A reservoir capacity dynamic assessment method and system
Through chaos theory and graph neural network analysis, a dynamic spatial interaction model was established, and the reservoir risk nodes were identified, which solved the lag problem of dynamic assessment of reservoir capacity in the existing technology, real-time monitoring and risk prediction of reservoir operation were realized. Improved efficiency and accuracy of flood control and water resource management.
Patent Information
- Application Number
- CN202411139478.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-20
- Publication Date
- 2024-11-05
- Estimated Expiration
- 2044-08-20
AI Technical Summary
The existing dynamic assessment methods for reservoir capacity rely on static or lagging water level data, making it difficult to deal with rapidly changing environmental conditions and emergencies, resulting in insufficient flood control and emergency response, and lack of flexibility and real-time ability to deal with complex dynamic systems.
Using a combination of chaos theory and graph neural network, the dynamic hydrological dependency graph is obtained through real-time hydrological data analysis, nonlinear patterns and numerical changes are identified, dynamic spatial interaction models are established, key risk nodes are identified, and random forests and gradient hoists are used. The model integration method conducts risk assessment to achieve real-time dynamic monitoring and prediction of reservoir capacity and water level changes.
It significantly improves the flexibility and responsiveness to reservoir operation strategies, enhances the safety and effectiveness of flood control and water resource allocation, and can respond to rapid environmental and climate changes more accurately.
Smart Images

Figure CN118657386B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of water level measurement, and in particular to a reservoir capacity dynamic evaluation method and system. Background Art
[0002] The field of water level measurement technology focuses on determining the height of water bodies, which is essential for water resource management, flood control, and the operation and management of reservoirs and dams. This technology includes various measurement tools and methods, such as water level gauges, radars, ultrasonic sensors, and more modern satellite remote sensing technology. Accurate water level data can be used to monitor and predict the water dynamics of reservoirs, rivers, lakes and seas, supporting hydrological research and environmental monitoring. In addition, water level data is also the basis for assessing water resource availability and planning water conservancy facilities.
[0003] Among them, the dynamic assessment method of reservoir capacity refers to a method of real-time monitoring and assessment of reservoir capacity, which is mainly used to ensure the effective operation of the reservoir, the optimal allocation of water resources, and flood control safety. Through dynamic assessment, managers can make quick decisions based on current and predicted water level changes, optimize reservoir water management and scheduling, and ensure the safety protection of downstream areas and the satisfaction of water supply needs.
[0004] Existing technologies mainly rely on static or lagged water level data, which makes it difficult to cope with rapidly changing environmental conditions and emergencies. For example, under extreme climate events, traditional methods may not be able to reflect the rapid changes in water levels in a timely manner, resulting in inadequate flood control and emergency response. Existing technologies generally do not have the flexibility and real-time performance required to handle complex dynamic systems in data processing and analysis, which limits the accuracy and efficiency of reservoir management. For example, when a reservoir encounters continuous rainfall or long-term drought, traditional technologies are often unable to adjust the reservoir's operating strategy in real time, which may lead to waste or shortage of water resources. Summary of the invention
[0005] The purpose of the present invention is to solve the shortcomings in the prior art and to propose a reservoir capacity dynamic evaluation method and system.
[0006] In order to achieve the above object, the present invention adopts the following technical scheme: a reservoir capacity dynamic evaluation method, comprising the following steps:
[0007] S1: Based on the real-time hydrological data collected from the reservoir basin and inflow, chaos theory is used to analyze the sensitive dependence of the data. By calculating the Lyapunov exponent of each data point, nonlinear patterns and numerical changes in the data are identified to obtain the hydrological dynamic dependence diagram;
[0008] S2: Input the data in the hydrological dynamic dependency graph into the graph neural network, update and adjust the information of the reservoir dam and the nodes between the river basins, analyze the interactive impact of the real-time data between the nodes, adjust the network weights and connection strategies of the nodes, and obtain a dynamic spatial interaction model;
[0009] S3: Analyze the current reservoir water storage capacity data and real-time water level change information using the dynamic spatial interaction model, identify key risk nodes in the reservoir, conduct risk point analysis, and obtain risk identification results;
[0010] S4: Perform data analysis on the risk identification results, use the model integration method of random forest and gradient boosting machine, compare the risk assessment values output by multiple models, select model prediction results based on model performance, and establish risk assessment results.
[0011] As a further solution of the present invention, the steps of obtaining the hydrological dynamic dependency graph are specifically as follows:
[0012] S111: Extract time series from real-time hydrological data and calculate the moving weighted average of the data using the formula:
[0013] ;
[0014] Generate time series analysis results;
[0015] in, represents the data value at time point t, represents the sequence analysis results at time point t, is the weight coefficient of the time series, The number of historical data points for reference;
[0016] S112: Apply chaos theory to analyze the time series analysis results, using the formula:
[0017] ;
[0018] Generate chaos analysis results;
[0019] in, is the chaos analysis result at time point t, To adjust the parameters, is the standard deviation of the squared values of the time series, is the average value of the time series;
[0020] S113: Using the chaos analysis results, the nonlinear patterns and numerical changes in the data are identified by calculating the Lyapunov exponent of each data point, using the formula:
[0021] ;
[0022] Get the hydrological dynamic dependency graph;
[0023] in, is the hydrological dynamic dependency graph, is the weight of each node in the dependency graph, is the total number of time points.
[0024] As a further solution of the present invention, the step of acquiring the dynamic space interaction model is specifically as follows:
[0025] S211: extracting node data from the hydrological dynamic dependency graph, calculating the initial influence of multiple nodes, and performing initial calculation of node influence using the formula:
[0026] ;
[0027] Generate node influence analysis results;
[0028] in, is the influence analysis result of the node, is the set of neighbor nodes of the node, is the distance between nodes i and j, is the connection weight between nodes;
[0029] S212: According to the node influence analysis results, update the information of reservoir dams and river basin nodes, using the formula:
[0030] ;
[0031] Generate node information update results;
[0032] in, The node information update result, , To adjust the parameters, is the information exchange coefficient, Neighbor node The impact results;
[0033] S213: Using the node information update result, adjust the node's network weight and connection strategy using the formula:
[0034] ;
[0035] Generate network weight adjustment results;
[0036] in, is the adjusted network weight between nodes i and j, , are the information update results of nodes i and j respectively, , is the adjustment factor;
[0037] S214: Analyze the interactive impact of real-time data between nodes according to the network weight adjustment result, using the formula:
[0038] ;
[0039] Get the dynamic spatial interaction model;
[0040] in, is a dynamic spatial interaction model. Real-time data interaction volume.
[0041] As a further solution of the present invention, the steps of obtaining the risk identification results are specifically as follows:
[0042] S311: Analyze the current reservoir water storage capacity data using the dynamic spatial interaction model, using the formula:
[0043] ;
[0044] Generate reservoir storage capacity analysis results;
[0045] in, The water storage capacity analysis results of the representative node, represents the node weights in the model, represents the water storage capacity of the node, represents the baseline water storage capacity shift;
[0046] S312: Based on the analysis results of the reservoir's water storage capacity and combined with real-time water level change information, the formula is used:
[0047] ;
[0048] Generate node risk assessment results;
[0049] in, The risk assessment result of the representative node, Represents the real-time water level of the node, represents the adjustment factor, represents the smoothing parameter to prevent division by zero errors;
[0050] S313: Based on the node risk assessment results, by identifying the key risk nodes in the reservoir, marking the high-risk nodes, using the formula:
[0051] ;
[0052] Get the key risk node marking results;
[0053] in, Represents the key risk node marking results, represents a high-risk node set, Represents the marking function;
[0054] S314: Combining the reservoir storage capacity analysis results, the node risk assessment results and the key risk node marking results, the formula is used:
[0055] ;
[0056] Obtain risk identification results;
[0057] Among them, Z represents the risk identification result, Represents the value of the node in the key risk node marking result, Represents the total number of nodes, Represents a weighted index.
[0058] As a further solution of the present invention, the steps of obtaining the risk assessment results are specifically as follows:
[0059] S411: Perform data analysis on the risk identification results and use the random forest model integration method to make preliminary risk prediction using the formula:
[0060] ;
[0061] Generating Random Forest Predictions ;
[0062] in, represents the number of trees in the model, The risk identification result is data points;
[0063] S412: Analyze the risk identification results using a gradient boosting model, using the formula:
[0064] ;
[0065] Generate Gradient Boosting Machine predictions ;
[0066] in, represents the number of iterations in the model, is the learning rate for each iteration;
[0067] S413: Compare the random forest prediction result and the gradient boosting machine prediction result, using the formula:
[0068] ;
[0069] Select the model prediction result with the best performance to obtain the optimal model prediction result P;
[0070] S414: Based on the prediction result of the preferred model, the formula is used:
[0071]
[0072] Get risk assessment results ;
[0073] in, Represents the total number of models referenced in the evaluation.
[0074] A reservoir capacity dynamic assessment system, the reservoir capacity dynamic assessment system is used to execute the reservoir capacity dynamic assessment method, the system comprises:
[0075] The hydrological data analysis module collects reservoir inflow and rainfall data, performs time series analysis on each data point, calculates dynamic response values, extracts key dependencies from the data, and generates a hydrological dynamic dependency graph;
[0076] The graph network update module extracts node information from the hydrological dynamic dependency graph, updates the properties of nodes and edges in the graph, adjusts the weight of each node, matches the actual changes in hydrological conditions, and forms a dynamic spatial interaction model;
[0077] The risk assessment module uses a dynamic spatial interaction model to conduct real-time monitoring of water levels and water storage at multiple key nodes of the reservoir, identify risk nodes, and combine data interaction between nodes to form key risk analysis results;
[0078] The model integrated decision module evaluates the key risk analysis results, compares the risk prediction data of differentiated models, selects model outputs based on accuracy, and combines the data of all participating models to generate risk assessment results.
[0079] Compared with the prior art, the advantages and positive effects of the present invention are:
[0080] In the present invention, by analyzing the sensitive dependencies of real-time hydrological data, for example, by identifying nonlinear patterns and numerical changes, it is possible to more accurately grasp the subtle changes in hydrological data. Compared with traditional methods, this analysis method is more adaptable to rapid changes in the environment and climate. By processing these data using graph neural networks, the information between nodes in the reservoir system can be updated and adjusted in real time, thereby enhancing the overall responsiveness and flexibility of the system. In addition, accurate risk point analysis and model prediction improve the accuracy of predicting potential risks, significantly improving the safety and effectiveness of flood control and water resource allocation. BRIEF DESCRIPTION OF THE DRAWINGS
[0081] Figure 1 It is a schematic diagram of the workflow of the present invention;
[0082] Figure 2 A flow chart of the steps for obtaining the hydrological dynamic dependency graph of the present invention;
[0083] Figure 3 A flowchart of the steps for obtaining the dynamic spatial interaction model of the present invention;
[0084] Figure 4 A flow chart of the steps for obtaining the risk identification results of the present invention;
[0085] Figure 5 The figure is a flow chart of the steps for obtaining the risk assessment results of the present invention. DETAILED DESCRIPTION
[0086] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.
[0087] In the description of the present invention, it should be understood that the terms "length", "width", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, in the description of the present invention, "multiple" means two or more, unless otherwise clearly and specifically defined.
[0088] Embodiment 1
[0089] See also Figure 1 The present invention provides a technical solution: a reservoir capacity dynamic assessment method, comprising the following steps:
[0090] S1: Based on the real-time hydrological data collected from the reservoir basin and inflow, chaos theory is used to analyze the sensitive dependence of the data. By calculating the Lyapunov exponent of each data point, nonlinear patterns and numerical changes in the data are identified to obtain the hydrological dynamic dependence diagram;
[0091] S2: Input the data in the hydrological dynamic dependency graph into the graph neural network, update and adjust the information of the reservoir dam and the nodes between the river basins, analyze the interactive impact of the real-time data between the nodes, adjust the network weights and connection strategies of the nodes, and obtain a dynamic spatial interaction model;
[0092] S3: Use the dynamic spatial interaction model to analyze the current reservoir water storage capacity data and real-time water level change information, identify the key risk nodes in the reservoir, conduct risk point analysis, and obtain risk identification results;
[0093] S4: Perform data analysis on the risk identification results, use the model integration method of random forest and gradient boosting machine, compare the risk assessment values output by multiple models, select model prediction results based on model performance, and establish risk assessment results.
[0094] The hydrological dynamic dependency diagram includes sensitivity data points, pattern recognition results, and change index. The dynamic spatial interaction model specifically includes node interaction data, weight adjustment records, and connection efficiency. The risk identification results specifically refer to key risk nodes, risk point analysis, and water level change data. The risk assessment results include model prediction results, risk assessment values, and selected model performance.
[0095] See also Figure 2 ,The specific steps for obtaining the hydrological dynamic dependency graph are:
[0096] S111: Extract time series from real-time hydrological data and calculate the moving weighted average of the data using the formula:
[0097] ;
[0098] Generate time series analysis results;
[0099] in, represents the data value at time point t, represents the sequence analysis results at time point t, is the weight coefficient of the time series, The number of historical data points for reference;
[0100] S112: Apply chaos theory to analyze the time series analysis results, using the formula:
[0101] ;
[0102] Generate chaos analysis results;
[0103] in, is the chaos analysis result at time point t, To adjust the parameters, is the standard deviation of the squared values of the time series, is the average value of the time series;
[0104] S113: Using the chaos analysis results, the nonlinear patterns and numerical changes in the data are identified by calculating the Lyapunov exponent of each data point, using the formula:
[0105] ;
[0106] Get the hydrological dynamic dependency graph;
[0107] in, is the hydrological dynamic dependency graph, is the weight of each node in the dependency graph, is the total number of time points.
[0108] formula:
[0109] ;
[0110] Detailed explanation and derivation:
[0111] : represents the data point k units before time t.
[0112] : Represents the weight coefficient, which should be assigned according to the distance of the data point from the current time point (generally speaking, the closer the distance, the greater the weight).
[0113] : Indicates the number of historical data points for reference.
[0114] Calculation example:
[0115] Assume the number of historical data points to refer to , and the weight distribution is , data points ,but:
[0116] ;
[0117] formula:
[0118]
[0119] Detailed explanation and derivation:
[0120] The standard deviation of the squared values.
[0121] mean The average value of .
[0122] : An adjustment parameter that controls the degree of influence of the ratio.
[0123] Calculation example:
[0124] Assumptions =2, The value is 24.73, assuming the standard deviation is 5 and the mean is 20, then:
[0125] ;
[0126] formula:
[0127] ;
[0128] Detailed explanation and derivation:
[0129] : The weight of the node.
[0130] : Chaos analysis results at time point t.
[0131] : Total number of nodes.
[0132] Calculation example:
[0133] Assume there are 3 nodes , the weight is , The value is 2.43, then:
[0134] ;
[0135] The hydrological dynamic dependency graph is gradually established through calculations at each step.
[0136] See also Figure 3 ,The specific steps for obtaining the dynamic spatial interaction model are:
[0137] S211: Extract node data from the hydrological dynamic dependency graph, calculate the initial influence of multiple nodes, and perform initial calculation of node influence using the formula:
[0138] ;
[0139] Generate node influence analysis results;
[0140] in, is the influence analysis result of the node, is the set of neighbor nodes of the node, is the distance between nodes i and j, is the connection weight between nodes;
[0141] S212: Update the information of reservoir dams and river basin nodes according to the results of node influence analysis, using the formula:
[0142] ;
[0143] Generate node information update results;
[0144] in, The node information update result, , To adjust the parameters, is the information exchange coefficient, Neighbor node The impact results;
[0145] S213: Use the node information update result to adjust the node's network weight and connection strategy using the formula:
[0146] ;
[0147] Generate network weight adjustment results;
[0148] in, is the adjusted network weight between nodes i and j, , are the information update results of nodes i and j respectively, , is the adjustment factor;
[0149] S214: According to the network weight adjustment result, the interactive impact of real-time data between nodes is analyzed, and the formula is used:
[0150] ;
[0151] Get the dynamic spatial interaction model;
[0152] in, is a dynamic spatial interaction model. Real-time data interaction volume.
[0153] formula:
[0154] ;
[0155] Assumptions:
[0156] Assume that node i has two neighbor nodes j and k.
[0157] distance unit, unit.
[0158] Connection weight .
[0159] Calculation process:
[0160] Calculate the denominator: ;
[0161] Calculate the numerator part:
[0162] ;
[0163] Divide the numerator by the denominator to get ;
[0164] Result analysis:
[0165] Represents the influence index of the node. The higher the value, the greater the influence of the node in the network.
[0166] formula:
[0167] ;
[0168] Assumptions:
[0169] Use the calculation in the previous step .
[0170] , assuming .
[0171] Calculation process:
[0172] calculate ;
[0173] calculate ;
[0174] Adding these two results gives ;
[0175] Result analysis:
[0176] The update information index of the representative node affects its interaction intensity with other nodes.
[0177] formula:
[0178] ;
[0179] Assumptions:
[0180] ,use and hypothetical .
[0181] Calculation process:
[0182] calculate ;
[0183] calculate ;
[0184] calculate ;
[0185] Result analysis:
[0186] Represents the adjusted connection weight between nodes i and j, and its value affects the interaction strength between nodes.
[0187] formula:
[0188] ;
[0189] Assumptions:
[0190] , assuming the real-time data interaction volume .
[0191] Calculation process:
[0192] calculate ;
[0193] calculate ;
[0194] Result analysis:
[0195] It represents the total intensity of interactions between nodes in the entire network and is an indicator of network activity.
[0196] See also Figure 4 , the specific steps for obtaining risk identification results are:
[0197] S311: Use the dynamic spatial interaction model to analyze the current reservoir storage capacity data using the formula:
[0198] ;
[0199] Generate reservoir storage capacity analysis results;
[0200] in, The water storage capacity analysis results of the representative node, represents the node weights in the model, represents the water storage capacity of the node, represents the baseline water storage capacity shift;
[0201] S312: Based on the analysis results of the reservoir's water storage capacity and combined with real-time water level change information, the formula is used:
[0202] ;
[0203] Generate node risk assessment results;
[0204] in, The risk assessment result of the representative node, Represents the real-time water level of the node, represents the adjustment factor, represents the smoothing parameter to prevent division by zero errors;
[0205] S313: Based on the node risk assessment results, the key risk nodes in the reservoir are identified and the high-risk nodes are marked using the formula:
[0206] ;
[0207] Get the key risk node marking results;
[0208] in, Represents the key risk node marking results, represents a high-risk node set, Represents the marking function;
[0209] S314: Combine the reservoir storage capacity analysis results, node risk assessment results and key risk node marking results, and use the formula:
[0210] ;
[0211] Obtain risk identification results;
[0212] Among them, Z represents the risk identification result, Represents the value of the node in the key risk node marking result, Represents the total number of nodes, Represents a weighted index.
[0213] formula:
[0214] ;
[0215] in:
[0216] : The weight of the node;
[0217] : water storage capacity of the node;
[0218] : Baseline water storage capacity offset;
[0219] :Used to stabilize the impact of high water storage nodes and avoid excessive impact of extreme values on the model;
[0220] Calculation process and examples:
[0221] Assume that the weight of a node , water storage capacity Ten thousand cubic meters, baseline shift .
[0222] ;
[0223] The value represents the water storage capacity assessment result of node i under given model weights and real-time data.
[0224] formula:
[0225] ;
[0226] in:
[0227] It is the water storage capacity.
[0228] Represents the real-time water level.
[0229] , To adjust the coefficients and smoothing parameters, avoid zero denominator.
[0230] Calculation process and example: Assuming real-time water level Meter, adjustment factor , smoothing parameter .
[0231] ;
[0232] The value represents the risk level assessment of the node under given real-time conditions.
[0233] formula:
[0234] ;
[0235] in:
[0236] It is a collection of high-risk nodes.
[0237] is a function that converts risk value into a more manageable token.
[0238] Calculation process and examples:
[0239] Assume the threshold is , the node is high risk, .
[0240] ;
[0241] This represents the cumulative risk score for the set of high-risk nodes.
[0242] formula:
[0243]
[0244] in:
[0245] is the total number of nodes.
[0246] is a weighted index.
[0247] Calculation process and examples:
[0248] Assuming the total number of nodes , weighted index .
[0249] ;
[0250] ;
[0251] ;
[0252] ;
[0253] This indicates the risk level of the entire system, with 1.622 indicating a lower risk on the risk scale.
[0254] See also Figure 5 , the specific steps for obtaining risk assessment results are:
[0255] S411: Perform data analysis on the risk identification results and use the random forest model ensemble method to make preliminary risk predictions using the formula:
[0256] ;
[0257] Generating Random Forest Predictions ;
[0258] in, represents the number of trees in the model, is the vth data point in the risk identification result;
[0259] S412: Use the gradient boosting model to analyze the risk identification results, using the formula:
[0260] ;
[0261] Generate Gradient Boosting Machine predictions ;
[0262] in, represents the number of iterations in the model, is the learning rate for each iteration;
[0263] S413: Compare the prediction results of random forest and gradient boosting machine, using the formula:
[0264] ;
[0265] Select the model prediction result with the best performance to obtain the optimal model prediction result P;
[0266] S414: Based on the prediction results of the optimal model, the formula is:
[0267] ;
[0268] Get risk assessment results ;
[0269] in, Represents the total number of models referenced in the evaluation.
[0270] Detailed explanation and derivation of formulas and parameters
[0271] Assumptions:
[0272] is the risk identification result of the jth data point.
[0273] is the number of trees in the random forest model.
[0274] is the number of iterations of the gradient boosting machine model.
[0275] is the learning rate for each iteration.
[0276] Assume the following values:
[0277] -Risk identification result array
[0278] - Number of trees
[0279] - Number of iterations
[0280] - The learning rate for each iteration
[0281] formula:
[0282] ;
[0283] Calculation process:
[0284] For each risk identification result ,calculate and .
[0285] Add these two results.
[0286] Sum all results.
[0287] Divide the sum by the number of trees .
[0288] Calculation example:
[0289] for :
[0290] ;
[0291] ;
[0292] ;
[0293] ;
[0294] formula:
[0295] ;
[0296] Calculation process:
[0297] For each ,calculate and .
[0298] Add these two results.
[0299] Multiply by the learning rate .
[0300] Sum all results.
[0301] Divide the sum by the number of iterations .
[0302] Calculation example:
[0303] for :
[0304] ;
[0305] ;
[0306] ;
[0307] ;
[0308] formula:
[0309] ;
[0310] Calculation process:
[0311] square and .
[0312] Sum.
[0313] Take the square root.
[0314] Calculation example:
[0315] ;
[0316] formula:
[0317]
[0318] Calculation process:
[0319] Cubic P.
[0320] cube and , and sum.
[0321] use Divide by the sum.
[0322] Calculation example:
[0323] (Two models)
[0324] ;
[0325] A reservoir capacity dynamic assessment system, the reservoir capacity dynamic assessment system is used to execute the above reservoir capacity dynamic assessment method, the system comprises:
[0326] The hydrological data analysis module collects reservoir inflow and rainfall data, performs time series analysis on each data point, calculates dynamic response values, extracts key dependencies from the data, and generates a hydrological dynamic dependency graph;
[0327] The graph network update module extracts node information from the hydrological dynamic dependency graph, updates the properties of nodes and edges in the graph, adjusts the weight of each node, matches the actual changes in hydrological conditions, and forms a dynamic spatial interaction model;
[0328] The risk assessment module uses a dynamic spatial interaction model to conduct real-time monitoring of water levels and water storage at multiple key nodes of the reservoir, identify risk nodes, and combine data interaction between nodes to form key risk analysis results;
[0329] The model integrated decision module evaluates the key risk analysis results, compares the risk prediction data of differentiated models, selects model outputs based on accuracy, and combines the data of all participating models to generate risk assessment results.
[0330] The above are only preferred embodiments of the present invention and are not intended to limit the present invention in other forms. Any technician familiar with the profession may use the technical contents disclosed above to change or modify them into equivalent embodiments with equivalent changes and apply them to other fields. However, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention without departing from the technical solution of the present invention still falls within the protection scope of the technical solution of the present invention.
Claims
1. A reservoir capacity dynamic assessment method, characterized in that: The following steps are involved: Based on the real-time hydrological data collected from the reservoir basin and inflow, chaos theory is used to analyze the sensitive dependence of the data. By calculating the Lyapunov exponent of each data point, nonlinear patterns and numerical changes in the data are identified to obtain the hydrological dynamic dependence diagram; The steps for obtaining the hydrological dynamic dependency graph are specifically as follows: Extract the time series from the real-time hydrological data and calculate the moving weighted average of the data using the formula: , Generate time series analysis results; in, represents the data value at time point t, represents the sequence analysis results at time point t, is the weight coefficient of the time series, The number of historical data points for reference; Chaos theory is applied to analyze the time series analysis results, using the formula: , Generate chaos analysis results; in, is the chaos analysis result at time point t, To adjust the parameters, is the standard deviation of the squared values of the time series, is the average value of the time series; Using the chaos analysis results, the nonlinear patterns and numerical changes in the data are identified by calculating the Lyapunov exponent of each data point, using the formula: , Get the hydrological dynamic dependency graph; in, is the hydrological dynamic dependency graph, is the weight of each node in the dependency graph, is the total number of time points; Input the data in the hydrological dynamic dependency graph into the graph neural network, update and adjust the information of the reservoir dam and the nodes between the river basins, analyze the interactive impact of the real-time data between the nodes, adjust the network weights and connection strategies of the nodes, and obtain a dynamic spatial interaction model; The steps for obtaining the dynamic spatial interaction model are specifically as follows: Node data is extracted from the hydrological dynamic dependency graph, the initial influence of multiple nodes is calculated, and the initial calculation of node influence is performed using the formula: Generate node influence analysis results; in, is the influence analysis result of the node, is the set of neighbor nodes of the node, is the distance between nodes i and j, is the connection weight between nodes; According to the node influence analysis results, the information of reservoir dams and river basin nodes is updated using the formula: , Generate node information update results; in, The node information update result, , To adjust the parameters, is the information exchange coefficient, Neighbor node The impact results; The node information is updated using the result to adjust the node's network weight and connection strategy using the formula: , Generate network weight adjustment results; in, is the adjusted network weight between nodes i and j, , Node and The information update results, , is the adjustment factor; According to the network weight adjustment results, the interactive impact of real-time data between nodes is analyzed, and the formula is used: , Get the dynamic spatial interaction model; in, is a dynamic spatial interaction model. The amount of real-time data interaction; The dynamic spatial interaction model is used to analyze the current reservoir water storage capacity data and real-time water level change information, and by identifying the key risk nodes in the reservoir, risk point analysis is performed to obtain risk identification results; The steps for obtaining the risk identification results are specifically as follows: The dynamic spatial interaction model is used to analyze the current reservoir water storage capacity data using the formula: , Generate reservoir storage capacity analysis results; in, The water storage capacity analysis results of the representative node, represents the node weights in the model, represents the water storage capacity of the node, represents the baseline water storage capacity shift; Based on the analysis results of the reservoir's water storage capacity and combined with real-time water level change information, the formula is adopted: , Generate node risk assessment results; in, The risk assessment result of the representative node, Represents the real-time water level of the node, represents the adjustment factor, represents the smoothing parameter to prevent division by zero errors; Based on the node risk assessment results, the key risk nodes in the reservoir are identified and the high-risk nodes are marked using the formula: , Get the key risk node marking results; in, Represents the key risk node marking results, represents a high-risk node set, Represents the marking function; Combining the reservoir storage capacity analysis results, the node risk assessment results and the key risk node marking results, the formula is adopted: , Obtain risk identification results; Among them, Z represents the risk identification result, Represents the value of the node in the key risk node marking result. Represents the total number of nodes, represents the weighted index; The risk identification results are analyzed, and the risk assessment values output by multiple models are compared using the model integration method of random forest and gradient boosting machine. The model prediction results are selected based on the model performance to establish the risk assessment results.
2. The reservoir capacity dynamic assessment method according to claim 1, characterized in that: The steps for obtaining the risk assessment results are specifically as follows: The risk identification results were analyzed and a random forest model integration method was used to make a preliminary risk prediction using the formula: , Generating Random Forest Predictions ; in, represents the number of trees in the model, The risk identification result is data points; The risk identification results are analyzed using the gradient boosting model, using the formula: Generate Gradient Boosting Machine predictions ; in, represents the number of iterations in the model, is the learning rate for each iteration; Compare the random forest prediction results with the gradient boosting machine prediction results using the formula: , Select the model prediction result with the best performance to obtain the optimal model prediction result P; Based on the prediction results of the preferred model, the formula is adopted: , Get risk assessment results ; in, Represents the total number of models referenced in the evaluation.
3. A reservoir capacity dynamic assessment system, characterized in that: According to the reservoir capacity dynamic assessment method according to any one of claims 1-2, the system comprises: The hydrological data analysis module collects reservoir inflow and rainfall data, performs time series analysis on each data point, calculates dynamic response values, extracts key dependencies from the data, and generates a hydrological dynamic dependency graph; The graph network update module extracts node information from the hydrological dynamic dependency graph, updates the properties of nodes and edges in the graph, adjusts the weight of each node, matches the actual changes in hydrological conditions, and forms a dynamic spatial interaction model; The risk assessment module uses a dynamic spatial interaction model to conduct real-time monitoring of water levels and water storage at multiple key nodes of the reservoir, identify risk nodes, and combine data interaction between nodes to form key risk analysis results; The model integrated decision module evaluates the key risk analysis results, compares the risk prediction data of differentiated models, selects model outputs based on accuracy, and combines the data of all participating models to generate risk assessment results.
Citation Information
Patent Citations
Modeling method and system based on digital twin hydraulic power plant operation and maintenance data processing
CN117494547A
Real-time metering data processing platform
CN117725537A
Yellow River dam bank monitoring alarm system and method based on network communication platform
CN117892882A
