River lake-oasis-wetland cross-system water resource communication scheduling method based on intelligent water network

Abstract

The invention relates to the technical field of water resource allocation, and discloses a river lake-oasis-wetland cross-system water resource communication scheduling method based on an intelligent water network, and the method comprises the following steps: building a multi-level three-dimensional perception network covering three types of ecological systems of a river lake, an oasis and a wetland, obtaining space-time heterogeneous data of water level, flow, water quality, soil moisture content and vegetation coverage; constructing a cross-system water resource dynamic coupling model considering a water volume-water quality-ecology coupling effect, wherein the dynamic coupling model comprises a river lake-oasis interface hydraulic contact sub-model, an oasis-wetland moisture conversion sub-model and a wetland-river lake ecology feedback sub-model; and a multi-target collaborative scheduling model with optimal ecological benefits, maximum water resource utilization efficiency and minimum system risk as targets is constructed. By constructing a multi-target collaborative scheduling model and comprehensively considering ecological benefits, water resource utilization efficiency and system risks, optimal configuration of water resources among three types of ecological systems of rivers, lakes, oasis and wetlands is realized.

Description

Technical Field

[0001] This invention relates to the field of water resource allocation technology, specifically to a method for cross-system water resource interconnection and scheduling based on a smart water network, encompassing rivers, lakes, oases, and wetlands. Background Technology

[0002] In the field of water resource management, traditional methods for water resource allocation in rivers, lakes, oases, and wetlands primarily focus on the allocation of water resources within a single system, lacking comprehensive consideration of the coupled relationships between multiple ecosystems. For example, river and lake system allocation often aims at water balance and flood and drought control, oasis irrigation allocation emphasizes agricultural water use efficiency, and wetland protection relies on the simple guarantee of ecological flow. While these methods have achieved certain results in their respective fields, they have exposed numerous problems when facing complex and ever-changing natural environments and increasing demands for ecological protection. First, the allocation of a single system struggles to balance ecological benefits and water resource utilization efficiency, leading to frequent water waste and ecosystem degradation. Second, traditional allocation methods lack the ability to respond quickly to sudden water events, making it difficult for ecosystems to maintain key ecological functions under extreme conditions. Furthermore, existing technologies have limitations in data acquisition and processing, relying heavily on a limited number of ground monitoring stations, making it difficult to achieve real-time monitoring of large areas and multiple elements, resulting in insufficient scientific rigor and dynamic adaptability in allocation decisions. With the development of smart water management technology, although technologies such as the Internet of Things and satellite remote sensing have been applied in water resource monitoring, a systematic cross-ecosystem water resource connectivity and scheduling framework has not yet been formed.

[0003] Therefore, there is an urgent need for an intelligent water resource management method that can comprehensively consider the coupling relationship between water quantity, water quality, and ecology, and achieve multi-system coordinated scheduling, so as to improve water resource utilization efficiency, enhance ecosystem resilience, and promote the deep integration of water resource management and ecological protection. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a method for cross-system water resource interconnection and scheduling based on smart water networks, which solves the problems mentioned in the background.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a method for cross-system water resource interconnection and scheduling based on a smart water network, comprising the following steps: Establish a multi-level three-dimensional sensing network covering three types of ecosystems: rivers, lakes, oases, and wetlands, to acquire spatiotemporally heterogeneous data on water level, flow rate, water quality, soil moisture, and vegetation coverage; A cross-system dynamic coupling model of water resources is constructed, which considers the coupling effect of water quantity, water quality and ecology. The dynamic coupling model includes a sub-model of hydraulic connection at the river-lake-oasis interface, a sub-model of water transformation at the oasis-wetland interface, and a sub-model of ecological feedback between wetland and river-lake. Construct a multi-objective collaborative scheduling model with the goals of optimal ecological benefits, maximum water resource utilization efficiency, and minimum system risk; An improved deep deterministic strategy gradient algorithm is used to solve the multi-objective cooperative scheduling model and generate a cross-system connectivity scheduling scheme. The improved deep deterministic strategy gradient algorithm integrates physical law constraints and a self-learning mechanism of reward and punishment functions. The operation status of the gate pumping station group is controlled according to the cross-system interconnection scheduling scheme to realize the bidirectional orderly flow of water resources among the three types of ecosystems: rivers, lakes, oases, and wetlands. The operation status of the gate pumping station group is controlled according to the cross-system interconnection scheduling scheme to realize the bidirectional orderly flow of water resources among the three types of ecosystems: rivers, lakes, oases, and wetlands. Establish a cross-system water resource scheduling digital twin. Based on real-time and historical data collected by a multi-level three-dimensional sensing network, construct a 1:1 high-precision virtual simulation scenario to conduct pre-playing, simulation and risk prediction of scheduling schemes. Through real-time mapping between the virtual scenario and the physical system, dynamically correct scheduling parameters. A dynamic water purification and coordination module is added, embedding an ecological purification unit simulation sub-model into the oasis-wetland water transformation pathway. This quantifies the degradation efficiency of nitrogen and phosphorus pollutants by aquatic plants and microorganisms, incorporates the water purification effect into the constraints of the multi-objective collaborative scheduling model, and achieves simultaneous optimization of water resource allocation and water quality improvement. Establish a dynamic adaptation mechanism for scheduling schemes. Based on the differences in hydrological characteristics in different seasons, preset a set of scheduling parameters specific to each season. Combine this with real-time monitoring data on vegetation growth stages and wetland biological activity cycles to automatically match the optimal scheduling strategy, ensuring that the scheduling schemes are precisely aligned with the dynamic needs of the ecosystem.

[0006] Preferably, the multi-level stereo perception network includes: Satellite remote sensing layer acquires macro-scale information on surface water bodies, vegetation indices, and land use types; UAV aerial survey layer to obtain mesoscale channel networks, field moisture, and wetland inundation extent; The Internet of Things (IoT) sensing layer acquires microscale data such as water level, flow rate, water quality parameters, and soil moisture content. A network of hydrological and meteorological stations is used to acquire meteorological elements such as precipitation, evaporation, and temperature. Adding ground-based mobile monitoring units, using mobile monitoring vehicles equipped with rapid water quality detection sensors and soil moisture profile monitoring instruments, enables supplementary monitoring of key monitoring points in remote areas and complex terrains, solving the problem of blind spots in the coverage of fixed monitoring equipment; By adding a biosensing module to the IoT sensing layer, deploying aquatic organism activity sensors and vegetation physiological state monitoring sensors, real-time data on biological indicators such as fish activity intensity and plant photosynthetic efficiency are collected, providing a direct data source for the ecological benefit objective function. A data quality control subsystem is established to remove noisy data from the monitoring data through outlier identification algorithms and fill in missing data values ​​using spatiotemporal interpolation algorithms, thereby ensuring the consistency and reliability of multi-source heterogeneous data.

[0007] Preferably, the cross-system water resource dynamic coupling model is specifically as follows: The hydraulic connection sub-model of the river-lake-oasis interface adopts the coupling form of Saint-Venant's equations and Darcy's law to describe the vertical and lateral exchange process between surface water and groundwater. The oasis-wetland water transformation sub-model constructs a vegetation transpiration-soil infiltration-groundwater recharge continuum equation to characterize the water transformation pathway between the oasis vegetation system and the wetland system. A wetland-river-lake ecological feedback sub-model was established to develop a biodiversity index response function based on water depth, quantifying the feedback effect of wetland hydrological conditions on river-lake ecosystems. The sub-model of hydraulic connection at the river-lake-oasis interface is supplemented with a module of human activity influencing factors, which incorporates dynamic change data of oasis irrigation water intake, industrial water consumption, and urban and rural domestic water consumption, corrects the surface water-groundwater exchange coefficient, and improves the model's simulation accuracy of hydrological processes under human interference. The oasis-wetland water transformation sub-model adds a soil texture classification calculation module. Based on the differences in permeability coefficients of different soil types such as sandy loam, loam, and clay, a regional water transformation equation is established to accurately depict the water migration path under different soil conditions. The ecological response index system in the wetland-river-lake ecological feedback sub-model is expanded. In addition to the biodiversity index, three secondary response functions are added: wetland carbon sink, river and lake self-purification capacity, and vegetation cover stability. A comprehensive ecological feedback index is constructed by weighted summation to fully quantify the multidimensional impact of wetland hydrological conditions on river and lake ecosystems.

[0008] Preferably, the multi-objective cooperative scheduling model includes: The objective function for ecological benefits is expressed as the product of the normalized vegetation index-weighted area and the species diversity index. The objective function for water resource utilization efficiency adopts the ratio of economic output to water consumption and incorporates a water quality impact coefficient. The objective function for system risk adopts a combination of negative correlations between the severity of water shortage and the ecological flow guarantee rate. The three objective functions are transformed into a single-objective optimization problem by weighted summation.

[0009] Preferably, the improved depth deterministic strategy gradient algorithm specifically comprises: Embed a physical constraint layer in the Actor network to ensure that the output action satisfies the water balance equation and engineering operation constraints. Construct a multi-level reward and punishment function in the Critic network, including immediate rewards and punishments, process rewards and punishments, and terminal rewards and punishments; An experience replay priority mechanism is introduced to dynamically adjust the sample sampling probability based on the differences in the ecological benefits of the scheduling schemes; A multi-agent collaborative training framework was adopted to train river and lake scheduling agents, oasis water use agents, and wetland protection agents respectively. The three objective functions are transformed into a single-objective optimization problem by weighted summation; The ecological benefit objective function introduces the ecosystem service value coefficient, quantifies the carbon sink value, soil and water conservation value, and tourism and sightseeing value into specific weights, and forms a three-dimensional product with the normalized vegetation index weighted area and species diversity index to improve the comprehensiveness of ecological benefit assessment. The objective function for water resource utilization efficiency is supplemented with water use structure optimization factors. Weight coefficients are set according to the priority differences of agricultural, industrial, and ecological water use. A dynamic efficiency evaluation model is constructed in combination with the water quality impact coefficient to achieve differentiated optimization for different water use types. The system risk objective function is augmented with a probability factor for extreme hydrological events. Based on historical hydrological data, a probability prediction model for drought and flood events is constructed. The probability of events is coupled with the severity of water shortage and the ecological flow guarantee rate to improve the foresight and accuracy of risk assessment. A module for dynamically adjusting the weights of the objective functions is added. Based on the entropy weight method and the analytic hierarchy process, the weight ratios of the three objective functions are automatically optimized according to the real-time ecological conditions, water resource supply and demand, and regional development needs, so as to avoid scheduling imbalances caused by fixed weights.

[0010] Preferably, the cross-system connectivity scheduling scheme generation process includes: During the pre-scheduling phase, monthly water resource allocation plans are generated based on medium- and long-term hydrological forecasts; During the real-time scheduling phase, daily scheduling instructions are generated based on short-term hydrological forecasts and real-time monitoring data. During the emergency response phase, the rapid response mechanism of the contingency plan database is triggered in the event of extreme drought or flood events. In the post-evaluation phase, the parameters of the dynamic coupling model and the weights of the reward and punishment functions are optimized in reverse using actual operating data; Add a sub-layer for engineering equipment performance constraints to the Actor network, embedding equipment parameter thresholds such as maximum gate opening, pump station rated flow, and pipeline water conveyance capacity to ensure that the generated scheduling actions match the actual operating capacity of the engineering equipment and avoid exceeding the equipment's carrying capacity. An ecological threshold violation penalty factor is added to the reward and penalty function of the Critic network. When the scheduling scheme causes the ecological flow to fall below the minimum threshold or rise above the maximum threshold, the gradient penalty mechanism is triggered to strengthen the algorithm's compliance with ecological constraints. Optimize the experience playback priority mechanism, construct a three-dimensional evaluation index of ecological benefits, efficiency and risk, and dynamically adjust the sample sampling weight according to the comprehensive performance of the scheduling scheme in the three dimensions to avoid algorithm bias caused by a single indicator. In the multi-agent collaborative training framework, an agent negotiation mechanism is added. By introducing a federated learning algorithm, parameter sharing and collaborative decision-making of river and lake scheduling agents, oasis water use agents, and wetland protection agents are realized on the basis of local training, so as to solve the problem of multi-agent goal conflict. An algorithm convergence optimization module is added, adopting an adaptive learning rate adjustment strategy. The learning rate is dynamically adjusted according to the rate of change of the objective function during the iteration process. Combined with an early stopping mechanism, the algorithm is prevented from overfitting, thereby improving the stability and reliability of the scheduling scheme.

[0011] Preferably, the emergency response phase specifically includes: When it is identified that there has been no effective precipitation for 15 consecutive days and the soil moisture is below the wilting coefficient, a drought emergency response will be initiated, prioritizing water supply for the core wetland ecological area and the critical growth period of oasis economic crops; When it is detected that the upstream water level exceeds the warning level and the wetland water level exceeds the maximum ecological water level, the flood emergency response is activated to increase the wetland's drainage capacity into rivers and lakes and restrict water diversion from the oasis.

[0012] Preferably, the method further includes: Establish ecological flow threshold systems for three types of ecosystems: rivers and lakes, oases, and wetlands. These ecological flow threshold systems cover the minimum ecological flow, suitable ecological flow, and maximum ecological water level. The ecological flow threshold system is embedded as a hard constraint into the multi-objective collaborative scheduling model, and the out-of-bounds behavior is handled in the form of a penalty function.

[0013] Preferably, the smart water network architecture includes: The perception layer acquires multi-source heterogeneous spatiotemporal data; The model layer runs the dynamically coupled model and the multi-objective cooperative scheduling model; The decision-making level generates dispatch instructions and emergency plans; The control layer executes control over gate opening, pump station flow, and valve status. The application layer provides functions for water resource management, ecological monitoring, risk early warning, and benefit assessment.

[0014] This invention provides a method for cross-system water resource connectivity and scheduling based on a smart water network, encompassing rivers, lakes, oases, and wetlands. It offers the following advantages: 1. This invention constructs a multi-objective collaborative scheduling model, comprehensively considering ecological benefits, water resource utilization efficiency, and system risks, to achieve optimal allocation of water resources among three types of ecosystems: rivers and lakes, oases, and wetlands. This method effectively improves water resource utilization efficiency, reduces water waste, and simultaneously considers ecosystem protection and restoration, promoting sustainable ecosystem development. Through real-time monitoring and intelligent decision-making via a smart water network, the scientific nature and dynamic adaptability of water resource scheduling schemes are ensured, providing strong support for the sustainable utilization of regional water resources.

[0015] 2. This invention employs an improved deep deterministic policy gradient algorithm, integrating physical constraints and a self-learning mechanism for reward and penalty functions, enabling rapid response to sudden events such as extreme droughts or floods. By establishing an ecological flow threshold system and embedding it as a hard constraint into the scheduling model, it ensures the maintenance of key ecological functions of the ecosystem under different hydrological conditions. Furthermore, the multi-level sensing network of the smart water network architecture can monitor changes in the ecosystem in real time, providing data support for emergency response, thereby significantly enhancing the resilience and risk resistance of the entire ecosystem.

[0016] 3. This invention, based on smart water network technology, integrates various sensing methods such as satellite remote sensing, UAV aerial surveying, IoT sensing, and hydrological and meteorological station networks to construct a multi-level, three-dimensional sensing network covering rivers, lakes, oases, and wetlands. Digital twin technology enables the pre-simulation and optimization of scheduling schemes, forming a closed-loop management system of scheduling-simulation-feedback. This technological architecture not only improves the intelligence level of water resource scheduling but also promotes the deep integration of smart water management and ecological management, providing a new technological means and management model for regional water resource management and ecological protection. Attached Figure Description

[0017] Figure 1 This is the overall flowchart of the present invention; Figure 2 This is a flowchart of the phased scheduling and emergency response process of the present invention. Detailed Implementation

[0018] The technical solutions in 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.

[0019] Please see the appendix Figure 1 - Appendix Figure 2 This invention provides a method for cross-system water resource connectivity and scheduling based on a smart water network, comprising the following steps: Establish a multi-level, three-dimensional sensing network covering three types of ecosystems: rivers, lakes, oases, and wetlands, to acquire spatiotemporally heterogeneous data on water level, flow rate, water quality, soil moisture, and vegetation cover. The multi-level, three-dimensional sensing network includes: Satellite remote sensing layer acquires macro-scale information on surface water bodies, vegetation indices, and land use types; UAV aerial survey layer to obtain mesoscale channel networks, field moisture, and wetland inundation extent; The Internet of Things (IoT) sensing layer acquires microscale data such as water level, flow rate, water quality parameters, and soil moisture content. The hydrological and meteorological station network layer acquires meteorological elements such as precipitation, evaporation, and temperature.

[0020] Adding ground-based mobile monitoring units, using mobile monitoring vehicles equipped with rapid water quality detection sensors and soil moisture profile monitoring instruments, enables supplementary monitoring of key monitoring points in remote areas and complex terrains, solving the problem of blind spots in the coverage of fixed monitoring equipment; By adding a biosensing module to the IoT sensing layer, deploying aquatic organism activity sensors and vegetation physiological state monitoring sensors, real-time data on biological indicators such as fish activity intensity and plant photosynthetic efficiency are collected, providing a direct data source for the ecological benefit objective function. A data quality control subsystem is established to remove noisy data from the monitoring data through outlier identification algorithms and fill in missing data values ​​using spatiotemporal interpolation algorithms, thereby ensuring the consistency and reliability of multi-source heterogeneous data.

[0021] The operation status of the gate pumping station group is controlled according to the cross-system interconnection scheduling scheme to realize the bidirectional orderly flow of water resources among the three types of ecosystems: rivers, lakes, oases, and wetlands. Establish a cross-system water resource scheduling digital twin. Based on real-time and historical data collected by a multi-level three-dimensional sensing network, construct a 1:1 high-precision virtual simulation scenario to conduct pre-playing, simulation and risk prediction of scheduling schemes. Through real-time mapping between the virtual scenario and the physical system, dynamically correct scheduling parameters. A dynamic water purification and coordination module is added, embedding an ecological purification unit simulation sub-model into the oasis-wetland water transformation pathway. This quantifies the degradation efficiency of nitrogen and phosphorus pollutants by aquatic plants and microorganisms, incorporates the water purification effect into the constraints of the multi-objective collaborative scheduling model, and achieves simultaneous optimization of water resource allocation and water quality improvement. Establish a dynamic adaptation mechanism for scheduling schemes. Based on the differences in hydrological characteristics in different seasons, preset a set of scheduling parameters specific to each season. Combine this with real-time monitoring data on vegetation growth stages and wetland biological activity cycles to automatically match the optimal scheduling strategy, ensuring that the scheduling schemes are precisely aligned with the dynamic needs of the ecosystem.

[0022] Specifically, the satellite remote sensing layer utilizes the high coverage and high resolution of satellite sensors to acquire information on surface water bodies, vegetation indices, and land use types at the macro scale, providing fundamental data for the status of large-scale ecosystems. The UAV aerial survey layer uses high-precision sensors mounted on UAVs to acquire information on canal networks, field moisture, and wetland inundation extent at the meso scale, compensating for the limitations of satellite remote sensing in monitoring detailed local areas. The Internet of Things (IoT) sensing layer uses sensors distributed in key locations to acquire real-time data on water levels, flow rates, water quality parameters, and soil moisture content at the micro scale, providing data support for precise monitoring and rapid response. The hydrological and meteorological station network layer monitors meteorological elements such as precipitation, evaporation, and temperature, providing meteorological background data for water resource allocation. This multi-level, three-dimensional sensing network works collaboratively to achieve comprehensive monitoring from macro to micro and from static to dynamic perspectives, providing a high-precision and timely data foundation for cross-system water resource interconnection and allocation, meeting the requirements of modern smart water management for comprehensive and accurate data collection.

[0023] A cross-system dynamic coupling model of water resources is constructed, considering the coupling effects of water quantity, water quality, and ecology. This dynamic coupling model includes a river-lake-oasis interface hydraulic connection sub-model, an oasis-wetland water transformation sub-model, and a wetland-river-lake ecological feedback sub-model. Specifically, the cross-system dynamic coupling model of water resources is as follows: The hydraulic connection sub-model of the river-lake-oasis interface adopts the coupling form of Saint-Venant's equations and Darcy's law to describe the vertical and lateral exchange process between surface water and groundwater. The oasis-wetland water transformation sub-model constructs a vegetation transpiration-soil infiltration-groundwater recharge continuum equation to characterize the water transformation pathway between the oasis vegetation system and the wetland system. A wetland-river-lake ecological feedback sub-model was established, which uses a biodiversity index response function based on water depth to quantify the feedback effect of wetland hydrological conditions on river and lake ecosystems.

[0024] The sub-model of hydraulic connection at the river-lake-oasis interface is supplemented with a module of human activity influencing factors, which incorporates dynamic change data of oasis irrigation water intake, industrial water consumption, and urban and rural domestic water consumption, corrects the surface water-groundwater exchange coefficient, and improves the model's simulation accuracy of hydrological processes under human interference. The oasis-wetland water transformation sub-model adds a soil texture classification calculation module. Based on the differences in permeability coefficients of different soil types such as sandy loam, loam, and clay, a regional water transformation equation is established to accurately depict the water migration path under different soil conditions. The ecological response index system in the wetland-river-lake ecological feedback sub-model is expanded. In addition to the biodiversity index, three secondary response functions are added: wetland carbon sink, river and lake self-purification capacity, and vegetation cover stability. A comprehensive ecological feedback index is constructed by weighted summation to fully quantify the multidimensional impact of wetland hydrological conditions on river and lake ecosystems.

[0025] Specifically, a hierarchical sub-model is used to describe the hydrological processes and ecological feedback between different ecosystem interfaces. The river-lake-oasis interface hydraulic connection sub-model employs a coupling of Saint-Venant's equations and Darcy's law. Saint-Venant's equations describe the movement characteristics of surface water, while Darcy's law characterizes the flow patterns of groundwater. This coupling accurately simulates the vertical and lateral exchange processes between surface water and groundwater, thus achieving dynamic interaction simulation of water resources between rivers, lakes, and oases. The oasis-wetland water transformation sub-model constructs a continuum equation of vegetation transpiration, soil infiltration, and groundwater recharge, linking water consumption in the oasis vegetation system with water recharge in the wetland system. This characterizes the water transformation path between the two ecosystems, providing a theoretical basis for the coordinated scheduling of oasis irrigation and wetland water replenishment. The wetland-river-lake ecological feedback sub-model establishes a biodiversity index response function based on water depth. By quantifying the impact of wetland hydrological changes on the biodiversity of river and lake ecosystems, it simulates the ecological feedback mechanism from wetlands to rivers and lakes, providing scientific support for the optimal allocation of water resources under ecological protection goals. The construction of this dynamic coupling model can comprehensively reflect the physical, chemical and ecological processes of cross-system water resources, providing a solid theoretical foundation for the scientific allocation of water resources and the protection of ecosystems.

[0026] Construct a multi-objective collaborative scheduling model with the goals of optimal ecological benefits, maximum water resource utilization efficiency, and minimum system risk; An improved deep deterministic policy gradient algorithm is used to solve the multi-objective cooperative scheduling model, generating a cross-system connectivity scheduling scheme. The improved deep deterministic policy gradient algorithm integrates physical law constraints and a self-learning mechanism of reward and punishment functions. The operation status of the gate pumping station group is controlled according to the cross-system interconnection scheduling scheme to realize the bidirectional orderly flow of water resources among the three types of ecosystems: rivers, lakes, oases, and wetlands.

[0027] Specifically, by quantifying the value of ecosystem services, the positive impact of water resource allocation on the ecosystem is ensured; water resource allocation is optimized to maximize water resource utilization efficiency and reduce waste; and the impact of extreme water events on the system is reduced to minimize system risk. To solve this multi-objective model, an improved deep deterministic strategy gradient algorithm is adopted. This algorithm integrates physical constraints to ensure that the allocation scheme conforms to the basic laws of hydrology and engineering operation, while introducing a self-learning mechanism for reward and penalty functions to dynamically adjust the allocation strategy to adapt to complex and changing hydrological conditions. Through the cross-system connectivity allocation scheme generated by the algorithm, the operation of the gate pumping station group is controlled, realizing the bidirectional orderly flow of water resources among the three major ecosystems. This ensures ecological needs are met while improving the overall water resource utilization efficiency and the system's resilience.

[0028] The process of generating a cross-system connectivity scheduling scheme includes: During the pre-scheduling phase, monthly water resource allocation plans are generated based on medium- and long-term hydrological forecasts; During the real-time scheduling phase, daily scheduling instructions are generated based on short-term hydrological forecasts and real-time monitoring data. During the emergency response phase, the rapid response mechanism of the contingency plan database is triggered in the event of extreme drought or flood events. Specifically, in the pre-scheduling phase, monthly water resource allocation plans are planned in advance based on medium- and long-term hydrological forecast data, providing macro-level guidance for the entire scheduling cycle and ensuring the rational allocation and storage of water resources over time. In the real-time scheduling phase, short-term hydrological forecasts and real-time monitoring data are combined to generate more precise daily scheduling instructions. By dynamically adjusting scheduling strategies, timely responses to short-term hydrological changes are made to ensure the efficient utilization of daily water resources and meet ecological needs. In the emergency response phase, for sudden events such as extreme droughts or floods, pre-set plans in the contingency plan database are quickly triggered to enable immediate handling of emergencies, ensuring the stable operation of the system and ecological safety under extreme conditions. This phased, multi-level scheduling plan generation mechanism effectively integrates information at different time scales, improves the scientific nature and adaptability of scheduling decisions, and provides a flexible and reliable management framework for cross-system water resource interconnection scheduling.

[0029] The emergency response phase is specifically as follows: When it is identified that there has been no effective precipitation for 15 consecutive days and the soil moisture is below the wilting coefficient, a drought emergency response will be initiated, prioritizing water supply for the core wetland ecological area and the critical growth period of oasis economic crops; When it is detected that the upstream water level exceeds the warning level and the wetland water level exceeds the maximum ecological water level, the flood emergency response is activated to increase the wetland's drainage capacity into rivers and lakes and restrict water diversion from the oasis.

[0030] In the post-evaluation phase, the parameters of the dynamic coupling model and the weights of the reward and punishment function are optimized in reverse using actual operating data.

[0031] Specifically, the emergency response and post-evaluation mechanism of this invention is based on the principle of dynamic decision-making driven by real-time monitoring data to ensure the scientific allocation and system optimization of water resources under extreme hydrological conditions. In drought emergency response, the mechanism is triggered when there is no effective precipitation for 15 consecutive days and soil moisture is below the wilting coefficient. Priority is given to ensuring water supply to the core wetland ecological area and the critical growth period of oasis economic crops, thereby maintaining ecosystem stability and agricultural sustainability during periods of extreme water scarcity. In flood emergency response, the mechanism is activated when upstream water levels exceed warning levels and wetland water levels exceed the maximum ecological water level. This effectively controls flood risk, protects wetland ecological functions, and prevents flooding in oasis areas by increasing the drainage capacity of wetlands to rivers and lakes and restricting water diversion from oases. The post-evaluation phase utilizes actual operational data to perform reverse optimization of the dynamic coupling model parameters and reward / penalty function weights. Through a feedback mechanism, the model and algorithm are continuously adjusted and improved to more accurately reflect actual operational conditions, thereby enhancing the scientific rigor and adaptability of the scheduling scheme and providing continuous optimization technical support for cross-system water resource interconnection scheduling.

[0032] Multi-objective cooperative scheduling models include: The objective function for ecological benefits is expressed as the product of the normalized vegetation index-weighted area and the species diversity index. The objective function for water resource utilization efficiency adopts the ratio of economic output to water consumption and incorporates a water quality impact coefficient. The objective function for system risk adopts a combination of negative correlations between the severity of water shortage and the ecological flow guarantee rate. The three objective functions are transformed into a single-objective optimization problem by weighted summation.

[0033] The ecological benefit objective function introduces the ecosystem service value coefficient, quantifies the carbon sink value, soil and water conservation value, and tourism and sightseeing value into specific weights, and forms a three-dimensional product with the normalized vegetation index weighted area and species diversity index to improve the comprehensiveness of ecological benefit assessment. The objective function for water resource utilization efficiency is supplemented with water use structure optimization factors. Weight coefficients are set according to the priority differences of agricultural, industrial, and ecological water use. A dynamic efficiency evaluation model is constructed in combination with the water quality impact coefficient to achieve differentiated optimization for different water use types. The system risk objective function is augmented with a probability factor for extreme hydrological events. Based on historical hydrological data, a probability prediction model for drought and flood events is constructed. The probability of events is coupled with the severity of water shortage and the ecological flow guarantee rate to improve the foresight and accuracy of risk assessment. A module for dynamically adjusting the weights of the objective functions is added. Based on the entropy weight method and the analytic hierarchy process, the weight ratios of the three objective functions are automatically optimized according to the real-time ecological conditions, water resource supply and demand, and regional development needs, so as to avoid scheduling imbalances caused by fixed weights.

[0034] Specifically, this study constructs three objective functions—ecological benefits, water resource utilization efficiency, and systemic risk—to achieve comprehensive optimization of cross-system water resource scheduling. The ecological benefits objective function quantifies the value of ecosystem services by multiplying the weighted area of ​​the normalized vegetation index (NDI) with the species diversity index, reflecting the contribution of vegetation cover and biodiversity to ecological benefits. The water resource utilization efficiency objective function comprehensively assesses the economic efficiency of water resource utilization and water quality assurance capabilities by using the ratio of economic output to water consumption and incorporating a water quality impact coefficient. The systemic risk objective function measures the risk level in the water resource scheduling process by using a negative correlation between the severity of water shortage and the ecological flow guarantee rate. By weighted summing of these three objective functions, the multi-objective optimization problem is transformed into a single-objective optimization problem, thereby achieving dynamic balance and coordinated optimization among different objectives. This model design effectively considers ecological, economic, and risk control needs, providing a scientific and rational decision-making basis for cross-system water resource interconnection and scheduling.

[0035] The improved gradient algorithm for deep deterministic policies is as follows: Embed a physical constraint layer in the Actor network to ensure that the output action satisfies the water balance equation and engineering operation constraints. Construct a multi-level reward and punishment function in the Critic network, including immediate rewards and punishments, process rewards and punishments, and terminal rewards and punishments; An experience replay priority mechanism is introduced to dynamically adjust the sample sampling probability based on the differences in the ecological benefits of the scheduling schemes; A multi-agent collaborative training framework was adopted to train river and lake scheduling agents, oasis water use agents, and wetland protection agents respectively.

[0036] Add a sub-layer for engineering equipment performance constraints to the Actor network, embedding equipment parameter thresholds such as maximum gate opening, pump station rated flow, and pipeline water conveyance capacity to ensure that the generated scheduling actions match the actual operating capacity of the engineering equipment and avoid exceeding the equipment's carrying capacity. An ecological threshold violation penalty factor is added to the reward and penalty function of the Critic network. When the scheduling scheme causes the ecological flow to fall below the minimum threshold or rise above the maximum threshold, the gradient penalty mechanism is triggered to strengthen the algorithm's compliance with ecological constraints. Optimize the experience playback priority mechanism, construct a three-dimensional evaluation index of ecological benefits, efficiency and risk, and dynamically adjust the sample sampling weight according to the comprehensive performance of the scheduling scheme in the three dimensions to avoid algorithm bias caused by a single indicator. In the multi-agent collaborative training framework, an agent negotiation mechanism is added. By introducing a federated learning algorithm, parameter sharing and collaborative decision-making of river and lake scheduling agents, oasis water use agents, and wetland protection agents are realized on the basis of local training, so as to solve the problem of multi-agent goal conflict. An algorithm convergence optimization module is added, adopting an adaptive learning rate adjustment strategy. The learning rate is dynamically adjusted according to the rate of change of the objective function during the iteration process. Combined with an early stopping mechanism, the algorithm is prevented from overfitting, thereby improving the stability and reliability of the scheduling scheme.

[0037] Specifically, firstly, a physical constraint layer is embedded in the Actor network. By embedding the water balance equation and engineering operation constraints into the network structure, it is ensured that the scheduling actions generated by the algorithm strictly conform to the basic physical laws of hydrology and water conservancy engineering, thereby guaranteeing the feasibility and reliability of the scheduling scheme. Secondly, a multi-level reward and punishment function is constructed in the Critic network, including immediate rewards and punishments, process rewards and punishments, and terminal rewards and punishments. Through a multi-dimensional feedback mechanism, the scheduling scheme is dynamically evaluated, enabling the algorithm to optimize and adjust the scheduling strategy based on short-term and long-term benefits. In addition, an experience replay priority mechanism is introduced, dynamically adjusting the sample sampling probability according to the differences in the ecological benefits of the scheduling schemes, prioritizing the learning of scheduling experiences with significant ecological benefits, improving the learning efficiency of the algorithm and its responsiveness to ecological goals. Finally, a multi-agent collaborative training framework is adopted to train river and lake scheduling agents, oasis water use agents, and wetland protection agents respectively. Through the collaborative cooperation among agents, efficient solutions to complex cross-system water resource scheduling problems are achieved, improving the optimization degree and adaptability of the overall scheduling scheme.

[0038] The method further includes: Establish ecological flow threshold systems for three types of ecosystems: rivers and lakes, oases, and wetlands. The ecological flow threshold systems cover the minimum ecological flow, suitable ecological flow, and maximum ecological water level. An ecological flow threshold system is embedded as a hard constraint into a multi-objective collaborative scheduling model, and a penalty function is used to handle out-of-bounds behavior.

[0039] Specifically, by constructing an ecological flow threshold system and embedding it into the scheduling model, the aim is to delineate ecological security boundaries for cross-system water resource scheduling and ensure the feasibility of scheduling schemes. Specifically, threshold systems covering minimum ecological flow, suitable ecological flow, and maximum ecological water level are established for three types of ecosystems: rivers and lakes, oases, and wetlands. Based on ecological principles and hydrological laws, this system defines the range of hydrological conditions required to maintain the healthy operation of each type of ecosystem and clarifies the upper and lower limits of water volume that the ecosystem can withstand. Embedding this threshold system as a hard constraint into the multi-objective collaborative scheduling model means that during the model solution process, any scheduling scheme that violates the ecological flow threshold constraint will be considered an infeasible solution, thus fundamentally eliminating the possibility of irreversible damage to the ecosystem caused by the scheduling scheme. Simultaneously, a penalty function is used to handle boundary violations. A mathematical penalty term is applied to scheduling strategies that approach or exceed the constraint boundary, guiding the optimization algorithm to actively avoid infeasible areas and ensuring that the generated scheduling scheme strictly meets ecological protection requirements. This mechanism, while ensuring ecological security, achieves efficient water resource utilization and multi-objective collaborative optimization, providing scientific and reliable decision support for cross-system interconnected scheduling.

[0040] The smart water network architecture includes: The perception layer acquires multi-source heterogeneous spatiotemporal data; The model layer runs the dynamically coupled model and the multi-objective cooperative scheduling model; The decision-making level generates dispatch instructions and emergency plans; The control layer executes control over gate opening, pump station flow, and valve status. The application layer provides functions for water resource management, ecological monitoring, risk early warning, and benefit assessment.

[0041] Specifically, through the collaborative operation of the perception layer, model layer, decision-making layer, control layer, and application layer, intelligent management of the entire process of cross-system water resource interconnection and scheduling is achieved. The perception layer is responsible for collecting multi-source heterogeneous spatiotemporal data, including key information such as water level, flow rate, water quality, and meteorological conditions, providing a data foundation for the upper layers. The model layer runs a dynamic coupling model and a multi-objective collaborative scheduling model based on the data from the perception layer, generating optimized scheduling schemes through scientific calculations. The decision-making layer generates specific scheduling instructions and emergency plans based on the output results of the model layer and combined with real-time monitoring data. The control layer directly executes the instructions from the decision-making layer, precisely controlling gate opening, pump station flow, and valve status to achieve bidirectional and orderly flow of water resources among rivers, lakes, oases, and wetlands. The application layer provides managers with water resource management, ecological monitoring, risk warning, and benefit assessment functions, enabling visualized display and post-evaluation of scheduling effects. A closed-loop feedback mechanism is formed between each layer through data flow and control flow, ensuring the scientific, real-time, and reliable nature of scheduling decisions, providing solid technical support for cross-system water resource interconnection and scheduling.

[0042] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for cross-system water resource connectivity and scheduling based on smart water networks, characterized in that, Includes the following steps: Establish a multi-level three-dimensional sensing network covering three types of ecosystems: rivers, lakes, oases, and wetlands, to acquire spatiotemporally heterogeneous data on water level, flow rate, water quality, soil moisture, and vegetation coverage; A cross-system dynamic coupling model of water resources is constructed, which considers the coupling effect of water quantity, water quality and ecology. The dynamic coupling model includes a sub-model of hydraulic connection at the river-lake-oasis interface, a sub-model of water transformation at the oasis-wetland interface, and a sub-model of ecological feedback between wetland and river-lake. Construct a multi-objective collaborative scheduling model with the goals of optimal ecological benefits, maximum water resource utilization efficiency, and minimum system risk; An improved deep deterministic strategy gradient algorithm is used to solve the multi-objective cooperative scheduling model and generate a cross-system connectivity scheduling scheme. The improved deep deterministic strategy gradient algorithm integrates physical law constraints and a self-learning mechanism of reward and punishment functions. The operation status of the gate pumping station group is controlled according to the cross-system interconnection scheduling scheme to realize the bidirectional orderly flow of water resources among the three types of ecosystems: rivers, lakes, oases, and wetlands. Establish a cross-system water resource scheduling digital twin. Based on real-time and historical data collected by a multi-level three-dimensional sensing network, construct a 1:1 high-precision virtual simulation scenario to conduct pre-playing, simulation and risk prediction of scheduling schemes. Through real-time mapping between the virtual scenario and the physical system, dynamically correct scheduling parameters. A dynamic water purification and coordination module is added, embedding an ecological purification unit simulation sub-model into the oasis-wetland water transformation pathway. This quantifies the degradation efficiency of nitrogen and phosphorus pollutants by aquatic plants and microorganisms, incorporates the water purification effect into the constraints of the multi-objective collaborative scheduling model, and achieves simultaneous optimization of water resource allocation and water quality improvement. Establish a dynamic adaptation mechanism for scheduling schemes. Based on the differences in hydrological characteristics in different seasons, preset a set of scheduling parameters specific to each season. Combine this with real-time monitoring data on vegetation growth stages and wetland biological activity cycles to automatically match the optimal scheduling strategy, ensuring that the scheduling schemes are precisely aligned with the dynamic needs of the ecosystem.

2. The method for cross-system water resource connectivity and scheduling based on a smart water network according to claim 1, characterized in that, The multi-level stereo perception network includes: Satellite remote sensing layer acquires macro-scale information on surface water bodies, vegetation indices, and land use types; UAV aerial survey layer to obtain mesoscale channel networks, field moisture, and wetland inundation extent; The Internet of Things (IoT) sensing layer acquires microscale data such as water level, flow rate, water quality parameters, and soil moisture content. A network of hydrological and meteorological stations is used to acquire meteorological elements such as precipitation, evaporation, and temperature. Adding ground-based mobile monitoring units, using mobile monitoring vehicles equipped with rapid water quality detection sensors and soil moisture profile monitoring instruments, enables supplementary monitoring of key monitoring points in remote areas and complex terrains, solving the problem of blind spots in the coverage of fixed monitoring equipment; By adding a biosensing module to the IoT sensing layer, deploying aquatic organism activity sensors and vegetation physiological state monitoring sensors, real-time data on biological indicators such as fish activity intensity and plant photosynthetic efficiency are collected, providing a direct data source for the ecological benefit objective function. A data quality control subsystem is established to remove noisy data from the monitoring data through outlier identification algorithms and fill in missing data values ​​using spatiotemporal interpolation algorithms, thereby ensuring the consistency and reliability of multi-source heterogeneous data.

3. The method for cross-system water resource connectivity and scheduling based on a smart water network according to claim 1, characterized in that, The cross-system water resource dynamic coupling model is specifically as follows: The hydraulic connection sub-model of the river-lake-oasis interface adopts the coupling form of Saint-Venant's equations and Darcy's law to describe the vertical and lateral exchange process between surface water and groundwater. The oasis-wetland water transformation sub-model constructs a vegetation transpiration-soil infiltration-groundwater recharge continuum equation to characterize the water transformation pathway between the oasis vegetation system and the wetland system. A wetland-river-lake ecological feedback sub-model was established to develop a biodiversity index response function based on water depth, quantifying the feedback effect of wetland hydrological conditions on river-lake ecosystems. The sub-model of hydraulic connection at the river-lake-oasis interface is supplemented with a module of human activity influencing factors, which incorporates dynamic change data of oasis irrigation water intake, industrial water consumption, and urban and rural domestic water consumption, corrects the surface water-groundwater exchange coefficient, and improves the model's simulation accuracy of hydrological processes under human interference. The oasis-wetland water transformation sub-model adds a soil texture classification calculation module. Based on the differences in permeability coefficients of different soil types such as sandy loam, loam, and clay, a regional water transformation equation is established to accurately depict the water migration path under different soil conditions. The ecological response index system in the wetland-river-lake ecological feedback sub-model is expanded. In addition to the biodiversity index, three secondary response functions are added: wetland carbon sink, river and lake self-purification capacity, and vegetation cover stability. A comprehensive ecological feedback index is constructed by weighted summation to fully quantify the multidimensional impact of wetland hydrological conditions on river and lake ecosystems.

4. A method for cross-system water resource connectivity and scheduling based on a smart water network, according to claim 1, is characterized in that... The multi-objective cooperative scheduling model includes: The objective function for ecological benefits is expressed as the product of the normalized vegetation index-weighted area and the species diversity index. The objective function for water resource utilization efficiency adopts the ratio of economic output to water consumption and incorporates a water quality impact coefficient. The objective function for system risk adopts a combination of negative correlations between the severity of water shortage and the ecological flow guarantee rate. The three objective functions are transformed into a single-objective optimization problem by weighted summation; The ecological benefit objective function introduces the ecosystem service value coefficient, quantifies the carbon sink value, soil and water conservation value, and tourism and sightseeing value into specific weights, and forms a three-dimensional product with the normalized vegetation index weighted area and species diversity index to improve the comprehensiveness of ecological benefit assessment. The objective function for water resource utilization efficiency is supplemented with water use structure optimization factors. Weight coefficients are set according to the priority differences of agricultural, industrial, and ecological water use. A dynamic efficiency evaluation model is constructed in combination with the water quality impact coefficient to achieve differentiated optimization for different water use types. The system risk objective function is augmented with a probability factor for extreme hydrological events. Based on historical hydrological data, a probability prediction model for drought and flood events is constructed. The probability of events is coupled with the severity of water shortage and the ecological flow guarantee rate to improve the foresight and accuracy of risk assessment. A module for dynamically adjusting the weights of the objective functions is added. Based on the entropy weight method and the analytic hierarchy process, the weight ratios of the three objective functions are automatically optimized according to the real-time ecological conditions, water resource supply and demand, and regional development needs, so as to avoid scheduling imbalances caused by fixed weights.

5. A method for cross-system water resource connectivity and scheduling based on a smart water network, according to claim 1, characterized in that, The improved depth deterministic strategy gradient algorithm is specifically as follows: Embed a physical constraint layer in the Actor network to ensure that the output action satisfies the water balance equation and engineering operation constraints. Construct a multi-level reward and punishment function in the Critic network, including immediate rewards and punishments, process rewards and punishments, and terminal rewards and punishments; An experience replay priority mechanism is introduced to dynamically adjust the sample sampling probability based on the differences in the ecological benefits of the scheduling schemes; A multi-agent collaborative training framework was adopted to train river and lake scheduling agents, oasis water use agents, and wetland protection agents respectively. Add a sub-layer for engineering equipment performance constraints to the Actor network, embedding equipment parameter thresholds such as maximum gate opening, pump station rated flow, and pipeline water conveyance capacity to ensure that the generated scheduling actions match the actual operating capacity of the engineering equipment and avoid exceeding the equipment's carrying capacity. An ecological threshold violation penalty factor is added to the reward and penalty function of the Critic network. When the scheduling scheme causes the ecological flow to fall below the minimum threshold or rise above the maximum threshold, the gradient penalty mechanism is triggered to strengthen the algorithm's compliance with ecological constraints. Optimize the experience playback priority mechanism, construct a three-dimensional evaluation index of ecological benefits, efficiency and risk, and dynamically adjust the sample sampling weight according to the comprehensive performance of the scheduling scheme in the three dimensions to avoid algorithm bias caused by a single indicator. In the multi-agent collaborative training framework, an agent negotiation mechanism is added. By introducing a federated learning algorithm, parameter sharing and collaborative decision-making of river and lake scheduling agents, oasis water use agents, and wetland protection agents are realized on the basis of local training, thus solving the problem of multi-agent goal conflict. An algorithm convergence optimization module is added, adopting an adaptive learning rate adjustment strategy. The learning rate is dynamically adjusted according to the rate of change of the objective function during the iteration process. Combined with an early stopping mechanism, the algorithm is prevented from overfitting, thereby improving the stability and reliability of the scheduling scheme.

6. A method for cross-system water resource connectivity and scheduling based on a smart water network, according to claim 1, characterized in that, The process of generating the cross-system connectivity scheduling scheme includes: During the pre-scheduling phase, monthly water resource allocation plans are generated based on medium- and long-term hydrological forecasts; During the real-time scheduling phase, daily scheduling instructions are generated based on short-term hydrological forecasts and real-time monitoring data. During the emergency response phase, the rapid response mechanism of the contingency plan database is triggered in the event of extreme drought or flood events. In the post-evaluation phase, the parameters of the dynamic coupling model and the weights of the reward and punishment function are optimized in reverse using actual operating data.

7. A method for cross-system water resource connectivity and scheduling based on a smart water network, according to claim 6, is characterized in that... The emergency response phase specifically includes: When it is identified that there has been no effective precipitation for 15 consecutive days and the soil moisture is below the wilting coefficient, a drought emergency response will be initiated, prioritizing water supply for the core wetland ecological area and the critical growth period of oasis economic crops; When it is detected that the upstream water level exceeds the warning level and the wetland water level exceeds the maximum ecological water level, the flood emergency response is activated to increase the wetland's drainage capacity into rivers and lakes and restrict water diversion from the oasis.

8. A method for cross-system water resource connectivity and scheduling based on a smart water network, according to claim 1, characterized in that, The method further includes: Establish ecological flow threshold systems for three types of ecosystems: rivers and lakes, oases, and wetlands. These ecological flow threshold systems cover the minimum ecological flow, suitable ecological flow, and maximum ecological water level. The ecological flow threshold system is embedded as a hard constraint into the multi-objective collaborative scheduling model, and the out-of-bounds behavior is handled in the form of a penalty function.

9. A method for cross-system water resource connectivity and scheduling based on a smart water network, according to claim 1, characterized in that, The smart water network architecture includes: The perception layer acquires multi-source heterogeneous spatiotemporal data; The model layer runs the dynamically coupled model and the multi-objective cooperative scheduling model; The decision-making level generates dispatch instructions and emergency plans; The control layer executes control over gate opening, pump station flow, and valve status. The application layer provides functions for water resource management, ecological monitoring, risk early warning, and benefit assessment.

Images