Ecological risk assessment and joint regulation method and system for river supplied by reclaimed water

Abstract

The application discloses an ecological risk assessment and joint regulation method and system for river water supply by reclaimed water, and comprises the following steps: constructing a river water supply digital twin body fused with human interference to realize full-factor real-time synchronous simulation and prediction of the reclaimed water supply process; the application adopts a scheme of constructing a full-factor digital twin body fused with a human interference quantitative module, adds quantitative calculation units of high-frequency human interference such as ship navigation, coastal water intake and recreational fishing on the basis of integrating a hydrodynamic water quality and water ecological coupling model, establishes a real-time data mapping channel of the physical river and the virtual twin body, and drives the dynamic evolution of the twin body through Internet of Things sensing data, so that the application has the advantages of being capable of accurately reproducing the real operation state of the river ecological system and greatly improving the simulation and prediction precision of the reclaimed water supply process, and provides reliable basic support for subsequent risk assessment and regulation decision.

Description

Technical Field

[0001] This invention relates to the field of environmental protection technology, and in particular to a method and system for ecological risk assessment and joint regulation of rivers replenished by reclaimed water. Background Technology

[0002] Reclaimed water, as a stable unconventional water resource for cities, has become an important way to alleviate water shortages and restore damaged river ecosystems globally through large-scale river replenishment. However, with the continuous expansion of reclaimed water replenishment, the potential ecological risks it may bring are becoming increasingly prominent. How to scientifically assess and precisely regulate the reclaimed water replenishment process has become a critical issue that urgently needs to be addressed in the field of water environment ecological protection.

[0003] Existing technologies for ecological risk assessment and regulation of rivers replenished by reclaimed water still have significant shortcomings: First, traditional hydrological and water quality simulations are mostly based on ideal assumptions and fail to systematically quantify the impact of high-frequency human disturbances such as boat navigation, water intake along the banks, and recreational fishing, resulting in significant discrepancies between simulation results and reality; Second, risk assessments rely excessively on conventional water quality indicators, failing to identify irreversible abrupt changes in the ecosystem from quantitative to qualitative transformation, leading to severely delayed early warnings; Third, the use of uniform water quality standards and replenishment thresholds fails to consider the differences between different river types and ecological protection goals, resulting in both insufficient protection and resource waste; Fourth, the regulation model is mainly reactive and lacks forward-looking prediction, making it difficult to proactively prevent ecological risks.

[0004] The aforementioned deficiencies severely restrict the ecological security of reclaimed water replenishing rivers and fail to meet the current needs for the coordinated development of efficient water resource utilization and water ecosystem protection. Summary of the Invention

[0005] To overcome the problems mentioned in the background art, the present invention proposes a method and system for ecological risk assessment and joint regulation of reclaimed water replenishing rivers.

[0006] To achieve the above-mentioned technical effects, the present invention is implemented through the following technical solution: a method for ecological risk assessment and joint regulation of rivers replenished by reclaimed water, characterized by comprising the following steps:

[0007] S1. Construct a digital twin of river reclaimed water replenishment that integrates human interference, to achieve real-time synchronous simulation and prediction of all elements of the reclaimed water replenishment process, as detailed below:

[0008] S1.1 Collect basic geographic data, historical hydrological and water quality data, water ecological survey data, and effluent water quality and water supply capacity data of the target river to construct a three-dimensional geographic information base for the river.

[0009] S1.2 Integrate a hydrodynamic module, a water quality transport module, and a water ecological evolution module on the three-dimensional geographic information base, and add a human interference quantification module;

[0010] S1.3 Establish a real-time data mapping channel between the digital twin and the physical river, and drive the dynamic update of the twin through real-time data obtained by IoT sensing devices;

[0011] S2. Based on the ecological critical point theory, a mutation risk prediction model is constructed by integrating multi-source heterogeneous ecological signals. This model outputs the distance of the ecosystem from the critical point and the probability of mutation, and issues an ultra-early warning. Details are as follows:

[0012] S2.1 Construct a multi-source heterogeneous index system that includes hydrological situation variability signals, key species behavior anomaly signals, and microbial network complexity signals;

[0013] S2.2 Based on historical ecological mutation event data, extract the critical slowdown features before the mutation occurs, and construct a multimodal machine learning mutation risk prediction model;

[0014] S2.3. Real-time input of multi-source heterogeneous monitoring data, model output of the current distance of the ecosystem from the critical point and the probability of mutation, and issuance of different levels of early warning signals when the probability exceeds the set threshold;

[0015] S3. For different ecological functional units and protection objectives, determine differentiated and tiered ecological security thresholds for key risk factors in reclaimed water; specifically as follows:

[0016] S3.1. Ecological zoning of the target watershed: Based on river type, hydrological characteristics, ecosystem structure and main ecological protection objectives, the river is divided into different ecological functional units.

[0017] S3.2. For each ecological functional unit, conduct indoor control experiments and in-situ field observations to determine the concentration of key risk factors in reclaimed water that have no visible effect on key protected species, the minimum visible effect concentration, and the effect threshold at the ecosystem level.

[0018] S3.3 Based on the results of the ecosystem resilience assessment, the threshold of each risk factor is divided into three levels: safe level, warning level, and dangerous level.

[0019] S3.4 Establish the response relationship between different reclaimed water replenishment ratios and risk factor concentrations, and form differentiated graded control standards for reclaimed water replenishment for different ecological functional units.

[0020] S4. Combining meteorological and hydrological forecasts with risk prediction models, generate multi-scenario optimized scheduling plans and update them dynamically to achieve proactive joint control of reclaimed water replenishment; details are as follows:

[0021] S4.1 Obtain meteorological and hydrological forecast data for the next 7-30 days and input them into the digital twin;

[0022] S4.2 Generate multiple reclaimed water replenishment scenarios that include different combinations of replenishment flow rate, replenishment time, replenishment inlet location, and replenishment water quality level;

[0023] S4.3 For each scenario, simulate the changes in hydrology and water quality, water ecological response and sudden change risk level at different future time periods, and evaluate the ecological benefits, water resource utilization efficiency and economic costs of each scenario.

[0024] S4.4. Use a multi-objective optimization algorithm to obtain the Pareto optimal solution set and generate multiple optimized scheduling plans;

[0025] S4.5 Establish a dynamic adjustment mechanism for the contingency plan. Based on real-time monitoring data and updated forecast results, re-simulate and optimize the dispatch plan every 24 hours.

[0026] Furthermore, in S1.2, the artificial interference quantification module includes:

[0027] Vessel Disturbance Quantification Unit: Based on AIS Automatic Identification System data, this unit quantifies the intensity of water disturbance, the amount of suspended solids resuspended, and the degree of damage to benthic habitats of vessels of different tonnages and speeds.

[0028] Water intake disturbance quantification unit: integrates real-time water intake data from industrial, agricultural, and domestic water intakes along the coast to simulate the effects of water intake on flow rate changes, velocity field alterations, and reduced pollutant dilution capacity.

[0029] Fishing Disturbance Quantification Unit: Through video monitoring and big data statistics, the fishing density, bait placement, and catch intensity in different areas are quantified to assess their impact on fish population structure and food web.

[0030] Furthermore, in S2.1, the hydrological situation variability signal includes the flow variation coefficient, flood-dry ratio, pulse frequency, and duration of low flow; the key species behavior anomaly signal includes the activity range, feeding frequency, gregarious behavior, and escape response of fish; and the microbial network complexity signal includes the number of nodes, connectivity, modularity index, and average path length of the microbial co-occurrence network.

[0031] Furthermore, in S2.2, the training process of the multimodal machine learning mutation risk prediction model is as follows: collect data on all ecological mutation events that have occurred in the target river in history, extract critical slowdown features within 6 months before the mutation occurs, including increased variance, increased autocorrelation coefficient, and decreased resilience, and use the Transformer multimodal fusion architecture to fuse and train time-series hydrological data, video image data, and microbial sequencing data.

[0032] Furthermore, in S3.3, based on the results of the ecosystem resilience assessment, the threshold for each risk factor is divided into three levels: safe, warning, and dangerous, as detailed below:

[0033] Safety level: Mutation probability <30%, stable ecosystem structure and function, no obvious risk;

[0034] Warning level: 30% ≤ mutation probability < 70%, the ecosystem is slightly disturbed, key species begin to behave abnormally, and there is a potential risk of mutation;

[0035] Danger level: Mutation probability ≥70%, the ecosystem is close to the critical point and irreversible mutation may occur at any time.

[0036] Furthermore, in S4.4, the objective function of the multi-objective optimization algorithm is: to minimize ecological risk, maximize water resource utilization efficiency, and minimize regulation cost as optimization objectives, to obtain a Pareto optimal solution set, and to select and generate multiple executable scheduling plans from the optimal solution set according to different priority requirements.

[0037] Another objective of this invention is to provide an ecological risk assessment and joint regulation system for reclaimed water-recharged rivers, characterized in that the system comprises:

[0038] Multi-source sensing layer: Composed of hydrological and water quality sensors, underwater acoustic monitoring equipment, underwater high-definition cameras, AIS ship receivers, water intake metering devices, video monitoring equipment, and microbial sampling and analysis units, it is responsible for real-time collection of all elements of river data;

[0039] Digital twin layer: Deployed on a cloud server, it includes a 3D geographic information base, a hydrodynamic-water quality-water ecology coupling model, and a human disturbance quantification module to achieve high-precision simulation and prediction of the reclaimed water replenishment process;

[0040] Risk assessment and early warning layer: It has a built-in multimodal machine learning mutation risk prediction model and a differentiated graded ecological security threshold library, which is responsible for assessing the ecological risk level in real time and issuing ultra-early warnings;

[0041] Optimized scheduling layer: Integrates meteorological and hydrological forecasting interfaces and multi-objective optimization algorithms, and is responsible for generating multi-scenario optimized scheduling plans and updating them dynamically;

[0042] Decision support layer: Provides a visual display interface to present digital twin simulation results, risk warning information, and scheduling plans to managers in an intuitive way, and supports one-click issuance and execution tracking of plans.

[0043] The beneficial effects of this invention are:

[0044] This invention constructs a full-element digital twin that integrates a module for quantifying human interference. Based on the integrated hydrodynamic, water quality, and water ecology coupling model, it adds a quantitative calculation unit for high-frequency human interference such as boat navigation, water intake along the riverbank, and recreational fishing. It establishes a real-time data mapping channel between the physical river and the virtual twin, and drives the dynamic evolution of the twin through IoT sensing data. It has the advantages of accurately reproducing the real operating state of the river ecosystem and significantly improving the simulation and prediction accuracy of the reclaimed water replenishment process, providing a reliable foundation for subsequent risk assessment and regulation decisions.

[0045] This invention introduces the theory of ecosystem tipping points and integrates multi-dimensional ecological signals to construct a mutation risk prediction model. This approach breaks through the limitations of traditional water quality indicators, integrates multi-source heterogeneous information such as hydrological situation variability, key species behavioral anomalies, and microbial network complexity, extracts critical slowing characteristics when the ecosystem approaches its tipping point, and uses multimodal machine learning algorithms to train the prediction model. It has the advantages of issuing ultra-early warnings before irreversible mutations occur in the ecosystem and significantly extending the risk response time window, fundamentally changing the traditional post-event remedial ecological risk management model.

[0046] This invention establishes a differentiated hierarchical ecological security threshold system based on ecological functional zoning. First, the watershed is divided into different ecological functional units according to river characteristics and protection objectives. Then, through a combination of indoor control experiments and in-situ field observations, the species-level and ecosystem-level effect thresholds of key risk factors in each unit are determined. Combined with the results of ecosystem resilience assessment, three levels of safety warning danger are classified, and a quantitative response relationship between water replenishment ratio and risk factor concentration is established. This invention has the advantages of providing precise regulation basis for different regions, maximizing water resource utilization efficiency while ensuring ecological security, and achieving synergistic optimization of ecological protection and water resource utilization.

[0047] This invention combines meteorological and hydrological forecasting with risk prediction models to proactively optimize scheduling schemes across multiple scenarios. It obtains meteorological and hydrological data for a period of time in advance, inputs it into a digital twin to simulate the ecological response under different water replenishment schemes, and uses a multi-objective optimization algorithm to solve for the Pareto optimal solution set. This generates multiple executable scheduling plans and updates them dynamically based on real-time data. It has the advantage of being able to identify potential ecological risks in advance and take proactive intervention measures, achieving a fundamental shift from passive response to proactive prevention and control, and significantly improving the stability and safety of the reclaimed water replenishment river ecosystem. Attached Figure Description

[0048] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0049] Figure 1 This is a flowchart illustrating the ecological risk assessment and joint regulation method for reclaimed water replenishing rivers according to the present invention.

[0050] Figure 2 This is a schematic diagram of the ecological risk assessment and joint regulation system for reclaimed water replenishing rivers according to the present invention.

[0051] Figure 3 This is a schematic diagram of the digital twin structure of river reclaimed water replenishment that incorporates human interference, as described in the ecological risk assessment and joint regulation method for reclaimed water replenishment rivers of the present invention. Detailed Implementation

[0052] 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.

[0053] Example 1

[0054] This embodiment describes the ecological risk assessment and joint regulation method for rivers replenished by reclaimed water, as detailed below:

[0055] The ecological risk assessment and joint regulation method for rivers replenished by reclaimed water is applicable to all urban rivers, plain rivers and some mountain rivers that use reclaimed water as their main source of replenishment.

[0056] The basic hardware environment required for implementing the ecological risk assessment and joint regulation method for reclaimed water replenishment rivers includes: cloud server clusters, IoT sensing devices, data transmission networks, and field execution terminals; the basic software environment includes: Linux operating system, Python 3.8+, geographic information system software, numerical simulation software, and machine learning framework.

[0057] like Figure 1 As shown, the specific process of the ecological risk assessment and joint regulation method for reclaimed water replenishing rivers in this invention is as follows:

[0058] S1. Construct a digital twin of river reclaimed water replenishment that integrates human-induced interference, as detailed below:

[0059] S1.1 Constructing a 3D Geographic Information Base for Rivers

[0060] Collect basic geographic data for the entire basin of the target river, including: digital elevation model (DEM) data with a resolution of 5m, river topographic map at a scale of 1:2000, river cross-sectional data measured every 50-100m, and coordinate and parameter data of levee engineering and hydraulic structures (gates, pumping stations, and water inlets); collect historical hydrological and water quality data, including daily flow, water level, flow velocity, water temperature, pH, dissolved oxygen, COD, ammonia nitrogen, total phosphorus, and total nitrogen data for the past 10 years; collect aquatic ecological survey data, including community composition, abundance, and distribution data of fish, benthic organisms, and plankton; and collect data on reclaimed water plants, including daily water supply, effluent water quality indicators, treatment process parameters, and maximum water supply capacity for the past 3 years.

[0061] The above data was standardized and spatially registered. A 1:1 scale 3D geographic information base for the river was constructed using the Cesium open-source 3D geographic information engine, enabling 3D visualization of river topography, hydraulic structures, and riverside land use. The base supports basic geographic functions such as spatial query, distance measurement, and profile analysis, and reserves data interfaces for subsequent model integration and real-time data access.

[0062] S1.2, Integrated Multiphysics Coupling Model and Artificial Interference Quantification Module

[0063] like Figure 3 As shown, a hydrodynamic-water quality-water ecology coupled model is integrated on a three-dimensional geographic information base:

[0064] The hydrodynamic module uses the MIKE21 two-dimensional hydrodynamic model, with a time step of 1 minute, a spatial grid resolution of 5-10m, and boundary conditions including upstream inflow, downstream water level, and reclaimed water replenishment flow.

[0065] The water transport module uses the WASP water quality model to simulate the convection diffusion, degradation and transformation processes of 12 pollutants, including COD, ammonia nitrogen, total phosphorus, total nitrogen, and heavy metals.

[0066] The aquatic ecosystem evolution module uses the AQUATOX model to simulate the nutritional relationships and population dynamics among phytoplankton, zooplankton, benthic organisms, and fish.

[0067] A new artificial interference quantization module has been added, which contains three independent quantization units:

[0068] Vessel Disturbance Quantification Unit: Based on real-time data of vessel tonnage, speed, position, and heading collected by the AIS Automatic Identification System, the intensity of vessel disturbance is quantified using the following formula:

[0069] ;

[0070] In the formula, For the intensity of ship disturbance, This is the disturbance coefficient (typically 0.001-0.005, related to river depth). For ship tonnage, The speed of the vessel is used as the reference. The amount of suspended solids resuspended and the area of ​​benthic habitat destruction are calculated based on the disturbance intensity. When the disturbance intensity exceeds the threshold (typically 500), the benthic mortality rate is calculated at a rate of 5% for every 100 increase in disturbance intensity.

[0071] Water Intake Disturbance Quantification Unit: Integrates real-time metering data from all industrial, agricultural, and domestic water intakes along the coast to perform real-time corrections to the flow field of the hydrodynamic model. For each water intake, based on its water intake volume... Based on the location of the water intake, the corresponding flow rate is deducted at the corresponding grid nodes, and the change in the velocity field is calculated. At the same time, the pollutant dilution coefficient is corrected according to the flow rate change. When the water intake causes a reduction in the river section flow rate of more than 20%, the pollutant dilution capacity decreases by 1.2 times the proportion of flow rate reduction.

[0072] Fishing Disturbance Quantification Unit: High-definition video monitoring equipment is installed at major fishing spots, and the YOLOv8 target detection algorithm is used to identify the number of anglers, fishing rods, and baiting behavior in real time. A quantitative relationship is established between fishing density, catch intensity, and baiting amount through big data statistics: Catch intensity = 0.3 × number of anglers (person / 100m), baiting amount = 0.5kg / (person·day). The baiting amount is converted into nitrogen and phosphorus inputs (nitrogen content approximately 5%, phosphorus content approximately 1%), and input into the water quality model; the catch intensity is converted into fish mortality rate, and input into the aquatic ecosystem model.

[0073] S1.3 Establish a real-time data mapping and dynamic update mechanism

[0074] The project involves deploying IoT sensing devices along the target riverbanks, including: installing one integrated hydrological and water quality sensor every 2-3 km to monitor flow rate, water level, flow velocity, water temperature, pH, dissolved oxygen, COD, ammonia nitrogen, total phosphorus, and total nitrogen; deploying underwater acoustic monitoring equipment (ADCP + fish finder) and underwater high-definition cameras in areas with concentrated fish populations; deploying AIS ship receivers in navigable sections; installing smart metering devices at all water intakes; and installing video surveillance equipment at major fishing spots.

[0075] All sensing devices transmit real-time data to the cloud server via 4G / 5G networks using the MQTT protocol, with a data transmission frequency of once every 15 minutes. A two-way data mapping channel is established between the digital twin and the physical river. Real-time data automatically drives the operation of the hydrodynamic-water quality-aquatic ecology coupled model, updating the state of the digital twin and achieving synchronous evolution between the physical river and the virtual twin. When the deviation between the real-time monitoring data and the model simulation values ​​exceeds 15%, the model parameter calibration process is automatically triggered, and the model parameters are dynamically optimized using the Kalman filter algorithm.

[0076] S2. Ultra-early warning based on multi-source heterogeneous signal fusion according to ecological critical point theory, as detailed below:

[0077] S2.1 Constructing a multi-source heterogeneous ecological mutation risk indicator system

[0078] Construct a 26-dimensional indicator system containing three core signals:

[0079] Hydrological variability signals (12 dimensions): including daily flow variation coefficient, monthly flow variation coefficient, annual flood-dry ratio, flood pulse frequency, duration of low flow, duration of high flow, flow change rate, baseflow index, runoff depth variation coefficient, velocity variation coefficient, water level variation coefficient, and water temperature variation coefficient. All indicators are calculated using the sliding window method, with a window size of 30 days and a step size of 1 day.

[0080] Key Species Behavioral Anomalies (8-Dimensional): Data was collected using underwater acoustic monitoring equipment and underwater high-definition cameras. Target detection and tracking algorithms were employed to extract fish activity range, average swimming speed, feeding frequency, school density, school size, vertical migration frequency, escape reaction time, and diurnal activity rhythm deviation. Escape reaction time was defined as the average time it takes for a fish to begin escaping after being stimulated by external factors, and diurnal activity rhythm deviation was defined as the cosine similarity between the current activity rhythm and the natural rhythm.

[0081] Microbial network complexity signal (6 dimensions): River water and sediment samples were collected every 15 days, and microbial community composition data were obtained using 16S rRNA high-throughput sequencing technology. A microbial co-occurrence network was constructed based on species abundance data, and network topological features, including the number of nodes, number of edges, average connectivity, modularity index, average path length, and clustering coefficient, were calculated using the igraph package in R.

[0082] S2.2 Constructing a multimodal machine learning mutation risk prediction model

[0083] Historical data from the target river over the past 10 years were collected, and all ecological mutation events that have occurred were identified (such as algal blooms, mass mortality of benthic organisms, replacement of dominant species, and sudden drop in dissolved oxygen). For each mutation event, all indicator data within the 6 months prior to the mutation were extracted and labeled as "pre-mutation sample"; an equal amount of data from the period before the mutation occurred was extracted and labeled as "normal sample".

[0084] Critical slowing features were extracted from all samples, including: the sliding variance, sliding autocorrelation coefficient (lag 1), and system resilience index (calculated through impulse response experiments) for each indicator. The original indicator data and critical slowing features were combined as model input.

[0085] A mutation risk prediction model is constructed using the Transformer multimodal fusion architecture:

[0086] Input layer: It is divided into three branches, which respectively receive 12-dimensional hydrological time series data, 8-dimensional fish behavior time series data and 6-dimensional microbial network feature data;

[0087] Feature extraction layer: Each branch uses an independent Transformer encoder to extract features, with 3 encoder layers and 4 attention heads;

[0088] Fusion layer: Employs an attention mechanism to fuse the features from the three branches to obtain a global feature vector;

[0089] Output layer: Using the Sigmoid activation function, outputs the probability of a mutation occurring in the ecosystem within the next 30 days.

[0090] The dataset was divided into training and test sets in an 8:2 ratio. The Adam optimizer was used for training with a learning rate of 0.001, a batch size of 32, and 100 training epochs. An early stopping mechanism was employed to prevent overfitting; training was stopped when the validation set loss no longer decreased after 10 consecutive epochs. After training, the model achieved an accuracy ≥90% and a recall ≥85% on the test set.

[0091] S2.3 Real-time risk assessment and ultra-early warning

[0092] The multi-source heterogeneous data collected in real time is input into the trained mutation risk prediction model. The model outputs the distance of the current ecosystem from the critical point (normalized value, ranging from 0 to 1, the smaller the value, the closer to the critical point) and the probability of mutation occurring in the next 30 days once every hour.

[0093] Three warning levels are defined based on the probability of mutation:

[0094] Safety level: Mutation probability <30%, stable ecosystem structure and function, no obvious risk;

[0095] Warning level: 30% ≤ mutation probability < 70%, the ecosystem is slightly disturbed, key species begin to behave abnormally, and there is a potential risk of mutation;

[0096] Danger level: Mutation probability ≥70%, the ecosystem is close to the critical point and irreversible mutation may occur at any time.

[0097] When the warning level reaches the warning level or above, the system will automatically send a warning signal to the management personnel via SMS, email, system pop-up window, etc., and display the location of the river section where the risk occurs, the main risk factors, and the preliminary measures to be taken.

[0098] S3. Determination of Differentiated and Graded Ecological Security Thresholds

[0099] S3.1, Ecological Function Zoning of the Target Watershed

[0100] The target watershed was ecologically zoned using GIS spatial overlay analysis. The zoning criteria included: river type (mountain rivers, plain rivers, urban rivers), hydrological characteristics (annual runoff, flow variability, flow velocity), ecosystem structure (community composition, biodiversity), major ecological protection targets (rare and endangered species, native species, drinking water sources), and riparian land use types.

[0101] The target river is divided into several independent ecological functional units, typical of which include:

[0102] Upstream ecological protection zone: The main goal is to protect rare and endangered fish and native biodiversity, with high water quality requirements and minimal human interference;

[0103] Midstream scenic and recreational area: The main goal is to provide scenic and recreational functions. The water should be clear, without obvious odor, and free from algal blooms.

[0104] Downstream industrial water use areas: The main objective is to meet industrial water demand, and the water quality requirements are relatively low, but it is necessary to prevent pollution from heavy metals and toxic and harmful substances.

[0105] S3.2 Determination of Thresholds for the Effect of Key Risk Factors

[0106] For each ecological functional unit, key risk factors in reclaimed water were screened, including conventional pollutants (COD, ammonia nitrogen, total phosphorus, total nitrogen), heavy metals (cadmium, lead, mercury, chromium), emerging pollutants (antibiotics, microplastics, endocrine disruptors) and salinity.

[0107] The effect threshold was determined by combining indoor controlled experiments with in-situ field observations.

[0108] Indoor controlled experiments: Key protected species of this ecological functional unit were selected as experimental organisms, and exposure experiments were set up with 5-7 concentration gradients, with 3 parallel samples for each gradient. The survival rate, growth rate, reproduction rate, and behavioral abnormalities of the experimental organisms were observed, and the No-Effect Concentration (NOEC) and Lowest-Effect Concentration (LOEC) were calculated using the probability unit method.

[0109] In-situ field observation: Multiple in-situ observation points are set up along the target river to continuously monitor risk factor concentrations and ecosystem structure changes for 1-2 years. A piecewise regression method is used to determine the ecosystem-level effect threshold; that is, when the risk factor concentration exceeds this threshold, a significant change in the ecosystem community structure occurs.

[0110] S3.3, Classification of Hierarchical Ecological Security Thresholds

[0111] Based on the results of the ecosystem resilience assessment, the threshold for each risk factor is divided into three levels:

[0112] Safety threshold: 0.6 times NOEC, at which point the ecosystem structure and function are stable, risk factors have no significant adverse effects on organisms, and the ecosystem has strong resilience;

[0113] Warning threshold: The NOEC value is used. At this point, the ecosystem is slightly disturbed, the physiological indicators of key species begin to change, but there is no obvious population decline yet, and the ecosystem can still recover through self-regulation.

[0114] Hazard threshold: The LOEC value is used. At this point, the ecosystem is close to the critical point, the behavior of key species begins to be abnormal, and the population shows a downward trend. If intervention measures are not taken in time, irreversible mutations will occur.

[0115] S3.4 Establish the response relationship between water replenishment ratio and risk factor concentration.

[0116] Based on simulation results from digital twins, a quantitative response relationship was established between different reclaimed water replenishment ratios and the concentrations of risk factors in various ecological functional units. A nonlinear regression model was used to fit the data, yielding a response function of the following form:

[0117] ;

[0118] In the formula, For the first The concentration of each risk factor The ratio of reclaimed water to replenishment water, The concentration of this risk factor in the reclaimed water. This represents the concentration of the risk factor in the upstream water.

[0119] Based on the response function and the graded ecological security threshold, the maximum allowable water replenishment ratio for each ecological functional unit under different reclaimed water effluent quality conditions is calculated, forming a differentiated graded control standard for reclaimed water replenishment for different ecological functional units.

[0120] S4. Multi-scenario proactive optimization scheduling and joint control

[0121] S4.1 Obtain meteorological and hydrological forecast data

[0122] Meteorological and hydrological forecasts for the next 7-30 days are obtained through the public API interfaces of the National Meteorological Information Center and local hydrological bureaus. Meteorological forecasts include hourly rainfall, temperature, evaporation, wind speed, and wind direction; hydrological forecasts include daily upstream inflow, tributary inflow, and downstream water levels. Data is updated daily, and every 6 hours in the event of significant weather changes.

[0123] The acquired forecast data is preprocessed, including data cleaning, outlier removal, and spatiotemporal interpolation, and converted into a format recognizable by the digital twin. This data is then input into the digital twin as the boundary conditions for the simulation.

[0124] S4.2, Generate multi-dimensional reclaimed water replenishment scenarios

[0125] Based on the water supply capacity of the reclaimed water plant, the distribution of water replenishment points, and the current ecological risk status, multi-dimensional reclaimed water replenishment scenarios are generated. Each scenario includes different combinations of the following adjustable parameters:

[0126] Makeup water flow rate: ranging from 0 to the maximum water supply capacity of the reclaimed water plant, with a step size of 0.1 m³ / s;

[0127] Water replenishment time: including different modes such as continuous water replenishment, intermittent water replenishment, and staggered water replenishment;

[0128] Water inlet location: including single water inlet water supply, combined water supply from multiple water inlets, and flow distribution from different water inlets;

[0129] Water quality grades for replenishment water include different reclaimed water effluent quality grades such as Class A standard, near Class IV standard, and near Class III standard.

[0130] An orthogonal experimental design method is used to generate a representative set of water replenishment scenarios, ensuring coverage of all possible parameter combinations while avoiding excessive computation due to an excessive number of scenarios. A typical scenario set contains 10-20 different water replenishment scenarios.

[0131] S4.3 Scenario Simulation and Multi-Objective Assessment

[0132] Each water replenishment scenario is input into a digital twin to simulate the changes in river hydrology and water quality, aquatic ecological response, and abrupt change risk levels over the next 7-30 days. The simulation time step is 1 hour, and the output includes hourly hydrological and water quality data for each grid node and daily ecological risk index for each ecological functional unit.

[0133] Each scenario is comprehensively evaluated from three dimensions:

[0134] Ecological benefits: The Ecological Risk Index (ERI) is used for assessment. ERI is the weighted average of the probability of mutation risk of each ecological functional unit, and the weights are determined according to the importance of the ecological functional unit.

[0135] Water resource utilization efficiency: The ecological improvement effect per unit of water replenishment is evaluated, namely the ratio of the reduction value of the ecological risk index to the amount of water replenishment.

[0136] Economic costs include the cost of reclaimed water treatment, water transmission, and gate scheduling, and are calculated based on the quality grade of the reclaimed water effluent and the amount of water replenished.

[0137] S4.4 Multi-objective optimization solution and scheduling plan generation

[0138] The NSGA-II multi-objective optimization algorithm was adopted, with the objective functions of "minimizing ecological risk, maximizing water resource utilization efficiency, and minimizing regulation costs" to obtain the Pareto optimal solution set. The algorithm parameters were set as follows: population size of 100, number of iterations of 200, crossover probability of 0.9, and mutation probability of 0.1.

[0139] Based on different priority requirements, 3-5 executable optimization scheduling plans are generated from the Pareto optimal solution set. Each plan contains detailed scheduling instructions, including: water supply flow rate, water supply start time, water supply duration, reclaimed water effluent quality requirements, and gate opening and closing time and degree.

[0140] S4.5 Dynamic Adjustment and Implementation Tracking of Contingency Plans

[0141] A dynamic adjustment mechanism for the contingency plan is established. Every 24 hours, the system re-runs the digital twin simulation based on the latest real-time monitoring data and updated meteorological and hydrological forecast data, and re-solves the multi-objective optimization problem to generate an updated dispatch plan. When the deviation between the real-time monitoring data and the simulated values ​​exceeds 20%, or when the ecological risk level changes, the contingency plan adjustment process is immediately triggered.

[0142] After the dispatch plan is confirmed by management personnel, it is issued to the reclaimed water plant, water supply gate, and relevant execution terminals through the system with one click. The system tracks the execution of the plan in real time, compares the deviation between the actual execution parameters and the plan parameters, and issues an alarm when the deviation exceeds 10%, ensuring that the dispatch instructions are executed accurately.

[0143] Example 2

[0144] Based on the aforementioned method for ecological risk assessment and joint regulation of rivers replenished by reclaimed water, an ecological risk assessment and joint regulation system for rivers replenished by reclaimed water is constructed as follows:

[0145] like Figure 2 As shown, the system adopts a layered architecture design, and the specific implementation of each layer is as follows:

[0146] Multi-source sensing layer: Composed of various IoT sensing devices deployed along the riverbanks, including hydrological and water quality sensors, underwater acoustic monitoring equipment, underwater high-definition cameras, AIS ship receivers, water intake metering devices, video surveillance equipment, and microbial sampling and analysis units. All devices are uniformly connected to the IoT platform to achieve unified data collection and transmission.

[0147] Digital twin layer: Deployed on a cloud server cluster, it includes a 3D geographic information base, a hydrodynamic-water quality-aquatic ecology coupled model, and a module for quantifying human disturbance. It is deployed using containerization technology, supporting parallel computing and dynamic scaling of the model to ensure real-time simulation.

[0148] Risk assessment and early warning layer: Includes a pre-trained multimodal machine learning mutation risk prediction model and a differentiated, tiered ecological security threshold library. Employs real-time stream computing technology to process multi-source heterogeneous data, enabling real-time assessment and ultra-early warning of ecological risks.

[0149] Optimized scheduling layer: Integrates meteorological and hydrological forecasting interfaces and the NSGA-II multi-objective optimization algorithm to automatically generate multi-scenario optimized scheduling plans and dynamically update them based on real-time data. Supports automatic plan generation, manual modification, and version management.

[0150] Decision Support Layer: Developed using web technology, this layer features a visual interface that provides functions such as digital twin 3D visualization, risk warning visualization, and dispatch plan visualization. It allows administrators to access the system remotely via computers, tablets, mobile phones, and other terminals, enabling remote decision-making and command.

[0151] The embodiments of the present invention have been described in detail above with reference to the accompanying drawings. However, the present invention is not limited to the above embodiments. Within the scope of knowledge possessed by those skilled in the art, various changes can be made without departing from the spirit of the present invention.

Claims

1. A method for ecological risk assessment and joint regulation of rivers replenished by reclaimed water, characterized in that, Includes the following steps: S1. Construct a digital twin of river reclaimed water replenishment that integrates human interference, to achieve real-time synchronous simulation and prediction of all elements of the reclaimed water replenishment process, as detailed below: S1.1 Collect basic geographic data, historical hydrological and water quality data, water ecological survey data, and effluent water quality and water supply capacity data of the target river to construct a three-dimensional geographic information base for the river. S1.2 Integrate a hydrodynamic module, a water quality transport module, and a water ecological evolution module on the three-dimensional geographic information base, and add a human interference quantification module; S1.3 Establish a real-time data mapping channel between the digital twin and the physical river, and drive the dynamic update of the twin through real-time data obtained by IoT sensing devices; S2. Based on the ecological critical point theory, a mutation risk prediction model is constructed by integrating multi-source heterogeneous ecological signals. This model outputs the distance of the ecosystem from the critical point and the probability of mutation, and issues an ultra-early warning. Details are as follows: S2.1 Construct a multi-source heterogeneous index system that includes hydrological situation variability signals, key species behavior anomaly signals, and microbial network complexity signals; S2.2 Based on historical ecological mutation event data, extract the critical slowdown features before the mutation occurs, and construct a multimodal machine learning mutation risk prediction model; S2.

3. Real-time input of multi-source heterogeneous monitoring data, model output of the current distance of the ecosystem from the critical point and the probability of mutation, and issuance of different levels of early warning signals when the probability exceeds the set threshold; S3. For different ecological functional units and protection objectives, determine differentiated and tiered ecological security thresholds for key risk factors in reclaimed water; specifically as follows: S3.

1. Ecological zoning of the target watershed: Based on river type, hydrological characteristics, ecosystem structure and main ecological protection objectives, the river is divided into different ecological functional units. S3.

2. For each ecological functional unit, conduct indoor control experiments and in-situ field observations to determine the concentration of key risk factors in reclaimed water that have no visible effect on key protected species, the minimum visible effect concentration, and the effect threshold at the ecosystem level. S3.3 Based on the results of the ecosystem resilience assessment, the threshold of each risk factor is divided into three levels: safe level, warning level, and dangerous level. S3.4 Establish the response relationship between different reclaimed water replenishment ratios and risk factor concentrations, and form differentiated graded control standards for reclaimed water replenishment for different ecological functional units; S4. Combining meteorological and hydrological forecasts with risk prediction models, generate multi-scenario optimized scheduling plans and update them dynamically to achieve proactive joint control of reclaimed water replenishment; details are as follows: S4.1 Obtain meteorological and hydrological forecast data for the next 7-30 days and input them into the digital twin; S4.2 Generate multiple reclaimed water replenishment scenarios that include different combinations of replenishment flow rate, replenishment time, replenishment inlet location, and replenishment water quality level; S4.3 For each scenario, simulate the changes in hydrology and water quality, water ecological response and sudden change risk level at different future time periods, and evaluate the ecological benefits, water resource utilization efficiency and economic costs of each scenario. S4.

4. Use a multi-objective optimization algorithm to obtain the Pareto optimal solution set and generate multiple optimized scheduling plans; S4.5 Establish a dynamic adjustment mechanism for the contingency plan. Based on real-time monitoring data and updated forecast results, re-simulate and optimize the dispatch plan every 24 hours.

2. The method for ecological risk assessment and joint regulation of rivers replenished by reclaimed water according to claim 1, characterized in that, In S1.2, the artificial interference quantification module includes: Vessel Disturbance Quantification Unit: Based on AIS Automatic Identification System data, this unit quantifies the intensity of water disturbance, the amount of suspended solids resuspended, and the degree of damage to benthic habitats of vessels of different tonnages and speeds. Water intake disturbance quantification unit: integrates real-time water intake data from industrial, agricultural, and domestic water intakes along the coast to simulate the effects of water intake on flow rate changes, velocity field alterations, and reduced pollutant dilution capacity. Fishing Disturbance Quantification Unit: Through video monitoring and big data statistics, the fishing density, bait placement, and catch intensity in different areas are quantified to assess their impact on fish population structure and food web.

3. The method for ecological risk assessment and joint regulation of rivers replenished by reclaimed water according to claim 1, characterized in that, In S2.1, The hydrological situation variability signal includes the flow variation coefficient, flood-dry ratio, pulse frequency, and duration of low flow. The key species behavioral abnormality signals include the fish's activity range, feeding frequency, schooling behavior, and escape response; The microbial network complexity signal includes the number of nodes, connectivity, modularity index, and average path length of the microbial co-occurrence network.

4. The method for ecological risk assessment and joint regulation of rivers replenished by reclaimed water according to claim 1, characterized in that, In S2.2, the training process of the multimodal machine learning mutation risk prediction model is as follows: collect data on all ecological mutation events that have occurred in the target river in history, extract critical slowdown features within 6 months before the mutation occurs, including increased variance, increased autocorrelation coefficient, and decreased resilience, and use the Transformer multimodal fusion architecture to fuse and train time-series hydrological data, video image data, and microbial sequencing data.

5. The method for ecological risk assessment and joint regulation of rivers replenished by reclaimed water according to claim 1, characterized in that, In S3.3, based on the results of the ecosystem resilience assessment, the threshold of each risk factor is divided into three levels: safe, warning, and dangerous, as detailed below: Safety level: Mutation probability <30%, stable ecosystem structure and function, no obvious risk; Warning level: 30% ≤ mutation probability < 70%, the ecosystem is slightly disturbed, key species begin to behave abnormally, and there is a potential risk of mutation; Danger level: Mutation probability ≥70%, the ecosystem is close to the critical point and irreversible mutation may occur at any time.

6. The method for ecological risk assessment and joint regulation of rivers replenished by reclaimed water according to claim 1, characterized in that, In S4.4, the objective function of the multi-objective optimization algorithm is: to minimize ecological risk, maximize water resource utilization efficiency, and minimize regulation cost as optimization objectives, to obtain a Pareto optimal solution set, and to select and generate multiple executable scheduling plans from the optimal solution set according to different priority requirements.

7. An ecological risk assessment and joint regulation system for reclaimed water replenishing rivers, used to implement the method described in any one of claims 1-6, characterized in that the system... include: Multi-source sensing layer: Composed of hydrological and water quality sensors, underwater acoustic monitoring equipment, underwater high-definition cameras, AIS ship receivers, water intake metering devices, video monitoring equipment, and microbial sampling and analysis units, it is responsible for real-time collection of all elements of river data; Digital twin layer: Deployed on a cloud server, it includes a 3D geographic information base, a hydrodynamic-water quality-water ecology coupling model, and a human disturbance quantification module to achieve high-precision simulation and prediction of the reclaimed water replenishment process; Risk assessment and early warning layer: It has a built-in multimodal machine learning mutation risk prediction model and a differentiated graded ecological security threshold library, which is responsible for assessing the ecological risk level in real time and issuing ultra-early warnings; Optimized scheduling layer: Integrates meteorological and hydrological forecasting interfaces and multi-objective optimization algorithms, and is responsible for generating multi-scenario optimized scheduling plans and updating them dynamically; Decision support layer: Provides a visual display interface to present digital twin simulation results, risk warning information, and scheduling plans to managers in an intuitive way, and supports one-click issuance and execution tracking of plans.

Images