points are deployed along multiple cross sections of the target river section to form a three-dimensional sensing grid in space. The cross sections refer to representative monitoring locations selected at certain intervals along the longitudinal direction of the river. They should typically include at least the upstream inflow control section, the midstream key habitat section, and the downstream outflow control section to cover the ecological gradient changes of the river's upstream and downstream. Each cross section is further divided into three types of measuring points: deep water area, shallow water area, and near-shore area. These three types of areas have significant differences in water depth, flow velocity, and light conditions, corresponding to different micro-habitat types, and their ecological processes and water requirements are also different.

[0042] In terms of sensor configuration, multiple types of monitoring devices are integrated to simultaneously acquire the aforementioned data; the water temperature sensor is used to measure the water temperature profile at different depths; the dissolved oxygen sensor works based on the optical fluorescence quenching principle and can accurately determine the dissolved oxygen concentration in the water; the conductivity sensor uses a four-electrode measurement principle to obtain the conductivity value of the water; the chlorophyll α fluorescence probe measures the fluorescence intensity of chlorophyll α through specific excitation and detection wavelengths to infer its concentration; the flow profiler uses the acoustic Doppler principle to coherently process the scattered echoes of suspended particles in the water and measures the three-dimensional flow field at a 15-minute cycle to obtain the eastward, northward, and vertical flow velocity components; the hydrophone, or passive underwater acoustic monitoring sensor, receives underwater sound wave signals and records the soundscape information generated by biological sounds and physical processes in the aquatic environment; the above sensors form a complete system for simultaneous monitoring of water physicochemical and hydrodynamic processes.

[0043] In a preferred embodiment, the aquatic biological behavioral indicators include fish behavior telemetry data and aquatic vegetation remote sensing image data; wherein, the fish behavior telemetry data is acquired by implanting acoustic transmitters into target fish and using a hydrophone array deployed along the route to quantify the fish's behavioral response to flow velocity, water depth, and temperature fields; the aquatic vegetation remote sensing image data is acquired by deploying hyperspectral imaging terminals to quantify the response of aquatic vegetation cover and growth to hydrological changes.

[0044] In this embodiment, fish behavior telemetry data is acquired through active acoustic telemetry technology: First, a certain number of target fish individuals are selected, and a miniature acoustic transmitter is implanted into their abdominal cavity. The transmitter emits coded acoustic pulse signals at a preset frequency. Multiple acoustic receivers are deployed at appropriate intervals along the target river section to form a passive hydrophone array. When a fish carrying the transmitter swims through the detection range of the receiver, the receiver records the fish's identification code and the time of passage. By statistically analyzing the temporal and spatial distribution of a large number of passage records, the directional and avoidance behavioral responses of fish to environmental factors such as flow velocity, water depth, and temperature field can be quantified. For example, under what flow velocity conditions do fish tend to gather, and how do they migrate when the water temperature changes?

[0045] Remote sensing image data of aquatic vegetation was acquired using hyperspectral imaging terminals deployed along the riverbanks. These terminals image at high spectral resolution across a broad band from visible to near-infrared, scanning the riverbanks and shallow areas daily to acquire hyperspectral image data cubes containing dozens to hundreds of consecutive narrow bands. The image data is processed using a pre-trained deep convolutional semantic segmentation network, which automatically identifies and extracts the distribution areas and coverage of different types of aquatic vegetation, including submerged, emergent, and wetland plants, and calculates the Normalized Difference Vegetation Index (NDVI) using the formula: , where R NIRR represents the near-infrared reflectance. Red The reflectance is in the red band. By measuring the temporal variation of NDVI, the response of aquatic vegetation cover and growth to changes in hydrological conditions (such as water level fluctuations and inundation duration) can be quantified.

[0046] In a preferred embodiment, the ecological water demand response analytical model includes multiple parallel feature extraction branches, which are used to extract water quality evolution features, hydrodynamic change features, and biological response features from the time series of water quality elements, the time series of hydrodynamic elements, and the time series of biological response elements, respectively. The outputs of the multiple feature extraction branches are interactively fused by the feature fusion layer and then input into the regression layer to output the dynamic ecological water demand index.

[0047] In this embodiment, the model adopts a design of multiple parallel feature extraction branches to adapt to the multi-source heterogeneous characteristics of the input data. Since water quality elements, hydrodynamic elements and biological response elements have different time variation characteristics, different data structures and different physical meanings, it is difficult to effectively extract the unique features of each type of element using a single-structure network. Therefore, the model sets up at least three parallel feature extraction branches, each adopting a network structure suitable for the characteristics of its input data.

[0048] The first branch specifically processes time series data of water quality elements, namely, time-varying data of water temperature, dissolved oxygen, electrical conductivity, and chlorophyll α concentration, extracting abstract feature vectors that can characterize the process and trend of water quality evolution. The second branch specifically processes time series data of hydrodynamic elements, namely, time-varying data of three-dimensional flow velocity in the flow field, extracting abstract feature vectors that can characterize changes in hydrodynamic state and flow regime transitions. The third branch specifically processes time series data of biological response elements, namely, time-varying data of underwater acoustic signals, fish telemetry data, and aquatic vegetation remote sensing data, extracting abstract feature vectors that can characterize the response patterns and intensity of aquatic organisms to changes in the aquatic environment.

[0049] The feature vectors output from the three branches are then fed into a feature fusion layer for interactive fusion. The purpose of feature fusion is to capture the coupling relationship between different ecological processes, such as how water quality changes and hydrodynamic changes synergistically affect biological responses. The fused comprehensive features are input into a regression layer, which outputs the final dynamic ecological water demand index. The role of the regression layer is to map the high-dimensional fused features to a one-dimensional scalar output space, completing the end-to-end mapping from multi-source sensing data to a single water demand index.

[0050] In a preferred embodiment, the feature fusion layer employs a multi-head attention mechanism to perform cross-attention operations on the outputs of each feature extraction branch and then fuse them.

[0051] In this embodiment, the core idea of ​​the multi-head attention mechanism is to allow the model to learn the correlation between different features in different representation subspaces.

[0052] Specifically, the fusion layer first concatenates the feature vectors output from the first, second, and third branches into a comprehensive feature matrix F; then, it performs three different linear transformation matrices... , , The comprehensive feature matrix F is projected into the query matrix Q, the key matrix K, and the value matrix V, respectively; then the cross-attention output A is calculated using the following formula: ;in, This means that the matrix obtained by multiplying the query matrix by the transpose of the key matrix represents the relevance score between each feature position. Scaling factor It is the dimension of the key vector, divided by This is to prevent the gradient of the softmax function from vanishing due to excessively large dot product values; the softmax function is normalized on the last dimension, converting the relevance score into attention weights; finally, the attention weights are multiplied by the value matrix V to obtain the weighted attention output; this output is processed by residual connections and layer normalization to obtain the final fused feature, as shown in the formula. In this context, LayerNorm represents the layer normalization operation, and residual connections F+A help to preserve the original feature information and alleviate the gradient vanishing problem in deep networks. The multi-head mechanism refers to the simultaneous execution of the above attention operations in multiple parallel subspaces. Each head has independent linear transformation matrix parameters, and the outputs of each head are finally concatenated together, enabling the model to capture the complex interaction relationships between features from different perspectives.

[0053] In a preferred embodiment, the dynamic ecological water demand index is composed of a baseflow maintenance index, a pulse replenishment index, and a habitat abundance index. The baseflow maintenance index represents the minimum flow requirement to maintain the survival and basic metabolism of current aquatic organisms. The pulse replenishment index represents the requirement for short-term high-flow pulses to simulate natural hydrological rhythms, trigger fish spawning, or inhibit excessive algal growth. The habitat abundance index represents the requirement for specific flow processes to submerge floodplains or restore aquatic vegetation through water level changes.

[0054] In this embodiment, it is clearly defined that it consists of three components: baseflow maintenance index, pulse replenishment index, and habitat abundance index. These three components correspond to three different dimensions of water demand in river and lake ecosystems, and together they constitute a complete ecological water demand profile.

[0055] The baseflow maintenance index characterizes the minimum flow required to maintain the survival and basic metabolic activities of aquatic organisms. The higher the value of the index, the closer the current flow level of rivers and lakes is to the lower limit of survival that aquatic organisms can tolerate, the greater the risk of baseflow deficit in the system, and the more urgent it is to replenish water to maintain the most basic ecological functions. The level of this component is highly correlated with water quality factors such as water temperature and dissolved oxygen, because biological metabolism increases oxygen consumption under high temperature conditions, and the demand for baseflow flow increases accordingly.

[0056] The pulse replenishment index characterizes the degree of demand for short-duration, high-flow pulse events required to simulate natural hydrological rhythms. In natural rivers, seasonal flood pulses play an irreplaceable role in triggering ecological processes such as fish spawning migration, scouring riverbed sediments, and inhibiting excessive algal growth. When these natural pulse events are weakened or eliminated by human intervention such as reservoir regulation, it is necessary to artificially create similar pulse replenishment events. The higher the index, the closer the current system is to the timing point when these key ecological processes need to be triggered. This component is highly correlated with the frequency of fish acoustic telemetry behavior and flow field change characteristics.

[0057] The habitat abundance index characterizes the degree of demand for specific flow processes required to submerge floodplains or restore aquatic vegetation by changing water levels; periodic floodplain submersion is a key mechanism for maintaining riparian wetland ecosystems and biodiversity in floodplains; when floodplains are exposed for a long period due to insufficient flow, specific water replenishment processes are needed to raise the water level and achieve submersion; the higher the index, the more urgent the demand for flooding by floodplains or aquatic vegetation; this component is highly correlated with the characteristics of vegetation cover changes caused by water depth changes.

[0058] The three components can be decomposed from the comprehensive dynamic ecological water demand index through a linear decomposition model, minimizing the reconstruction error under the constraint that the sum of the weights of each component is 1.

[0059] In a preferred embodiment, S3: The upper layer of the two-layer flexible allocation decision model is a global decision-maker based on Monte Carlo tree search. This global decision-maker takes the dynamic ecological water demand index and its components, reservoir storage capacity, and medium-term runoff forecast statistical characteristics as its states, and takes the floating value of the ecological water demand satisfaction rate as its actions. During the search process, a dynamic exploration factor based on the uncertainty of medium-term runoff forecast is introduced, and a surrogate model trained based on historical data is used to simulate and extrapolate the state changes of multiple future steps under the water demand satisfaction strategy. The return is calculated based on the instantaneous benefit function based on the baseflow maintenance degree, pulse timing accuracy, and habitat inundation area, generating the elastic space of the water demand satisfaction strategy.

[0060] The lower layer of the two-layer elastic dispatch decision model is a real-time scheduler based on mixed integer linear programming. It uses reservoir outflow, gate opening and unit load as decision variables, and the weighted sum of pump station energy consumption, water abandonment and head loss as the cost function. It solves to generate the instruction set under hydraulic balance constraints, equipment operation constraints and the water demand satisfaction rate interval constraints corresponding to the elastic space.

[0061] In this embodiment, the upper-level global decision-maker uses Monte Carlo tree search as its core algorithm framework. Monte Carlo tree search is a heuristic search algorithm based on stochastic simulation, which is particularly suitable for sequential decision problems. The state input of the decision-maker is a multi-dimensional feature vector, which specifically includes the following types of information: the dynamic ecological water demand index value calculated by step S2 in the current period, the base flow maintenance index, pulse water replenishment index and habitat abundance index obtained after the decomposition of the index, the current water storage data of the reservoir, and the statistical characteristics of the medium-term runoff forecast probability distribution for a future preset period (e.g., 72 hours) (usually including the forecast mean and standard deviation, where the standard deviation reflects the uncertainty of the forecast). The action space is defined as a series of discrete actions consisting of the ecological water demand satisfaction rate fluctuating up and down from the current benchmark value by a preset step size (e.g., 5%), such as keeping it unchanged, increasing by 5%, increasing by 10%, decreasing by 5%, etc.

[0062] In each simulation iteration of Monte Carlo tree search, the algorithm goes through four stages: selection, expansion, simulation, and backtracking. The selection stage starts from the root node and selects the optimal child node layer by layer until the leaf node or a node that has not been fully expanded is reached. In this stage, the present invention adopts an improved confidence interval upper bound tree strategy and creatively introduces a dynamic exploration factor based on the uncertainty of medium-term runoff forecast into the standard node value evaluation formula. This factor makes the algorithm tend to increase the weight of the exploration space when the uncertainty of hydrological forecast is large, thereby prioritizing the evaluation of relatively conservative water demand satisfaction strategies to avoid decision risks caused by forecast deviations. In the simulation and extrapolation stage, the internally integrated surrogate model is called. This surrogate model is a gated recurrent unit network trained based on historical scheduling data. It takes the current state and the selected action as input and extrapolates the state change trajectory within multiple future steps in an autoregressive manner to evaluate the expected evolution of reservoir water level and dynamic ecological water demand index under the water demand satisfaction strategy.

[0063] For each step in the simulated trajectory, an immediate reward evaluation is performed based on an instantaneous benefit function; this benefit function consists of three weighted terms: the first term corresponds to the base current maintenance benefit, using... In the form of, Let t be the actual discharge flow rate at step t. To maintain the target flow rate for the base flow, The function maps the ratio to the interval [0,1], and the benefit approaches 1 when the actual flow rate reaches or exceeds the target; the second term corresponds to the benefit of the accuracy of the pulse water replenishment timing, using... In the form of, The actual pulse generation time step, The desired pulse generation time step, The time tolerance parameter shows that the benefit value decreases exponentially with the absolute value of the deviation between the actual and expected time, with a benefit of 1 when perfectly matched; the third term corresponds to the habitat inundation area benefit, using... In the form of, Let t be the area of ​​the floodplain that is actually submerged in step t. For the target inundation area, the benefits are calculated using the upper limit of the actual proportion; the three benefits are weighted according to coefficients. The instantaneous benefit value of this step is obtained by weighted summation, and the total return is the weighted average of the instantaneous benefits of all simulation steps. In the backtracking phase, the total return value is propagated back along the search path to update the number of visits and cumulative action value of each node. After a large number of simulations, the action with the highest number of visits is selected as the baseline of the optimal water demand satisfaction rate, and the elasticity space is generated accordingly.

[0064] The lower-level real-time scheduler is modeled and solved using a mixed-integer linear programming method; its decision variables include: reservoir outflow Q. out (Continuous variable), arc gate opening θ (integer or discrete variable), and unit flow rate Q turb (Semi-continuous variables); the objective is to minimize the cost function. Where α is the economic weighting coefficient for pump station energy consumption. Energy consumption of the pumping station For local time-of-use electricity pricing, As a water abandonment penalty factor, For the amount of water discarded, This is the economic weighting coefficient for head loss. This refers to the head loss during power generation. The equivalent electrical energy value conversion factor corresponding to head loss; constraints include: reservoir water balance constraints. ,in and These represent the reservoir water storage at the beginning and end of the i-th time period, respectively. For inbound flow, This refers to the inflow or precipitation within the region. Evaporation and leakage losses; piecewise linearization constraints on the nonlinear relationship between gate opening and flow rate, approximating linear interpolation between specific opening nodes using a special ordered set method; minimum technical output and vibration zone avoidance constraints; and water demand satisfaction rate interval constraints corresponding to the elastic space. ,in To meet ecological water demand, This refers to the reference ecological flow demand determined by the dynamic ecological water demand index. and These represent the lower and upper limits of the satisfaction rate allowed by the elastic space, respectively.

[0065] As a preferred implementation, triggering the iterative correction of the dynamic ecological water demand index in S4 includes: collecting snapshots of ecological and hydrological elements at multiple preset time points after the release of allocated water and generating a new dynamic ecological water demand index, and calculating the ecological water demand gap; using a fuzzy logic regulator, taking the rate of change and acceleration of change of the ecological water demand gap within a sliding window as input, and outputting the adjustment range of the satisfaction rate through fuzzy inference; when multiple consecutive snapshots show that the ecological water demand gap is continuously narrowing, sending a control message to the two-layer elastic allocation decision model to lower the water demand satisfaction rate, and sending a control message to raise the water demand satisfaction rate when the ecological water demand gap expands.

[0066] In this embodiment, firstly, at multiple preset time points after the water allocation is performed, a set of complete multi-source eco-hydrological element snapshots are collected through a distributed sensing network. The multiple time points are set to track the temporal evolution of the river and lake ecosystem's response to water replenishment. For example, data are collected at 2 hours, 6 hours, 24 hours, and 72 hours after water replenishment to cover different stages from immediate response to delayed response. Each set of snapshots contains complete water quality, hydrodynamic, and biological behavior data.

[0067] Each set of snapshots is sequentially input into the ecological water demand response analytical model. After multi-factor fusion and nonlinear mapping as described above, a new dynamic ecological water demand index E is generated. new ; E at each time point new The dynamic ecological water demand index benchmark value E recorded and saved before the allocation action is carried out. baseline By comparing the results, the ecological water demand gap was calculated using the formula Gap = 1 - E. new The meaning of this gap is: when E new When E is relatively high and close to 1, the Gap approaches zero, indicating that the ecological state of rivers and lakes has approached the target functional level after water replenishment, and the water demand gap has been basically filled; conversely, if E is low... new The readings remain low, with a significant gap, indicating that the water demand shortfall remains substantial.

[0068] Subsequently, an adaptive regulator based on fuzzy logic is used to make inference decisions regarding the adjustment range of the satisfaction rate; the regulator has two input variables: one is the rate of change of the ecological water demand gap within the sliding window. The first is the difference between gap values ​​at adjacent time points, reflecting whether the gap is widening or narrowing; the second is the acceleration of gap change. The rate of change itself reflects whether the trend is accelerating or decelerating. The introduction of these two variables allows the regulator to identify the dynamic characteristics of the gap change, rather than relying solely on a single numerical value. These two input variables are fuzzified and mapped to different fuzzy subsets and corresponding membership functions. The fuzzified inputs are then used for inference based on a pre-set fuzzy control rule base, compiled by water ecology experts based on their experience in water replenishment response. For example, if the gap change rate is negative (i.e., the gap is rapidly narrowing) and the acceleration is stable, the water demand satisfaction rate is significantly reduced. The inference uses the Mamdani fuzzy inference model, obtaining the output fuzzy set through a min-max synthesis rule, and then defuzzifying it using the centroid method to calculate the precise adjustment range of the satisfaction rate. Its formula is ,in To output the membership function of the fuzzy set, the integral is performed over the universe of discourse of the output variable.

[0069] If the inference results of multiple consecutive snapshots all show that the ecological water demand gap continues to narrow, indicating that the current water replenishment efforts have led to a continuous improvement in the ecological state of rivers and lakes, and there is room for appropriate reduction, then a decision containing a downward adjustment action and... should be sent to the two-layer flexible allocation decision model. Adjust the percentage control message; conversely, if multiple consecutive snapshots show that the ecological water demand gap is widening again, indicating insufficient water replenishment or an intensifying trend of ecological degradation, then send an upward adjustment message. The upper-level decision-maker receives the control message and adjusts the water demand satisfaction rate benchmark value for the next decision cycle, completing a complete closed-loop feedback iteration.

[0070] On the other hand, this application provides a flexible water allocation system for river and lake recovery coupled with ecological water demand response, including: A distributed eco-hydrological sensing subsystem is used to collect multi-source eco-hydrological elements that characterize the response of river and lake ecosystems to hydrological processes in real time. An ecological water demand response analysis engine is used to input the multi-source ecological hydrological elements into a pre-constructed ecological water demand response analysis model and output a dynamic ecological water demand index. A two-layer flexible allocation decision platform is used to generate an elastic space for water demand satisfaction strategy in the upper layer, and generate a set of instructions for discharge flow and gate opening in the lower layer within the elastic space, taking the dynamic ecological water demand index and real-time available water quantity index as input. The water volume execution and ecological feedback closed-loop control module is used to execute the instruction set, monitor changes in eco-hydrological elements and feed them back to the ecological water demand response analysis engine to trigger the iterative correction of the dynamic ecological water demand index; In this embodiment, the four component subsystems correspond to the four steps in the execution method; The distributed eco-hydrological sensing subsystem is the front-end data acquisition part of the system. It is deployed on-site in the target river section of the river and lake to be restored. It consists of multiple monitoring sections, multiple types of sensors and their supporting data acquisition, preprocessing and communication equipment. The subsystem collects multi-source eco-hydrological element data that can characterize the response state of the river and lake ecosystem to hydrological processes in real time, and sends the data to the core computing layer through the network transmission layer. The ecological water demand response analysis engine is one of the core algorithm components of the system. It runs on the computing server and internally stores the ecological water demand response analysis model that has been trained and optimized offline. The engine receives multi-source data from the sensing subsystem, performs forward inference calculations of the model, outputs a dynamic ecological water demand index, and publishes the index to the message middleware for subsequent modules to call. The two-layer flexible allocation decision-making platform is the core of the system's decision-making, comprising two functional modules: an upper-layer global decision-maker and a lower-layer real-time scheduler. The upper-layer global decision-maker runs the Monte Carlo tree search algorithm, taking dynamic ecological water demand index and hydrological data as input, and outputs the elastic space of the water demand satisfaction strategy. The lower-layer real-time scheduler runs a mixed-integer linear programming model, solving for the optimal scheduling scheme under the elastic space and various constraints, and generating a specific instruction set for the discharge flow and gate opening. The water quantity execution and ecological feedback closed-loop control module is a key link connecting the decision-making level and the physical execution equipment, and also undertakes the task of driving the closed-loop feedback. On the one hand, this module converts the instructions generated by the decision-making platform into control signals that can be recognized by the gate programmable logic controller, and sends them out for execution through a secure encrypted tunnel. On the other hand, after the instructions are executed, it collects response snapshot data according to a preset time triggering mechanism and feeds it back to the ecological water demand response analysis engine, driving the iterative update of the dynamic ecological water demand index and forming a complete closed-loop coupling.

[0071] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

[0072] Furthermore, it should be understood that although this specification describes embodiments, not every embodiment contains only one independent technical solution. This narrative style is merely for clarity. Those skilled in the art should consider the specification as a whole, and the technical solutions in each embodiment can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.

Claims

1. A method for elastically allocating river and lake recovery water volume coupled with ecological water demand response, characterized in that, include: S1. Construct a distributed sensing network in the target river sections of rivers and lakes to be revitalized, and collect multi-source eco-hydrological elements that characterize the response of river and lake ecosystems to hydrological processes in real time. S2. Input the multi-source eco-hydrological elements into the pre-constructed ecological water demand response analytical model, and output the dynamic ecological water demand index through multi-element fusion and nonlinear mapping; the dynamic ecological water demand index represents the real-time water demand fluctuation range required by rivers and lakes to maintain or approach the target ecological function under the current hydraulic, water quality and biological conditions. S3. Using the dynamic ecological water demand index and the real-time available water volume index characterizing the watershed's water resource supply capacity as inputs, a two-layer elastic allocation decision model is used to generate an instruction set for downstream flow and gate opening; wherein, the upper layer of the two-layer elastic allocation decision model aims to maximize the comprehensive ecological benefits within the scheduling cycle and generates an elastic space for water demand satisfaction strategies; the lower layer, within the elastic space, aims to minimize water allocation costs and generates the instruction set. S4. Execute the instruction set to release the allocated water volume to the target river section, and continuously monitor the changes in eco-hydrological elements after execution through the distributed sensing network. Feed back the monitoring results to the ecological water demand response analysis model to trigger the iterative correction of the dynamic ecological water demand index, forming a closed-loop dynamic coupling between water allocation and ecological response.

2. The method according to claim 1, characterized in that, The multi-source eco-hydrological elements include water body physicochemical indicators, hydrological and hydrodynamic indicators, and aquatic biological behavioral indicators.

3. The method according to claim 2, characterized in that, The construction of the distributed sensing network described in S1 includes: setting up measuring points in deep water, shallow water and near shore areas at multiple cross sections along the target river section, integrating water temperature sensors, dissolved oxygen sensors, conductivity sensors, chlorophyll α fluorescence probes, flow profilers and hydrophones to simultaneously collect water temperature, dissolved oxygen, conductivity, chlorophyll α concentration, three-dimensional flow velocity of the flow field and underwater sound signals.

4. The method according to claim 3, characterized in that, The aquatic organism behavioral indicators include fish behavior telemetry data and aquatic vegetation remote sensing image data. The fish behavior telemetry data is acquired by implanting acoustic transmitters into target fish and using a hydrophone array deployed along the route to quantify the fish's behavioral response to flow velocity, water depth, and temperature fields. The aquatic vegetation remote sensing image data is acquired by deploying hyperspectral imaging terminals to quantify the response of aquatic vegetation cover and growth to hydrological changes.

5. The method according to claim 1, characterized in that, The ecological water demand response analytical model includes multiple parallel feature extraction branches, which are used to extract water quality evolution features, hydrodynamic change features, and biological response features from the time series of water quality elements, the time series of hydrodynamic elements, and the time series of biological response elements, respectively. The outputs of the multiple feature extraction branches are interactively fused by the feature fusion layer and then input into the regression layer to output the dynamic ecological water demand index.

6. The method according to claim 5, characterized in that, The feature fusion layer employs a multi-head attention mechanism, performing cross-attention operations on the outputs of each feature extraction branch and then fusing them.

7. The method according to claim 1 or 5, characterized in that, The dynamic ecological water demand index consists of the baseflow maintenance index, the pulse replenishment index, and the habitat abundance index. The baseflow maintenance index represents the minimum flow requirement to maintain the survival and basic metabolism of current aquatic organisms. The pulse replenishment index represents the requirement for short-term high-flow pulses to simulate natural hydrological rhythms, trigger fish spawning, or inhibit excessive algal growth. The habitat abundance index represents the requirement for specific flow processes to submerge floodplains or restore aquatic vegetation through water level changes.

8. The method according to claim 1, characterized in that, In S3: The upper layer of the two-layer flexible allocation decision model is a global decision-maker based on Monte Carlo tree search. This global decision-maker takes the dynamic ecological water demand index and its components, reservoir storage capacity, and medium-term runoff forecast statistical characteristics as its state, and takes the floating value of the ecological water demand satisfaction rate as its action. During the search process, a dynamic exploration factor based on the uncertainty of medium-term runoff forecast is introduced, and a surrogate model trained based on historical data is used to simulate and extrapolate the state changes of multiple future steps under the water demand satisfaction strategy. The return is calculated based on the instantaneous benefit function based on the baseflow maintenance degree, pulse timing accuracy, and habitat inundation area, and the elastic space of the water demand satisfaction strategy is generated. The lower layer of the two-layer elastic dispatch decision model is a real-time scheduler based on mixed integer linear programming. It uses reservoir outflow, gate opening and unit load as decision variables, and the weighted sum of pump station energy consumption, water abandonment and head loss as the cost function. It solves to generate the instruction set under hydraulic balance constraints, equipment operation constraints and the water demand satisfaction rate interval constraints corresponding to the elastic space.

9. The method according to claim 1, characterized in that, The iterative correction of the dynamic ecological water demand index in S4 includes: collecting snapshots of ecological and hydrological elements at multiple preset time points after the release of allocated water and generating a new dynamic ecological water demand index to calculate the ecological water demand gap; using a fuzzy logic regulator, taking the rate of change and acceleration of change of the ecological water demand gap within a sliding window as input, and outputting the adjustment range of the satisfaction rate through fuzzy inference; when multiple consecutive snapshots show that the ecological water demand gap is continuously narrowing, sending a control message to the two-layer elastic allocation decision model to lower the water demand satisfaction rate, and sending a control message to raise the water demand satisfaction rate when the ecological water demand gap is widening.

10. A flexible water allocation system for river and lake recovery coupled with ecological water demand response, characterized in that, include: A distributed eco-hydrological sensing subsystem is used to collect multi-source eco-hydrological elements that characterize the response of river and lake ecosystems to hydrological processes in real time. An ecological water demand response analysis engine is used to input the multi-source ecological hydrological elements into a pre-constructed ecological water demand response analysis model and output a dynamic ecological water demand index. A two-layer flexible allocation decision platform is used to generate an elastic space for water demand satisfaction strategy in the upper layer, and generate a set of instructions for discharge flow and gate opening in the lower layer within the elastic space, taking the dynamic ecological water demand index and real-time available water quantity index as input. The water volume execution and ecological feedback closed-loop control module is used to execute the instruction set, monitor changes in eco-hydrological elements, and feed them back to the ecological water demand response analysis engine to trigger the iterative correction of the dynamic ecological water demand index.

Images