Power grid flood prevention risk early warning and auxiliary decision making system based on flood deduction

Abstract

The invention discloses a power grid flood prevention risk early warning and auxiliary decision making system based on flood deduction, and relates to the technical field of flood storage area power grid flood prevention. Compared with a previous flood storage area power grid flood prevention method, the problem that an existing flood deduction model is limited in simulation precision and accuracy is solved; the coupling degree of power grid equipment and the flood model is low; cross-department collaboration and information sharing are difficult; and the problems of system visualization and insufficient interactivity are solved. Accurate flood deduction is carried out through the flood deduction model, information such as the flood development trend, the flood submerging range and the water depth of power equipment can be predicted in advance, and prospective guidance is provided for power grid flood prevention work; meanwhile, a power grid operation department adopts a coping scheme in advance based on a risk early warning strategy; when a flood disaster occurs, the system can quickly provide an emergency decision scheme, and the emergency response speed is improved; and VR and AR technologies are combined, so that the visualization degree is greatly improved, and the first-aid repair accuracy and speed are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of flood control for power grids in flood storage areas, and in particular to a flood control risk early warning and auxiliary decision-making system for power grids based on flood simulation. Background Art

[0002] Frequent flood disasters pose a serious threat to the safe and stable operation of the power grid, so more advanced technical means are needed for risk warning and decision support. With the development of the economy and the continuous growth of social demand for electricity, the scale of the power grid has continued to expand and the structure has become increasingly complex. Substations, transmission lines and other equipment in the power grid are widely distributed and face varying degrees of risk in the face of flood disasters.

[0003] The rapid development of sensor technology, Internet of Things technology, satellite remote sensing technology, etc., has made it possible to collect and monitor flood-related data and power grid equipment operating status data in real time. The rapid development of computer technology and the continuous improvement of numerical simulation methods have made it possible to perform high-precision numerical simulation of flood evolution processes. By establishing a flood evolution model, it is possible to predict the flood inundation range, water depth, flow rate and other information, providing an important basis for the assessment of flood control risks in power grids. The emergence of artificial intelligence technology and big data technology has provided new means for processing and analyzing massive amounts of flood and power grid data. These technologies can be used to mine and analyze data, extract valuable information, and achieve more accurate risk warnings and auxiliary decision-making.

[0004] However, the existing flood simulation models have limited simulation accuracy and precision in complex terrain, urban environments, and the interaction of multiple factors. In addition, the connection and interaction between the existing flood simulation models and the power grid equipment models are not tight enough to accurately reflect the actual impact of floods on power grid equipment. The information sharing mechanism between relevant departments such as meteorology, water conservancy, and electricity is imperfect, and there are obstacles to data circulation and sharing, which is not conducive to the timely release and implementation of emergency decisions. In addition, the visualization of risk warning and auxiliary decision-making results in existing technologies is not good, and it is difficult to intuitively present information such as the scope of flood inundation and the damage to power grid equipment.

[0005] In order to solve the above problems, the present invention proposes a power grid flood prevention risk warning and auxiliary decision-making system based on flood simulation. Summary of the invention

[0006] The purpose of the present invention is to propose a power grid flood prevention risk warning and auxiliary decision-making system based on flood simulation to solve the problems raised in the background technology: Existing flood simulation models have certain limitations, resulting in limited simulation precision and accuracy; the coupling between power grid equipment and flood models is low; cross-departmental collaboration and information sharing are difficult, resulting in obstacles to subsequent rescue and disaster relief; and the system lacks visualization and interactivity.

[0007] In order to achieve the above object, the present invention adopts the following technical solutions: The power grid flood prevention risk warning and auxiliary decision-making system based on flood simulation includes: 3D model building module: used to build 3D twin models of distribution network equipment and external environment; Flood dynamic simulation module: used to perform flood dynamic simulation and calculate the flood depth of power equipment; Flood prevention risk warning module: used for risk warning according to the calculation results of the flood dynamic simulation module; Equipment auxiliary decision module: used to support auxiliary decision-making based on various sensing devices in the flood prevention and emergency repair of distribution network; The flood dynamic simulation module includes a flood dynamic simulation unit and a flood depth calculation unit, which are respectively used for performing flood dynamic simulation and calculating the flood depth of power equipment; the flood dynamic simulation unit divides dynamic grids based on the terrain data of the study area where the distribution network equipment is located, and the dynamic grids are adaptively adjusted according to the changes in water flow velocity and water depth during the flood dynamic simulation process; each dynamic grid builds a one-dimensional hydrodynamic model and a two-dimensional hydrodynamic model, and performs flood simulation based on a multi-scale fusion network fusion model; the flood depth calculation unit calculates the flood depth of different areas based on the simulation results of the flood dynamic simulation unit, and performs corresponding flood depth calculations for different power equipment.

[0008] Preferably, the steps of the three-dimensional model building module to build a three-dimensional twin model are as follows: S1: Based on high-precision sensors and monitoring equipment, collect the physical parameters of distribution network equipment and the external environment data of the area where the distribution network equipment is located; S2: Build an initial 3D model based on the collected data, and associate the equipment model with real-time operation data based on digital twin technology to achieve dynamic simulation of the model; S3: Integrate the distribution network equipment model with the external environment model, and establish an interaction mechanism between the equipment and the environment to achieve the simulation of two-way impact; S4: Conduct simulation analysis of the operating conditions of the fused 3D twin model, and train the model based on historical data and machine learning algorithms to predict future operating conditions; S5: Based on the display interface of visualization effect, VR and AR technology are combined to present the three-dimensional twin model; according to the analysis results of the model, decision support is provided for the equipment auxiliary decision module.

[0009] Preferably, the one-dimensional hydrodynamic model is specifically constructed as follows:

[0010] in,r It is the side inflow; Q is the cross-sectional discharge of the river; S is the water-passing area; D For the width of the river; H is the water level; R x is the component of the lateral inflow velocity in the water flow direction; M is the flow modulus; l is the momentum correction factor; t For time; x It represents the distance along the direction of water flow; g is the acceleration due to gravity; represents partial derivative; When the outflow is free outflow, the cross-sectional flow of the river Q 1 is:

[0011] When the outflow is submerged, the flow rate of the river section Q 2 is:

[0012] in, h u is the node water level on the building; h is the building top elevation; h d is the water level of the node below the building; α is the free outflow coefficient; β is the flooding outflow coefficient; The two-dimensional hydrodynamic model is as follows:

[0013]

[0014]

[0015] in, h 0 is water depth; v x For water flowing x Direction, i.e. the velocity of the water in the direction it flows; v y For water flowing y Direction, that is, the flow velocity vertical to the direction of water flow; u x , u y is the source term; Based on the multi-scale fusion network, the built one-dimensional model and two-dimensional model are fused, and the flood is simulated through the fused model; The multi-scale fusion network is specifically as follows: Extract one-dimensional model feature set G a With the 2D model feature set G b The features of are used as input, and are changed to the same number of channels through a 1×1 convolution. The two-dimensional model feature set is concatenated with the one-dimensional model feature set that has been upsampled by bilinear interpolation. The concatenated feature set is passed through a 3×3 convolution layer to obtain an offset field of the same size as the two-dimensional model feature. ,in, R represents the field of real numbers, 2 and N b Both represent dimensions. N b × N b " represents the spatial size of the offset field; specifically: F = conv ( con ( up ( G a ),( G b ))) in, up ( · ) is bilinear interpolation upsampling; con ( · ) is a splicing operation; conv ( · ) is a 3×3 convolutional layer; After obtaining the offset field, the one-dimensional model features are distorted into two-dimensional model features, and each point (X, Y) of the offset field is represented by the one-dimensional model feature set. G a Point (X a ,Y a ) mapping, as follows:

[0016] in, x and y is the learnable 2-D transformation offset of point (X,Y); γ is the mapping parameter; Through the differentiable bilinear sampling mechanism using (X a ,Y a ) to obtain the output of the distorted two-dimensional model feature set, and the final feature fusion result W as follows:

[0017] Among them, (X b ,Y b ) is the two-dimensional model feature set G b The point in The outflow coefficient in one-dimensional hydrodynamic model based on improved Grey Wolf optimization algorithm α、β and the fusion coefficient in the multi-scale fusion network n Optimize The improved gray wolf optimization algorithm is as follows: The coefficient value interval is divided into the first interval, the second interval and the third interval, which correspond to the first level, the second level and the third level of the three social levels of the gray wolf group. The optimization process of the coefficient corresponds to the hunting behavior of the wolf pack. The value of the coefficient corresponds to the position vector of the individual wolf pack, and the value of the optimal coefficient corresponds to the position vector of the prey individual, as follows:

[0018]

[0019]

[0020] in, U 1. U 2. U 3 correspond to the updated coefficient values ​​of the first interval, the second interval and the third interval of the coefficient value interval respectively; V 1. V 2. V 3 correspond to the optimal coefficient values ​​to be updated in the first interval, the second interval and the third interval of the coefficient value interval respectively; r 11 and r 12 , r 21 and r 22 , r 31 and r 32 A random vector between [0,1] corresponding to the first interval, the second interval and the third interval of the value interval of the coefficient respectively; t is the number of updates; T is the maximum number of updates; The error calculation of the flood simulation model is performed based on the optimal coefficient values ​​of the three intervals currently obtained, and the parameter value with the smallest error is selected as the optimal parameter value for the next iterative update:

[0021] in, U ( t+ 1) For the t+ 1 updated coefficient value; U best,j Indicates the current j An optimal solution; r 3 is a random vector between [0,1]; s It is an adjustment parameter used to prevent the denominator from being 0; x is the control parameter; max j 、min j For the j The upper and lower bounds of the optimal solution; d Is a sensitive factor.

[0022] Preferably, the flood dynamic simulation module also feeds back the dynamic simulation results to the three-dimensional model building module for real-time visual display.

[0023] Preferably, the flood risk warning module studies and trains the historical calculation results of the flood dynamic deduction module based on the LSTM model with optimized parameters based on the improved Gray Wolf optimization algorithm, and also integrates meteorological forecast data, geographic information data and socio-economic data to establish a comprehensive risk assessment indicator system, and performs weighted and comprehensive analysis on different types of data; and inputs new deduction results into the trained model in real time to quickly predict risk levels and areas at risk of disasters.

[0024] Preferably, the flood risk warning module also divides the study area into different risk zones, and sets corresponding risk indicators and warning levels for each zone according to flood simulation results and geographical features.

[0025] Preferably, the equipment auxiliary decision-making module collects perception data from different perception devices and performs deep fusion based on a multi-scale fusion network to construct a knowledge graph for flood prevention and emergency repair of distribution networks for information integration.

[0026] Preferably, the equipment auxiliary decision-making module also simulates flood disaster scenarios of different degrees to conduct emergency drills based on the data of the sensing device.

[0027] Preferably, the equipment auxiliary decision module also combines GPS and GIS technology to locate the positions of emergency repair personnel and materials in real time, and plans the optimal emergency repair route according to the flood inundation situation and road traffic conditions.

[0028] Compared with the prior art, the present invention provides a power grid flood prevention risk warning and auxiliary decision-making system based on flood simulation, which has the following beneficial effects: The present invention uses a flood simulation model to perform accurate flood simulation, and can predict in advance the development trend of the flood, the scope of inundation, the water depth of power equipment and other information, provide forward-looking guidance for the flood prevention work of the power grid, and enable the power grid operation department to take proactive preventive measures instead of passively responding to disasters; at the same time, based on the risk warning strategy, the power grid operation department can reinforce, protect or transfer equipment in flood risk areas in advance; when a flood disaster occurs, the system can quickly provide emergency decision-making plans and improve the emergency response speed; it also combines VR and AR technologies to greatly improve the degree of visualization and improve the accuracy and speed of emergency repairs. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 This is a system block diagram mentioned in Embodiment 1 of the present invention; Figure 2 This is a schematic diagram of the dynamic grid mentioned in Example 1 of the present invention; Figure 3 A schematic diagram of the flooded water depth of the area mentioned in Example 1 of the present invention; Figure 4 This is a schematic diagram of different water level risk warnings mentioned in Example 1 of the present invention.

[0030] Meaning of the symbols in the figure: 100. Three-dimensional model building module; 200. Flood dynamic simulation module; 210. Flood dynamic simulation unit; 220. Inundation depth calculation unit; 300. Flood risk warning module; 400. Equipment auxiliary decision module. DETAILED DESCRIPTION

[0031] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0032] The present invention uses a flood simulation model to accurately simulate floods, and can predict the development trend of floods, the scope of inundation, the water depth of power equipment and other information in advance, providing forward-looking guidance for power grid flood prevention work, so that power grid operation departments can take proactive preventive measures instead of passively responding to disasters; at the same time, based on risk warning strategies, power grid operation departments can reinforce, protect or transfer equipment in flood risk areas in advance; when flood disasters occur, the system can quickly provide emergency decision-making plans to improve emergency response speed; it also combines VR and AR technologies to greatly improve visualization, and improve the accuracy and speed of emergency repairs. Specifically including the following contents.

[0033] Embodiment 1: See also Figure 1-4 The present invention provides a power grid flood prevention risk warning and auxiliary decision-making system based on flood simulation, including: The three-dimensional model building module 100 is used to build a three-dimensional twin model of the distribution network equipment and the external environment. The specific steps are as follows: S1: Data collection and integration Use high-precision sensors and monitoring equipment to collect data on the physical parameters of distribution network equipment, such as size, shape, material, operating status, etc. For example, use a laser scanner to obtain the precise geometry of the equipment. Collect external environmental data, including topography, meteorological conditions (temperature, humidity, wind speed, etc.), geographic information, etc. For example, obtain real-time meteorological data through meteorological satellites.

[0034] S2: Model building and optimization Based on the collected data, use professional 3D modeling software to create the initial 3D model of the distribution network equipment. For example, use 3DMAX or Maya and other software. Apply digital twin technology to associate the equipment model with real-time operation data to achieve dynamic simulation. For example, when the equipment operation status changes, the corresponding parameters in the model are updated in real time.

[0035] S3: Environmental Integration and Interaction The distribution network equipment model is integrated with the external environment model to consider the impact of environmental factors on the equipment. For example, the effect of strong winds on electric poles can be simulated. An interaction mechanism between the equipment and the environment can be established to simulate two-way impacts. For example, the impact of equipment heating on the surrounding temperature, and the feedback of temperature changes on equipment performance.

[0036] S4: Simulation Analysis and Prediction Use simulation software to simulate and analyze the operating conditions of the fused 3D twin model. For example, predict the probability of equipment failure under different environmental conditions. Train the model based on historical data and machine learning algorithms to predict future operating conditions. For example, predict the maintenance cycle of equipment through neural networks.

[0037] S5: Visualization and decision support Develop a highly visual display interface to present the 3D twin model to users in an intuitive way. For example, use virtual reality (VR) or augmented reality (AR) technology to allow users to experience the model immersively. Based on the analysis results of the model, provide decision support for the planning, operation, maintenance and management of the distribution network. For example, arrange maintenance plans in advance based on the predicted failure risks.

[0038] Flood dynamic simulation module 200: used for performing flood dynamic simulation and calculating the flood depth of power equipment; the flood dynamic simulation module 200 includes a flood dynamic simulation unit 210 and a flood depth calculation unit 220, which are used for performing flood dynamic simulation and calculating the flood depth of power equipment respectively.

[0039] Reference Figure 2 The flood dynamic simulation unit 210 divides the dynamic grid based on the terrain data of the study area where the distribution network equipment is located. The dynamic grid is adaptively adjusted according to the water flow velocity and water depth changes during the flood dynamic simulation process, and the resolution of the grid can be automatically adjusted. In areas with turbulent water flow and drastic changes in water depth, the grid is encrypted to improve the calculation accuracy; in areas with stable water flow, a coarser grid is used to save computing resources. Moving boundary technology can also be used to more accurately simulate the flooding and retreat process of the flood; each dynamic grid builds a one-dimensional hydrodynamic model and a two-dimensional hydrodynamic model, and performs flood simulation based on a multi-scale fusion network fusion model; the one-dimensional hydrodynamic model is specifically constructed as follows:

[0040] in, r It is the side inflow; Q is the cross-sectional discharge of the river; S is the water-passing area; D For the width of the river; H is the water level; R x is the component of the lateral inflow velocity in the water flow direction; M is the flow modulus; l is the momentum correction factor; t For time; x It represents the distance along the direction of water flow; g is the acceleration due to gravity; represents partial derivative.

[0041] When the outflow is free outflow, the cross-sectional flow of the river Q 1 is:

[0042] When the outflow is submerged, the flow rate of the river section Q 2 is:

[0043] in, h u is the node water level on the building; h is the building top elevation; h d is the water level of the node below the building; α is the free outflow coefficient; β is the flooding outflow coefficient; The two-dimensional hydrodynamic model is as follows:

[0044]

[0045]

[0046] in, h 0 is water depth; v x For water flowing x Direction, i.e. the velocity of the water in the direction it flows; v y For water flowing y Direction, that is, the flow velocity vertical to the direction of water flow; u x , u y is the source term; Based on the multi-scale fusion network, the built one-dimensional model and two-dimensional model are fused, and the flood is simulated through the fused model; The multi-scale fusion network is as follows: Extract one-dimensional model feature set G a With the 2D model feature set G b The features of are used as input, and are changed to the same number of channels through a 1×1 convolution. The two-dimensional model feature set is concatenated with the one-dimensional model feature set that has been upsampled by bilinear interpolation. The concatenated feature set is passed through a 3×3 convolution layer to obtain an offset field of the same size as the two-dimensional model feature. ,in, R represents the field of real numbers, 2 and N b Both represent dimensions. N b × N b " represents the spatial size of the offset field; specifically: F = conv ( con ( up ( G a ),( G b ))) in, up ( · ) is bilinear interpolation upsampling; con ( · ) is a splicing operation; conv ( · ) is a 3×3 convolutional layer; After obtaining the offset field, the one-dimensional model features are distorted into two-dimensional model features, and each point (X, Y) of the offset field is represented by the one-dimensional model feature set. G a Point (X a ,Y a) mapping, as follows:

[0047] in, x and y is the learnable 2-D transformation offset of point (X,Y); γ is the mapping parameter; Through the differentiable bilinear sampling mechanism using (X a ,Y a ) to obtain the output of the distorted two-dimensional model feature set, and the final feature fusion result W as follows:

[0048] Among them, (X b ,Y b ) is the two-dimensional model feature set G b The point in The outflow coefficient in one-dimensional hydrodynamic model based on improved Grey Wolf optimization algorithm α、β and the fusion coefficient in the multi-scale fusion network n Optimize The improved gray wolf optimization algorithm is as follows: The coefficient value interval is divided into the first interval, the second interval and the third interval, which correspond to the first level, the second level and the third level of the three social levels of the gray wolf group. The optimization process of the coefficient corresponds to the hunting behavior of the wolf pack. The value of the coefficient corresponds to the position vector of the individual wolf pack, and the value of the optimal coefficient corresponds to the position vector of the prey individual, as follows:

[0049]

[0050]

[0051] in, U 1. U 2. U 3 correspond to the updated coefficient values ​​of the first interval, the second interval and the third interval of the coefficient value interval respectively; V 1. V 2. V 3 correspond to the optimal coefficient values ​​to be updated in the first interval, the second interval and the third interval of the coefficient value interval respectively; r 11 and r 12 , r 21 and r22 , r 31 and r 32 A random vector between [0,1] corresponding to the first interval, the second interval and the third interval of the value interval of the coefficient respectively; t is the number of updates; T is the maximum number of updates; The error calculation of the flood simulation model is performed based on the optimal coefficient values ​​of the three intervals currently obtained, and the parameter value with the smallest error is selected as the optimal parameter value for the next iterative update:

[0052] in, U ( t+ 1) For the t+ 1 updated coefficient value; U best,j Indicates the current j An optimal solution; r 3 is a random vector between [0,1]; s It is an adjustment parameter used to prevent the denominator from being 0; x is the control parameter; max j 、min j For the j The upper and lower bounds of the optimal solution; d is a sensitive factor; Reference Figure 3 , based on the fusion of different types of hydrodynamic models, for example, using a two-dimensional water flow model in the main channel of flood evolution, and using a one-dimensional shallow water equation model in the vast floodplain, the advantages of both can be combined to calculate the flood depth, and the flood depth at this time can be calculated according to the location of the power equipment. This can improve the accuracy and reliability of flood depth calculation in flood simulation, and provide stronger support for the assessment and response of flood disasters.

[0053] The water levels at different locations based on the above flood simulation model can be found in Table 1: Table 1 Flood simulation results at different locations

[0054] Comparing the measured water levels at different locations with the deduced water levels, the maximum error in the water level does not exceed 20 cm, proving that the model in this embodiment has a high deduction accuracy.

[0055] Flood risk warning module 300: used to issue risk warning according to the calculation results of flood dynamic simulation module 200; Reference Figure 4In addition to the calculation results of the flood dynamic deduction module 200, the flood risk warning module 300 also integrates meteorological forecast data (such as rainfall, wind speed), geographic information data (such as land use type, population distribution) and socio-economic data (such as building value, distribution of important infrastructure). For example, combined with real-time rainfall forecasts, the development trend and potential risks of floods can be more accurately judged. The flood risk warning module 300 uses an optimized LSTM model to learn and train the historical calculation results of the flood dynamic deduction module 200. For example, flood deduction data under different rainfall, terrain conditions and water conservancy facilities are input to allow the model to learn the complex relationship between flood inundation range, water depth and flow rate and risk level. Establish a comprehensive risk assessment indicator system, perform weighted and comprehensive analysis on different types of data, input new deduction results into the trained model in real time, and quickly predict risk levels and possible disaster-stricken areas.

[0056] The study area is also divided into different risk zones, and unique risk indicators and warning levels are set for each zone based on flood simulation results and geographical features. For example, the city is divided into different areas such as commercial areas, residential areas and industrial areas, and different warning strategies are formulated according to the importance and vulnerability of each area. Customized warning information is sent to residents and relevant departments in different zones, including evacuation routes, emergency resource distribution, etc.

[0057] The flood risk warning module 300 also feeds back the dynamic simulation results to the three-dimensional model building module 100 for real-time visualization. This includes the flow of floods, submerged areas, and changes in water depth. For example, using virtual reality (VR) or augmented reality (AR) technology, decision makers can understand the flood risk situation in an immersive way. An interactive operation interface can also be provided, allowing users to select different parameters and scenarios for simulation and adjust the warning strategy in real time.

[0058] Equipment auxiliary decision module 400: used to support auxiliary decision-making based on various sensing devices in the flood prevention and emergency repair of the distribution network; The equipment decision-making support module 400 integrates data from different sensing devices (such as water level sensors, meteorological sensors, power equipment monitoring sensors, etc.) and performs deep fusion. For example, the water level data is associated with the local rainfall and wind speed data to analyze its comprehensive impact on the distribution network equipment. A correlation model between the data is established to explore potential laws and trends to provide a more comprehensive basis for decision-making support. In addition, a knowledge graph for flood prevention and emergency repair of the distribution network is constructed to integrate equipment information, fault types, emergency repair methods and other knowledge. For example, the vulnerable parts, common faults and solutions of different types of power equipment are included in the knowledge graph. It is also possible to combine expert experience and data from sensing devices to provide intelligent suggestions and solutions for emergency repair decisions through expert systems.

[0059] The equipment auxiliary decision module 400 can also build a virtual simulation environment based on the data collected by the sensing device to simulate flood disaster scenarios of different degrees. For example, the impact of floods on distribution network equipment can be simulated in a virtual environment to evaluate the effects of different emergency repair plans. Emergency drills based on virtual simulation can be carried out to familiarize emergency repair personnel with the response process and strategy in advance, thereby improving the efficiency of actual emergency repairs.

[0060] The equipment auxiliary decision module 400 combines the global positioning system (GPS) and the geographic information system (GIS) to locate the positions of the repair personnel and materials in real time. For example, the distribution of the repair team can be accurately displayed on the map. According to the flood inundation situation and road traffic conditions, the optimal travel path is planned for the repair personnel to ensure that they can quickly reach the repair site.

[0061] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A power grid flood risk warning and decision-making support system based on flood simulation, characterized by: include: A three-dimensional model building module (100): used to build a three-dimensional twin model of the distribution network equipment and the external environment; Flood dynamic simulation module (200): used for performing flood dynamic simulation and calculating the flood depth of power equipment; A flood prevention risk warning module (300): used to issue a risk warning based on the calculation results of the flood dynamic simulation module (200); Equipment auxiliary decision module (400): used to support auxiliary decision-making based on multiple sensing devices in flood prevention and emergency repair of distribution network; The flood dynamic deduction module (200) comprises a flood dynamic deduction unit (210) and a flood depth calculation unit (220), which are respectively used for performing flood dynamic deduction and calculating the flood depth of power equipment; wherein the flood dynamic deduction unit (210) divides dynamic grids based on the terrain data of the research area where the distribution network equipment is located, and the dynamic grids are adaptively adjusted according to the changes in water flow velocity and water depth during the flood dynamic deduction process; each dynamic grid constructs a one-dimensional hydrodynamic model and a two-dimensional hydrodynamic model, and performs flood deduction based on a multi-scale fusion network fusion model; the flood depth calculation unit (220) calculates the flood depth of different areas based on the deduction results of the flood dynamic deduction unit (210), and performs corresponding flood depth calculations for different power equipment.

2. The power grid flood prevention risk warning and auxiliary decision-making system based on flood simulation according to claim 1 is characterized in that: The three-dimensional model building module (100) performs the following steps to build a three-dimensional twin model: S1: Based on high-precision sensors and monitoring equipment, collect the physical parameters of distribution network equipment and the external environment data of the area where the distribution network equipment is located; S2: Build an initial 3D model based on the collected data, and associate the equipment model with real-time operation data based on digital twin technology to achieve dynamic simulation of the model; S3: Integrate the distribution network equipment model with the external environment model, and establish an interaction mechanism between the equipment and the environment to achieve the simulation of two-way impact; S4: Conduct simulation analysis of the operating conditions of the fused 3D twin model, and train the model based on historical data and machine learning algorithms to predict future operating conditions; S5: Based on the display interface of the visualization effect, the three-dimensional twin model is presented in combination with VR and AR technologies; and according to the analysis results of the model, decision support is provided for the equipment auxiliary decision module (400).

3. The power grid flood prevention risk warning and auxiliary decision-making system based on flood simulation according to claim 2 is characterized in that: The one-dimensional hydrodynamic model is specifically constructed as follows: in, r It is the side inflow; Q is the cross-sectional discharge of the river; S is the water-passing area; D For the width of the river; H is the water level; R x is the component of the lateral inflow velocity in the water flow direction; M is the flow modulus; λ is the momentum correction factor; t For time; x It represents the distance along the direction of water flow; g is the acceleration due to gravity; represents partial derivative; When the outflow is free outflow, the cross-sectional flow of the river Q 1 is: When the outflow is submerged, the flow rate of the river section Q 2 is: in, h u is the node water level on the building; h is the building top elevation; h d is the water level of the node below the building; α is the free outflow coefficient; β is the flooding outflow coefficient; The two-dimensional hydrodynamic model is as follows: in, h 0 is water depth; v x For water flowing x Direction, i.e. the velocity of the water in the direction it flows; v y For water flowing y Direction, that is, the flow velocity vertical to the direction of water flow; u x , u y is the source term; Based on the multi-scale fusion network, the built one-dimensional model and two-dimensional model are fused, and the flood is simulated through the fused model; The multi-scale fusion network is specifically as follows: Extract one-dimensional model feature set G a With the 2D model feature set G b The features of are used as input, and are changed to the same number of channels through a 1×1 convolution. The two-dimensional model feature set is concatenated with the one-dimensional model feature set that has been upsampled by bilinear interpolation. The concatenated feature set is passed through a 3×3 convolution layer to obtain an offset field of the same size as the two-dimensional model feature. ,in, R represents the field of real numbers, 2 and N b Both represent dimensions, N b × N b " represents the spatial size of the offset field; specifically: F = conv ( con ( up ( G a ),( G b ))) in, up ( · ) is bilinear interpolation upsampling; con ( · ) is a splicing operation; conv ( · ) is a 3×3 convolutional layer; After obtaining the offset field, the one-dimensional model features are distorted into two-dimensional model features, and each point (X, Y) of the offset field is represented by the one-dimensional model feature set. G a Point (X a ,Y a ) mapping, as follows: in, x and y is the learnable 2-D transformation offset of point (X,Y); γ is the mapping parameter; Through the differentiable bilinear sampling mechanism using (X a ,Y a ) to obtain the output of the distorted two-dimensional model feature set, and the final feature fusion result W as follows: Among them, (X b ,Y b ) is the two-dimensional model feature set G b The point in The outflow coefficient in one-dimensional hydrodynamic model based on improved Grey Wolf optimization algorithm α、β and the fusion coefficient in the multi-scale fusion network n Optimize The improved gray wolf optimization algorithm is as follows: The coefficient value interval is divided into the first interval, the second interval and the third interval, which correspond to the first level, the second level and the third level of the three social levels of the gray wolf group. The optimization process of the coefficient corresponds to the hunting behavior of the wolf pack. The value of the coefficient corresponds to the position vector of the individual wolf pack, and the value of the optimal coefficient corresponds to the position vector of the prey individual, as follows: in, U 1. U 2. U 3 correspond to the updated coefficient values ​​of the first interval, the second interval and the third interval of the coefficient value interval respectively; V 1. V 2. V 3 correspond to the optimal coefficient values ​​to be updated in the first interval, the second interval and the third interval of the coefficient value interval respectively; r 11 and r 12 , r 21 and r 22 , r 31 and r 32 A random vector between [0,1] corresponding to the first interval, the second interval and the third interval of the value interval of the coefficient respectively; τ is the number of updates; T is the maximum number of updates; The error calculation of the flood simulation model is performed based on the optimal coefficient values ​​of the three intervals currently obtained, and the parameter value with the smallest error is selected as the optimal parameter value for the next iterative update: in, U ( τ+ 1) For the τ+ 1 updated coefficient value; U best,j Indicates the current j An optimal solution; r 3 is a random vector between [0,1]; σ It is an adjustment parameter used to prevent the denominator from being 0; ξ is the control parameter; max j 、min j For the j The upper and lower bounds of the optimal solution; δ Is a sensitive factor.

4. The power grid flood prevention risk warning and auxiliary decision-making system based on flood simulation according to claim 3 is characterized in that: The flood dynamic simulation module (200) also feeds back the dynamic simulation results to the three-dimensional model building module (100) for real-time visual display.

5. The power grid flood prevention risk warning and auxiliary decision-making system based on flood simulation according to claim 4 is characterized in that: The flood risk warning module (300) learns and trains the historical calculation results of the flood dynamic deduction module (200) based on the LSTM model with parameter optimization based on the improved gray wolf optimization algorithm, and also integrates meteorological forecast data, geographic information data and socio-economic data to establish a comprehensive risk assessment indicator system, and performs weighted and comprehensive analysis on different types of data; and inputs new deduction results into the trained model in real time to quickly predict risk levels and areas at risk of disaster.

6. The power grid flood prevention risk warning and auxiliary decision-making system based on flood simulation according to claim 5 is characterized in that: The flood risk warning module (300) further divides the study area into different risk zones, and sets corresponding risk indicators and warning levels for each zone according to flood simulation results and geographical features.

7. The power grid flood prevention risk warning and auxiliary decision-making system based on flood simulation according to claim 1 is characterized in that: The equipment auxiliary decision module (400) collects perception data from different perception devices and performs deep fusion based on a multi-scale fusion network, thereby constructing a knowledge graph for flood prevention and emergency repair of the distribution network for information integration.

8. The power grid flood prevention risk warning and auxiliary decision-making system based on flood simulation according to claim 7 is characterized in that: The equipment auxiliary decision module (400) also simulates flood disaster scenarios of different degrees to conduct emergency drills based on the data of the sensing device.

9. The power grid flood prevention risk warning and auxiliary decision-making system based on flood simulation according to claim 8 is characterized in that: The equipment auxiliary decision module (400) also combines GPS and GIS technology to locate the positions of emergency repair personnel and materials in real time, and plans the optimal emergency repair route based on the flood inundation situation and road traffic conditions.

Images