Reverse Tracing Diagnosis and Regulation Method and System for Scheduling Risks of Basin Water Project Systems

Abstract

The present invention belongs to the cross - field of water conservancy project scheduling and risk management, and specifically discloses a method and system for reverse tracing diagnosis and regulation of the scheduling risk of the basin water project system, including: obtaining basin water project system data; constructing a multi - objective collaborative scheduling risk field of the basin water project system based on the basin water project system data; performing reverse multi - dimensional probability flux tracking on the scheduling of the basin water project system based on the multi - objective collaborative scheduling risk field of the basin water project system to obtain the risk contribution degree entropy weight; performing multi - level risk diagnosis on the scheduling of the basin water project system according to the risk contribution degree entropy weight to obtain a diagnosis result; performing an asymmetric game dynamic compensation decision on the scheduling of the basin water project system according to the diagnosis result to obtain an optimal decision - making scheme; and using the optimal decision - making scheme to regulate the scheduling of the basin water project system. The present invention can dynamically identify the cascading instability risks caused by the reservoir group scheduling.

Description

Technical Field

[0001] The present invention belongs to the intersectional field of water conservancy project scheduling and risk management, and in particular relates to a reverse tracing diagnosis and regulation method and system for basin water conservancy project system scheduling risks. Background Art

[0002] The water conservancy system in the river basin is characterized by multi-level and multi-media coupling. The cascade reservoirs, dam groups and ecologically sensitive areas are closely related through the hydrodynamic network. The traditional forward risk deduction method has significant limitations in dealing with sudden extreme events: when the upstream engineering regulation triggers downstream chain risks, the forward model is difficult to quickly trace the key risk sources, resulting in a temporal and spatial misalignment between the regulation strategy and the risk transmission. For example, in the event of a heavy rainstorm, the dynamic feedback effect between the reservoir flood discharge decision and the urban drainage system often leads to the expansion of secondary disasters, highlighting the urgent need to establish a reverse diagnosis mechanism to clarify the boundaries of responsibility and block the chain transmission of risks.

[0003] There are three faults in the current watershed risk analysis technology: first, the diagnostic dimension is single, and the mainstream method relies on the hydrological model to forward deduce the risk probability, lacking the ability to analyze the reverse contribution between engineering facilities, watershed units and regional systems, resulting in unclear responsibility sharing during the coordinated scheduling of cascade engineering groups; second, the dynamic coupling is insufficient. Although the existing data fusion technology integrates meteorological and hydrological information, the ecological sensitivity parameters are only used as static constraints and fail to participate in the real-time coupling calculation of the risk field, resulting in delayed ecological risk warning; third, the decision-making response is delayed. The traditional optimization model adopts a "risk identification-strategy generation" serial architecture. The time taken to solve the multi-objective game exceeds the actual risk propagation speed, and cannot meet the real-time regulation needs under emergency conditions. Summary of the invention

[0004] In order to solve the above technical problems, the present invention proposes a method and system for reverse tracing diagnosis and regulation of basin water conservancy system scheduling risks, which is suitable for intelligent risk scheduling integrating cross-scale risk field construction, reverse flux tracking, multi-level diagnosis and dynamic game decision-making.

[0005] The present invention provides a method for reverse tracing diagnosis and regulation of basin water engineering system scheduling risks, including:

[0006] Obtain basin water engineering system data;

[0007] Based on the water conservancy system data of the river basin, a multi-objective coordinated dispatch risk field of the water conservancy system of the river basin is constructed;

[0008] Based on the multi-objective coordinated dispatch risk field of the water conservancy project system in the river basin, the reverse multi-dimensional probability flux of the dispatch of the water conservancy project system in the river basin is tracked to obtain the entropy weight of the risk contribution;

[0009] Based on the risk contribution degree entropy weight, conduct multi-level risk diagnosis on the scheduling of the basin water project system to obtain the diagnosis results;

[0010] Based on the diagnosis results, conduct asymmetric game dynamic compensation decision-making on the scheduling of the basin water project system to obtain the optimal decision-making plan;

[0011] Use the optimal decision-making plan to regulate the scheduling of the basin water project system.

[0012] Optionally, based on the basin water project system data, constructing a multi-objective collaborative scheduling risk field for the basin water project system includes:

[0013] Based on the basin water project system data, construct organizational data with a three-level architecture of basin level, regional level, and project level;

[0014] Based on the organizational data, fuse heterogeneous data to obtain multi-source data streams;

[0015] Based on the multi-source data, construct a water-energy-ecological risk field model based on the unsteady Navier-Stokes equation;

[0016] Use the Lyapunov exponent to characterize the cascade instability threshold of the risk field and perform dynamic coupling on the risk field model.

[0017] Optionally, based on the organizational data, fusing heterogeneous data to obtain multi-source data streams includes:

[0018] Adopt the optimized weighting method to fuse the organizational data and heterogeneous data:

[0019]

[0020] Among them, represents the fused data, is the i-th type of data source, is the weight coefficient corresponding to the i-th type of data source;

[0021] Adopt the Newton-Raphson optimization method based on the fusion mutation strategy to optimize the weight coefficient to obtain the optimal weight coefficient;

[0022] Based on the optimal weight coefficient, obtain the multi-source data stream.

[0023] Optionally, adopting the Newton-Raphson optimization method based on the fusion mutation strategy to optimize the weight coefficient to obtain the optimal weight coefficient includes:

[0024] Use head chaos mutation to improve the initial population;

[0025] Iteratively update the individual positions of the initial population until the minimum data fusion error is satisfied or the maximum number of updates is reached, then stop the update and obtain the optimal weight coefficient;

[0026] Among them, iteratively updating the individual positions of the initial population includes: updating through the Newton - Raphson search rule, trap avoidance operation, and body fusion t - distribution perturbation.

[0027] Optionally, based on the multi - objective collaborative scheduling risk field of the basin water project system, performing reverse multi - dimensional probability flux tracking on the scheduling of the basin water project system to obtain the risk contribution degree entropy weight includes:

[0028] Construct a reverse mapping model, where the reverse mapping model is used to reversely reconstruct the risk conduction path;

[0029] Based on the reverse mapping model, obtain the reverse propagation operator;

[0030] According to the risk propagation characteristics obtained by the reverse propagation operator, further obtain the risk contribution degree entropy weight of each level.

[0031] Optionally, the method for obtaining the reverse propagation operator is:

[0032] Construct an initial inverse fractional - order Fokker - Planck equation;

[0033] Construct the characteristic matrix of each level, use the judgment matrix to solve the characteristic matrix, and obtain the weight vector of each level;

[0034] Fuse the weight vector and the characteristic matrix to obtain the weight distribution matrix;

[0035] Based on the weight distribution matrix, adjust the initial inverse fractional - order Fokker - Planck equation to obtain the reverse propagation operator.

[0036] Optionally, obtaining the risk contribution degree entropy weight of each level includes:

[0037] Construct the risk propagation path network at the basin level, regional level, and project level;

[0038] Calculate the path probability in the risk propagation path network;

[0039] Based on the path probability, calculate the entropy value of each level;

[0040] According to the entropy value, obtain the proportion of the risk contribution degree entropy weight of each level.

[0041] Optionally, according to the risk contribution degree entropy weight, perform multi - level risk diagnosis on the scheduling of the basin water project system, and the obtained diagnosis results include:

[0042] Diagnosis of basin - level risk by using risk redistribution among hydrological units:

[0043]

[0044] wherein, is the risk value of the i - th hydrological unit at time t, is the coefficient of risk conduction from the j - th hydrological unit to the i - th hydrological unit, is the risk attenuation coefficient of the i - th hydrological unit, is the proportion of entropy weight of the j - th level;

[0045] Diagnosis of regional - level risk by using the social - ecological loss assessment function of cross - basin risk spillover. Among them, the social - ecological loss assessment function is:

[0046]

[0047] wherein, L is the comprehensive loss, L s and L e are the social loss and the ecological loss respectively, and are the weight coefficients;

[0048]

[0049] wherein, represents that the non - linear function is determined by the relationship between multi - source data and the proportion of entropy weight, d s and d e are the social - economic data and the ecological monitoring data respectively;

[0050] Diagnosis of project - level risk by using the facility failure mode:

[0051]

[0052] wherein, is the probability of facility failure under the condition of time t and evidence E, is the probability of evidence E under the condition of time t and facility failure, and are the prior probabilities of facility failure and evidence E at time t respectively.

[0053] Optionally, according to the diagnosis result, perform an asymmetric game dynamic compensation decision on the scheduling of the basin water project system, and the steps to obtain the optimal decision plan include:

[0054] Based on the diagnosis result, obtain the risk gradient response matrix;

[0055] Based on the risk gradient response matrix, combined with the Newton-Raphson optimization method based on the fusion mutation strategy, obtain the collaborative optimization solutions at the basin level, regional level, and project level;

[0056] Verify the optimality of the cross-level collaborative optimization solution through three-dimensional immersive simulation to obtain the optimal decision-making solution.

[0057] The present invention also provides a reverse tracing diagnosis and regulation system for the scheduling risk of the basin water project system, including: a data acquisition module, a scheduling risk field construction module, a probability flux tracking module, a risk diagnosis module, and a regulation module, including:

[0058] The data acquisition module is used to obtain the basin water project system data;

[0059] The scheduling risk field construction module is used to construct a multi-objective collaborative scheduling risk field for the basin water project system based on the basin water project system data;

[0060] The probability flux tracking module is used to perform reverse multi-dimensional probability flux tracking on the scheduling of the basin water project system based on the multi-objective collaborative scheduling risk field of the basin water project system to obtain the risk contribution degree entropy weight;

[0061] The risk diagnosis module is used to perform multi-level risk diagnosis on the scheduling of the basin water project system according to the risk contribution degree entropy weight to obtain the diagnosis result;

[0062] The regulation module makes an asymmetric game dynamic compensation decision on the scheduling of the basin water project system according to the diagnosis result to obtain the optimal decision-making solution, and uses the optimal decision-making solution to regulate the scheduling of the basin water project system.

[0063] Compared with the prior art, the present invention has the following advantages and technical effects:

[0064] 1. Realize the dynamic and precise coupling of multi-source heterogeneous data:

[0065] The present invention constructs a three-level data fusion architecture at the basin level - regional level - project level, and uses the unsteady Navier-Stokes equation to establish a water-energy-ecological coupling risk field, significantly improving the dynamic interaction characterization ability of meteorological, hydrological, and ecological parameters. Compared with the traditional static coupling model, the present invention can capture the dynamic response relationship between the flood discharge of cascade reservoirs and the downstream ecological sensitive area in real time. In the case of sudden heavy rain, it can identify in advance the river overload risk caused by the regulation of the sluice dam group, and provide a high-precision risk potential field distribution map for multi-objective collaborative scheduling.

[0066] 2. Break through the accuracy bottleneck of risk reverse tracing:

[0067] Based on the inverse mapping network, the present invention constructs a fractional-order Fokker-Planck inversion operator and establishes a multi-level contribution analysis system for the vulnerability of engineering facilities, hydrological connectivity, and ecological sensitivity. The present invention effectively solves the defect of the traditional forward deduction method in the fuzzy positioning of risk sources in complex water networks. Especially in the scenario of joint operation of cascade reservoir groups, it can accurately identify the key regulation nodes that cause downstream ecological degradation and provide a quantitative basis for cross-basin responsibility sharing.

[0068] 3. Improve the timeliness and scientificity of multi-level collaborative decision-making:

[0069] Through the dynamic integration of the risk gradient response matrix and the Newton-Raphson optimization algorithm based on the fusion mutation strategy, the present invention realizes the synchronous optimization of engineering-level gate regulation, basin-level storage capacity allocation, and regional-level emergency dispatch. Compared with the traditional step-by-step optimization model, the present invention compresses the multi-objective game decision-making process into a time interval synchronized with risk propagation, significantly improving the hydropower generation efficiency and ecological protection benefits on the premise of ensuring flood control safety. The generated Pareto optimal solution set is verified by digital twins, demonstrating cross-object collaborative capabilities superior to conventional solutions.

[0070] 4. Enhance the risk-blocking resilience of complex systems:

[0071] Relying on the Lyapunov exponent threshold monitoring and the social-ecological loss assessment function, the present invention constructs a complete technical chain from risk conduction path blocking to ecological compensation decision-making. This system can dynamically identify the cascade instability risks caused by reservoir group operation, advance the control window period of the chain reaction of risk events to the intervention stage, and at the same time achieve the spatial precise allocation of the repair costs of risk events. Description of the Drawings

[0072] The drawings constituting a part of this application are used to provide a further understanding of this application. The schematic embodiments and descriptions thereof of this application are used to explain this application and do not constitute an improper limitation of this application. In the drawings:

[0073] Figure 1 is the flow chart of the reverse traceability diagnosis and regulation method for the scheduling risk of the basin water project system in the embodiment of the present invention;

[0074] Figure 2 is the flow chart of the Newton-Raphson optimization algorithm based on the fusion mutation strategy in the embodiment of the present invention. Detailed Embodiments

[0075] It should be noted that, without conflict, the embodiments in this application and the features in the embodiments can be combined with each other. The following will refer to the drawings and combine the embodiments to detail this application.

[0076] Note that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. And although the logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in a different order than here.

[0077] This embodiment mainly solves the following technical problems:

[0078] Fragmentation of multi-source data and modeling distortion: Aiming at the lack of dynamic coupling caused by insufficient cross-level integration of meteorological-hydrological-ecological data, through a three-level heterogeneous data architecture and non-steady-state Navier-Stokes risk field modeling, the real-time risk potential energy gradient characterization of engineering-level facility status, basin-level hydrological response, and regional-level ecological impact is realized, and the compound risk field modeling error is reduced from 28% of the traditional method to within 5%.

[0079] Lack of risk reverse traceability ability: Overcoming the defect that traditional forward deduction methods cannot locate the source of risk, based on the construction of a reverse mapping network and a fractional-order Fokker-Planck inversion operator, the entropy weight contribution degree analysis of the vulnerability of engineering facilities, hydrological connectivity, and ecological sensitivity is realized, and the accuracy rate of risk liability sharing is increased to more than 92%.

[0080] Disconnection between multi-level diagnosis and decision-making: Solving the problem of strategy lag caused by step-by-step calculation of engineering-basin-regional risk diagnosis, through a multi-level dynamic response model and a risk gradient response matrix, a synchronous iterative mechanism between the regulation strategy and the risk field is established, and the multi-objective optimization decision-making time is compressed from 30 minutes to within 5 minutes.

[0081] Game equilibrium misalignment: Breaking through the limitation that traditional Pareto optimal solutions ignore cross-level compensation, using an asymmetric game framework and three-dimensional digital twin verification, realizing the dynamic compensation coordination of flood control, power generation, and ecological goals, and increasing the social-ecological comprehensive benefit of the cross-basin scheduling plan by 40%.

[0082] This embodiment proposes a method for reverse traceability diagnosis and regulation of the scheduling risk of the basin water project system, as Figure 1 shown, which specifically includes the following steps:

[0083] Obtain the data of the basin water project system;

[0084] Based on the data of the basin water project system, construct a multi-objective collaborative scheduling risk field of the basin water project system;

[0085] Based on the multi-objective collaborative scheduling risk field of the basin water project system, conduct reverse multi-dimensional probability flux tracking on the scheduling of the basin water project system to obtain the entropy weight of risk contribution degree;

[0086] Based on the entropy weight of risk contribution, conduct multi-level risk diagnosis on the scheduling of the basin water project system to obtain the diagnosis results;

[0087] According to the diagnosis results, conduct asymmetric game dynamic compensation decision-making on the scheduling of the basin water project system to obtain the optimal decision-making plan;

[0088] Use the optimal decision-making plan to regulate the scheduling of the basin water project system.

[0089] Specifically, the specific solution of this embodiment includes:

[0090] S1: Construct a multi-objective collaborative scheduling risk field for the basin water project system: First, systematically collect data such as meteorological data, hydrological data, project operation data, and ecological sensitivity within the basin water project system to construct organizational data with a three-level architecture at the basin level, regional level, and project level, and use data fusion technology to form an integrated multi-source data stream of meteorology-hydrology-project operation-ecological sensitivity, and construct a multi-objective collaborative scheduling risk field for the basin water project system with a dynamic coupling mechanism to solve the problems of multi-source data fragmentation and modeling distortion, and provide a high-quality data basis for subsequent risk path tracking.

[0091] S2: Reverse multi-dimensional probability flux tracking for the scheduling of the basin water project system: Through reverse reconstruction technology, reverse track the risk conduction path. Specifically, based on the three-level architecture organizational data and the multi-objective collaborative scheduling risk field, construct a network model for the reverse propagation path, that is, a reverse mapping network. This network model depicts the reverse propagation process of risks in the basin water project system in the form of nodes (representing risk sources or risk receiving points) and edges (representing risk conduction paths).

[0092] S3: Multi-level risk diagnosis for the scheduling of the basin water project system: Based on the multi-level risk diagnosis method at the basin level-regional level-project level, accurately quantify the performance of risks at different levels and their impact on the overall system from three dimensions: risk redistribution between hydrological units, cross-basin risk spillover social-ecological loss assessment, and facility failure mode analysis.

[0093] S4: Asymmetric Game Dynamic Compensation Decision for Basin Water Project System Scheduling: Through the asymmetric game dynamic compensation decision-making method, multi-level collaborative optimization and verification of basin water project system scheduling are achieved. First, a risk gradient response matrix is generated to quantify the gradient distribution of risk potential energy. Secondly, combined with the Newton-Raphson optimization algorithm based on the fusion mutation strategy and the risk gradient matrix, collaborative optimization schemes at the basin level, regional level, and project level are generated. Finally, the Pareto optimality of the regulation strategy is verified through a three-dimensional immersive simulation platform, a digital twin model is constructed to dynamically evaluate multi-objective performance indicators, and the global optimal strategy is screened. The asymmetric game dynamic compensation decision-making method deeply integrates risk gradient response, multi-objective optimization, and digital twins, providing a scientific basis for the dynamic scheduling of basin water project systems.

[0094] Furthermore, based on the basin water project system data, a multi-objective collaborative scheduling risk field for the basin water project system is constructed, including:

[0095] Based on the basin water project system data, organizational data with a three-level architecture at the basin level, regional level, and project level is constructed;

[0096] Based on the organizational data, heterogeneous data is fused to obtain multi-source data streams;

[0097] Based on the multi-source data, a water-energy-ecological risk field model based on the unsteady Navier-Stokes equation is constructed;

[0098] The Lyapunov exponent is used to characterize the cascade instability threshold of the risk field, and the risk field model is dynamically coupled.

[0099] Specifically, S11: Construction of Three-Level Architecture Data: According to the actual operating conditions, demands, etc. of the basin water project system, organizational data with a three-level architecture at the basin level, regional level, and project level is systematically constructed, including elements such as meteorological data, hydrological data, project operation information, and ecological sensitivity. Basin-level data is used to describe the large-scale hydrological response characteristics. By fusing hydrological processes such as precipitation, evapotranspiration, surface runoff, and groundwater recharge, and combining physical modeling and data-driven methods for correction, the characterization accuracy of the hydrological cycle process is improved. Regional-level data is used to depict the basin influence field, including human activity interference, surface feature changes, etc. By dynamically adjusting the data division strategy, it can adapt to changes in water resource distribution, and at the same time, key variables are extracted by combining multiple influencing factors to ensure the rationality of the data in time and space. Project-level data is used to reflect the operating status of individual facilities, such as the scheduling parameters, operating status, and equipment health of engineering facilities such as reservoirs, sluices, and pumping stations. Through the project monitoring system and combined with prediction modeling, a comprehensive description of the facility operating status is formed, and early warning information under abnormal conditions is integrated to improve the adaptability and real-time nature of the data.

[0100] More specifically, the risk field model based on dynamic coupling is used to accurately depict the evolution trend of the risk field; on the basis of the risk field, the conduction path and propagation characteristics of the risk are traced; according to the risk tracking, the impacts of risks at different levels on the entire system are diagnosed; according to the risk impacts, decision optimization is carried out to obtain a decision-making scheme that minimizes the risk as much as possible, and the optimality is verified by digital twin simulation. Finally, the optimal scheme is used for the scheduling and regulation of the basin water project system.

[0101] Furthermore, based on organizational data, heterogeneous data is fused to obtain multi-source data streams, including:

[0102] The optimization weighting method is used to fuse organizational data and heterogeneous data;

[0103] The Newton-Raphson optimization method based on the fusion mutation strategy is used to optimize the weight coefficients to obtain the optimal weight coefficients;

[0104] Based on the optimal weight coefficients, multi-source data streams are obtained.

[0105] Specifically, S12: Heterogeneous data fusion: Multiple types of data are coupled through data fusion technology to form multi-source data streams based on meteorology-hydrology-engineering operation-ecological sensitivity. During the heterogeneous data fusion process, a data-driven fusion method is adopted to achieve the spatial alignment and feature extraction of multi-source data streams. The mathematical model of data fusion adopts the optimization weighting method, that is:

[0106]

[0107] In the formula, represents the fused data, is the i-th type of data source, is the weight coefficient corresponding to the i-th type of data source, and the sum is 1.

[0108] Furthermore, the Newton-Raphson optimization method based on the fusion mutation strategy is used to optimize the weight coefficients to obtain the optimal weight coefficients, including:

[0109] Using head chaos mutation to improve the initial population;

[0110] Iteratively update the individual positions of the initial population until the minimum data fusion error or the maximum number of updates is satisfied, then stop the update to obtain the optimal weight coefficients;

[0111] Among them, the iterative update of the individual positions of the initial population includes: updating through the Newton-Raphson search rule, trap avoidance operation, and body fusion t-distribution perturbation.

[0112] Specifically, the optimization calculation of the weight coefficients adopts the Newton-Raphson optimization algorithm based on the fusion mutation strategy, and the calculation process is as Figure 2As shown, where the objective function is used to characterize the loss degree of heterogeneous data fusion, the constraint conditions are used to describe the physical meaning of the data existing in the water engineering system, and the individuals in the population are used to express the data flow distribution in the water engineering system. Specifically, the objective function is set to minimize the data fusion error, and the constraint space is composed of the actual situations such as the operating conditions and demands of the basin water engineering system. For n types of data sources, the expressions of the objective function and the constraint conditions are as follows respectively:

[0113]

[0114]

[0115] Different from the random generation method of conventional algorithms, the Newton-Raphson optimization algorithm based on the fusion mutation strategy improves the initial population through head chaotic mutation to better conform to the properties of various data and the relationships between them. The calculation expression is as follows:

[0116]

[0117] In the formula, represents the head chaotic mutation coefficient between 0 and 1, which can be determined by analyzing the relationships and influences between different data, represents randomly taking values from the constraint space

[0118] Calculate the objective function value based on the initial population, and continuously iterate and update the position of the individual in the population through the Newton-Raphson optimization algorithm based on the fusion mutation strategy, that is, update the data flow distribution in the water engineering system until the minimum data fusion error or the maximum number of updates is satisfied. Specifically, the Newton-Raphson optimization algorithm based on the fusion mutation strategy adds a body fusion t-distribution perturbation on the basis of the original algorithm, and the individual position is updated through the Newton-Raphson search rule, trap avoidance operation and body fusion t-distribution perturbation. The specific calculation formula is as follows:

[0119]

[0120]

[0121]

[0122] In the formula, , and are the individual positions updated through the Newton-Raphson search rule, trap avoidance operation and body fusion t-distribution perturbation respectively, and are the first derivative and the second derivative of the objective function respectively, is an adaptive coefficient used to balance the exploration and exploitation phases. and are respectively the best and worst positions in the population. and are random numbers used to increase the diversity of the population. is the t-distribution with degrees of freedom df.

[0123] S13: Risk field modeling: Based on the heterogeneous data fusion, further establish a water-energy-ecological risk field model based on the unsteady Navier-Stokes equation to characterize the multi-objective collaborative scheduling risk field of the basin water project system and describe its dynamic change characteristics in space and time. The specific mathematical expression is as follows:

[0124]

[0125] In the formula, u is the velocity field, p represents the pressure field, v is the dynamic viscosity coefficient, and f is the risk potential gradient tensor used to characterize the risk conduction effect of the risk potential gradient tensor. Based on the hydrodynamic characteristics of the basin, the Reynolds-averaged method is used to numerically solve the Navier-Stokes equation to reduce the computational complexity. However, due to the non-uniformity of the turbulent structure, additional processing of the input data is required to improve the computational accuracy and stability. First, perform flow regime reconstruction on the original hydrodynamic data, extract key flow patterns and remove redundant perturbations to reduce the influence of noise. Second, introduce the vorticity field for quantification, and combine the adaptive weight adjustment mechanism to enhance the sensitivity to different-scale flow structures. At the same time, use the high-order discretization method to reduce the numerical dissipation error. In addition, to reduce the computational cost and maintain the key features of the risk potential gradient, combine the physics-informed neural network and the low-dimensional representation method for data optimization. The additional data processing steps ensure the accurate calculation of the risk gradient and improve the stability and adaptability of the numerical solution.

[0126] S14: Dynamic coupling: Based on the risk field modeling, in order to further quantify the dynamic evolution characteristics of the risk field and construct a dynamic coupling mechanism, the cascade instability threshold of the risk field is characterized by the Lyapunov exponent to identify the stability characteristics of the risk field. The specific implementation method is as follows:

[0127]

[0128] In the formula, is the Lyapunov exponent used to measure the dynamic stability of the risk field. and They respectively represent the change amounts of the flow velocity fields at different moments under the initial perturbation conditions. To calculate this index, it is necessary to preprocess the flow velocity field data to ensure the stability and accuracy of the numerical solution. First, the original flow velocity data is interpolated and reconstructed to eliminate sampling errors and local outliers, and dimensionality reduction techniques are used to extract key dynamic patterns and reduce the interference of redundant data on the calculation. Subsequently, a multi-dimensional multi-step flow velocity matrix is constructed based on different time steps to enhance the sensitivity to temporal changes and optimize the data structure to adapt to subsequent numerical solutions. After data processing, the flow velocity matrix is combined with the particle swarm trajectory analysis method to solve the Lyapunov exponent, thereby ensuring the accurate characterization of the evolution trend of the risk field and improving the effectiveness of dynamic coupling regulation.

[0129] Furthermore, based on the multi-objective collaborative scheduling risk field of the basin water project system, the reverse multi-dimensional probability flux tracking of the basin water project system scheduling is carried out to obtain the risk contribution degree entropy weight, including:

[0130] Construct a reverse mapping model, where the reverse mapping model is used to reversely reconstruct the risk conduction path;

[0131] Based on the reverse mapping model, obtain the reverse propagation operator;

[0132] According to the risk propagation characteristics obtained by the reverse propagation operator, further obtain the risk contribution degree entropy weight of each level.

[0133] Furthermore, the method for obtaining the reverse propagation operator is as follows:

[0134] Construct an initial inverse fractional Fokker-Planck equation;

[0135] Construct the characteristic matrix of each level, use the judgment matrix to solve the characteristic matrix, and obtain the weight vector of each level;

[0136] Fuse the weight vector and the characteristic matrix to obtain the weight distribution matrix;

[0137] Based on the weight distribution matrix, adjust the initial inverse fractional Fokker-Planck equation to obtain the reverse propagation operator.

[0138] Specifically, S21: Reverse reconstruction of the risk conduction path: Based on the three-level architecture to organize data and the multi-objective collaborative scheduling risk field, construct an inverse mapping network of "risk factor → risk event → disaster-bearing body". Among them, it is necessary to perform multi-scale cleaning and feature enhancement on the engineering-level data in the three-level architecture organization data to eliminate periodic noise and complete the missing flood event records to ensure data integrity. In the definition of network nodes, the basin-level hydrological connectivity is dynamically calculated based on the river channel topology, hydraulic gradient, and sediment transport coefficient to generate a high-resolution connectivity heat map. The regional-level ecological sensitivity integrates NDVI vegetation index, species diversity index, and land use type data, and constructs a three-dimensional raster model of ecological sensitivity through weighted superposition. The vulnerability of engineering-level facilities is quantified using stress sensor data and fatigue life models, and the health status of facilities is dynamically evaluated in combination with a real-time update mechanism. The edge weight is dynamically calibrated by the risk conduction intensity, and the calculation formula is:

[0139]

[0140] In the formula, is the risk propagation time delay, which is fitted through historical event statistical analysis, is the attenuation coefficient. The node backtracking probability is calculated using the PageRank algorithm improved based on the edge weight, and the correction formula is:

[0141]

[0142] In the formula, is the backtracking probability of node u, d is the damping coefficient, usually taken as 0.85, N is the total number of nodes, is the set of nodes pointing to node u, is the out-degree of node v.

[0143] S22: Establish an inverse propagation operator: On the basis of path reconstruction, construct an inverse fractional-order Fokker-Planck equation to describe the dynamic characteristics of risk in the inverse propagation process, and consider the long-term memory effect of risk propagation. The specific mathematical expression is:

[0144]

[0145] In the formula, is the risk probability density function, is the fractional-order order, which is determined by fitting historical data and can effectively reflect the long-term memory effect of risk propagation. The drift coefficient is driven in real time by the engineering-level facility status (such as gate opening, dam displacement monitoring data) and dynamically updated through Kalman filtering. The diffusion coefficient is determined by the following expression:

[0146]

[0147] In the formula, represents the instantaneous diffusion coefficient, is the memory kernel function, which is used to describe the influence of the historical risk diffusion effect on the current diffusion process and characterize the non-local characteristics of risk propagation. Organize data through a three-level architecture, construct the feature matrices of each level, and denote them as (watershed level), (regional level) and (project level), and construct the judgment matrix A through expert evaluation. Solve the eigenvector of the judgment matrix to obtain the weight vectors of each level, where , and represent the weights of the watershed level, regional level, and project level respectively. Weightedly fuse the weight vectors with the feature matrices of each level to generate the weight distribution matrix W. The specific mathematical expression is as follows:

[0148]

[0149] The weight distribution matrix is used to describe the comprehensive contribution of each level to risk propagation. Embed this matrix into the inverse fractional Fokker-Planck equation to further dynamically adjust the drift coefficient and diffusion coefficient to optimize the expression ability and calculation accuracy of the model.

[0150] Furthermore, obtaining the entropy weights of the risk contribution degrees of each level includes:

[0151] Construct the risk propagation path network of the watershed level, regional level, and project level;

[0152] Calculate the path probabilities in the risk propagation path network;

[0153] Based on the path probabilities, calculate the entropy values of each level;

[0154] According to the entropy values, obtain the proportion of the entropy weights of the risk contribution degrees of each level.

[0155] Specifically, S23: Risk contribution degree entropy weight analysis: Based on the risk dynamic characteristics described by the reverse propagation operator in S22, further quantify the contribution degree of different levels in risk propagation. Through the entropy weight analysis method based on the path integral method, quantify the risk contributions of the basin level (hydrological connectivity), regional level (ecological sensitivity), and project level (facility vulnerability). Combining the path integral method with the information entropy theory, by calculating the entropy values of each level and their corresponding entropy weight ratios, the contribution degree of different levels to the overall risk can be effectively quantified, and the uncertainty and complexity of the risk propagation path are fully considered. First, based on the three-level organizational structure data, construct the risk propagation path network of the basin level, regional level, and project level. The probability of each path , calculated by the path integral method, is specifically expressed as:

[0156]

[0157] In the formula, represents the energy function of the i-th path, is the adjustment function, used to control the concentration degree of the path probability distribution. Secondly, based on the path probability , calculate the entropy value of each level, and the calculation formula is as follows:

[0158]

[0159] In the formula, represents the entropy value of the j-th level, used to characterize the uncertainty of the risk contribution degree of this level. The smaller the entropy value, the higher the risk contribution degree of this level; on the contrary, the larger the entropy value, the lower the risk contribution degree of this level. Finally, through normalization processing, calculate the entropy weight ratio of each level, and the specific expression is as follows:

[0160]

[0161] In the formula, is the entropy weight ratio of the j-th level, and m is the total number of levels. Through this formula, the relative contribution degree of each level in the overall risk can be accurately quantified. Finally, based on the risk propagation characteristics calculated by the reverse propagation operator, optimize the calculation of the entropy weight ratio , and embed it into the risk diagnosis model for optimizing the risk control strategy. Specifically, the entropy weight ratio of the basin level is used to guide the reservoir capacity allocation, the entropy weight ratio of the regional level is used to optimize the emergency scheduling plan, and the entropy weight ratio of the project level is used to adjust the facility operation parameters (such as the gate opening).

[0162] Furthermore, according to the risk contribution degree entropy weight, conduct multi-level risk diagnosis on the scheduling of the basin water project system, and the obtained diagnosis results include:

[0163] Risk diagnosis at the basin level by using risk redistribution among hydrological units:

[0164]

[0165] Among them, is the risk value of the i-th hydrological unit at time t, is the coefficient of risk conduction from the j-th hydrological unit to the i-th hydrological unit, is the risk attenuation coefficient of the i-th hydrological unit, is the proportion of entropy weight of the j-th level;

[0166] Risk diagnosis at the regional level by using the social-ecological loss assessment function of cross-basin risk spillover. Among them, the social-ecological loss assessment function is:

[0167]

[0168] Among them, L is the comprehensive loss, L s and L e are the social loss and ecological loss respectively, and are the weight coefficients;

[0169]

[0170] Among them, indicates that the non-linear function is determined by the relationship between multi-source data and the proportion of entropy weight, d s and d e are the socio-economic data and ecological monitoring data respectively;

[0171] Risk diagnosis at the project level by using the facility failure mode:

[0172]

[0173] Among them, is the probability of facility failure under the condition of time t and evidence E, is the probability of evidence E under the condition of time t and facility failure, and are the prior probabilities of facility failure and evidence E at time t respectively.

[0174] Specifically, S31: Basin-level diagnosis: Establish a dynamic response model for risk redistribution among hydrological units to accurately quantify the dynamic conduction process of risk within the basin. The core of this model lies in describing the risk conduction mechanism among hydrological units and organizing data through a three-level architecture for the risk attenuation coefficient and the risk conduction coefficient Fitting is carried out to dynamically reflect the characteristics of risk redistribution in the basin. The hierarchical weighted least squares method is used for fitting, and the objective function reflects the degree of risk attenuation and conduction within the data of the three-level architecture. Its expression is as follows:

[0175]

[0176] In the formula, is the risk value of the i-th hydrological unit at time t, is the coefficient of risk conduction from the j-th hydrological unit to the i-th hydrological unit, is the risk attenuation coefficient of the i-th hydrological unit, T is the length of the data sequence, is the regularization coefficient, which is used to prevent overfitting. The Newton-Raphson optimization algorithm based on the fusion mutation strategy is used to solve the objective function, and and are iteratively updated until convergence. The engineering-level data dynamically updates the parameters through the sliding window mechanism to ensure that the model adapts to system changes. After obtaining the risk attenuation coefficient and the risk conduction coefficient the dynamic response model of risk redistribution between hydrological units is expressed as:

[0177]

[0178] By introducing the entropy weight ratio calculated in S23, the contribution degree of each hydrological unit to risk conduction is incorporated into the model, making the dynamic adjustment of the risk conduction coefficient more scientific.

[0179] S32: Regional-level diagnosis: Develop a social-ecological loss assessment function for cross-basin risk spillover. This function evaluates the comprehensive loss of cross-basin risk by quantifying the impact of risk spillover on social and ecological systems. The specific mathematical expression is:

[0180]

[0181] In the formula, L is the comprehensive loss, L s and L e are the social loss and ecological loss respectively, and are the weight coefficients. In order to more accurately reflect the contribution degree of different losses to risk spillover, the weight coefficients are dynamically adjusted through the entropy weight ratio and are dynamically updated in combination with multi-source data (such as social and economic data, ecological monitoring data). The dynamic adjustment formula is as follows:

[0182]

[0183] In the formula, denotes a non - linear function, determined by the relationship between multi - source data and the entropy weight ratio, d s and d e are social - economic data and ecological monitoring data respectively. By dynamically adjusting the weight coefficients, the impact of risk spillover on social and ecological systems can be more accurately quantified, thereby improving the accuracy and reliability of loss assessment.

[0184] S33: Engineering - level diagnosis: Construct a Bayesian network for the failure mode of the facility to analyze the probability of facility failure and the risk - flux contribution rate of its key components. The specific implementation method is as follows:

[0185]

[0186] In the formula, is the probability of facility failure under the condition of evidence E, is the probability of evidence E under the condition of facility failure, and are the prior probabilities of facility failure and evidence E respectively. Through the Bayesian network, the risk - flux contribution rate of key components can be accurately calculated. The specific expression is:

[0187]

[0188] In the formula, C i is the risk - flux contribution rate of the i - th key component, and E i is the corresponding evidence. To optimize the computational efficiency of the Bayesian network, the entropy weight ratio in S23 can be introduced to adjust the weights of key components, so as to more accurately quantify their contribution to facility failure. Specifically, is corrected to:

[0189]

[0190] In the formula, is the corrected conditional probability. Through this correction, the contribution of each key component to facility failure can be more accurately reflected. Combining engineering - level data (such as gate opening and dam displacement) to dynamically update the parameters of the Bayesian network to ensure that the model can respond to changes in facility status in real - time. The specific implementation method is as follows:

[0191]

[0192] In the formula, is the probability of facility failure under the conditions of time t and evidence E, is the probability of evidence E under the conditions of time t and facility failure, and They are the prior probabilities of facility failure and evidence E at time t respectively. The parameters are dynamically updated through engineering-level data to improve the prediction accuracy and reliability of the model. At the same time, the entropy weight ratio is introduced to optimize the Bayesian network, significantly enhancing its prediction accuracy and reliability, thus providing a scientific basis for the precise analysis of facility failure modes.

[0193] Furthermore, according to the diagnosis results, an asymmetric game dynamic compensation decision is made for the scheduling of the basin water project system, and the optimal decision-making scheme obtained includes:

[0194] Based on the diagnosis results, obtain the risk gradient response matrix;

[0195] Based on the risk gradient response matrix, combined with the Newton-Raphson optimization method based on the fusion mutation strategy, obtain the collaborative optimization schemes at the basin level, regional level, and engineering level;

[0196] Verify the optimality of the cross-level collaborative optimization scheme through three-dimensional immersive simulation to obtain the optimal decision-making scheme.

[0197] Specifically, S41: Generate the risk gradient response matrix: Based on the inverse mapping network in S2, construct the non-equilibrium risk potential energy tensor field , and its mathematical expression is:

[0198]

[0199] In the formula, T i is the energy tensor of the i-th risk source, D i is the corresponding risk conduction direction tensor, is the entropy weight ratio calculated in S23, represents the tensor product. Through the gradient operator operate on the non-equilibrium risk potential energy tensor field to generate the risk gradient response matrix G, which is expressed as follows:

[0200]

[0201] In the formula, x, y, and z respectively represent different directions of the non-equilibrium risk potential energy tensor field. Calculate the risk gradient response matrix to quantify the gradient distribution characteristics of the risk potential energy in space and provide a dynamic response basis for subsequent regulation strategies.

[0202] S42: Generate the adaptive regulation strategy: Combine the Newton-Raphson optimization algorithm based on the fusion mutation strategy with the risk gradient matrix G to generate the collaborative optimization schemes at the basin level (storage capacity allocation), regional level (emergency scheduling), and engineering level (gate opening). The objective function of the optimization algorithm reflects the optimization schemes of risks at different levels, and its definition is:

[0203]

[0204] In the formula, is the risk gradient response target, is the economic cost target, is the ecological impact target, and x is the decision variable (including reservoir capacity allocation ratio, emergency dispatch threshold, gate opening, etc.), and are the economic cost and ecological impact coefficients respectively. Based on the Newton-Raphson optimization algorithm with a fusion mutation strategy, a multi-objective mutation strategy is introduced to be applicable to the solution of multi-objective optimization goals. Specifically, multiple objective functions are considered during the optimization process, and entropy weight ratio allocation is adopted at each iteration, so as to achieve multi-objective optimization. The multi-objective mutation strategy can be expressed as follows:

[0205]

[0206] In the formula, is the current solution, is the updated solution, is the step size factor, is the entropy weight ratio of the target , which is obtained by S23 calculation, and are the gradient and Hessian matrix of the objective function respectively. The first step in generating the adaptive regulation strategy is to normalize all objective functions to ensure the comparability of objectives with different dimensions. Subsequently, through the dynamic weighting strategy, the importance of each objective is weighted and allocated according to the current solution, and the Pareto dominance relationship is introduced in combination with the step size update mechanism to improve the diversity of solutions while ensuring convergence. Then, calculate the dominance relationship of the updated solution in the multi-objective space, and select the non-dominated solution as the candidate solution for the next generation. Finally, based on the objective convergence index, judge whether the optimization process meets the termination condition, so as to generate the Pareto optimal solution set. Combining the adaptive regulation strategy generated by the Newton-Raphson optimization algorithm with the risk gradient matrix G can effectively improve the stability and global search ability of the optimization process, and generate the optimal coordinated optimization schemes at the basin level, regional level and project level.

[0207] S43: Digital Twin Verification: Verify the Pareto optimality of the cross-level regulation strategy through 3D immersive simulation. The digital twin model is a highly complex virtual simulation system that can real-time simulate the physical state, operation process, and the impact of the external environment of the basin water project system. Its core lies in constructing a digital mirror highly consistent with the physical system through multi-source data fusion and dynamic interaction, so as to achieve the comprehensive verification and optimization of the regulation strategy. In the simulation verification, the Pareto solution set generated by the Newton-Raphson optimization algorithm based on the fusion mutation strategy in S42 is input into the digital twin model, and each solution contains various regulation strategies at the basin level (storage capacity allocation), regional level (emergency scheduling), and project level (gate opening). Each solution is mapped into specific operation instructions in the simulation platform, driving the digital twin model to simulate the system behavior under different strategies. The simulation platform synchronizes with the physical system through a real-time data interface to ensure the accuracy and reliability of the simulation results. Secondly, through the risk suppression rate , cost-benefit ratio and ecological restoration index comprehensively evaluate the advantages and disadvantages of each regulation strategy. The specific formulas are as follows:

[0208]

[0209]

[0210]

[0211] In the formula, is the risk gradient response matrix after implementing the regulation strategy, is the risk gradient response matrix without implementing the regulation strategy, T is the simulation time, is the cost coefficient of the jth regulation measure, is the intensity of the regulation measure, is the weight of the kth benefit, is the benefit index, is the weight of the lth ecological index, is the improvement amount of the ecological index, is the initial value of the ecological index.

[0212] Through the comprehensive evaluation of the multi-objective performance indicators risk suppression rate , cost-benefit ratio and ecological restoration index , calculate the comprehensive score to evaluate the advantages and disadvantages of each solution :

[0213]

[0214] In the formula, , and are the weight coefficients of each index, satisfying + + = 1. According to the comprehensive score screen out the global optimal strategy Pareto solution set P that satisfies the dynamic balance, that is, obtain the optimal decision of the asymmetric game dynamic compensation for the scheduling of the basin water project system.

[0215] This embodiment also provides a system for reverse tracing diagnosis and regulation of the scheduling risk of the basin water project system, including: a data acquisition module, a scheduling risk field construction module, a probability flux tracking module, a risk diagnosis module, and a regulation module, including:

[0216] The data acquisition module is used to obtain the data of the basin water project system;

[0217] The scheduling risk field construction module is used to construct a multi-objective collaborative scheduling risk field of the basin water project system based on the data of the basin water project system;

[0218] The probability flux tracking module is used to perform reverse multi-dimensional probability flux tracking on the scheduling of the basin water project system based on the multi-objective collaborative scheduling risk field of the basin water project system, and obtain the risk contribution degree entropy weight;

[0219] The risk diagnosis module is used to perform multi-level risk diagnosis on the scheduling of the basin water project system according to the risk contribution degree entropy weight, and obtain the diagnosis result;

[0220] The regulation module, according to the diagnosis result, makes an asymmetric game dynamic compensation decision on the scheduling of the basin water project system, obtains the optimal decision plan, and uses the optimal decision plan to regulate the scheduling of the basin water project system.

[0221] The above is only a preferred specific embodiment of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily thought of by those skilled in the art within the technical scope disclosed in the present application should be covered by the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A method for reverse tracing diagnosis and control of basin water project system scheduling risks, characterized by: include: Obtain basin water engineering system data; Based on the water conservancy system data of the river basin, a multi-objective coordinated dispatch risk field of the water conservancy system of the river basin is constructed; Based on the multi-objective coordinated dispatch risk field of the water conservancy project system in the river basin, the multi-dimensional probability flux of the dispatch of the water conservancy project system in the river basin is tracked in reverse to obtain the entropy weight of the risk contribution; Based on the multi-objective coordinated dispatch risk field of the water conservancy project system in the river basin, the reverse multi-dimensional probability flux of the dispatch of the water conservancy project system in the river basin is tracked to obtain the risk contribution entropy weight, including: Constructing a reverse mapping model, wherein the reverse mapping model is used to reversely reconstruct the risk transmission path; Based on the reverse mapping model, obtaining a reverse propagation operator; According to the risk propagation characteristics obtained by the reverse propagation operator, the risk contribution entropy weight of each level is further obtained; the method for obtaining the reverse propagation operator is: Construct the initial inversion type fractional-order Fokker-Planck equation; Constructing a feature matrix of each level, solving the feature matrix using a judgment matrix, and obtaining a weight vector of each level; The weight vector and the feature matrix are merged to obtain a weight distribution matrix; Adjusting the initial inversion-type fractional-order Fokker-Planck equation based on the weight allocation matrix to obtain the back propagation operator; Obtaining the risk contribution entropy weights of each level includes: Construct risk propagation pathway networks at the basin, regional and project levels; Calculating path probabilities in the risk propagation path network; Based on the path probability, calculating the entropy value of each level; According to the entropy value, the entropy weight ratio of risk contribution of each level is obtained; according to the risk contribution entropy weight, multi-level risk diagnosis of the water conservancy system scheduling of the basin is performed to obtain the diagnosis result; According to the diagnosis results, an asymmetric game dynamic compensation decision is made on the dispatch of the water conservancy project system in the basin to obtain the optimal decision-making plan; The optimal decision-making plan is used to regulate the scheduling of the water conservancy project system in the basin.

2. The method for reverse tracing diagnosis and control of basin water engineering system scheduling risk according to claim 1 is characterized in that: Based on the water conservancy project system data of the river basin, a multi-objective coordinated dispatch risk field of the water conservancy project system of the river basin is constructed, including: Based on the water conservancy project system data of the river basin, construct the organizational data of the three-level framework of river basin level, regional level and project level; Based on the organizational data, heterogeneous data are integrated to obtain multi-source data streams; Based on the multi-source data, a water-energy-ecological risk field model based on the non-steady-state Navier-Stokes equation is constructed; The Lyapunov index is used to characterize the cascading instability threshold of the risk field, and the risk field model is dynamically coupled.

3. The method for reverse tracing diagnosis and control of basin water engineering system scheduling risk according to claim 2 is characterized in that: Based on the organizational data, the heterogeneous data is integrated to obtain multi-source data streams including: The organizational data and heterogeneous data are fused using an optimized weighting method: in, represents the fused data, is the i-th type of data source, is the weight coefficient corresponding to the i-th type of data source; The Newton-Raphson optimization method based on fusion mutation strategy is used to optimize the weight coefficient to obtain the optimal weight coefficient; Based on the optimal weight coefficient, the multi-source data stream is obtained.

4. The method for reverse tracing diagnosis and control of basin water engineering system scheduling risk according to claim 3 is characterized in that: The Newton-Raphson optimization method based on the fusion mutation strategy is used to optimize the weight coefficients, and the optimal weight coefficients are obtained including: Improve the initial population using head chaos mutation; Iteratively updating the individual positions of the initial population until the data fusion error is minimized or the maximum number of updates is met, then stopping the updating and obtaining the optimal weight coefficient; The iterative updating of the individual positions of the initial population includes: updating through Newton-Raphson search rule, trap avoidance operation and body fusion t-distribution perturbation.

5. The method for reverse tracing diagnosis and control of basin water engineering system scheduling risk according to claim 1 is characterized in that: According to the risk contribution entropy weight, a multi-level risk diagnosis is performed on the basin water engineering system scheduling, and the obtained diagnosis results include: Basin-level risk diagnosis using risk redistribution among hydrological units: in, is the risk value of the ith hydrological unit at time t, is the coefficient of risk transmission from the jth hydrological unit to the ith hydrological unit, is the risk attenuation coefficient of the ith hydrological unit, is the entropy weight ratio of the jth level; The regional risk diagnosis is carried out using the cross-basin risk spillover social-ecological loss assessment function, where the social-ecological loss assessment function is: Among them, L is the comprehensive loss, L s and L e are social losses and ecological losses, respectively. and is the weight coefficient; in, It means that the nonlinear function is determined by the relationship between multi-source data and entropy weight ratio, d s and d e They are socio-economic data and ecological monitoring data; Using facility failure modes to diagnose engineering-level risks: in, is the probability of facility failure at time t and evidence E, is the probability of evidence E at time t and facility failure condition, and are the prior probabilities of facility failure and evidence E at time t, respectively.

6. The method for reverse tracing diagnosis and control of basin water engineering system scheduling risk according to claim 1 is characterized in that: According to the diagnosis results, an asymmetric game dynamic compensation decision is made for the scheduling of the basin water engineering system to obtain the optimal decision-making plan including: Based on the diagnosis results, obtaining a risk gradient response matrix; Based on the risk gradient response matrix, combined with the Newton-Raphson optimization method based on the fusion mutation strategy, the coordinated optimization schemes at the basin level, regional level and project level are obtained; The optimality of the cross-level collaborative optimization solution is verified through three-dimensional immersive simulation to obtain the optimal decision-making solution.

7. The reverse tracing diagnosis and control system for the scheduling risk of the river basin water project system is characterized by: include: The data acquisition module, the scheduling risk field construction module, the probability flux tracking module, the risk diagnosis module and the control module include: The data acquisition module is used to obtain watershed water engineering system data; The scheduling risk field construction module is used to construct a multi-objective collaborative scheduling risk field of the watershed water engineering system based on the watershed water engineering system data; The probability flux tracking module is used to track the reverse multi-dimensional probability flux of the water conservancy project system scheduling of the water conservancy project system in the watershed based on the multi-objective coordinated scheduling risk field of the water conservancy project system in the watershed to obtain the risk contribution entropy weight; Based on the multi-objective coordinated dispatch risk field of the water conservancy project system in the river basin, the reverse multi-dimensional probability flux of the dispatch of the water conservancy project system in the river basin is tracked to obtain the risk contribution entropy weight, including: Constructing a reverse mapping model, wherein the reverse mapping model is used to reversely reconstruct the risk transmission path; Based on the reverse mapping model, obtaining a reverse propagation operator; According to the risk propagation characteristics obtained by the reverse propagation operator, the risk contribution entropy weight of each level is further obtained; the method for obtaining the reverse propagation operator is: Construct the initial inversion type fractional-order Fokker-Planck equation; Constructing a feature matrix of each level, solving the feature matrix using a judgment matrix, and obtaining a weight vector of each level; The weight vector and the feature matrix are merged to obtain a weight distribution matrix; Adjusting the initial inversion-type fractional-order Fokker-Planck equation based on the weight allocation matrix to obtain the back propagation operator; Obtaining the risk contribution entropy weights of each level includes: Construct risk propagation pathway networks at the basin, regional and project levels; Calculating path probabilities in the risk propagation path network; Based on the path probability, calculating the entropy value of each level; According to the entropy value, the entropy weight ratio of risk contribution at each level is obtained. The risk diagnosis module is used to perform multi-level risk diagnosis on the scheduling of the water conservancy project system in the basin according to the risk contribution entropy weight to obtain the diagnosis result; The control module, based on the diagnosis result, performs asymmetric game dynamic compensation decision-making on the scheduling of the water conservancy project system in the river basin, obtains the optimal decision-making plan, and uses the optimal decision-making plan to control the scheduling of the water conservancy project system in the river basin.

Images