A real-time control system and method for operation risk of a water conservancy project

Abstract

The application provides a water conservancy hub operation risk real-time control system and method, and belongs to the technical field of water power and water conservancy, and the system comprises: a risk monitoring subsystem, which is used for constructing a risk transmission link, laying a risk monitoring network, and monitoring the change of an external environment of a hydropower station and the operation of an internal unit in real time through terminal monitoring equipment and manual inspection; a risk analysis subsystem, which is used for analyzing the risk transmission link, determining a risk occurrence point, formulating and screening an optimal risk cutting scheme according to the risk occurrence point, feeding back the cutting situation of the risk to the risk analysis subsystem through a risk control subsystem, and synchronously analyzing the propagation position of the risk by using a parallel simulation method; and the risk control subsystem, which is used for cutting the risk and feeding back the cutting situation of the risk to the risk analysis subsystem. The application realizes multi-dimensional risk perception and control of the hydropower station and improves the risk perception and risk response capability of the hydropower station.

Description

Technical Field

[0001] This invention belongs to the field of hydropower and water conservancy technology, and in particular relates to a real-time control system and method for the operation risk of water conservancy projects. Background Technology

[0002] Numerous and diverse risk sources affect the safe operation of water conservancy projects. Any malfunction in any critical component of a water conservancy project—such as the reservoir, dam, power station, or generator gates—or any event such as natural disasters, equipment failures, operational errors, or management loopholes, can threaten its safe operation. For example, torrential rains or upstream flood discharges can cause floods exceeding standard levels downstream, drastically increasing the dam's flood control pressure and potentially endangering its safety, leading to dam failure. Earthquakes can cause deformation, cracking, or even destruction of dams and other structures. Landslide dams can cause sudden rises in reservoir water levels; if breached, they can create downstream flood peaks, causing dangers such as dam overflow and dam failure. Debris flows can destroy transportation facilities and villages, causing river blockages. Sudden water pollution incidents at reservoirs can rapidly spread, escalating into a water crisis across the entire basin. Inherent quality problems in the project itself can lead to landslides, deformation, cracks, and seepage damage, potentially triggering dam failures or other major emergencies. Insufficient emergency flood discharge and emptying capacity of high dams and large reservoirs, prolonged operation of hydropower station turbines in unstable zones, loopholes in safety management, and poor coordination of basin-wide scheduling during operation all pose threats to the safety of water conservancy projects. Furthermore, the complex structures of large-scale water conservancy projects involve numerous planning, construction, and implementation stages, with interdependent constraints among these stages. These early deficiencies accumulate and become potential risks for the later operation of the water conservancy projects.

[0003] Currently, when risks arise during the operation of water conservancy projects, the management unit typically identifies the risks and, based on the location of the risk, requires professional safety inspectors to implement risk control measures. This process often requires no cooperation from other professionals. If multiple departments are involved, the management unit decides on comprehensive countermeasures based on the location and type of the risk, with relevant professionals proposing solutions. This risk control approach is generally effective for single-type, common risks, and many risks can be correctly addressed and effectively managed. However, the risks are not eradicated; latent risks may still spread to other parts of the water conservancy project and potentially affect the entire cascade, representing a systemic risk. It is precisely because of our limited understanding of these potential risks and the connections between them, and the lack of a systematic approach to their control, that extreme or sudden events can lead to a series of derivative disasters, causing systemic risks to spiral out of control.

[0004] In fact, mismanagement of a single risk at a single site can trigger a butterfly effect, causing a series of risk events. Traditional risk control methods, primarily based on risk observation, data analysis, and manual decision-making, represent an independent control approach for a single risk at a single water conservancy project. This approach fails to consider the correlation between this risk and other risks or the safe operation of other water conservancy projects, making it difficult to effectively identify potential risks in cascade water conservancy projects, let alone monitor and control the propagation of such risks in real time. Advances in information and intelligent technologies have made real-time control of operational risks in complex water conservancy projects possible. When a single water conservancy project requires risk control by a specific professional, the control results only need to meet the risk management requirements of that department. However, the risk is often not completely eradicated and may still cause secondary hazards, such as disasters affecting other departments of the water conservancy project and upstream and downstream cascades. The biggest drawback of this risk control approach is that it fails to recognize the disaster correlation that can occur when risks occur at large water conservancy projects. Without completely cutting off the risk, there is a possibility of cascading disasters as the risk propagates within the water conservancy project, leading to subsequent uncontrollable impacts.

[0005] The difficulty in risk control for cascade hydropower projects lies in their broad temporal and spatial scale. Every point in the basin can potentially become the starting point for basin-wide risks, evolving into a series of risk events that ultimately affect the operational safety of the entire cascade hydropower project. Risk control may also generate new secondary impacts, posing dangers to other hydropower stations in the basin and affecting the normal operation of the cascade hydropower projects. Adjustments to the benefits of one hydropower station within the cascade hydropower project may also have secondary impacts, negatively affecting the operation of other hydropower stations.

[0006] These risk events, when applied to operating water conservancy projects, can be categorized into four main areas: flood control, power generation, navigation, and ecology, as well as other less frequent events. When considering a single power station, there are numerous solutions available to address these risks, maximizing the benefits of that station. However, when considering the overall benefits of a cascade water conservancy project, the available experience is scarce. Currently, there is no clear network regarding the risk transmission relationships among cascade water conservancy projects in a river basin. The joint scheduling of these projects for risk control is rigid and cannot be adjusted in real time according to actual conditions. Therefore, there is an urgent need for a method to effectively address risk events occurring during the operation of cascade water conservancy projects. This method should improve the risk response capabilities of these projects while minimizing the loss of benefits during risk control. Space and time are the sources of risk, and dimensions are the key to locating risk chains. Previous risk control methods may effectively control risks at a specific time or location within a water conservancy project, but they cannot accurately grasp the risk correlations and their derivative impacts. Even with increased manpower and resources, real-time control of operational risks at water conservancy projects cannot be achieved. This invention cleverly utilizes information and intelligent technologies to establish a new real-time risk control method for cascade water conservancy projects. It forms a closed loop from causal monitoring, risk source identification, risk evolution simulation to real-time risk control, enabling real-time tracking and complete eradication of risks in cascade water conservancy projects. Summary of the Invention

[0007] To address the aforementioned shortcomings in existing technologies, this invention provides a real-time control system and method for the operational risks of hydropower stations, enabling multi-dimensional risk perception and control, improving the risk perception and response capabilities of hydropower stations, and reducing economic losses.

[0008] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0009] This solution provides a real-time control system for the operational risks of water conservancy projects, including:

[0010] The risk monitoring subsystem is used to construct a risk transmission link, find key monitoring points based on the risk transmission link, deploy a risk monitoring network, and, based on the risk monitoring network, monitor the changes in the external environment of the hydropower station and the operation of internal units in real time through terminal monitoring equipment and manual inspections.

[0011] The risk analysis subsystem is used to analyze the risk transmission chain based on real-time monitoring results, determine the risk occurrence point, formulate and screen the optimal risk interception scheme based on the risk occurrence point, and feed back the risk interception status to the risk analysis subsystem through the risk control subsystem, and use parallel simulation method to analyze the risk propagation location synchronously.

[0012] The risk control subsystem is used to cut off risks using the optimal risk cutoff scheme and to feed back the risk cutoff results to the risk analysis subsystem.

[0013] Furthermore, the risk monitoring subsystem includes a normal monitoring mode that uses both data-driven and event-driven methods to monitor the operation of the hydropower station in real time, and an emergency monitoring mode that supplements the risk area with aerial remote sensing and satellite remote sensing when the normal monitoring mode cannot fully monitor the risk area.

[0014] The data-driven approach is used to provide real-time monitoring of the operation of internal units of the water conservancy project using terminal monitoring equipment, and to promptly reflect the operational status of the water conservancy project.

[0015] The event-driven mechanism is used to monitor changes in the external environment of the hydropower station through manual inspections and to report any discovered risk events.

[0016] Furthermore, the risk monitoring subsystem includes:

[0017] The identification module is used to identify defects in the planning and construction phases of water conservancy projects and historical problems that occurred in the early operation phases, establish a risk event set for water conservancy projects, and classify the location of risk triggers into external and internal factors based on the risk event set.

[0018] The risk transmission link drawing module is used to analyze the relationship between various risk sources based on the risk triggers involved in the risk event set of the water conservancy project, draw the risk transmission link, and find key monitoring points based on the risk transmission link to deploy a risk monitoring network.

[0019] The monitoring module is used to monitor changes in the external environment of the hydropower station and the operation of its internal units in real time through terminal monitoring equipment and manual inspections, based on the risk monitoring network, and to update the real-time monitoring information.

[0020] Furthermore, the risk event set of the water conservancy project includes power generation risk, flood control risk, navigation risk, dam and reservoir area risk, ecological risk, and public emergency risk;

[0021] The aforementioned power generation risks include the impact of non-electricity system risks on reservoir operation and scheduling, vibration of hydropower plant units, impact of tailrace turbulence on gates during flood discharge, risk of large-scale power outages at hydropower stations, risk of flooding of plant buildings, and risk of failure of major equipment and facilities and units at hydropower stations.

[0022] The flood control risks include the impact of reservoir discharge on low-frequency vibrations of the dam site and surrounding structures, the impact of reservoir discharge vibrations on equipment and facilities, the impact of density currents entering the dam front on generating units and sediment discharge, the impact of water conservancy hub discharge on water conservancy hub structures and near-dam bank slopes, the impact of water conservancy hub discharge on ship lifts, and cavitation problems in stilling basins.

[0023] The shipping risks mentioned include the impact of flood discharge from water conservancy projects on downstream waterways, the impact of the operation and scheduling of cascade water conservancy projects on downstream shipping, and the backwater effect of tributaries on the downstream water level of the main stream.

[0024] The risks in the reservoir area include the impact of floods exceeding the design standard on the safety of the water conservancy project structures, the impact of earthquakes exceeding the design standard on the safety of the water conservancy project structures, the impact of dam deformation on the normal operation of the water conservancy project structures, the impact of landslides on the safety of the water conservancy project, the long-term stability of dam foundation seepage control measures and their impact on dam safety, and the surge problems caused by landslides.

[0025] The ecological risks mentioned include the impact of hydropower station flood discharge and power generation tailwater on downstream fish, and the impact of sewage discharge and reduced hydrodynamic force in the reservoir area of ​​water conservancy hubs on water quality;

[0026] The aforementioned public health emergencies include shipwrecks and the appearance of large floating objects on the water.

[0027] Furthermore, the risk transmission link drawing module includes:

[0028] The risk source location search submodule is used to find the risk source locations involved in the risk evolution and development based on the risk triggers involved in the risk event set of the water conservancy hub, starting with the hidden risks formed by defects in the planning and construction of the water conservancy hub, and combining the real risks existing in the actual operation process.

[0029] The risk transmission link construction submodule is used to analyze the risk propagation and evolution relationship between individual water conservancy projects and cascade water conservancy projects, connect various risk sources to obtain risk transmission links, and find key monitoring points based on risk transmission links to deploy a risk monitoring network. The risk transmission of individual water conservancy projects includes structural force transmission, and the risk transmission of cascade water conservancy projects includes hydraulic transmission and power transmission.

[0030] Furthermore, the risk analysis subsystem includes:

[0031] The positioning module is used for data-driven analysis, based on received anomaly monitoring information, to monitor the corresponding risk occurrence point in the risk transmission chain, and to locate the risk occurrence point within the risk transmission chain; and

[0032] For event-driven scenarios, after a risk occurs, the system connects to the real-time control system for operational risks of water conservancy hubs, locates the risk occurrence point through received early warning or monitoring signals, and positions it within the risk transmission chain based on that risk occurrence point.

[0033] The risk chain simulation module is used to simulate and analyze the risk transmission process based on the location results, using mechanism models, intelligent learning models, and artificial knowledge and experience models, and to obtain the risk transmission results.

[0034] The risk adaptability simulation module is used to formulate risk cutoff schemes based on the risk transmission results, and to select the optimal risk cutoff scheme using a fuzzy evaluation method based on probabilistic intuition. It also feeds back the risk cutoff status to the risk analysis subsystem through the risk control subsystem, and uses parallel simulation methods to perform synchronous analysis on the risk propagation location.

[0035] Furthermore, the simulation analysis of the risk transmission process using artificial knowledge and experience models specifically includes:

[0036] In the risk transmission chain, select the risk location of the water conservancy project, collect historical risk data related to the risk location, and sort out the direction of risk propagation when the risk occurs in the location.

[0037] Based on the location of the risk, the direction of risk propagation is analyzed, and a Bayesian network of risk transmission is constructed based on the propagation of risk between various risk occurrence points.

[0038] When the probability of risk transmission cannot be extracted from historical risk data, a language measurement scale is established to divide the probability of risk transmission into seven levels. Experts provide evaluation opinions, and the fuzziness score of the probability of risk occurrence is obtained by integrating the expert evaluation opinions through fuzzy set theory. The probability of risk transmission is then calculated by defuzzification.

[0039] Based on the risk propagation probability, the Bayesian network is updated using real risk data, and the updated Bayesian network is used to simulate and analyze the risk transmission process to obtain the risk transmission results.

[0040] Furthermore, the calculation yields the probability of risk propagation, specifically as follows:

[0041] In the risk transmission chain, the propagation probability between adjacent nodes is calculated:

[0042]

[0043] Where P(U) represents the propagation probability between adjacent nodes, P(·) represents the joint probability, n represents the total number of nodes, i represents the number of nodes, and Pa(X) represents the number of nodes. i ) represents node X i The parent set;

[0044] The propagation probability between adjacent nodes is used to calculate node X. i Marginal probability:

[0045]

[0046] Among them, X i ,P(X i ) represents node X i The marginal probability, X j Let j represent a node, and j represent the number of nodes;

[0047] According to node X i The marginal probability is used to calculate the updated risk transmission probability when updating the risk location:

[0048]

[0049] Where P(U / E) represents the updated risk transmission probability, P(U,E) is the joint risk transmission probability of risk variable U and risk variable E, and P(E) represents the risk transmission probability of risk variable E occurring.

[0050] Based on the updated risk propagation probability, fuzzy set theory is used to assign probability values ​​to the risk probabilities that are missing in the Bayesian network.

[0051] A language measurement scale is established, which divides the probability of risk occurrence into seven levels. Each expert judges the probability of risk occurrence at each node. The expert judgments are integrated using fuzzy set theory to obtain a fuzzy score of the probability of risk occurrence. Defuzzification is then used to convert the fuzzy score into a probability number. Based on the probability number, the expert judgments are added to the risk transmission chain. The expression for defuzzification is as follows.

[0052]

[0053] Where X* represents the risk probability, ∫ represents the integral over the membership function, and μ i (x) represents the membership function, x represents the variable, i.e., the fuzzy number, and d represents the integral sign, which is the integral of the fuzzy function;

[0054] Based on the risk transmission chain incorporating expert opinions, and using fuzzy numbers... The probability of risk propagation is calculated as follows:

[0055]

[0056] Furthermore, the specific process of the fuzziness scoring is as follows:

[0057] The average consensus AA(E) of the M experts was calculated. u ):

[0058]

[0059] E u (u = 1, 2, ..., M)

[0060] Where u represents the number of experts, and v represents the number of experts. Indicates the similarity between two fuzzy numbers;

[0061] Based on the average degree of consistency AA(E) u The relative degree of agreement (RA) of the experts was calculated. u ):

[0062]

[0063] According to the relative degree of agreement among experts (RA(E)) u The expert consensus coefficient CC(E) was calculated. u ):

[0064] CC(E u )=β·w(E u )+(1-β)·RA(E u )

[0065] Where β represents the relaxation factor, w(E) u ) represents the weighting factor of expert u;

[0066] According to the expert consensus coefficient, CC(E) u The summative value of expert opinions was calculated. Complete the scoring process for fuzziness:

[0067]

[0068] in, This represents M experts, CC(E) M ) represents the consensus coefficient of M experts.

[0069] Furthermore, the risk adaptation simulation module includes:

[0070] The risk cutoff plan formulation submodule is used to analyze the loss situation of each propagation path based on the risk transmission results and formulate a risk cutoff plan.

[0071] The optimal risk cutoff scheme selection submodule is used to select the optimal risk cutoff scheme based on the risk cutoff scheme using a fuzzy evaluation method based on probability intuition, and upload the optimal risk cutoff scheme to the risk control subsystem.

[0072] The first feedback submodule is used to feed back the risk cutoff status to the risk analysis subsystem through the risk control subsystem according to the optimal risk cutoff scheme, and to perform synchronous analysis on the risk propagation location using parallel simulation methods.

[0073] Furthermore, the optimal risk truncation scheme includes: a starting point truncation scheme, a midpoint control scheme, and a tail point recovery scheme.

[0074] Furthermore, the optimal risk cutoff scheme selection submodule includes:

[0075] The weighting allocation unit is used to allocate weights based on the decision-maker's opinion.

[0076] The Aggregate Intuitive Fuzzy Decision Matrix Construction Unit is used to construct an aggregate intuitive fuzzy decision matrix based on the assigned weights and the decision-maker's evaluation criteria.

[0077] The weight calculation unit is used to calculate the weight corresponding to each evaluation criterion of the level based on the aggregated intuitionistic fuzzy decision matrix.

[0078] The aggregated weighted intuitionistic fuzzy decision matrix construction unit is used to construct the aggregated weighted intuitionistic fuzzy decision matrix based on the weights corresponding to each evaluation criterion of the level and the aggregated intuitionistic fuzzy decision matrix.

[0079] The ideal solution calculation unit is used to calculate the intuitionistic fuzzy positive ideal solution and the intuitionistic fuzzy negative ideal solution based on the aggregated weighted intuitionistic fuzzy decision matrix.

[0080] The probability separation metric calculation unit is used to calculate the probability separation metric based on the intuitionistic fuzzy positive ideal solution and the intuitionistic fuzzy negative ideal solution.

[0081] Evaluation unit, used to evaluate the relative proximity coefficient of intuitionistic fuzzy positive ideal solutions based on a probability separation metric;

[0082] The filtering unit is used to rank the risk cutoff schemes based on the relative proximity coefficient and select the risk cutoff scheme with the highest relative proximity as the optimal risk cutoff scheme.

[0083] This invention also provides a control method for a real-time control system for operational risks of water conservancy projects, comprising the following steps:

[0084] S1. Risk Monitoring Phase: Construct a risk transmission link, identify key monitoring points based on the risk transmission link, deploy a risk monitoring network, and, based on the risk monitoring network, monitor in real time the changes in the external environment of the hydropower station and the operation of internal units through terminal monitoring equipment and manual inspections.

[0085] S2. Risk Analysis Phase: Analyze the risk transmission chain based on real-time monitoring results, determine the risk occurrence point, formulate and screen the optimal risk interception plan based on the risk occurrence point, and feed back the risk interception status to the risk analysis phase through the risk control phase. Use parallel simulation method to analyze the risk propagation location synchronously.

[0086] S3. Risk Control Phase: Use the optimal risk cutoff scheme to cut off risks and feed the risk cutoff results back to the risk analysis phase.

[0087] The beneficial effects of this invention are:

[0088] (1) Existing risk control technologies are effective in controlling risks in single-unit water conservancy projects and single-professional risks, but they are difficult to handle the correlation analysis and control of risks in different parts of water conservancy projects and across different professions, and even more so in handling the risk cutoff of cascade water conservancy projects. The problem solved by this invention is focused on the real-time prevention and control of associated risks in cascade water conservancy projects, solving the problem that existing technologies cannot solve from a system perspective. To solve this problem, this invention has made innovative breakthroughs in risk point sources, risk transmission links, risk correlation identification (i.e., structural force transmission, hydraulic transmission, and power transmission), as well as the construction of a real multi-dimensional risk network space for water conservancy projects and real-time risk control. This enables multi-dimensional risk perception and control of hydropower stations, improves the risk perception and risk response capabilities of hydropower stations, and reduces the economic benefits of hydropower stations.

[0089] (2) This invention proposes a classification method that combines function and event, namely: reservoir dam, spillway, power generation, navigation, ecological environment and public emergencies. This avoids the shortcomings of water conservancy hubs being classified according to function or location, which result in overlap and blurred boundaries, and lays the foundation for subsequent risk classification and prevention.

[0090] (3) This invention proposes a method to sort out the risk sources and transmission links of water conservancy hubs by taking risk events as the starting point, clarifying the specific links of risk transmission, which not only avoids a lot of ineffective work, but also achieves accurate risk identification and helps to make precise prevention and control.

[0091] (3) This invention proposes a method for creating and analyzing the risk transmission link space of cascade water conservancy hubs. It can not only identify the correlation between different parts of a single water conservancy hub and cross-professional risks, but also realistically reproduce all links of risk propagation along water flow, power flow and mechanical flow of cascade water conservancy hubs.

[0092] (4) This invention provides a quantitative analysis method for the risk transmission link space, that is, it creates a common analysis mechanism of mechanism model, data statistical analysis and empirical probability estimation, which can still make a relatively accurate analysis even in the case of lack of data and unclear mechanism.

[0093] (5) The present invention proposes a triggering mode based on data-driven and event-driven approaches, which can cover all risks that water conservancy projects may encounter.

[0094] (6) This invention innovatively constructs a multi-layered risk control mechanism, establishing a three-level encirclement for risk control. Level 1: Risks can be intercepted by individual water conservancy projects without affecting the operation of upstream and downstream water conservancy projects. Level 2: Risks require comprehensive scheduling of upstream and downstream cascade hydropower stations to be intercepted, affecting the normal operation of cascade hydropower stations. Level 3: Risks affect the normal operation of water conservancy projects in the entire basin, requiring comprehensive scheduling and control of the risk across the entire basin. By mapping the three-level encirclement to the risk network space, when a risk occurs, the risk core, i.e., the source of risk diffusion, is found. Using this point as the center, the locations where the risk may spread are divided according to the three-level risk standard, forming three encirclements. Multi-layered protection can ensure the complete interruption of the risk link.

[0095] (7) The real-time risk feedback system proposed in this invention can track the risk control process in real time, provide real-time feedback on the risk control situation and make corresponding adjustments. The dynamic control of risk can not only effectively reduce the probability of errors in static risk control, but also significantly improve the efficiency of risk rescue and reduce rescue costs.

[0096] (8) In this invention, the risk analysis subsystem adopts a parallel simulation method to analyze all possible risk links for the same level of risk and block all possible risk transmission.

[0097] (9) This invention achieves a closed-loop risk control system for water conservancy projects for the first time. After identifying the risk transmission relationship, the optimal location for risk interception is determined, a risk interception plan is formulated, and the risk interception plan is evaluated. The preferred plan is then used for risk control. Based on the risk interception situation, the system is updated in real time, the risk propagation location is adjusted, and a risk interception plan is formulated again, thus achieving closed-loop management of risk control.

[0098] (10) This invention proposes a risk scheme optimization method based on probabilistic intuition Bayesian method and a risk scheme optimization method based on probabilistic intuition fuzzy evaluation algorithm, which is particularly suitable for dealing with occasional risks when the frequency of risk occurrence is low and the data is insufficient to support decision-making, and further improves the risk transmission link. Attached Figure Description

[0099] Figure 1 This is a schematic diagram of the system structure of the present invention.

[0100] Figure 2 This is a schematic diagram of the real-time feedback mechanism for risk cutoff and risk control points in this embodiment.

[0101] Figure 3 This is a flowchart of the method of the present invention. Detailed Implementation

[0102] The specific embodiments of the present invention are described below to enable those skilled in the art to understand the present invention. However, it should be understood that the present invention is not limited to the scope of the specific embodiments. For those skilled in the art, various changes are obvious as long as they are within the spirit and scope of the present invention as defined and determined by the appended claims. All inventions utilizing the concept of the present invention are protected.

[0103] Example 1

[0104] This invention proposes a real-time control system for the operational risks of water conservancy projects. Its key feature is the establishment of a real-time control system integrating risk monitoring, risk analysis, and risk control for cascade water conservancy projects. This system tracks and traces the sources of risks, taking a step-by-step approach during risk evolution analysis. It monitors the operational status and risk evolution of each part of the water conservancy project, and adjusts the risk control plan based on feedback information after risk control. Crucially, this invention analyzes the systemic risks of cascade water conservancy projects from a holistic perspective, employing a dynamic risk control mechanism to achieve multi-dimensional and multi-layered safety protection during the operation of the water conservancy projects.

[0105] like Figure 1 As shown, the present invention provides a real-time control system for operational risks of water conservancy projects, comprising:

[0106] The risk monitoring subsystem is used to construct a risk transmission link, find key monitoring points based on the risk transmission link, deploy a risk monitoring network, and, based on the risk monitoring network, monitor the changes in the external environment of the hydropower station and the operation of internal units in real time through terminal monitoring equipment and manual inspections.

[0107] The risk analysis subsystem is used to analyze the risk transmission chain based on real-time monitoring results, determine the risk occurrence point, formulate and screen the optimal risk interception scheme based on the risk occurrence point, and feed back the risk interception status to the risk analysis subsystem through the risk control subsystem, and use parallel simulation method to analyze the risk propagation location synchronously.

[0108] In this embodiment, parallel simulation involves synchronously analyzing all possible propagation links, while simultaneously strengthening the collection of monitoring data. Based on the effectiveness of risk control implementation, selected paths are retained, other options are excluded, and the simulation is advanced to the next stage.

[0109] The risk control subsystem is used to cut off risks using the optimal risk cutoff scheme and to feed back the risk cutoff results to the risk analysis subsystem.

[0110] In this embodiment, the risk monitoring subsystem includes a normal monitoring mode that uses both data-driven and event-driven methods to monitor the operation of the hydropower station in real time, and an emergency monitoring mode that supplements the risk area with aerial remote sensing and satellite remote sensing when the normal monitoring mode cannot fully monitor the risk area.

[0111] The data-driven approach is used to provide real-time monitoring of the operation of internal units of the water conservancy project using terminal monitoring equipment, and to promptly reflect the operational status of the water conservancy project.

[0112] The event-driven mechanism is used to monitor changes in the external environment of the hydropower station through manual inspections and to report any discovered risk events.

[0113] In this embodiment, the risk monitoring subsystem includes:

[0114] The identification module is used to identify defects in the planning and construction phases of water conservancy projects and historical problems that occurred in the early operation phases, establish a risk event set for water conservancy projects, and classify the location of risk triggers into external and internal factors based on the risk event set.

[0115] The risk transmission link drawing module is used to analyze the relationship between various risk sources based on the risk triggers involved in the risk event set of the water conservancy project, draw the risk transmission link, and find key monitoring points based on the risk transmission link to deploy a risk monitoring network.

[0116] The monitoring module is used to monitor changes in the external environment of the hydropower station and the operation of its internal units in real time through terminal monitoring equipment and manual inspections, based on the risk monitoring network, and to update the real-time monitoring information.

[0117] In this embodiment, the risk event set of the water conservancy hub includes power generation risk, flood control risk, navigation risk, dam and reservoir area risk, ecological risk, and public emergency risk;

[0118] The aforementioned power generation risks include the impact of non-electricity system risks on reservoir operation and scheduling, vibration of hydropower plant units, impact of tailrace turbulence on gates during flood discharge, risk of large-scale power outages at hydropower stations, risk of flooding of plant buildings, and risk of failure of major equipment and facilities and units at hydropower stations.

[0119] The flood control risks include the impact of reservoir discharge on low-frequency vibrations of the dam site and surrounding structures, the impact of reservoir discharge vibrations on equipment and facilities, the impact of density currents entering the dam front on generating units and sediment discharge, the impact of water conservancy hub discharge on water conservancy hub structures and near-dam bank slopes, the impact of water conservancy hub discharge on ship lifts, and cavitation problems in stilling basins.

[0120] The shipping risks mentioned include the impact of flood discharge from water conservancy projects on downstream waterways, the impact of the operation and scheduling of cascade water conservancy projects on downstream shipping, and the backwater effect of tributaries on the downstream water level of the main stream.

[0121] The risks in the reservoir area include the impact of floods exceeding the design standard on the safety of the water conservancy project structures, the impact of earthquakes exceeding the design standard on the safety of the water conservancy project structures, the impact of dam deformation on the normal operation of the water conservancy project structures, the impact of landslides on the safety of the water conservancy project, the long-term stability of dam foundation seepage control measures and their impact on dam safety, and the surge problems caused by landslides.

[0122] The ecological risks mentioned include the impact of hydropower station flood discharge and power generation tailwater on downstream fish, and the impact of sewage discharge and reduced hydrodynamic force in the reservoir area of ​​water conservancy hubs on water quality;

[0123] The aforementioned public health emergencies include shipwrecks and the appearance of large floating objects on the water.

[0124] In this embodiment, the risk transmission link drawing module includes:

[0125] The risk source location search submodule is used to find the risk source locations involved in the risk evolution and development based on the risk triggers involved in the risk event set of the water conservancy hub, starting with the hidden risks formed by defects in the planning and construction of the water conservancy hub, and combining the real risks existing in the actual operation process.

[0126] The risk transmission link construction submodule is used to analyze the risk propagation and evolution relationship between individual water conservancy projects and cascade water conservancy projects, connect various risk sources to obtain risk transmission links, and find key monitoring points based on risk transmission links to deploy a risk monitoring network. The risk transmission of individual water conservancy projects includes structural force transmission, and the risk transmission of cascade water conservancy projects includes hydraulic transmission and power transmission.

[0127] In this embodiment, the risk analysis subsystem includes:

[0128] The positioning module is used for data-driven analysis, based on received anomaly monitoring information, to monitor the corresponding risk occurrence point in the risk transmission chain, and to locate the risk occurrence point within the risk transmission chain; and

[0129] For event-driven scenarios, after a risk occurs, the system connects to the real-time control system for operational risks of water conservancy hubs, locates the risk occurrence point through received early warning or monitoring signals, and positions it within the risk transmission chain based on that risk occurrence point.

[0130] The risk chain simulation module is used to simulate and analyze the risk transmission process based on the location results, using mechanism models, intelligent learning models, and artificial knowledge and experience models, and to obtain the risk transmission results.

[0131] The risk adaptability simulation module is used to formulate risk cutoff schemes based on the risk transmission results, and to select the optimal risk cutoff scheme using a fuzzy evaluation method based on probabilistic intuition. It also feeds back the risk cutoff status to the risk analysis subsystem through the risk control subsystem, and uses parallel simulation methods to perform synchronous analysis on the risk propagation location.

[0132] In this embodiment, the simulation analysis of the risk transmission process using a human knowledge and experience model specifically includes:

[0133] In the risk transmission chain, select the risk location of the water conservancy project, collect historical risk data related to the risk location, and sort out the direction of risk propagation when the risk occurs in the location.

[0134] Based on the location of the risk, the direction of risk propagation is analyzed, and a Bayesian network of risk transmission is constructed based on the propagation of risk between various risk occurrence points.

[0135] When the probability of risk transmission cannot be extracted from historical risk data, a language measurement scale is established to divide the probability of risk transmission into seven levels. Experts provide evaluation opinions, and the fuzziness score of the probability of risk occurrence is obtained by integrating the expert evaluation opinions through fuzzy set theory. The probability of risk transmission is then calculated by defuzzification.

[0136] Based on the risk propagation probability, the Bayesian network is updated using real risk data, and the updated Bayesian network is used to simulate and analyze the risk transmission process to obtain the risk transmission results.

[0137] In this embodiment, the calculated probability of risk propagation is specifically as follows:

[0138] In the risk transmission chain, the propagation probability between adjacent nodes is calculated:

[0139]

[0140] Where P(U) represents the propagation probability between adjacent nodes, P(·) represents the joint probability, n represents the total number of nodes, i represents the number of nodes, and Pa(X) represents the number of nodes. i ) represents node X i The parent set;

[0141] The propagation probability between adjacent nodes is used to calculate node X. i Marginal probability:

[0142]

[0143] Among them, X i ,P(X i) represents node X i The marginal probability, X j Let j represent a node, and j represent the number of nodes;

[0144] According to node X i The marginal probability is used to calculate the updated risk transmission probability when updating the risk location:

[0145]

[0146] Where P(U / E) represents the updated risk transmission probability, P(U,E) is the joint risk transmission probability of risk variable U and risk variable E, and P(E) represents the risk transmission probability of risk variable E occurring.

[0147] Based on the updated risk propagation probability, fuzzy set theory is used to assign probability values ​​to the risk probabilities that are missing in the Bayesian network.

[0148] A language measurement scale is established, which divides the probability of risk occurrence into seven levels. Each expert judges the probability of risk occurrence at each node. The expert judgments are integrated using fuzzy set theory to obtain a fuzzy score of the probability of risk occurrence. Defuzzification is then used to convert the fuzzy score into a probability number. Based on the probability number, the expert judgments are added to the risk transmission chain. The expression for defuzzification is as follows.

[0149]

[0150] Where X* represents the risk probability, ∫ represents the integral over the membership function, and μ i (x) represents the membership function, x represents the variable, i.e., the fuzzy number, and d represents the integral sign, which is the integral of the fuzzy function;

[0151] Based on the risk transmission chain incorporating expert opinions, and using fuzzy numbers... The probability of risk propagation is calculated as follows:

[0152]

[0153] In this embodiment, the specific process of fuzziness scoring is as follows:

[0154] The average consensus AA(E) of the M experts was calculated. u ):

[0155]

[0156] E u (u = 1, 2, ..., M)

[0157] Where u represents the number of experts, and v represents the number of experts. Indicates the similarity between two fuzzy numbers;

[0158] Based on the average degree of consistency AA(E) u The relative degree of agreement (RA) of the experts was calculated. u ):

[0159]

[0160] According to the relative degree of agreement among experts (RA(E)) u The expert consensus coefficient CC(E) was calculated. u ):

[0161] CC(E u )=β·w(E u )+(1-β)·RA(E u )

[0162] Where β represents the relaxation factor, w(E) u ) represents the weighting factor of expert u;

[0163] According to the expert consensus coefficient, CC(E) u The summative value of expert opinions was calculated. Complete the scoring process for fuzziness:

[0164]

[0165] in, This represents M experts, CC(E) M ) represents the consensus coefficient of M experts.

[0166] In this embodiment, the risk adaptability simulation module includes:

[0167] The risk cutoff plan formulation submodule is used to analyze the loss situation of each propagation path based on the risk transmission results and formulate a risk cutoff plan.

[0168] The optimal risk cutoff scheme selection submodule is used to select the optimal risk cutoff scheme based on the risk cutoff scheme using a fuzzy evaluation method based on probability intuition, and upload the optimal risk cutoff scheme to the risk control subsystem.

[0169] The first feedback submodule is used to feed back the risk cutoff status to the risk analysis subsystem through the risk control subsystem according to the optimal risk cutoff scheme, and to perform synchronous analysis on the risk propagation location using parallel simulation methods.

[0170] In this embodiment, the optimal risk truncation scheme includes: a starting point truncation scheme, a midpoint control scheme, and a tail point recovery scheme.

[0171] In this embodiment, the optimal risk truncation scheme selection submodule includes:

[0172] The weighting allocation unit is used to allocate weights based on the decision-maker's opinion.

[0173] The Aggregate Intuitive Fuzzy Decision Matrix Construction Unit is used to construct an aggregate intuitive fuzzy decision matrix based on the assigned weights and the decision-maker's evaluation criteria.

[0174] The weight calculation unit is used to calculate the weight corresponding to each evaluation criterion of the level based on the aggregated intuitionistic fuzzy decision matrix.

[0175] The aggregated weighted intuitionistic fuzzy decision matrix construction unit is used to construct the aggregated weighted intuitionistic fuzzy decision matrix based on the weights corresponding to each evaluation criterion of the level and the aggregated intuitionistic fuzzy decision matrix.

[0176] The ideal solution calculation unit is used to calculate the intuitionistic fuzzy positive ideal solution and the intuitionistic fuzzy negative ideal solution based on the aggregated weighted intuitionistic fuzzy decision matrix.

[0177] The probability separation metric calculation unit is used to calculate the probability separation metric based on the intuitionistic fuzzy positive ideal solution and the intuitionistic fuzzy negative ideal solution.

[0178] Evaluation unit, used to evaluate the relative proximity coefficient of intuitionistic fuzzy positive ideal solutions based on a probability separation metric;

[0179] The filtering unit is used to rank the risk cutoff schemes based on the relative proximity coefficient and select the risk cutoff scheme with the highest relative proximity as the optimal risk cutoff scheme.

[0180] The present invention will now be further described.

[0181] In this embodiment, regarding the risk monitoring subsystem: First, defects in the planning and construction phases of the water conservancy project and problems that occurred during its early operation are identified. A risk event set for the water conservancy project is established and categorized, divided into internal and external factors based on the location of the risk triggers. Internal factors are risks inherent to the water conservancy project itself and its important components, such as equipment malfunctions in navigation and flood discharge, and human error in the operation of power plant units. External factors are risks related to the water conservancy project, mainly natural disasters and social events occurring within the project's control area, which may force the project to adjust its operation. It is important to note that the investigation and identification of risk events is open; that is, once a new risk affecting the safe operation of the water conservancy project is discovered, it must be immediately added to enrich and improve the risk event set.

[0182] In this embodiment, the result of the risk investigation of the water conservancy project is a set of risk events. During the risk investigation, it is necessary to identify past risk events at the water conservancy project. Various techniques can be employed, such as discussions, brainstorming, questionnaires, literature reviews, prototype monitoring, and experimental analysis. A detailed investigation should be conducted into the defects in the early planning and construction processes of the water conservancy project, as well as risk events that have occurred in the operation of other similar projects both domestically and internationally. This will identify potential risks for the water conservancy project at this level. Based on this, all risk events are categorized and organized. The causes of each risk event, the potential damage to the water conservancy project's parts or / and functions, and related risk control measures are analyzed category by category and individually. A bow-tie diagram of the risk transmission chain is drawn. Simultaneously, it is important to expand the scope of the risk investigation, that is, for similar water conservancy projects, analyze the damage to the water conservancy project caused by similar risk events and all the risk-affected parts from international, domestic, intra-basin, and extra-basin perspectives.

[0183] In this embodiment, risk events can be classified in multiple ways. Besides the classification based on internal and external factors, they can also be categorized according to their impact on the functions of the water conservancy hub. For example, they can be categorized by flood control, power generation, navigation, sediment discharge, irrigation, water supply, and ecological functions; or by the location of the water conservancy hub, such as the reservoir area, dam, power station, waterway, and lock; or a combination of both, such as reservoir dam, spillway, power generation, navigation, ecological environment, and public emergencies. Within each major category, risk events that do not belong to each other are classified into subcategories. Each major category of risk events can serve as an independent dimension of the operational risk of the water conservancy hub.

[0184] In this embodiment, to facilitate risk management, the invention creatively classifies risk events into six categories: power generation risk, flood control risk, navigation risk, dam and reservoir area risk, ecological risk, and public emergency risk. Power generation risks include, but are not limited to, the impact of power transmission system risks on reservoir operation and scheduling, vibration of hydropower plant units, the impact of tailrace turbulence on gates during flood discharge, large-scale power outages at hydropower stations, flooding of plant buildings, and failures of major equipment and facilities at hydropower stations. Flood control risks include, but are not limited to, the impact of reservoir discharge on low-frequency vibrations of the dam site and surrounding structures, the impact of reservoir discharge vibrations on equipment and facilities, the impact of density currents entering the dam front on units and sediment discharge, the impact of hydropower plant discharge on hydropower structures and near-dam banks, the impact of hydropower plant discharge on ship lifts, and stilling basin cavitation problems. Navigation risks include, but are not limited to, the impact of hydropower plant discharge on downstream waterways, and the impact of cascade hydropower plant operation and scheduling on... Risks include: the impact on downstream shipping; the backwater effect of tributaries on the downstream water level of the main stream; risks in the dam body and reservoir area, including but not limited to: the impact of floods exceeding design standards on the safety of water conservancy structures; the impact of earthquakes exceeding design standards on the safety of water conservancy structures; the impact of dam deformation on the normal operation of water conservancy structures; the impact of landslides on the safety of water conservancy projects (reservoir area, dam body, generating units); the long-term stability of dam foundation seepage control measures and their impact on dam safety; and surge problems caused by landslides. Ecological risks include but not limited to: the impact of hydropower station discharge and tailwater on downstream fish; and the impact of sewage discharge and weakened hydrodynamics in the reservoir area on water quality. Risks of sudden public events include but not limited to: shipwrecks and the appearance of large floating objects on the water surface.

[0185] In this embodiment, all point sources involved in a risk event are identified. Based on the previous risk bowtie diagram, this invention identifies all triggering point sources (i.e., risk inducements) involved in a single risk event, and further analyzes the input-output relationships between each risk point source to draw a risk transmission chain. This risk transmission chain identification process can independently utilize or combine risk analysis methods such as Safety Assessment (FSA), Fault Tree Analysis, System Dynamics, and Bayesian methods. Starting with the implicit risks arising from defects in the planning and construction of water conservancy projects, and using the actual risks existing in the actual operation as a starting point, the location of the point sources involved in the evolution and development of such risks is identified. Furthermore, the risk propagation and evolution relationships between individual water conservancy projects and cascade water conservancy projects are analyzed, connecting each risk point source to obtain the complete risk transmission chain.

[0186] In this embodiment, regarding the identification of the correlation between operational risks of cascade hydropower projects, this invention proposes to analyze the evolution of risks from the perspective of energy transfer. The risks of individual hydropower projects are mainly transmitted through structural mechanics, while those between cascade hydropower projects are mainly transmitted along water flow and through electrical energy transfer. These three risk energy transmission chains intertwine at the connection points of reservoirs and power stations, linking risk events end to end. The connection points can be the same location / part, or they can be related in some way, such as water level, flow rate, or equivalent stress. By connecting risk events, an interconnected risk transmission network is formed, which creates a multi-dimensional, three-dimensional risk transmission link space.

[0187] In this embodiment, the network space of the risk transmission chain of the present invention is digital and can support quantitative analysis of risk transmission.

[0188] This embodiment further explains the process of digitizing the risk transmission chain. Based on existing mechanistic knowledge, human experience, or data analysis results, a risk transmission chain is created. The identification and analysis methods for related risk transmission relationships include the individual or combined use of causal analysis, comprehensive safety assessment (FSA), fault tree analysis, and / or Bayesian methods. These methods can all be used to connect risk sources and form a risk transmission chain, which is the basis for conducting quantitative analysis of risk transmission.

[0189] In this embodiment, the quantitative analysis methods for risk transmission can employ mechanistic models, statistical data analysis, and empirical probability estimation. These three types of models, methods, or techniques have different application scenarios. Mechanistic models can analyze and handle risks whose occurrence mechanisms are well understood and, in reality, are relatively well-controlled. Statistical data analysis primarily deals with risks that occur frequently, have a large amount of accumulated risk monitoring data, and whose causal relationships are clearly understood, but whose specific mechanisms are still unclear. Of course, it can also be applied to risks handled by mechanistic models. In some cases, the combined use of mechanistic models and data-driven models can significantly improve the accuracy of risk analysis. Risks handled by empirical probability methods occur infrequently and are difficult to control; once they occur, they usually affect the safe operation of water conservancy projects. This type of risk is the focus of this invention.

[0190] In this embodiment, the risk monitoring subsystem is used to monitor the risk status during the operation of the water conservancy project, mainly serving subsequent risk analysis. This invention, based on a basically determined risk transmission link space, identifies the key links requiring monitoring and deploys a risk monitoring network.

[0191] In this embodiment, risk monitoring serves risk analysis; therefore, the risk monitoring network layout needs to understand the data required by the risk analysis methods. In this invention, risk analysis employs two driving methods: data-driven and event-driven.

[0192] In this embodiment, data-driven monitoring utilizes pre-set monitoring equipment to provide real-time monitoring of the operation of the water conservancy project, promptly reflecting its operational status. The monitoring scope includes the dam body, dam foundation, dam abutments, reservoir banks near the dam, and the reservoir area, as well as various auxiliary equipment attached to the main dam structure. Installing monitoring equipment at these locations ensures real-time monitoring of the dam environment, providing monitoring information to support real-time risk control. In addition, comprehensive monitoring of the cascade water conservancy projects is conducted, allowing for real-time perception of their operational status. The monitoring scope includes water levels in each reservoir, rainfall in the reservoir area, and geological conditions within the reservoir area, forming an integrated monitoring network encompassing the internal operating conditions of the water conservancy project, the external environment, and the entire cascade system. This provides data support for subsequent analysis of the risk evolution process between different parts of the water conservancy project and for mapping risk transmission chains.

[0193] In this embodiment, event-driven risk events are discovered through manual inspections or external information. These risk events are beyond the scope of routine monitoring and cannot be alerted by automated monitoring equipment; examples include shipwrecks and earthquakes. Some events are discovered through manual inspections and reported, thus linking the risk to the risk transmission chain. Other events are outside the monitoring capabilities of the hydropower station but may still affect the safe operation of the water conservancy project. These require analysis of disaster information released by geological and meteorological departments to identify potential risk sources, followed by further processing through the risk transmission chain.

[0194] In this embodiment, the monitoring information of the risk monitoring subsystem includes, but is not limited to, meteorological, hydrological, and geological data. Furthermore, the risk monitoring subsystem includes two monitoring modes: routine monitoring and emergency monitoring. During normal operation of the hydropower station, the routine monitoring mode is used, employing both data-driven and event-driven methods to monitor risks arising during the station's operation. When a risk event is rare or occurs in a remote location, and existing monitoring information is insufficient for assessing the risk event, the corresponding emergency monitoring mode is adopted based on the type and location of the risk event. For example, if a shipwreck suddenly occurs in the reservoir area, and routine monitoring equipment cannot comprehensively monitor the risk area, aerial remote sensing and satellite remote sensing are used to focus on monitoring the risk area, supplement monitoring information, trace the risk source, and develop corresponding risk interception plans.

[0195] In this embodiment, the risk monitoring subsystem monitors changes in the external environment and the operational status of internal units of the hydropower station through terminal monitoring equipment and manual inspections. The collected data includes, but is not limited to, meteorological data, hydrological data, and geological data. Edge computing technology is used to process the collected information in a distributed manner at various collection points, extract key information, and upload it to the risk analysis subsystem.

[0196] In this embodiment, after receiving information uploaded by the risk monitoring subsystem, the risk analysis subsystem performs risk transmission analysis to determine the location of the risk occurrence. Based on the risk transmission chain, it estimates the next spread range of the risk and predicts the evolution or transformation process of the risk from a point to a line to a surface. When a risk occurs at a certain location in the risk transmission chain, the monitoring and defense efforts for adjacent links are strengthened, or the monitoring density and frequency are increased, and relevant air, space, and ground mobile monitoring methods such as drones, unmanned ships, global navigation satellite systems, 3D laser scanning, micro-core piles, and temporary sites are added.

[0197] In this embodiment, after a risk occurs or a variable change exceeding a set threshold is detected, the risk analysis subsystem, based on the aforementioned established risk transmission chain, utilizes a mechanistic model, an intelligent learning model, and / or a human knowledge and experience model. The detected variables are input into the model, and then the risk transmission process is quantitatively analyzed to fully deduce the entire risk transmission process and predict all possible consequences of the risk. The simulation analysis model for the risk transmission process mainly includes three categories: 1) mechanistic models, 2) data-driven or intelligent learning models, and 3) human knowledge and experience probability models. The models are used both independently and in conjunction with verification. In actual operation, firstly, a mechanistic model is considered. If there is already mature risk transmission mechanism experience for the risk location, the mechanistic model is used for analysis based on the specific location to find the consequences of risk transmission. Secondly, a data-driven intelligent learning model is considered, combining historical event data and real-time monitoring data to analyze and simulate the risk event and find the consequences of risk transmission. Finally, considering probabilistic models based on human knowledge and experience, when a risk event is rare, historical data is insufficient, or monitoring information for the risk event is inadequate, the transmission consequences of the risk are determined by combining probabilistic analysis and human experience. It should be noted that in this invention, the transmission of each risk process is simulated and cross-validated using at least two models to ensure the reliability of the results. These models may be of different types (e.g., mechanistic models and big data-driven models) or of the same type constructed using different methods (e.g., artificial intelligence models employing neural network learning, deep learning, knowledge vector machines, etc.).

[0198] In this embodiment, key risk monitoring locations also serve as nodes for risk analysis. The risk monitoring scope includes the dam body, dam foundation, dam abutments, reservoir bank near the dam, and the reservoir area. It also includes various auxiliary equipment attached to the main dam structure, on which a large network of sensors is deployed. During risk analysis, to meet the needs of risk evolution simulation analysis, some parts of the model calculation require densification, primarily achieved by setting virtual nodes, including manually setting computational nodes and dividing the computational grid.

[0199] In this embodiment, the risk analysis steps are as follows:

[0200] 1. Network Positioning: 1) If data-driven, upon receiving anomaly information from the monitoring module, the system monitors the corresponding node location in the risk transmission link space based on information such as monitoring location coordinates, changes in monitoring devices, monitoring variables, and risk types. 2) If event-driven, after a risk occurs, the system manually inputs or inputs through a dedicated channel to access the system. By analyzing the received warning or monitoring information, the system locates the risk node and quickly positions it in the risk transmission link space.

[0201] 2. Risk Chain Simulation: This function analyzes the risk transmission process using mechanistic models, intelligent learning models, and artificial knowledge and experience models to obtain the consequences of risk transmission. Based on information provided by the risk monitoring module, the location and type of risk are identified. A corresponding risk analysis model is selected to simulate the risk evolution. The simulation results are used to analyze the next propagation location of the risk. Then, based on the specific circumstances of the next location, a suitable analysis model is selected for further prediction. For example, when a flood risk occurs, a hydrodynamic model is used to simulate the locations of water conservancy projects that may be affected by the flood. The simulation results are used as the risk origin at these locations, and a corresponding analysis model is selected for risk analysis at the next propagation location. This process continues to obtain the risk transmission chain and evolution process. For each chain, the model type used for each chain, and the models constructed based on different algorithms for each type of model, the selection is either automatically recommended by the system (the system design presets the application conditions of the models, and the system performs feature matching based on the risk type, evolution / stage, and other risk simulation needs) or determined by professionals based on the actual situation.

[0202] In this embodiment, the mechanistic model is a mathematical model based on known physical mechanisms. Before being formally adopted, the mechanistic model usually needs to undergo rigorous parameter calibration and deviation verification. Mechanism models for hydraulic engineering projects include specialized models of dams, power stations, generating units, and gates, and can be one-dimensional, two-dimensional, or three-dimensional. They can simulate water flow, current, and mechanical transmission during the evolution of risks. Comparing the model calculation results with pre-set thresholds can provide a relatively accurate estimate of the consequences of risk accidents. If a mechanistic model exists, it is usually given priority.

[0203] In this embodiment, the intelligent learning model is a big data-driven model, including neural networks, deep learning, deep neural networks, knowledge vector machines (SVM), genetic algorithms, PSO, etc. If there is enough historical risk data, even if people do not understand the internal mechanism of risk, this method can be used to infer the consequences of risk occurrence. When used in conjunction with a mechanism model (if any), the results are cross-validated and more accurate.

[0204] In this embodiment, major risks occur infrequently, but often result in severe consequences. Because of the low frequency of risk occurrence, data is scarce. In such cases, transforming human experience into data variables becomes crucial. This transformation process requires assistance from common-sense reasoning and experiential reasoning, such as brainstorming. This invention provides a method for constructing and improving a risk Bayesian network based on expert probabilistic intuition.

[0205] In this embodiment, mechanistic models and big data-driven models have been extensively studied. The following focuses on the construction of the risk Bayesian network used in this invention. Determining the transmission probability of network nodes is crucial for Bayesian network modeling and risk probability estimation. When historical data is sufficient, prior probability and conditional probability tables can be created by organizing historical data to obtain the transmission probability of risk between nodes. Finally, the probability of risk propagating from the root node to the leaf node is calculated. As mentioned earlier, some risks in water conservancy projects occur infrequently, some even being once-in-a-century or once-in-a-millennium risks, with almost no historical monitoring data. Their occurrence frequency relies entirely on expert or experience estimation. To address this, this invention improves upon the basic Bayesian network model, proposing a risk Bayesian network based on expert probabilistic intuition. This addresses the challenge of determining node probabilities and transmission probabilities between nodes in situations where historical risk monitoring data is severely lacking in the basic Bayesian network. This represents a development and improvement of Bayesian network modeling methods. When using expert brainstorming to determine probabilities, to ensure the fairness and reliability of the final result, a probabilistic intuition-based fuzzy evaluation algorithm is employed to balance the evaluation results of various experts and guarantee the objectivity and validity of the result. Its risk Bayesian method based on probabilistic intuition is as follows:

[0206] First, based on the established risk transmission link space, a point in the network (i.e., the risk location studied in the water conservancy project) is selected, and historical risk data related to that location are collected to identify potential propagation points when a risk occurs at that location. Appropriate analysis methods are selected and variables are set according to the risk location to analyze the direction of risk propagation. Based on the propagation of risk among various nodes, a Bayesian network of risk transmission is constructed.

[0207] Then, when the risk transmission probability cannot be extracted from historical risk data, a language measurement scale is established to divide the risk transmission probability into seven levels. Experts provide evaluation opinions, and the fuzziness score of the risk occurrence probability is obtained by integrating the expert evaluation opinions through fuzzy set theory. The risk transmission probability is then calculated by defuzzification.

[0208] Expert opinions are usually descriptive and qualitative estimates. Quantifying fuzzy and qualitative language requires certain technical processing. The technical process of this invention for processing expert opinions is: fuzzy set theory -> expert estimation -> fuzzification -> aggregation -> defuzzification.

[0209] In a risk transmission chain, the propagation probability between adjacent nodes is given by the following variable U: U = {X1, X2, ..., X...} n The joint probability distribution of} is:

[0210]

[0211] Where P(U) represents the propagation probability between adjacent nodes, P(·) represents the joint probability, n represents the total number of nodes, i represents the number of nodes, and Pa(X) represents the number of nodes. i ) represents node X i Parent set:

[0212] Based on the propagation probability, node X is calculated. i Marginal probability:

[0213]

[0214] Among them, X i ,P(X i ) represents node X i The marginal probability, X j Let j represent a node, and j represent the number of nodes;

[0215] According to node X i The marginal probability is used to calculate the updated risk transmission probability when updating the risk location:

[0216]

[0217] Where P(U / E) represents the updated risk transmission probability, P(U,E) is the joint risk transmission probability of risk variable U and risk variable E, and P(E) represents the risk transmission probability of risk variable E occurring.

[0218] Based on the updated risk propagation probability, fuzzy set theory is used to assign probability values ​​to the risk probabilities that are missing in the Bayesian network.

[0219] A language measurement scale is established, which divides the probability of risk occurrence into seven levels. Each expert judges the probability of risk occurrence at each node. The fuzzy score of the probability of risk occurrence is obtained by integrating the expert opinions through fuzzy set theory. The expert scores are converted into probability numbers by defuzzification and the expert opinions are added to the risk transmission link. The expression of the defuzzification is as follows:

[0220]

[0221] Where X* represents the risk probability, ∫ represents the integral over the membership function, and μ i (x) represents the membership function, x represents the variable, i.e., the fuzzy number, and d represents the integral sign, which is the integral of the fuzzy function;

[0222] Based on the risk transmission chain incorporating expert opinions, and using fuzzy numbers The probability of risk propagation is calculated as follows:

[0223]

[0224] The fuzzy number 'a' describes the uncertainty of expert language. This invention divides it into seven levels, from low to high, namely levels one to seven, for expert knowledge heuristics. The correspondence between the fuzzy number and the expert language measure is shown in Table 1, which is a language measurement measure.

[0225] Table 1

[0226]

[0227] Fuzzy membership degrees represent the uncertainty of expert evaluation by taking numbers between 0 and 1, and trapezoidal and triangular fuzzy numbers are used to capture prior probability values ​​and conditional probability values.

[0228] The transmission probability between each node is calculated, and a risk transmission link is gradually built. Fuzzy set theory probabilities are calculated, and expert opinions are incorporated into the construction of the risk transmission link. A pair of expert opinions; The degree of agreement (similarity) between two different expert opinions; Similarity between two fuzzy numbers; AA(E u ): The average degree of agreement among experts; RA(E) u ): The relative degree of agreement among experts; CC(E) u ): Expert consensus coefficient; The summary results of expert decisions; calculating a pair of expert E u Opinions (u = 1 to M) and Consistency (Similarity)

[0229] The specific process of fuzziness scoring is as follows:

[0230] The average consensus AA(E) of the M experts was calculated. u ):

[0231]

[0232] E u (u = 1, 2, ..., M)

[0233] Based on the average degree of consistency AA(E) u The relative degree of agreement (RA) of the experts was calculated. u ):

[0234]

[0235] According to the relative degree of agreement among experts (RA(E)) u The expert consensus coefficient CC(E) was calculated. u ):

[0236] CC(E u )=β·w(E u )+(1-β)·RA(E u )

[0237] β (0 ≤ β ≤ 1) is the relaxation factor of this method. It shows that w(E) u (The weighting factor of expert u) on RA(E) u The importance of β is considered. When β = 0, the weighting of experts is ignored, and there is a uniform distribution among experts. When β = 1, the consensus coefficient (CC) of experts has the same significance as the weighting. This paper adopts β = 0.5.

[0238] According to the expert consensus coefficient, CC(E) u The summative value of expert opinions was calculated. Complete the scoring process for fuzziness, and summarize the expert opinions as RA(E). u The value is calculated as follows:

[0239]

[0240] The opinions of experts are reasonably adopted based on fuzzy probabilities, and the Bayesian risk transmission link is improved by using real risk data after the risk occurs.

[0241] It should be noted that Bayesian network models also need to be validated before real-world application. The method is as follows: After the Bayesian network is constructed, experts first conduct a qualitative review of the rationality of risk network propagation. To ensure the rationality of the expert's empirical probabilities, the calculation results after expert-assigned values ​​need to be validated. Given sufficient data, a risk propagation path is randomly selected from the Bayesian network, and experts assign probabilities based on their experience. These probabilities replace the conditional probabilities obtained based on data-driven approaches, and the probability of complete risk transmission is calculated. The new calculation structure is compared with the original calculation results. If the results are consistent, the expert-assigned probabilities are considered acceptable; otherwise, error analysis is required to identify the causes of the errors, summarize the experience, and correct the probabilities before re-assigning probabilities. If the data is insufficient and data-driven probability results cannot be used for validation, experts assign propagation probabilities for each historical event and calculate the final result. The probability distribution of each result is statistically analyzed, and the mean probability is calculated and compared with the current calculation result. If the two have a high degree of fit, the Bayesian network analysis and calculation are complete; otherwise, error analysis is performed, the probabilities are corrected, and further calculations are performed.

[0242] In this embodiment, adaptive simulation is employed: To meet the needs of real-time risk control, the risk analysis of this invention utilizes dynamic and adaptive simulation technology. After risk control measures are implemented, the risk transmission chain may be altered or extended. Therefore, it is necessary to re-simulate the risk evolution or development based on the progress of risk relief, using new monitoring data or experience-based judgments, to provide technical support for optimizing the risk control plan or formulating a new one. This process requires interaction between the risk analysis module and the subsequent control module. The interaction method is as follows:

[0243] 1) By analyzing the simulation results of risk analysis, analyze the loss situation of each transmission path, find the weak links in risk transmission, and formulate risk interception plan.

[0244] 2) Compare the input costs of the cutoff plan with the losses recovered, find the risk cutoff position with the highest benefit, upload the risk cutoff plan to the risk control subsystem, and carry out risk cutoff.

[0245] 3) The risk control subsystem feeds back the risk cutoff status to the risk analysis subsystem. Using parallel simulation, it analyzes the next propagation location of the risk. There are multiple paths to choose from for risk cutoff control. There may be intersections between the paths. After the control is implemented, the selected path is retained and the other options are excluded.

[0246] 4) After controlling the selected path, based on the risk cutoff situation, analyze the next propagation trend of the risk by referring to the risk transmission link, and feed the new information back to the risk analysis subsystem.

[0247] If the risk is successfully contained, the normal operation of the water conservancy project will be restored. If it is not contained, steps 2-4 will be repeated until the risk is completely contained.

[0248] In this embodiment, the outcome of the risk analysis phase is a three-layer encirclement: the risk phase is the fundamental purpose of risk control. Previously, this invention proposed dividing the risk into three layers to ensure the complete interruption of risk transmission. Building upon this, this invention creatively proposes an interruption for each layer. Risk interruption is divided into three parts: starting point interruption, midpoint control, and tail point recovery.

[0249] In this embodiment, the layered enclosure division and risk control principles are as follows:

[0250] The first level: the risk is limited to a single water conservancy project, affecting more than one of the functions such as flood control, power generation, ecology, navigation, water supply, and sediment removal. It requires the cooperation of various professional departments within the water conservancy project, and may require coordination between upstream and downstream cascade projects or the power grid department. The risk can be eliminated. Within this range, the formulation of risk cutoff plans should comprehensively consider the investment cost of risk control and the potential increase in losses caused by the delay in risk control.

[0251] The second level: Risks affect a single water conservancy project, impacting more than three of its operational functions, including flood control, power generation, ecology, navigation, water supply, and sediment removal. These risks can be eliminated through coordinated prevention and control measures at different levels. Within this scope, priority should be given to the safety of the water conservancy project, ensuring the complete interruption of the risk transmission chain.

[0252] The third level: Risks require joint scheduling of cascade water conservancy projects. Prior to risk control, the normal operation of the cascade projects will be affected, impacting basin management. Therefore, the basin management commission needs to conduct comprehensive scheduling and risk control. Within this scope, the entire basin's resources should be mobilized for risk control, ensuring the complete severing of risk triggers, source points, and links.

[0253] In this embodiment, starting point interception refers to preventing an accident by implementing a reasonable plan before the risk causes damage to the risk-bearing part. Midpoint control refers to preventing or reducing damage to the risk-bearing part by implementing a reasonable control plan after the risk event has reached it. Tail point restoration refers to repairing the affected location after the risk event has passed, restoring the water conservancy project to its normal operating state.

[0254] like Figure 2As shown, in adaptive simulations, risk control points are constantly changing. Intuitively, when a risk point stops deteriorating, it usually indicates that the risk has been effectively controlled. Real-time feedback on changes in risk points can reflect the difficulties and bottlenecks in the risk cutoff process, allowing for targeted solutions and avoiding a one-size-fits-all approach to risk cutoff. The risk cutoff status is uploaded to the risk analysis module to determine whether risk cutoff is complete. If the risk is successfully cut off, the risk warning is lifted, and the affected location and nearby propagation nodes are closely monitored for a period of time to ensure complete risk cutoff. If the risk is not successfully cut off at the designated location, the risk analysis subsystem re-formulates the risk cutoff plan and re-implements risk control until the risk is completely cut off.

[0255] In this embodiment, after receiving the risk simulation analysis and the delineation results of the three-layer control scope, the risk control subsystem formulates a risk cutoff plan and organizes professional personnel to begin implementing risk cutoff layer by layer.

[0256] It is important to note that during the risk containment process, the spread of risk must be constantly monitored. At this time, continuous simulations and evolution of the risk status and containment situation of the water conservancy project are necessary. If the location of risk spread matches the predicted location in the risk containment plan, the plan will continue to be implemented. If the risk spread location deviates, resulting in new damaged areas, the changes in risk spread will be uploaded to the risk analysis subsystem, and the risk containment plan will be adjusted. The original risk containment plan will need to be optimized, adjusted, or discontinued.

[0257] The risk control mentioned above is based on risk analysis. The following explains how this invention optimizes the risk control scheme in real time:

[0258] First, analyze each transmission link one by one to find the weak links (points) in the risk transmission, consider the consistent time from the risk source to the evolution of the link, and divide the area into encirclements.

[0259] Then, for each risk link, a risk interception scheme is proposed. For the defined encirclement (three-dimensional spatial network), a set of risk interception schemes is proposed, which may include multiple schemes. In particular, this invention employs a parallel simulation method. There may be multiple paths for the next propagation of risk, and these paths may intersect. To meet the needs of real-time risk control, this invention performs synchronous analysis of all possible propagation links, strengthens monitoring data collection, and retains selected paths and excludes others based on the effectiveness of risk control implementation, thus advancing to the next stage of simulation.

[0260] The next step is to establish a multi-option evaluation model or indicator system. This will involve calculating the human and material resource costs of risk cutoff, the increasing risk loss over time, and the growing difficulty in controlling the risk, taking into account various factors to determine the optimal ranking of risk cutoff options.

[0261] In this embodiment, the present invention proposes a fuzzy evaluation algorithm based on probabilistic intuition to address this problem, as follows: The probabilistic intuitionistic fuzzy evaluation method for formulating a risk cutoff scheme involves the following 8 steps:

[0262] A1. Assign weights based on the opinions of decision-makers. Express all decision-makers' opinions in the form of linguistic variables, which are modeled using intuitionistic fuzzy numbers.

[0263]

[0264] Where, λ l The weights represent the decision-makers' respective degrees, μ is the degree of membership, v is the degree of non-membership, and π is the degree of hesitation.

[0265] A2. Construct an aggregated intuitionistic fuzzy decision matrix by utilizing the evaluation criteria of all decision-makers.

[0266] A3. Calculate the weight corresponding to each criterion. Since it is impossible for all criteria to be equally important, a weighted decision matrix is ​​constructed, which utilizes the decision-maker's view of the importance of each criterion.

[0267] A4. Construction of aggregated weighted intuitionistic fuzzy decision matrix.

[0268] A5. Calculate the positive ideal solution and the negative ideal solution of intuitionistic fuzzy calculus.

[0269] A6. Calculate the probability separation metric.

[0270] A7. Evaluate the relative proximity coefficient of the intuitive fuzzy positive ideal solution.

[0271] A8. Rank the alternatives, sort them in descending order, and select the one with the highest relative similarity as the best alternative.

[0272] In this embodiment, the risk cutoff plan is uploaded to the risk control subsystem for risk cutoff. The risk control subsystem feeds back the risk cutoff status to the risk analysis subsystem, which uses a parallel decision-making method to analyze the next propagation location of the risk. This process of risk propagation analysis, risk plan formulation, and risk cutoff is repeated until the risk is completely cut off. It should be noted that insufficient monitoring information is a frequent occurrence after a risk occurs. In particularly urgent situations, satellite remote sensing and aerial remote sensing monitoring should be added to the routine monitoring by ground stations to focus on the risk area and supplement monitoring data, providing a data foundation for the formulation of the risk cutoff plan.

[0273] In this embodiment, the present invention is illustrated using a real-time risk control system for water conservancy projects (hereinafter referred to as the "risk control system"):

[0274] First, the causation value is determined based on the number of contributing factors to the accident. If a risky accident is caused by one contributing factor, the causation value for that factor is recorded as 1. If a risky accident is caused by n contributing factors, the causation value for each contributing factor is 1 / n. This calculation method uses the basic concepts of fractions and frequency distribution to represent the percentage of accident causes.

[0275] Different types of risks have vastly different consequences, and even the same type of risk event can produce significantly different results at different locations within a water conservancy project or at different times of operation. The level of risk consequences is crucial in determining the human, material, and other resource costs invested in risk control.

[0276] Then, a "risk index" was established to assess the impact of risk events on the safe operation of water conservancy projects, classifying the consequences of risk events into five levels. Level 1: Causes permanent damage to water conservancy project structures, equipment, and facilities, resulting in large-scale, mass casualties, significant property losses, and impacting the stable development of the surrounding society, regional economy, and environment; Level 2: May cause permanent damage to water conservancy project structures, equipment, and facilities, resulting in numerous casualties, significant property losses, and social impact; Level 3: Localized damage to water conservancy project structures, equipment, and facilities, resulting in casualties and a certain amount of property loss; Level 4: Temporary damage to water conservancy project structures, equipment, and facilities, but repairable, resulting in minor injuries and property loss; Level 5: Damage in all aspects is less than Level 4.

[0277] Risk events are categorized into five levels based on their frequency of occurrence: Level 1: Frequent occurrence; Level 2: Occurs several times within a certain period; Level 3: Possibly occurs within a certain period; Level 4: Possibly occurs within a certain period but is unlikely to occur; Level 5: Extremely unlikely to occur, and can be considered unlikely to occur. The timeframes are determined by professionals based on the specific circumstances of each water conservancy project.

[0278] The 5x5 risk matrix reflects potential variations where frequency is greater than outcome. Ranking and validation of ratings require the definition of both outcome and frequency indices on a logarithmic scale. The "risk index" is a combination of the frequency index and the severity index, using the following formula:

[0279] Risk Index = Frequency Index × Severity Index

[0280] In this embodiment, based on the previously identified risk sources, the probability of accidents occurring in each part of each type of risk is determined. The transmission chain is analyzed according to three transmission types: hydraulic transmission, structural force transmission, and electrical transmission. A conditional probability table is created using the Bayesian method. Using the probability of each risk event as the prior probability and the risk trigger as the parent node, the next propagation direction and probability are calculated until the risk transmission ends in a historical event. This operation is repeated to calculate the probability of all collected risk events of all types. The risk transmission links obtained for the same type of risk transmission method are two-dimensional. Repeated calculations yield directed acyclic risk transmission links for all risks. First, the two-dimensional risk transmission links of the same type are integrated and drawn. Then, the risk networks drawn for the three types of transmission methods are organically combined. Based on all risk sources, a multi-dimensional, three-dimensional risk transmission link space for water flow, electric current, and structural mechanical flow transmission is drawn.

[0281] Finally, the risk control scheme is enclosed in a multi-dimensional, three-dimensional network space, and the risk control solution proposed in this invention is a risk interception measure with enclosed layers. In the design, this system establishes three layers of enclosure; in actual operation, the number of layers can be simplified or increased. The first layer: the risk is limited to a single water conservancy project, affecting more than one function such as flood control, power generation, ecology, navigation, water supply, and sediment discharge. This requires cooperation among various professionals within the water conservancy project, and may require coordination between upstream and downstream cascade projects or the power grid. The risk can be eliminated. The second layer: the risk affects a single water conservancy project, impacting more than three functions such as flood control, power generation, ecology, navigation, water supply, and sediment discharge. This risk requires coordinated prevention and control by upstream and downstream cascade projects. The fourth layer: the risk requires joint scheduling of cascade water conservancy projects. Prior to risk control, it will affect the normal operation of the cascade and impact watershed management. Comprehensive scheduling and risk control by the watershed commission is required.

[0282] Risk encirclement is based on the risk source location and the aforementioned risk management scope division. It also takes into account the time it takes for risk to spread (the time it takes for risk to spread from the source to the risk encirclement layer is roughly the same). The encirclement range is displayed in the risk transmission link space. Within the encirclement range are resources that can be allocated for real-time risk control. The risk control plan is given based on the encirclement time and resources within the encirclement. At the same time, the control plan for subsequent encirclements is a reserve plan, prepared in advance. If the previous encirclement is breached, it is convenient to quickly allocate resources in the subsequent encirclement.

[0283] The present invention is achieved through the above design:

[0284] 1) Problems encountered during the planning, construction, and operation of hydropower stations were identified, and past risk events at water conservancy projects were collected. To clearly classify and handle these events, a creative classification method was proposed, encompassing six major categories: reservoir dam, spillway, power generation, navigation, ecological environment, and public emergencies. This method not only spatially encompasses the entire water conservancy project, including the dam body, dam foundation, dam abutments, near-dam reservoir bank, and reservoir area, but also includes various auxiliary facilities attached to the main dam structure. It also considers the flood control, power generation, water supply, irrigation, navigation, and ecological functions of the water conservancy project. The operational safety of the water conservancy project was considered from multiple dimensions, including physical damage to the project caused by risks and damage to its functional performance.

[0285] 2) The risk transmission mechanism of water conservancy projects was clarified, namely, transmission along water flow, current, and structural forces. First, the transmission links of risk events were analyzed one by one. Then, comprehensive safety assessment (FSA), fault tree analysis, and / or Bayesian methods were used to connect the various risk sources. Finally, with the reservoir power station as the core and the transmission of water flow, current, and structural forces as the links, an interconnected risk transmission network was formed, thus creating a multi-dimensional and three-dimensional risk transmission link space.

[0286] 3) Mechanism models, data statistical analysis, and empirical probability estimation are a category of models and methods, which include many professional models and algorithms. In the design of this invention, a set of models is provided. In practical applications, the selection of each model is made by professionals based on the actual situation and existing models.

[0287] 4) Based on the established risk transmission links, a quantitative analysis model of risk transmission is constructed using a combination of fuzzy set theory and Bayesian methods. Then, the risk transmission links are extracted using system dynamics methods, laying the foundation for subsequent risk transmission analysis and real-time risk control.

[0288] 5) Real-time risk interception is divided into three parts: initial interception, midpoint control, and tail-point recovery. Initial interception involves implementing a reasonable plan to cut off the risk before it damages the risk-bearing area, thus preventing the risk from spreading to the water conservancy project and affecting its normal operation, potentially causing an accident. Midpoint control occurs when the risk event has reached the risk-bearing area of ​​the water conservancy project. At this point, it is necessary to determine the location of the risk's impact point, analyze its position within the three-level containment system, and develop a reasonable control plan to avoid or reduce damage to the risk-bearing area. Tail-point recovery involves repairing the affected area after the risk event has passed, restoring the water conservancy project to normal operation. Throughout these three stages, risk control information is fed back to the risk management plan development team in real time. Based on the feedback, the control plan is adjusted promptly, and the process of initial interception, midpoint control, and tail-point recovery continues until the risk is completely intercepted and the water conservancy project returns to normal operation.

[0289] 6) Risk analysis is dynamic. Based on information provided by the risk monitoring subsystem, the location and type of risk are identified. A corresponding risk analysis subsystem is selected to simulate the risk evolution. The simulation results are used to analyze the next propagation point of the risk. Then, based on the specific circumstances of the next location, an appropriate analysis model is selected for further prediction. For example, when a flood risk occurs, a hydrodynamic model is used to simulate the locations of water conservancy projects that may be affected by the flood. The simulation results are used as the risk origin at these locations, and a corresponding analysis model is selected to analyze the next propagation point of the risk. This process continues to derive the risk transmission chain and evolution process.

[0290] 7) The risk transmission link space also continuously develops and improves with the operation of the hydropower project. The construction of the risk transmission link is based on the internal motion mechanism of the hydropower station and human knowledge and experience. Using an intelligent learning network as a platform, the risk transmission link adopts a method combining fuzzy set theory and Bayesian analysis to construct the risk transmission link for the hydropower project. The mechanism model is used to find the location of risk occurrence and its transmission mode, initially constructing the risk transmission chain; the human knowledge and experience model is used to analyze the propagation path of historical risk events to improve the risk transmission chain; by inputting the risk transmission link into the intelligent learning model, machine learning algorithms are used to deeply explore the relationships between the units in the risk transmission link, forming a complete risk transmission link. The risk transmission link is continuously updated and improved through automatic learning, including establishing relationships between existing risk units and adding new risk units.

[0291] 8) Based on basic Bayesian methods, a risk Bayesian method based on probabilistic intuition is created. This method can quantitatively analyze the probability of risk occurrence and identify risk propagation paths when risk monitoring data and historical data are scarce, providing a basis for the formulation of subsequent risk cutoff plans.

[0292] Example 2

[0293] like Figure 3 As shown, this invention provides a method for real-time control of operational risks in water conservancy projects, the implementation of which is as follows:

[0294] S1. Risk Monitoring Phase: Construct a risk transmission link, identify key monitoring points based on the risk transmission link, deploy a risk monitoring network, and, based on the risk monitoring network, monitor in real time the changes in the external environment of the hydropower station and the operation of internal units through terminal monitoring equipment and manual inspections.

[0295] S2. Risk Analysis Phase: Analyze the risk transmission chain based on real-time monitoring results, determine the risk occurrence point, formulate and screen the optimal risk interception plan based on the risk occurrence point, and feed back the risk interception status to the risk analysis phase through the risk control phase. Use parallel simulation method to analyze the risk propagation location synchronously.

[0296] S3. Risk Control Phase: Use the optimal risk cutoff scheme to cut off risks and feed the risk cutoff results back to the risk analysis phase.

[0297] like Figure 3 The real-time control method for operational risks of water conservancy projects provided in the embodiment shown can execute the technical solution shown in the real-time control system for operational risks of water conservancy projects in the above system embodiment. Its implementation principle and beneficial effects are similar, and will not be repeated here.

[0298] Those skilled in the art will recognize that the embodiments described herein are intended to help the reader understand the principles of the invention, and should be understood that the scope of protection of the invention is not limited to such specific statements and embodiments. Those skilled in the art can make various other specific modifications and combinations based on the technical teachings disclosed in this invention without departing from the spirit of the invention, and these modifications and combinations are still within the scope of protection of this invention.

Claims

1. A real-time control system for operational risks of a water conservancy project, characterized in that, include: The risk monitoring subsystem is used to construct a risk transmission link, find key monitoring points based on the risk transmission link, deploy a risk monitoring network, and, based on the risk monitoring network, monitor the changes in the external environment of the hydropower station and the operation of internal units in real time through terminal monitoring equipment and manual inspections. The risk monitoring subsystem includes: The identification module is used to identify defects in the planning and construction phases of water conservancy projects and historical problems that occurred in the early operation phases, establish a risk event set for water conservancy projects, and classify the location of risk triggers into external and internal factors based on the risk event set. The risk transmission link drawing module is used to analyze the relationships between various risk sources based on the risk triggers involved in the risk event set of the water conservancy project, draw the risk transmission link, and locate key monitoring points based on the risk transmission link to deploy a risk monitoring network; the risk transmission link drawing module includes: The risk source location identification submodule is used to identify the risk source locations involved in the risk evolution and development of the risk event set of the water conservancy project, starting with the hidden risks formed by defects in the planning and construction of the water conservancy project, and combining the actual risks existing in the actual operation process. The risk transmission link construction submodule is used to analyze the risk propagation and evolution relationship between individual water conservancy projects and cascade water conservancy projects, connect the various risk sources to obtain the risk transmission link, and find key monitoring points based on the risk transmission link to deploy a risk monitoring network. The risk transmission of individual water conservancy projects includes structural force transmission, and the risk transmission of cascade water conservancy projects includes hydraulic transmission and power transmission. The monitoring module is used to monitor changes in the external environment of the hydropower station and the operation of its internal units in real time through terminal monitoring equipment and manual inspections, based on the risk monitoring network, and to update the real-time monitoring information. The risk analysis subsystem is used to analyze the risk transmission chain based on real-time monitoring results, determine the risk occurrence point, formulate and screen the optimal risk interception scheme based on the risk occurrence point, and feed back the risk interception status to the risk analysis subsystem through the risk control subsystem, and use parallel simulation method to analyze the risk propagation location synchronously. The risk analysis subsystem includes: The positioning module is used for data-driven analysis, based on received anomaly monitoring information, to monitor the corresponding risk occurrence point in the risk transmission chain, and to locate the risk occurrence point within the risk transmission chain; and For event-driven scenarios, after a risk occurs, the system connects to the real-time control system for operational risks of water conservancy hubs, locates the risk occurrence point through received early warning or monitoring signals, and positions it within the risk transmission chain based on that risk occurrence point. The risk chain simulation module is used to simulate and analyze the risk transmission process based on the location results, using mechanism models, intelligent learning models, and artificial knowledge and experience models, and to obtain the risk transmission results. The risk adaptability simulation module is used to formulate risk cutoff schemes based on the risk transmission results, and to screen the optimal risk cutoff scheme using a fuzzy evaluation method based on probabilistic intuition. It also feeds back the risk cutoff status to the risk analysis subsystem through the risk control subsystem, and uses parallel simulation methods to perform synchronous analysis on the risk propagation location. The optimal risk interception scheme includes: a starting point interception scheme, a midpoint control scheme, and a tail point recovery scheme. A multi-layered risk control mechanism is constructed, establishing a three-tiered risk control perimeter. Level 1: Risks can be intercepted by individual water conservancy projects without affecting the operation of upstream and downstream water conservancy projects. Level 2: Risks require comprehensive scheduling of upstream and downstream cascade hydropower stations to intercept, affecting their normal operation. Level 3: Risks affect the normal operation of water conservancy projects throughout the basin, requiring comprehensive scheduling and control of the entire basin. The three-tiered perimeter is mapped to the risk network space. When a risk occurs, the risk core is identified, and based on this core, the potential locations of the risk are divided according to the three-tiered risk standard, forming three perimeters. This multi-layered protection ensures the complete interception of the risk link. The risk control subsystem is used to cut off risks using the optimal risk cutoff scheme and to feed back the risk cutoff results to the risk analysis subsystem.

2. The real-time control system for operational risks of water conservancy projects according to claim 1, characterized in that, The risk monitoring subsystem includes a normal monitoring mode that uses both data-driven and event-driven methods to monitor the operation of the hydropower station in real time, and an emergency monitoring mode that supplements the risk area with aerial remote sensing and satellite remote sensing when the normal monitoring mode cannot fully monitor the risk area. The data-driven approach is used to provide real-time monitoring of the operation of internal units of the water conservancy project using terminal monitoring equipment, and to promptly reflect the operational status of the water conservancy project. The event-driven mechanism is used to monitor changes in the external environment of the hydropower station through manual inspections and to report any discovered risk events.

3. The real-time control system for operational risks of water conservancy projects according to claim 2, characterized in that, The risk event set for the water conservancy project includes power generation risk, flood control risk, navigation risk, dam and reservoir area risk, ecological risk, and public emergency risk. The aforementioned power generation risks include the impact of non-electricity system risks on reservoir operation and scheduling, vibration of hydropower plant units, impact of tailrace turbulence on gates during flood discharge, risk of large-scale power outages at hydropower stations, risk of flooding of plant buildings, and risk of failure of major equipment and facilities and units at hydropower stations. The flood control risks include the impact of reservoir discharge on low-frequency vibrations of the dam site and surrounding structures, the impact of reservoir discharge vibrations on equipment and facilities, the impact of density currents entering the dam front on generating units and sediment discharge, the impact of water conservancy hub discharge on water conservancy hub structures and near-dam bank slopes, the impact of water conservancy hub discharge on ship lifts, and cavitation problems in stilling basins. The shipping risks mentioned include the impact of flood discharge from water conservancy projects on downstream waterways, the impact of the operation and scheduling of cascade water conservancy projects on downstream shipping, and the backwater effect of tributaries on the downstream water level of the main stream. The risks in the reservoir area include the impact of floods exceeding the design standard on the safety of the water conservancy project structures, the impact of earthquakes exceeding the design standard on the safety of the water conservancy project structures, the impact of dam deformation on the normal operation of the water conservancy project structures, the impact of landslides on the safety of the water conservancy project, the long-term stability of dam foundation seepage control measures and their impact on dam safety, and the surge problems caused by landslides. The ecological risks mentioned include the impact of hydropower station flood discharge and power generation tailwater on downstream fish, and the impact of sewage discharge and reduced hydrodynamic force in the reservoir area of ​​water conservancy hubs on water quality; The aforementioned public health emergencies include shipwrecks and the appearance of large floating objects on the water.

4. The real-time control system for operational risks of water conservancy projects according to claim 1, characterized in that, The specific steps of using artificial knowledge and experience models to simulate and analyze the risk transmission process are as follows: In the risk transmission chain, select the risk location of the water conservancy project, collect historical risk data related to the risk location, and sort out the direction of risk propagation when the risk occurs in the location. Based on the location of the risk, the direction of risk propagation is analyzed, and a Bayesian network of risk transmission is constructed based on the propagation of risk between various risk occurrence points. When the probability of risk transmission cannot be extracted from historical risk data, a language measurement scale is established to divide the probability of risk transmission into seven levels. Experts provide evaluation opinions, and the fuzziness score of the probability of risk occurrence is obtained by integrating the expert evaluation opinions through fuzzy set theory. The probability of risk transmission is then calculated by defuzzification. Based on the risk propagation probability, the Bayesian network is updated using real risk data, and the updated Bayesian network is used to simulate and analyze the risk transmission process to obtain the risk transmission results.

5. The real-time control system for operational risks of water conservancy projects according to claim 4, characterized in that, The calculated probability of risk propagation is as follows: In the risk transmission chain, the propagation probability between adjacent nodes is calculated: in, This represents the propagation probability between adjacent nodes. Denotes the joint probability. Indicates the total number of nodes. Indicates the number of nodes. Represents a node The parent set; The node is calculated based on the propagation probability between adjacent nodes. Marginal probability: in, Represents a node The marginal probability, Represents a node. Indicates the number of nodes; According to the node The marginal probability is used to calculate the updated risk transmission probability when updating the risk location: in, This represents the updated probability of risk propagation. It is the joint risk transmission probability of risk variable U and risk variable E. This represents the probability of risk propagation occurring in risk variable E; Based on the updated risk propagation probability, fuzzy set theory is used to assign probability values ​​to the risk probabilities that are missing in the Bayesian network. A language measurement scale is established, which divides the probability of risk occurrence into seven levels. Each expert judges the probability of risk occurrence at each node. The expert judgments are integrated using fuzzy set theory to obtain a fuzzy score of the probability of risk occurrence. Defuzzification is then used to convert the fuzzy score into a probability number. Based on the probability number, the expert judgments are added to the risk transmission chain. The expression for defuzzification is as follows. in, Indicates the probability of risk. This represents the integral over the membership function. Represents the membership function. This represents a variable, i.e., a fuzzy number. Indicates the integral sign, used for integrating fuzzy functions; Based on the risk transmission chain incorporating expert opinions, and using fuzzy numbers... The probability of risk propagation is calculated as follows: 。 6. The real-time control system for operational risks of water conservancy projects according to claim 5, characterized in that, The specific process of fuzziness scoring is as follows: Calculated M Average consensus among experts : in, Indicates the number of experts. Indicates the number of experts. Indicates the similarity between two fuzzy numbers; Based on average consistency The relative degree of agreement among experts was calculated. : ； Based on the relative degree of agreement among experts The expert consensus coefficient was calculated. : in, Represents the relaxation factor. Experts Weighting factors; Based on expert consensus coefficient The summative value of expert opinions was calculated. Complete the scoring process for fuzziness: in, This indicates M experts. This represents the consensus coefficient of M experts.

7. The real-time control system for operational risks of water conservancy projects according to claim 6, characterized in that, The risk adaptation simulation module includes: The risk cutoff plan formulation submodule is used to analyze the loss situation of each propagation path based on the risk transmission results and formulate a risk cutoff plan. The optimal risk cutoff scheme selection submodule is used to select the optimal risk cutoff scheme based on the risk cutoff scheme using a fuzzy evaluation method based on probability intuition, and upload the optimal risk cutoff scheme to the risk control subsystem. The first feedback submodule is used to feed back the risk cutoff status to the risk analysis subsystem through the risk control subsystem according to the optimal risk cutoff scheme, and to perform synchronous analysis on the risk propagation location using parallel simulation methods.

8. The real-time control system for operational risks of water conservancy projects according to claim 7, characterized in that, The optimal risk truncation scheme selection submodule includes: The weighting allocation unit is used to allocate weights based on the decision-maker's opinion. The Aggregate Intuitive Fuzzy Decision Matrix Construction Unit is used to construct an aggregate intuitive fuzzy decision matrix based on the assigned weights and the decision-maker's evaluation criteria. The weight calculation unit is used to calculate the weight corresponding to each evaluation criterion of the level based on the aggregated intuitionistic fuzzy decision matrix. The aggregated weighted intuitionistic fuzzy decision matrix construction unit is used to construct the aggregated weighted intuitionistic fuzzy decision matrix based on the weights corresponding to each evaluation criterion of the level and the aggregated intuitionistic fuzzy decision matrix. The ideal solution calculation unit is used to calculate the intuitionistic fuzzy positive ideal solution and the intuitionistic fuzzy negative ideal solution based on the aggregated weighted intuitionistic fuzzy decision matrix. The probability separation metric calculation unit is used to calculate the probability separation metric based on the intuitionistic fuzzy positive ideal solution and the intuitionistic fuzzy negative ideal solution. Evaluation unit, used to evaluate the relative proximity coefficient of intuitionistic fuzzy positive ideal solutions based on a probability separation metric; The filtering unit is used to rank the risk cutoff schemes based on the relative proximity coefficient and select the risk cutoff scheme with the highest relative proximity as the optimal risk cutoff scheme.

9. The control method for a real-time control system for operational risk of a water conservancy project according to any one of claims 1 to 8, characterized in that, Includes the following steps: S1. Risk Monitoring Phase: Construct a risk transmission link, identify key monitoring points based on the risk transmission link, deploy a risk monitoring network, and, based on the risk monitoring network, monitor in real time the changes in the external environment of the hydropower station and the operation of internal units through terminal monitoring equipment and manual inspections. S2. Risk Analysis Phase: Analyze the risk transmission chain based on real-time monitoring results, determine the risk occurrence point, formulate and screen the optimal risk interception plan based on the risk occurrence point, and feed back the risk interception status to the risk analysis phase through the risk control phase. Use parallel simulation method to analyze the risk propagation location synchronously. S3. Risk Control Phase: Use the optimal risk cutoff scheme to cut off risks and feed the risk cutoff results back to the risk analysis phase.

Images