Digital twinborn river and lake health phase space diagnosis, early warning and regulation method

Abstract

The invention relates to the technical field of intelligent management of river and lake ecosystems, and discloses a river and lake health phase space diagnosis, early warning and regulation method fusing digital twinning, and the method comprises the steps: constructing a five-dimensional dynamic health phase space of river and lake health, and generating an elastic health reference trajectory based on a cross-scale coupling mechanism model; acquiring a real state track based on multi-source data fusion, and performing multi-cause traceability diagnosis by calculating a dynamic weight deviation vector between the real state track and an elastic health reference track; based on the high-order dynamic characteristics of the dynamic weight deviation vector, constructing a multi-scale trend evolution matrix and carrying out stability analysis to realize multi-level risk early warning; and based on the dynamic weight deviation vector, calculating a restoring force vector through adaptive PID control, generating a precise regulation and control instruction through robust mapping, and issuing and executing the precise regulation and control instruction. According to the method, the limitation of staticizing, splitting and lagging in the prior art is broken through.

Description

Technical Field

[0001] This invention relates to the field of intelligent management technology for river and lake ecosystems, and more specifically, to a method for spatial diagnosis, early warning and regulation of river and lake health phases that integrates digital twins. Background Technology

[0002] Currently, river and lake health management mainly relies on three fundamentally flawed technological paradigms: First, monitoring and early warning systems based on static indicators and threshold alarms. Their core problems lie in the fragmentation of indicators and the lag in response, failing to capture the nonlinear synergistic effects between multiple factors. Furthermore, rigid threshold settings easily lead to false alarms or missed alarms, often issuing alarms only after the system has suffered irreversible damage. Second, data-driven machine learning and probabilistic models, while partially overcoming the limitations of static thresholds, suffer from a "black box" nature, resulting in a lack of physical meaning in the decision-making process. Diagnostic results are difficult to guide precise regulation, and they are severely constrained by the coverage and quality of historical data. Their extrapolation ability is poor when facing unprecedented extreme scenarios, and probabilistic outputs can easily delay decision-making. Third, regulation decision-making schemes based on multi-objective optimization fall into the trap of fuzzy objective function quantification, high computational complexity, poor real-time performance, neglect of system self-organization, and excessive sensitivity to initial parameters. This often results in "mathematically optimal" solutions rather than "ecologically optimal" ones, and the slow computational response makes them unsuitable for emergency management needs. In summary, existing technologies generally suffer from structural defects such as "staticity, fragmentation, and lag," making them unable to effectively cope with complex disturbances.

[0004] Therefore, it is necessary to provide a spatial diagnosis, early warning and control method for river and lake health that integrates digital twins to solve the problem that existing technologies have not broken through the limitations of static, fragmented and lagging, and cannot cope with the combined disturbances of climate change and human activities. Summary of the Invention

[0005] In view of this, the present invention proposes a method for phase spatial diagnosis, early warning and regulation of river and lake health that integrates digital twins, aiming to solve the problem that existing technologies have not broken through the limitations of static, fragmented and lagging, and cannot cope with the combined disturbances of climate change and human activities.

[0006] This invention proposes a method for phase-space diagnosis, early warning, and regulation of river and lake health that integrates digital twins, including: A five-dimensional dynamic health phase space for rivers and lakes is constructed, and an elastic health reference trajectory is generated based on a cross-scale coupling mechanism model. Based on the fusion of multi-source data to obtain the real-state trajectory, and by calculating the dynamic weight deviation vector between the real-state trajectory and the elastic health reference trajectory, multi-cause tracing diagnosis is carried out. Based on the high-order dynamic characteristics of the dynamic weight deviation vector, a multi-scale trend evolution matrix is ​​constructed and stability analysis is performed to achieve multi-level risk early warning. Based on the dynamic weight deviation vector, the restoring force vector is calculated through adaptive PID control, and a precise control command is generated through robust mapping and issued for execution.

[0007] Furthermore, the construction of the five-dimensional dynamic health phase space for rivers and lakes, and the generation of elastic health reference trajectories based on a cross-scale coupling mechanism model, includes: Define the state vector in the five-dimensional health phase space; Construct a cross-scale physical-ecological-socioeconomic coupling mechanism model as a digital twin; The ideal driving conditions for climate adaptation are input into the cross-scale physical-ecological-socioeconomic coupling mechanism model to simulate and generate an elastic health reference trajectory, and a dynamic elastic trajectory pipeline around the elastic health reference trajectory is defined based on the system's restoring force vector.

[0008] Furthermore, the dimensions of the five-dimensional dynamic health phase space include: hydrological variables X1: water level, flow velocity, flow rate, hydraulic retention time, and flood season frequency; water quality variables X2: water temperature, pH, dissolved oxygen, permanganate index, total phosphorus, total nitrogen, ammonia nitrogen, and heavy metal concentration; aquatic ecological variables X3: chlorophyll a, zooplankton density, submerged plant coverage, fish biomass index, and benthic biodiversity index; and socioeconomic variables X4: water extraction intensity, wastewater treatment rate, and ecological buffer zone coverage. r And fertilizer application intensity; climate change variables X5: frequency of extreme precipitation, annual mean temperature change rate and evaporation; The state vector is: ; Where X(t,τ) represents a variable with time t and hydrological periodicity factor τ; t represents time, τ represents hydrological periodicity factor, τ=1 represents flood season, and τ=0.5 represents dry season; Normalizing X(t,τ) yields X i '(t,τ).

[0009] Furthermore, the cross-scale physical-ecological-socioeconomic coupling mechanism model includes an integrated hydrodynamic sub-model, a sediment-water interface material exchange sub-model, a water ecological dynamics sub-model, a socioeconomic driving sub-model, and a climate change response sub-model.

[0010] Furthermore, the step of obtaining the real-state trajectory based on multi-source data fusion and performing multi-causal tracing diagnosis by calculating the dynamic weight deviation vector between the real-state trajectory and the elastic health reference trajectory includes: By integrating in-situ monitoring, remote sensing monitoring and socio-economic statistics, robust Kalman filtering is used for data assimilation and quality control to construct the real-state vector and its evolutionary real-state trajectory. Calculate the health deviation vector between the actual state trajectory and the elastic health reference trajectory in phase space, and introduce a dynamic weight matrix to calculate the dynamic weight deviation degree; The health deviation vector is projected onto a predefined library of typical pathological pattern basis vectors and fuzzy membership is calculated to quantify the contribution of single or compound causes, and the path of cause propagation is analyzed by time series clustering.

[0011] Furthermore, the health deviation vector is: ; In the above formula, D(t) represents the health deviation vector at time t; X R (t) represents the current state vector at time t; X H (t) represents the elastic health reference trajectory state vector at time t; d i (t) represents the deviation of the i-th variable at time t (i.e., the difference between the actual value and the health reference value of the i-th variable); n represents the total number of variables; T represents the transpose of the vector (converting the row vector into a column vector).

[0012] Furthermore, the dynamic weight deviation is: ; In the above formula, S D (t) represents the dynamic weight deviation, w i (t, τ) represents the i-th dynamic weight matrix, d i (t) represents the deviation of the i-th variable; If S D (t) < 0.1 indicates a healthy state; if 0.1 ≤ S D (t) < 0.3 indicates a sub-healthy state; if S D (t)≥0.3 indicates an ill-condition.

[0013] Furthermore, when constructing a multi-scale trend evolution matrix and performing stability analysis based on the higher-order dynamic features of the dynamic weight deviation vector to achieve multi-level risk early warning, the following steps are included: The first, second, and third derivatives of the dynamic weight deviation vector are extracted and used as the deviation velocity vector, deviation acceleration vector, and deviation Jerk vector, respectively. For different spatial and temporal scales, a multi-scale trend evolution matrix containing the dynamic weight deviation vector and its derivative is constructed; The eigenvalues ​​of the multi-scale trend evolution matrix are calculated, and the stability of the system at each scale is determined based on the difference between the eigenvalue magnitude and 1. The warning level is divided according to the number and degree of instability at each scale, and the warning confidence is evaluated by combining the Bayesian method.

[0014] Furthermore, the step of calculating the restoring force vector based on the dynamic weight deviation vector through adaptive PID control, and generating precise control commands via robust mapping, and issuing them for execution, includes: Based on the dynamic weight deviation vector, its integral and derivative, the restoring force vector is calculated using an adaptive PID control algorithm. A robust mapping matrix is ​​established between the restoring force vector and the control actuator vector. The robust mapping matrix is ​​solved using the minimax criterion to generate effective control commands within the uncertainty range. The control measures are then prioritized according to the magnitude of the components of the restoring force vector.

[0015] Furthermore, the control actuator vector includes multiple parameters such as reservoir ecological discharge flow, aerator power, chemical dosage, ecological buffer zone irrigation volume, sewage treatment plant upgrading and renovation efforts, bottom sediment dredging intensity, submerged plant planting area, and water intake restriction coefficient.

[0016] Compared with existing technologies, the beneficial effects of this invention are as follows: Based on a cross-scale coupled high-dimensional health phase space self-diagnosis-self-early warning-adaptive regulation method for river and lake ecosystems, this invention achieves a breakthrough in cognitive paradigms through the constructed process of "five-dimensional phase space construction → multi-causal diagnosis → multi-scale early warning → adaptive regulation": from "static fragmentation" to "dynamic coupling". This invention is the first to construct a five-dimensional health phase space of "hydrology-water quality-aquatic ecology-socio-economic-climate change", replacing static indicators with dynamic elastic trajectories to reveal the "life rhythm" of river and lake health. Application in a certain watershed shows that compared with the traditional static indicator method, this method greatly improves the accuracy of eutrophication identification and the identification rate of hydrological-ecological coupling problems, realizing a cognitive leap from "reductionism" to "systems theory".

[0017] This invention also achieves a breakthrough in diagnostic accuracy: from "single attribution" to "multiple etiological origins." Based on dynamic weighted deviation vectors and a typical pathological pattern library, this method can quantify the contribution of complex etiologies and locate the transmission path. In a case study of an urban river, traditional methods only identified "eutrophication," while this method identified "60% eutrophication + 30% hydrological imbalance + 10% heavy metal pollution," and traced the causal path to "industrial discharge → heavy metal accumulation → fish population decline → uncontrolled algae growth → eutrophication + insufficient flow velocity → pollution retention." The diagnostic results showed a 95% agreement with the field investigation, providing a clear target for precise regulation.

[0018] This invention also achieves a breakthrough in early warning timeliness: from "post-event alarm" to "pre-event prediction." Traditional threshold alarms are essentially confirmations of "accidents." This invention, based on an early warning mechanism of dynamic instability trends, can accurately identify the inherent tendency of a system to become unstable during the "latency period" before significant deterioration of system state parameters. This provides a valuable time window for taking preventative measures, achieving a fundamental shift from "locking the stable door after the horse has bolted" to "prevention is better than cure," with immeasurable social, economic, and ecological benefits. Based on multi-scale trend evolution matrix and eigenvalue stability analysis, this method can issue early warnings 48-72 hours in advance, far superior to the traditional threshold method's "after-the-failure alarm" (12-24 hours lag).

[0019] This invention also achieves a breakthrough in control efficiency: from "lagging optimization" to "real-time adaptation." By integrating adaptive PID and robust mapping, the control response time of this method is shortened from hours to minutes by traditional optimization algorithms, thus improving control accuracy. The invention abandons computationally expensive and slow-responding optimization algorithms, replacing them with a "recovery force" model based on physical feedback. This control method, like a biological reflex, can respond almost instantaneously and appropriately to external disturbances and changes in internal state. It does not require pre-setting complex optimization objectives and weights, but rather takes "returning to health" as its sole and clear guideline, achieving extremely simple, efficient, and robust control.

[0020] This invention also achieves a breakthrough in universality: shifting from "data-dependent" to "mechanism-driven." The health reference trajectory is based on mechanistic model deduction, rather than fitting historical data, making it suitable for data-sparse scenarios. Because the health benchmark comes from future-oriented, ideal-situation mechanistic model deduction, rather than fitting past (potentially unhealthy) historical data, this invention can set a truly guiding and forward-looking health target for rivers and lakes. This makes this method applicable not only to rivers and lakes with abundant historical data, but also to newly constructed or remediated water bodies with sparse data, demonstrating strong universality. Attached Figure Description

[0021] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings: Figure 1 A flowchart of a method for phase-space diagnosis, early warning and regulation of river and lake health that integrates digital twins, provided in an embodiment of the present invention. Detailed Implementation

[0022] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the disclosure to those skilled in the art. It should be noted that, unless otherwise specified, embodiments and features in the embodiments of the present invention can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0023] In some embodiments of this application, see Figure 1 As shown, this embodiment provides a method for spatial diagnosis, early warning, and regulation of river and lake health based on digital twins, including the following steps: S100, construct a five-dimensional dynamic health phase space for river and lake health, and generate an elastic health reference trajectory based on a cross-scale coupling mechanism model; S200: Based on the fusion of multi-source data, the real-state trajectory is obtained, and the dynamic weight deviation vector between the real-state trajectory and the elastic health reference trajectory is calculated to perform multi-cause tracing diagnosis. S300. Based on the high-order dynamic characteristics of the dynamic weight deviation vector, a multi-scale trend evolution matrix is ​​constructed and stability analysis is performed to achieve multi-level risk early warning. S400. Based on the dynamic weight deviation vector, the restoring force vector is calculated through adaptive PID control, and a precise control command is generated through robust mapping and issued for execution.

[0024] In some embodiments of this application, the construction of a five-dimensional dynamic health phase space for river and lake health, and the generation of an elastic health reference trajectory based on a cross-scale coupling mechanism model, includes: Define the state vector in the five-dimensional health phase space; Construct a cross-scale physical-ecological-socioeconomic coupling mechanism model as a digital twin; The ideal driving conditions for climate adaptation are input into the cross-scale physical-ecological-socioeconomic coupling mechanism model to simulate and generate an elastic health reference trajectory, and a dynamic elastic trajectory pipeline around the elastic health reference trajectory is defined based on the system's restoring force vector.

[0025] In some embodiments of this application, the dimensions of the five-dimensional dynamic health phase space include: hydrological variables X1: water level, flow velocity, flow rate, hydraulic retention time, and flood season frequency; water quality variables X2: water temperature, pH, dissolved oxygen, permanganate index, total phosphorus, total nitrogen, ammonia nitrogen, and heavy metal concentration; aquatic ecological variables X3: chlorophyll a, zooplankton density, submerged plant coverage, fish biomass index, and benthic biodiversity index; and socioeconomic variables X4: water extraction intensity, wastewater treatment rate, and ecological buffer zone coverage. r And fertilizer application intensity; climate change variables X5: frequency of extreme precipitation, annual mean temperature change rate and evaporation; The state vector is: ; Where X(t,τ) represents a variable with time t and hydrological periodicity factor τ; t represents time, τ represents hydrological periodicity factor, τ=1 represents flood season, and τ=0.5 represents dry season; Normalizing X(t,τ) yields X i '(t,τ).

[0026] In some embodiments of this application, the cross-scale physical-ecological-socioeconomic coupling mechanism model includes an integrated hydrodynamic sub-model, a sediment-water interface material exchange sub-model, a water ecological dynamics sub-model, a socioeconomic driving sub-model, and a climate change response sub-model.

[0027] Specifically, the normalization formula is: ; Where: Xᵢ(t,τ) represents the original variable value of the i-th dimension when time is t and the hydrological periodic factor is τ; Xᵢ'(t,τ) represents the normalized variable value of the i-th dimension; This represents the lower limit of the health range of the i-th variable under period τ (determined by a combination of expert experience, ecological benchmark values, and historical natural state data). This represents the upper limit of the healthy range of the i-th variable under period τ; for example, the lower limit of the healthy range of dissolved oxygen is: during the flood season: XDO,min (τ=1) = 4 mg / L, during the dry season: XDO,min (τ=0.5) = 5 mg / L. τ represents the hydrological periodic factor (τ=1 for the flood season, τ=0.5 for the dry season).

[0028] The cross-scale physical-ecological-socioeconomic coupling mechanism model is a "digital twin" of rivers and lakes, composed of six coupled sub-models. It describes the evolution of the state vector through a system of ordinary differential / partial differential equations: ; In the above formula, X(t) represents the five-dimensional health phase space state vector at time t (including variables in the dimensions of hydrology, water quality, aquatic ecology, socio-economics, and climate change); t represents time; dX(t) / dt represents the first derivative of the state vector with respect to time (describing the rate of state evolution); M(・) represents the function describing the evolution law of the state vector (composed of the coupling of various sub-models); P(t) represents the set of model parameters at time t (such as riverbed roughness, pollutant degradation coefficient, algal growth rate, etc., calibrated through laboratory experiments, field monitoring, and data assimilation methods); B(t) represents the external driving force vector at time t (such as upstream water inflow, solar radiation, sewage discharge, etc.); τ represents the hydrological periodicity factor (τ=1 for the flood season, τ=0.5 for the dry season), which adapts to the periodic dynamics of the sub-models.

[0029] The core equations of each sub-model of the cross-scale physical-ecological-socioeconomic coupling mechanism model are as follows: (1) Nested hydrodynamic sub-model: The main stream employs two-dimensional shallow-water equations (spatial accuracy 50m×50m), while the tributaries utilize one-dimensional Saint-Venant equations (spatial accuracy 100m), enabling cross-scale flow field simulation. The core of the two-dimensional shallow-water equations is: ; In the above formula, h represents water depth; u represents flow velocity in the x-direction; v represents flow velocity in the y-direction; t represents time; and g represents gravitational acceleration (approximately 9.8 m / s²). 2 η represents the water level (relative to the reference surface); , Represents wind stress in the x and y directions; , The x and y axes represent the bed shear stresses; ρ represents the density of water (approximately 1000 kg / m³). 3 f represents the Coriolis force coefficient (related to latitude, the higher the latitude, the larger the coefficient); q represents the net inflow per unit area (such as precipitation, infiltration, sewage, etc., unit: m / s).

[0030] (2) Sub-model of material exchange at the sediment-water interface: Simulation of the adsorption-release dynamics of total phosphorus (TP) and total nitrogen (TN) in sediments, core equation (taking TP as an example): ; In the above formula, v represents the concentration of total phosphorus (TP) in sediments (unit: mg / kg); v represents the concentration of total phosphorus (TP) in water (unit: mg / L); t represents time; The adsorption rate coefficient of phosphorus (unit: 1 / d). k represents the desorption rate coefficient of phosphorus (unit: 1 / d). TThe temperature coefficient is represented by T (usually 1.05-1.15); T represents the water temperature (unit: °C); T0 represents the reference temperature (usually 20 °C); Q in Indicates the flow rate into the water body (unit: m). 3 / s); C TP,in Indicates the concentration of total phosphorus in the inflowing water (unit: mg / L); Q out The flow rate of water flowing out of the body (unit: m³) 3 / s); The contact area of ​​the sediment-water interface (unit: m²) 2 V represents the volume of water (unit: m³). 3 ).

[0031] (3) Aquatic Ecological Dynamics Sub-model: The Lotka-Volterra model is extended to include interactions across four trophic levels: algae, zooplankton, fish, and submerged plants. The core equation is (algae-submerged plant competition): ; In the above formula, P represents the algae concentration (unit: mg / L); C represents the submerged plant biomass (unit: g / m³). 2 ); t represents time; d / dt represents the derivative with respect to time; This represents the maximum growth rate of algae (unit: 1 / d). I represents the maximum growth rate of submerged plants (unit: 1 / d); I represents photosynthetically active radiation (unit: μmol / (m²)). 2 •s)); Light saturation point of algae / submerged plants (unit: μmol / (m²)) 2 •s); N represents the nitrogen concentration in the water (unit: mg / L); The half-saturation constant of nitrogen (unit: mg / L); α represents the half-saturation constant of phosphorus (unit: mg / L); α represents the rate coefficient of algal ingestion by zooplankton (unit: L / (g・d)); M represents the zooplankton density (unit: g / m³). 3 ); β represents the competition coefficient between algae and submerged plants (dimensionless); γ represents the rate coefficient of submerged plants being consumed by fish (F) (unit: 1 / (g・d)); F represents fish biomass (unit: g / m³). 3 ).

[0032] (4) Socioeconomic driving sub-model: Quantifying the impact of water intake and sewage discharge on rivers and lakes, such as the relationship between sewage treatment rate and NH3-N input: ; In the above formula, This indicates the concentration of ammonia nitrogen (NH3-N) flowing into rivers and lakes (unit: mg / L). Indicates wastewater discharge volume (unit: m³) 3 / s); This indicates the concentration of ammonia nitrogen in untreated wastewater (unit: mg / L). This represents the wastewater treatment rate (dimensionless, between 0 and 1). This represents the discharge volume of non-sewage (such as surface runoff) (unit: m³). 3 / s); This indicates the concentration of ammonia nitrogen in non-wastewater (unit: mg / L). Indicates total inflow ( Unit: m 3 / s).

[0033] (5) Climate change response sub-model: Coupled with the IPCC intermediate emission scenario (SSP2-4.5), simulate the impact of extreme precipitation on pollution load: ; In the above formula, Lnon-point represents the source pollution load under extreme precipitation scenario (unit: kg / d); L0 represents the source pollution load under normal precipitation scenario (unit: kg / d); kf represents the extreme precipitation impact coefficient (calibrated from historical data, dimensionless); fextreme represents the extreme precipitation frequency (unit: times / year, deviation rate relative to historical average level, dimensionless).

[0034] (6) Ecosystem service quantification sub-model: Assessing purification services (nitrogen and phosphorus removal) and habitat services (fish suitability index) provides a quantitative basis for regulatory targets: ; in, This indicates the quantitative indicators of water purification services (unit: kg); k TN The nitrogen purification coefficient (unit: 1 / d); k TP The phosphorus purification coefficient (unit: 1 / d); C TN Indicates the total nitrogen concentration in the water (unit: mg / L); C TP V represents the total phosphorus concentration in the water (unit: mg / L); V represents the water volume (unit: m³). 3 ); t represents time; This represents the habitat service index (dimensionless, between 0 and 1). This represents the normalized dissolved oxygen concentration (dimensionless). This represents the normalized submerged plant biomass (dimensionless). 0.3 represents the normalized fish biomass index (dimensionless); 0.3, 0.2, and 0.5 represent the weighting coefficients of each indicator (determined by expert evaluation, dimensionless).

[0035] In the cross-scale physical-ecological-socioeconomic coupling mechanism model, climate-adaptive ideal driving conditions are input to simulate and generate an elastic health reference trajectory. A dynamic elastic trajectory pipeline is then defined around the elastic health reference trajectory based on the system's resilience vector, specifically: (1) Ideal driving conditions : Based on the scenario setting of "optimal natural-social synergy", including: Hydrological drivers: Upstream inflow is climate-adaptive natural runoff (2050 runoff process under SSP2-4.5 scenario); Pollution-driven: Zero emissions from point sources, and 90% reduction of non-point source pollution (30% coverage) through ecological buffer zones. Socioeconomic drivers: Wastewater treatment rate 95%, water intake intensity ≤ 80% of ecological flow; Climate-driven: The frequency of extreme precipitation is controlled within the historical natural fluctuation range (±10%).

[0036] Generate a resilient health reference trajectory : Will Input a multi-scale physical-ecological-socioeconomic coupling mechanism model, simulate three complete hydrological years, and output the evolution trajectory of the state vector: ; In the above formula, X H (t,τ) represents the elastic health reference trajectory state vector at time t and hydrological periodicity factor τ; Simulate(・) represents the model simulation function; M represents the cross-scale physical-ecological-socioeconomic coupling mechanism model; X0(τ) represents the initial health state vector at period τ (obtained from historical natural data inversion); P represents the model parameter set; B H (t,τ) represents the ideal driving condition vector for climate adaptation at time t and hydrological periodicity factor τ; τ represents the hydrological periodicity factor (τ=1 for flood season, τ=0.5 for dry season).

[0037] (3) Dynamic elastic trajectory pipeline: A health trajectory is not a fixed curve, but rather follows a pattern. The centered "elastic pipeline" has its boundaries dynamically adjusted by the system's resilience. ; In the above formula, X H,max (t,τ) represents the upper bound state vector of the elastic health trajectory pipeline; XH,min (t,τ) represents the lower bound state vector of the elastic health trajectory pipeline; X H (t,τ) represents the elastic health reference trajectory state vector; δ(t,τ) represents the elasticity coefficient at time t and hydrological periodicity factor τ (δ=0.2 during flood season and δ=0.1 during dry season, reflecting the system vulnerability during dry season, dimensionless); R(t) represents the system resilience vector at time t (determined by submerged plant coverage and benthic biodiversity, dimensionless); τ represents the hydrological periodicity factor (τ=1 for flood season and τ=0.5 for dry season).

[0038] In some embodiments of this application, the step of obtaining the real-state trajectory based on multi-source data fusion and performing multi-etiological tracing diagnosis by calculating the dynamic weight deviation vector between the real-state trajectory and the elastic health reference trajectory includes: By integrating in-situ monitoring, remote sensing monitoring and socio-economic statistics, robust Kalman filtering is used for data assimilation and quality control to construct the real-state vector and its evolutionary real-state trajectory. Calculate the health deviation vector between the actual state trajectory and the elastic health reference trajectory in phase space, and introduce a dynamic weight matrix to calculate the dynamic weight deviation degree; The health deviation vector is projected onto a predefined library of typical pathological pattern basis vectors and fuzzy membership is calculated to quantify the contribution of single or compound causes, and the path of cause propagation is analyzed by time series clustering.

[0039] In some embodiments of this application, the health deviation vector is: ; In the above formula, D(t) represents the health deviation vector at time t; X R (t) represents the current state vector at time t; X H (t) represents the elastic health reference trajectory state vector at time t; d i (t) represents the deviation of the i-th variable at time t (i.e., the difference between the actual value and the health reference value of the i-th variable); n represents the total number of variables; T represents the transpose of the vector (converting the row vector into a column vector).

[0040] In some embodiments of this application, the dynamic weight deviation is: ; In the above formula, S D (t) represents the dynamic weight deviation at time t; w i (t, τ) represents the dynamic weight of the i-th variable at time t and when the hydrological periodicity factor is τ (determined by the degree of influence of the variable on the health of rivers and lakes, dimensionless); d i(t) represents the deviation of the i-th variable at time t; τ represents the hydrological cycle factor (τ=1 for the flood season, τ=0.5 for the dry season).

[0041] If S D (t) < 0.1 indicates a healthy state; if 0.1 ≤ S D (t) < 0.3 indicates a sub-healthy state; if S D (t)≥0.3 indicates an ill-condition.

[0042] Specifically, the integration of in-situ monitoring, remote sensing monitoring, and socio-economic statistics includes: In-situ monitoring: Online sensors (every 15 minutes) collect hydrological (water level, flow velocity) and water quality (DO, TP, TN) data; Remote sensing monitoring: Sentinel-2 satellite (5 days / time) inversion of Chl-a and submerged plant coverage; Field survey: Quarterly sampling and measurement of fish biomass and benthic biodiversity; Socioeconomic data: Environmental protection departments provide statistics on sewage treatment rates and total water intake, while meteorological departments provide extreme precipitation data.

[0043] Data assimilation and quality control are performed using robust Kalman filtering, constructing the real-state vector and its evolving real-state trajectory, specifically as follows: Robust Kalman filtering is used to fuse multi-source data and eliminate outliers (such as sensor malfunctions). ; in, This represents the fused real-state vector at time t; This represents the predicted value of the actual state vector at time t based on the data at time t-1; This represents the Kalman filter gain matrix at time t; H represents the multi-source observation data vector at time t (including in-situ monitoring, remote sensing monitoring, etc.); H represents the observation matrix (mapping the state vector to the observation space). The fused data is normalized to form the actual state vector. Its time series is the trajectory of the actual state.

[0044] The health deviation vector is projected onto a predefined library of typical pathological pattern basis vectors, and fuzzy membership is calculated to quantify the contribution of single or multiple etiologies. Furthermore, time-series clustering analysis is used to analyze the etiology transmission path. Specifically: Predefined standardized unit vectors for 8 typical pathological patterns ( ), covering both single and complex issues: (Eutrophication): TP, TN, and Chl-a components were +0.6, +0.5, and +0.7, respectively, while DO component was -0.4. (Heavy metal pollution): Hg and Cd components were +0.8 and +0.7 respectively, and fish biomass component was -0.6; (Hydrological imbalance): The velocity and flow rate components are -0.7 and -0.8, respectively, and the hydraulic residence time component is +0.6; (Eutrophication-hydrological imbalance coupling): TP (+0.5), Chl-a (+0.6), flow velocity (-0.5); Other models include: climate change-induced degradation, compound pollution, habitat fragmentation, and socioeconomic-driven degradation.

[0045] Project D(t) onto the basis vectors of each pathological pattern and calculate the projection components. Fuzzy membership degree : ; In the above formula, C k (t) represents the health deviation vector D(t) at time t, and the basis vector e of the k-th pathological pattern. k The projection component on; D(t) represents the health deviation vector at time t; e k θ represents the standardized unit vector of the k-th typical pathological pattern; · represents the dot product of the vectors; ||·|| represents the magnitude (i.e., the length) of the vector; k Represent D(t) and e k The included angle between them; cosθ k μ represents the cosine of the included angle (reflecting the similarity between the two vectors); k (t) represents the fuzzy membership degree (i.e., contribution, dimensionless, between 0 and 1) of the k-th etiology at time t; ΣC k (t) represents the sum of the projection components of all pathological patterns (summing over k from 1 to m, where m is the total number of pathological patterns).

[0046] The initial disturbance source can be traced through time series clustering of offset vectors (DBSCAN algorithm). For example, in a certain watershed... =0.7, clustering showed that TP deviation preceded flow velocity deviation by 2 days, and the high TP value area coincided with the non-point source pollution inlet. The pathogenesis was determined to be "non-point source TP input → algal bloom → increased oxygen consumption → decreased DO + insufficient flow velocity → pollutant retention → aggravated eutrophication".

[0047] In some embodiments of this application, when constructing a multi-scale trend evolution matrix and performing stability analysis based on the higher-order dynamic characteristics of the dynamic weight deviation vector to achieve multi-level risk early warning, the following steps are included: The first, second, and third derivatives of the dynamic weight deviation vector are extracted and used as the deviation velocity vector, deviation acceleration vector, and deviation Jerk vector, respectively. For different spatial and temporal scales, a multi-scale trend evolution matrix containing the dynamic weight deviation vector and its derivative is constructed; The eigenvalues ​​of the multi-scale trend evolution matrix are calculated, and the stability of the system at each scale is determined based on the difference between the eigenvalue magnitude and 1. The warning level is divided according to the number and degree of instability at each scale, and the warning confidence is evaluated by combining the Bayesian method.

[0048] Specifically, the first to third derivatives of D(t) are calculated to capture the dynamic trend of deviation: (1) Deviation from velocity vector : Describes the rate of change of deviation, reflecting the "speed of deterioration / improvement"; ; (2) Deviation from acceleration vector : Describes the change in the rate of deviation, reflecting the "strength of the deterioration / improvement trend"; ; (3) Deviation from Jerk vector : Describes changes in deviation acceleration and captures "sudden disturbances" (such as sudden sewage discharge); ; like This indicates that the deviation is accelerating (worsening); if A sudden increase in absolute value (>0.1) indicates a sudden disturbance.

[0049] Furthermore, for different spatial and temporal scales, a multi-scale trend evolution matrix is ​​constructed, which includes the dynamic weight deviation vector and its derivative, as follows: (1) Multi-scale division: Spatial scale: local scale (river section, such as 1-5km), regional scale (watershed, such as 50-100km); Time scales: short-term (daily scale, Δt=1 day), medium-term (monthly scale, Δt=30 days), long-term (annual scale, Δt=365 days).

[0050] (2) Trend evolution matrix ( (where s is the scale) For each scale, construct a 3×3 augmented matrix (containing ): ; in, This represents the trend evolution matrix at time t and scale s; s represents the scale (including spatial scales such as river segment / basin, and temporal scales such as day / month / year); SD(t) represents the dynamic weight deviation at time t; express The first derivative with respect to time (rate of change of deviation); express The second derivative with respect to time (acceleration of the deviation); The coefficients at time t and scale s (from) Estimate, describing the system's restoring elasticity and damping characteristics, dimensionless.

[0051] (3) Eigenvalue stability criterion: calculate eigenvalues The difference between the eigenvalue modulus and 1 Determine stability: all The system is stable (deviation from convergence). exist System instability (deviation from divergence).

[0052] Furthermore, the eigenvalues ​​of the multi-scale trend evolution matrix are calculated, and the stability of the system at each scale is determined based on the difference between the eigenvalue magnitude and 1. The warning level is then classified according to the number and severity of instability at different scales. Simultaneously, the warning confidence level is evaluated using a Bayesian method. Specifically: The uncertainty of the model and data is quantified using the Bayesian method, and the warning confidence level is output: ; Wherein, P(Warning Correct|Data) represents the probability of a warning being correct based on the observed data (i.e., confidence level, dimensionless, between 0 and 1); P(Data|Warning) represents the probability of the observed data appearing given a warning result (the degree of matching between the data and the warning pattern, dimensionless); P(Warning) represents the historical warning accuracy rate (e.g., the historical accuracy rate of yellow warnings is 92%, dimensionless); and P(Data) represents the prior probability of the observed data itself appearing (obtained through statistical analysis of historical data, dimensionless).

[0053] In some embodiments of this application, the step of calculating the restoring force vector based on the dynamic weight deviation vector through adaptive PID control, generating precise control commands via robust mapping, and issuing them for execution includes: Based on the dynamic weight deviation vector, its integral and derivative, the restoring force vector is calculated using an adaptive PID control algorithm. A robust mapping matrix is ​​established between the restoring force vector and the control actuator vector. The robust mapping matrix is ​​solved using the minimax criterion to generate effective control commands within the uncertainty range. The control measures are then prioritized according to the magnitude of the components of the restoring force vector.

[0054] In some embodiments of this application, the control actuator vector includes multiple parameters such as reservoir ecological discharge flow, aerator power, chemical dosage, ecological buffer zone irrigation volume, sewage treatment plant upgrading and renovation efforts, bottom sediment dredging intensity, submerged plant planting area, and water intake restriction coefficient.

[0055] Specifically, based on the dynamic weight deviation vector, its integral, and its derivative, the restoring force vector is calculated using an adaptive PID control algorithm, as follows: Regulation objective: To apply restorative force, to Pull back Restoring force vector Based on adaptive PID control, the proportional (P), integral (I), and derivative (D) terms are integrated to eliminate steady-state deviation and overshoot. ; Where: FR(t) represents the restoring force vector at time t (used to pull the current state back to the healthy trajectory); the negative sign indicates that the direction of the restoring force is opposite to the direction of deviation; Kp(t) represents the proportional gain matrix at time t (and S... D (t) is positively correlated, S D The larger Kp(t) is, the larger it is, and it is dimensionless. hour, D(t) represents the health deviation vector at time t; K i (t) represents the integral gain matrix at time t (positively correlated with ∫D(τ)dτ, eliminating long-term bias, dimensionless, e.g., when the cumulative bias = 0.3). ); ∫D(τ)dτ represents the integral of the health deviation vector over time (the cumulative deviation from the initial time to time t, where τ is the integration variable); K d (t) represents the differential gain matrix at time t (positively correlated with ||VD(t)||, suppressing the rate of degradation, dimensionless, such as... hour, ); V D (t) represents the deviation velocity vector at time t; ||VD(t)|| represents the magnitude of VD(t).

[0056] Adaptive Update: Optimize the gain matrix and objective function using the Q-learning algorithm. Updated every 24 hours .

[0057] The component directly indicates the direction of control: such as This indicates that the DO (Domain) needs to be upgraded. This indicates that low TP is required.

[0058] A robust mapping matrix is ​​established between the restoring force vector and the control actuator vector. The robust mapping matrix is ​​solved using the minimax criterion to generate effective control commands within the uncertainty range. The control measures are then prioritized according to the magnitude of the components of the restoring force vector. Specifically: Controlling the actuator vector (U(t)): Define 8 types of executable control measures, forming a vector: ; Wherein, U(t) represents the control actuator vector at time t; u1(t) represents the reservoir ecological discharge flow at time t (dimensionless value, 0-1 represents 0 to maximum discharge flow); u2(t) represents the aerator power at time t (dimensionless value); u3(t) represents the chemical dosage at time t (dimensionless value); u4(t) represents the ecological buffer zone irrigation amount at time t (dimensionless value); u5(t) represents the upgrading and renovation efforts of the sewage treatment plant at time t (dimensionless value); u6(t) represents the dredging intensity of the bottom sediment at time t (dimensionless value); u7(t) represents the submerged plant planting area at time t (dimensionless value); u8(t) represents the water intake restriction coefficient at time t (dimensionless value); T represents the transpose of the vector (converting the row vector into a column vector).

[0059] Robust mapping matrix ( ): Establish U(t) and The mapping relationship is considered, taking into account the uncertainty of the control effect (e.g., an increase of 1kW in aerator power results in a ±10% fluctuation in DO increase). A mapping matrix containing the uncertainty interval is constructed through Monte Carlo simulation: ; In the above formula, The average influence coefficient of the j-th actuator on the i-th state variable (e.g.) This means that for every 1kW increase in power, the DO (displacement) increases by 0.05. The uncertainty range is (e.g., ±0.005).

[0060] Solving robust control instructions: Solve The robust solution employs the "minmax criterion" to ensure effectiveness within the range of uncertainty: ; Solving using the interior-point method, specific control commands are output. For example: "The ecological outflow of the reservoir has been increased to 60m." 3 / s(u1=0.8)”; "Aerator #1 is operating at 90% power (u2=0.9)"; "Sediment dredging intensity 0.5km" 2 / day (u6=0.5)".

[0061] Regulation priority ranking: When regulatory resources are limited, the components are sorted based on their resilience, such as... In this case, priority should be given to regulating DO (turning on the aerator), followed by regulating TP (adding phosphorus removal agents).

[0062] Furthermore, the technical system of the present invention achieves closed-loop operation through six core modules, and the modules interact in real time through data interfaces. Table 1 shows the module names and a brief description of the working process of the present invention. Table 1 ; Specifically, the system hardware support includes edge computing nodes (processing real-time data), server clusters (running digital twin models), SCADA control systems (executing control commands), and a visualization platform (displaying health status, diagnostic results, and early warning levels).

[0063] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program goods. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program goods embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0064] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program goods according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0065] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0066] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0067] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A method for spatial diagnosis, early warning, and regulation of river and lake health integrating digital twins, characterized in that, include: A five-dimensional dynamic health phase space for rivers and lakes is constructed, and an elastic health reference trajectory is generated based on a cross-scale coupling mechanism model. Based on the fusion of multi-source data to obtain the real-state trajectory, and by calculating the dynamic weight deviation vector between the real-state trajectory and the elastic health reference trajectory, multi-cause tracing diagnosis is carried out. Based on the high-order dynamic characteristics of the dynamic weight deviation vector, a multi-scale trend evolution matrix is ​​constructed and stability analysis is performed to achieve multi-level risk early warning. Based on the dynamic weight deviation vector, the restoring force vector is calculated through adaptive PID control, and a precise control command is generated through robust mapping and issued for execution.

2. The method for spatial diagnosis, early warning, and regulation of river and lake health phases integrating digital twins as described in claim 1, characterized in that, The construction of the five-dimensional dynamic health phase space for rivers and lakes, and the generation of elastic health reference trajectories based on a cross-scale coupling mechanism model, includes: Define the state vector in the five-dimensional health phase space; Construct a cross-scale physical-ecological-socioeconomic coupling mechanism model as a digital twin; The ideal driving conditions for climate adaptation are input into the cross-scale physical-ecological-socioeconomic coupling mechanism model to simulate and generate an elastic health reference trajectory, and a dynamic elastic trajectory pipeline around the elastic health reference trajectory is defined based on the system's restoring force vector.

3. The method for spatial diagnosis, early warning, and regulation of river and lake health phases integrating digital twins as described in claim 1, characterized in that, The five dimensions of the dynamic health phase space include: hydrological variables X1: water level, flow velocity, flow rate, hydraulic retention time, and flood season frequency; water quality variables X2: water temperature, pH, dissolved oxygen, permanganate index, total phosphorus, total nitrogen, ammonia nitrogen, and heavy metal concentration; aquatic ecological variables X3: chlorophyll a, zooplankton density, submerged plant coverage, fish biomass index, and benthic biodiversity index; and socioeconomic variables X4: water extraction intensity, wastewater treatment rate, and ecological buffer zone coverage. r And fertilizer application intensity; climate change variables X5: frequency of extreme precipitation, annual mean temperature change rate and evaporation; The state vector is: ; Where X(t,τ) represents a variable with time t and hydrological periodicity factor τ; t represents time, τ represents hydrological periodicity factor, τ=1 represents flood season, and τ=0.5 represents dry season; Normalizing X(t,τ) yields X i '(t,τ).

4. The method for spatial diagnosis, early warning, and regulation of river and lake health phases integrating digital twins as described in claim 2, characterized in that, The cross-scale physical-ecological-socioeconomic coupling mechanism model includes an integrated hydrodynamic sub-model, a sediment-water interface material exchange sub-model, a water ecological dynamics sub-model, a socioeconomic driving sub-model, and a climate change response sub-model.

5. The method for spatial diagnosis, early warning, and regulation of river and lake health phases integrating digital twins as described in claim 1, characterized in that, The process of obtaining a real-state trajectory based on multi-source data fusion and performing multi-etiological source tracing diagnosis by calculating the dynamic weight deviation vector between the real-state trajectory and the elastic health reference trajectory includes: By integrating in-situ monitoring, remote sensing monitoring and socio-economic statistics, robust Kalman filtering is used for data assimilation and quality control to construct the real-state vector and its evolutionary real-state trajectory. Calculate the health deviation vector between the actual state trajectory and the elastic health reference trajectory in phase space, and introduce a dynamic weight matrix to calculate the dynamic weight deviation degree; The health deviation vector is projected onto a predefined library of typical pathological pattern basis vectors and fuzzy membership is calculated to quantify the contribution of single or compound causes, and the path of cause propagation is analyzed by time series clustering.

6. The method for spatial diagnosis, early warning, and regulation of river and lake health phases integrating digital twins according to claim 5, characterized in that, The health deviation vector is: ； In the above formula, D(t) represents the health deviation vector at time t; X R (t) represents the current state vector at time t; X H (t) represents the elastic health reference trajectory state vector at time t; d i (t) represents the deviation of the i-th variable at time t (i.e., the difference between the actual value and the health reference value of the i-th variable); n represents the total number of variables; T represents the transpose of the vector (converting the row vector into a column vector).

7. The method for spatial diagnosis, early warning, and regulation of river and lake health phases integrating digital twins according to claim 5, characterized in that, The dynamic weight deviation is: ； In the above formula, S D (t) represents the dynamic weight deviation, w i (t, τ) represents the i-th dynamic weight matrix, d i (t) represents the deviation of the i-th variable; If S D (t) < 0.1 indicates a healthy state; if 0.1 ≤ S D (t) < 0.3 indicates a sub-healthy state; if S D (t)≥0.3 indicates an ill-condition.

8. The method for spatial diagnosis, early warning, and regulation of river and lake health phases integrating digital twins according to claim 5, characterized in that, When constructing a multi-scale trend evolution matrix and performing stability analysis based on the higher-order dynamic features of the dynamic weight deviation vector to achieve multi-level risk early warning, the following methods are included: The first, second, and third derivatives of the dynamic weight deviation vector are extracted and used as the deviation velocity vector, deviation acceleration vector, and deviation Jerk vector, respectively. For different spatial and temporal scales, a multi-scale trend evolution matrix containing the dynamic weight deviation vector and its derivative is constructed; The eigenvalues ​​of the multi-scale trend evolution matrix are calculated, and the stability of the system at each scale is determined based on the difference between the eigenvalue magnitude and 1. The warning level is divided according to the number and degree of instability at each scale, and the warning confidence is evaluated by combining the Bayesian method.

9. The method for spatial diagnosis, early warning, and regulation of river and lake health phases integrating digital twins according to claim 1, characterized in that, The process of calculating the restoring force vector based on the dynamic weight deviation vector using adaptive PID control, and generating precise control commands via robust mapping, and then issuing and executing them, includes: Based on the dynamic weight deviation vector, its integral and derivative, the restoring force vector is calculated using an adaptive PID control algorithm. A robust mapping matrix is ​​established between the restoring force vector and the control actuator vector. The robust mapping matrix is ​​solved using the minimax criterion to generate effective control commands within the uncertainty range. The control measures are then prioritized according to the magnitude of the components of the restoring force vector.

10. The method for spatial diagnosis, early warning, and regulation of river and lake health phases integrating digital twins according to claim 9, characterized in that, The control actuator vectors include multiple parameters such as reservoir ecological discharge flow, aerator power, chemical dosage, ecological buffer zone irrigation volume, sewage treatment plant upgrading and renovation efforts, bottom sediment dredging intensity, submerged plant planting area, and water intake restriction coefficient.

Images