Method and system for predicting influence of flood storage and detention area application on flood level of river section

Abstract

The invention provides a method and system for predicting the influence of storage and detention area application on the flood level of the river reach, and the method comprises the steps: generating a plurality of storage and detention area application samples according to a historical typical flood sequence or an artificial simulation flood sequence of storage and detention area application and the corresponding influence data of storage and detention area application on the flood level of the river reach; key factors influencing the application effect of the flood storage and detention area are identified and optimized through an optimal information set selection wrapper; training and verification are carried out based on an artificial neural network, and a flood storage and detention area application influence prediction model is constructed; and using the flood storage and detention area application influence prediction model to predict the flood level influence of the application of the flood storage and detention areas on the river channel of the current river section under the water regimen and work regimen conditions of the application of the flood storage and detention areas. The method is simple, convenient and easy to use, the calculation speed and precision are both considered, high-precision and rapid prediction of the influence of the application of the flood storage and detention areas on the flood level is achieved, and the method has important engineering application value.

Description

Technical Field

[0001] The present invention belongs to the technical field of water conservancy project calculation and management, and relates to a method and system for predicting the influence of the operation of flood storage and detention areas on the flood level of the river course. Background Art

[0002] The construction of basin flood control projects is an important means to resist the threat of flood disasters. Flood storage and detention areas are temporary flood storage places, an important part of the basin flood control project system, and the "bottom card" to ensure the flood control safety of the basin. When a large flood occurs in the basin, after the reservoir reasonably stores flood and the river course fully discharges flood, if the incoming flood volume still exceeds the river course discharge capacity, it is necessary to activate the flood storage and detention areas to store and divide the excess flood volume, so as to avoid the breach of the dike and the occurrence of devastating disasters. In other years, various restricted production activities and living can be carried out inside the flood storage and detention areas. Formulating a scientific flood storage and detention area operation plan is of great significance for maximizing the overall flood control safety of the basin and reducing the impact of flood diversion on social economy. Rapidly and accurately predicting the influence of the operation of flood storage and detention areas on the flood level of the river course is the basis for formulating the real-time operation plan of flood storage and detention areas.

[0003] At present, the influence of the operation of flood storage and detention areas is usually obtained through hydrodynamic model calculation. The hydrodynamic model obtains the water level and flow process of each section by generalizing the river network and flood storage and detention areas and solving the Saint-Venant equations. Although the hydrodynamic model has high calculation accuracy and a clear physical basis, it uses a large number of calculation grids and complex algorithms to describe the spatio-temporal evolution of the hydrodynamic system, has high requirements for data length, modeling paradigm, and computing power conditions, is complex to use, and has a slow calculation speed, which does not meet the requirements of real-time calculation. Compared with the process-based hydrodynamic model, data-driven models such as support vector machines and artificial neural networks obtain the correlation relationship between variables based on statistical methods and have high calculation efficiency. However, the calculation accuracy of data-driven models depends to a large extent on the sufficiency and representativeness of data. The input factors and input data volume of data-driven models are crucial for balancing the calculation accuracy and usability of the model. Since flood storage and detention areas are only activated during extremely large floods, the operation data is very limited, making it difficult for data-driven models to obtain sufficient input data volume, and reducing the calculation accuracy of data-driven models in the calculation of the influence of the operation of flood storage and detention areas. Summary of the Invention

[0004] To solve the problems in the calculation of the influence of the operation of flood storage and detention areas described in the background art, that is, the hydrodynamic model has high requirements for data length, modeling paradigm, and computing power conditions, slow calculation speed, does not meet the requirements of real-time calculation, and it is difficult for data-driven models to obtain sufficient input data volume, the present invention proposes a method and system for predicting the influence of the operation of flood storage and detention areas on the flood level of the river course in this reach.

[0005] The method of the present invention includes: Generate multiple samples of flood storage and detention area (FSDA) operations based on historical typical flood sequences or artificially simulated flood sequences for FSDA operations and the corresponding data on the impact of FSDA operations on the flood levels in this river section. Based on multiple samples of FSDA operations, identify and optimize the key factors affecting the effectiveness of FSDA operations through an optimal information set selection wrapper. Using multiple samples of FSDA operations and the key factors affecting the effectiveness of FSDA operations as input factors, and the reduction amplitude of the water levels at important control stations in each time period within the prediction period as the target factor, train and validate based on an artificial neural network to construct a prediction model for the impact of FSDA operations. Apply the prediction model for the impact of FSDA operations to predict the impact of FSDA operations on the flood levels in this river section under the hydrological and engineering conditions of each FSDA operation.

[0006] Furthermore, the samples of FSDA operations include the flood diversion operation status of the FSDA and the impact of FSDA operations. The flood diversion operation status of the FSDA is determined by the incoming water conditions and engineering operation strategies. The method for obtaining the incoming water conditions is as follows: According to the activation conditions of the FSDA, select floods with the water levels at important control stations reaching a certain magnitude from historical typical flood sequences or artificially simulated flood sequences to represent different incoming water conditions. The method for obtaining the engineering operation strategies is as follows: Obtain variables including the flood diversion timing, flood diversion flow rate, and flood diversion duration from historical typical flood sequences or artificially simulated flood sequences. Using the Latin hypercube sampling method, first consider the variation ranges of each variable separately, conduct global sampling of each variable between the upper and lower limits to initially generate the engineering operation strategies for the FSDA. Then, consider the correlation relationships between variables, and according to the storage capacity limit of the FSDA and the randomly generated flood diversion flow rate, correct the flood diversion duration in the initially generated flood diversion operation status of the FSDA to obtain the final engineering operation strategies for the FSDA. The method for obtaining the impact of FSDA operations is as follows: Use the water levels at key control stations to represent the flood levels in this river section. Under the flood diversion operation status of each FSDA, run a calibrated and verified one-dimensional hydrodynamic model to calculate the water level process at the key control stations in the case of flood diversion in the FSDA. Under each hydrological condition, run the same hydrodynamic model to calculate the water level process at the key control stations in the case of no flood diversion in the FSDA. Compare and analyze the impact of flood diversion in the FSDA on the flood levels in this river section to obtain the impact of FSDA operations.

[0007] Even further, the method for identifying and optimizing the key factors affecting the effectiveness of FSDA operations includes: Based on various flood storage and detention area application samples, the candidate factor set is constructed with the flow of each upstream control station in the river section before the flood storage and detention area is put into use, the water level of each downstream control station, the rise and fall rate, and the flood diversion flow, flood diversion days, and impact period of the flood storage and detention area as possible influencing factors. According to the data of the impact of the candidate factor set and the use of flood storage and detention areas on the flood level of the river section, the packager is selected using the optimal information set, and the key factors that affect the effect of reducing the flood level are selected from the candidate factor set.

[0008] Furthermore, the optimization objectives in the optimal information set selection wrapper include: the prediction accuracy of the factor set for the target variable f 1( S i ) Highest, factor set dimension f 2( S i ) minimum, factor set and target variable correlation f 3( S i ) Maximum, factor set redundancy f 4( S i ) is the lowest; the factor set has the lowest prediction accuracy for the target variable f 1( S i ) Maximum and factor set dimension f 2( S i ) is the main optimization goal, and the correlation between the factor set and the target variable f 3( S i ) Maximum sum factor set redundancy f 4( S i ) is the lowest secondary optimization goal; in the optimization process of the optimal information set selection wrapper, the decision variables are set to 0 and 1 variables, representing the factors in each candidate factor set X Whether it is selected, the factors whose decision variables are 1 in each non-dominated solution are the selected factors, forming the set , a number of mutually non-dominated candidate factor solutions are obtained through one-time optimization of the non-dominated sorting genetic algorithm to form a non-dominated solution set. Then, according to non-quantitative factors including whether the observation data is easy to obtain and whether the observation is reliable, an appropriate solution is selected from the non-dominated solution set as the optimal key factor affecting the effect of flood level reduction.

[0009] Furthermore, the factor set has an accuracy of predicting the target variable f 1( S i ) Use the target variable to predict the value The symmetric uncertainty SU between the measured value y represents that the factor set has the highest prediction accuracy for the target variable f 1( S i ) The highest expression is as follows: (1), (2), (3), wherein, S i is the factor set, x i is each factor in the factor set, X is the alternative factor set, ; H (‧) is the information entropy of the variable; M is the number of hidden layer nodes, represents the sigmoid activation function parameter of the randomly generated m-th hidden layer node parameter, represents the output weight to be calibrated; The dimension of the factor set f 2( S i ) that is, the number of factors | S i | in the factor set, and the dimension of the factor set f 2( S i ) The lowest expression is as follows: (4); The correlation between the factor set and the target variable f 3( S i ) is expressed by the sum of the symmetric uncertainties between each factor x i ( x i ∈ S i ) in the factor set and the target variable y . The expression with the largest correlation f 3( S i ) between the factor set and the target variable is as follows: (5), wherein, S i is the factor set, x i is each factor in the factor set, X is the alternative factor set, , y is the target variable; The redundancy of the factor set f 4( S i ) is the sum of the pairwise symmetric uncertainties of the factors in the factor set. The expression with the lowest redundancy of the factor set f 4( S i ) is as follows: (6).

[0010] Furthermore, the construction process of the flood detention area operation impact prediction model includes: Using multiple flood detention area operation samples and the key factors affecting the reduction effect of the flood level as input factors, and the reduction amplitude of the water level at important control stations in each time period within the prediction period as the target factor. According to the universal approximation theorem, determine the number of hidden layers and the form of the transfer function. Determine the number of hidden neural networks through repeated experiments. Adopt a two-layer feedforward network with the sigmoid function as the activation function of the hidden layer neurons and the linear function as the activation function of the output layer neurons to approximate the multi-dimensional mapping problem and determine the artificial neural network structure. The expression is as follows: (7), wherein, is the water level change of the control station caused by the flood diversion of the flood detention area predicted at time t + τ , , T is the prediction period, N is the number of hidden layer neurons with the sigmoid function as the activation function, is the input vector at time t , which is composed of the identified key influencing factors, are the parameters of the artificial neural network, and the number is N×(2 + D)+1, where D is the dimension of the input space, that is, the number of key influencing factors; Use the Levenberg-Marquardt training algorithm to optimize the parameters of the artificial neural network. In each training, set a training set, a smaller validation set and a test set. The training set is used to calculate the gradient and update the network; the validation set is used to determine whether the prediction accuracy stops improving and when to stop training; the test set has no impact on training and independently measures the performance of the trained network. When the error metric of the validation set is stable or the prediction error is acceptable, terminate the training iteration, determine the artificial neural network structure and the corresponding parameter values, and use it as the flood detention area operation impact prediction model.

[0011] Based on the above method, the present invention proposes a prediction system for the influence of flood detention and retarding area operation on the flood level of this river section, including a flood detention and retarding area operation sample generation module, a key factor identification and optimization module, a flood detention and retarding area operation influence prediction model construction module, and a flood detention and retarding area operation influence on flood level prediction module.

[0012] The flood detention and retarding area operation sample generation module is used to generate various flood detention and retarding area operation samples according to the historical typical flood sequence or artificial simulation flood sequence of flood detention and retarding area operation and the corresponding data of the influence of flood detention and retarding area operation on the flood level of this river section.

[0013] The key factor identification and optimization module is used to identify and optimize the key factors affecting the effect of flood detention and retarding area operation based on various flood detention and retarding area operation samples through the optimal information set selection wrapper.

[0014] The flood detention and retarding area operation influence prediction model construction module is used to take various flood detention and retarding area operation samples and the key factors affecting the effect of flood detention and retarding area operation as input factors, and take the reduction amplitude of the water level of each important control station during the prediction period as the target factor, and based on the artificial neural network for training and verification, construct a flood detention and retarding area operation influence prediction model.

[0015] The flood detention and retarding area operation influence on flood level prediction module is used to use the flood detention and retarding area operation influence prediction model to predict the influence of flood detention and retarding area operation on the flood level of this river section under the water regime and engineering conditions of each flood detention and retarding area operation.

[0016] The present invention also proposes a computer-readable storage medium, which stores a computer program, and when the computer program is executed by a processor, it realizes the prediction method and system for the influence of flood detention and retarding area operation on the flood level of this river section as described above.

[0017] Compared with the prior art, the present invention has the following advantages: (1) Combining the advantages of the hydrodynamic model and the data-driven model to achieve high-precision and fast prediction of the influence of flood level; (2) Obtaining a large amount of sample data of flood detention and retarding area operation through historical typical flood sequences or artificial simulation flood sequences, overcoming the limitation of insufficient measured data, and the constructed model is simple and easy to use; (3) Through the optimal information set selection wrapper, mining and clarifying the key factors that determine the effect of flood detention and retarding area operation, improving the prediction accuracy of the subsequent constructed model; (4) Providing an effective technical means for predicting the effect of flood detention and retarding area operation in the post-engineering period, and having high engineering application and popularization value.

[0018] In summary, the present invention solves the problems existing in the hydrodynamic model and data-driven model in the calculation of the impact of the flood detention and retention area. The present invention not only overcomes the limitation of insufficient measured data by artificially simulating a large number of sample data, but also excavates key influencing factors, constructs an efficient prediction model, which is simple, easy to use, and takes into account both the calculation speed and accuracy, realizes high-precision and fast prediction of the impact of flood levels, and has important engineering application value. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 It is a flowchart of the method of the present invention.

[0020] Figure 2 It is a selection matrix of key influencing factors for the flood diversion effect of a certain flood detention and retention area.

[0021] Figure 3 It is the correlation between the predicted value and the target value of the flood diversion effect of a certain flood detention and retention area on a certain hydrological station for different sample sets.

[0022] Figure 4 It is the cumulative distribution curve of the prediction error of the flood diversion effect of a certain flood detention and retention area on a certain hydrological station.

[0023] Figure 5 It is the first example of the change process of the water level drop of a certain hydrological station after the flood diversion of a certain flood detention and retention area calculated by different models under typical flood conditions.

[0024] Figure 6 It is the second example of the change process of the water level drop of a certain hydrological station after the flood diversion of a certain flood detention and retention area calculated by different models under typical flood conditions.

[0025] Figure 7 It is the third example of the change process of the water level drop of a certain hydrological station after the flood diversion of a certain flood detention and retention area calculated by different models under typical flood conditions.

[0026] Figure 8 It is the system architecture diagram of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0027] In order to make the technical problems, technical solutions and beneficial effects to be solved by the present application clearer, the present application will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application, and are not used to limit the present application.

[0028] Embodiment 1

[0029] The prediction method for the impact of the operation of the flood detention and retention area on the flood level of this river section, the flowchart is as Figure 1 shown, and is specifically described as follows.

[0030] First, according to the historical typical flood sequences or artificially simulated flood sequences of the flood detention and retarding areas and the corresponding data on the impact of the operation of the flood detention and retarding areas on the flood levels of the river section, multiple samples of the operation of the flood detention and retarding areas are generated.

[0031] Specifically, the samples of the operation of the flood detention and retarding areas include the flood diversion operation status of the flood detention and retarding areas and the impact of the operation of the flood detention and retarding areas.

[0032] Among them, the flood diversion operation status of the flood detention and retarding areas is determined by the incoming water conditions and the engineering operation strategy. The method for obtaining the incoming water conditions is as follows: According to the activation conditions of the flood detention and retarding areas, select floods with the water levels of important control stations reaching a certain magnitude from the historical typical flood sequences or artificially simulated flood sequences to represent different incoming water conditions. The method for obtaining the engineering operation strategy is as follows: Obtain variables including the flood diversion timing, flood diversion discharge, and flood diversion duration from the historical typical flood sequences or artificially simulated flood sequences. Using the Latin hypercube sampling method, first consider the change ranges of each variable separately, conduct global sampling of each variable between the upper and lower limits to initially generate the engineering operation strategy of the flood detention and retarding areas. Then, consider the correlation relationship between variables, and according to the storage capacity limit of the flood detention and retarding areas and the randomly generated flood diversion discharge, correct the flood diversion duration in the initially generated flood diversion operation status of the flood detention and retarding areas to obtain the final engineering operation strategy of the flood detention and retarding areas.

[0033] The method for obtaining the impact of the operation of the flood detention and retarding areas is as follows: Use the water levels of key control stations to represent the flood levels of the river section. Under the flood diversion operation status of each flood detention and retarding area, run the calibrated and verified one-dimensional hydrodynamic model to calculate the water level process of the key control stations under the condition of flood diversion in the flood detention and retarding areas. Under each hydrological condition, run the same hydrodynamic model to calculate the water level process of the key control stations under the condition of no flood diversion in the flood detention and retarding areas. Compare and analyze the impact of flood diversion in the flood detention and retarding areas on the flood levels of the river section to obtain the impact of the operation of the flood detention and retarding areas.

[0034] In this embodiment, for a certain flood detention and retarding area, under different hydrological conditions and operation strategies, a large number of samples of the impact of the operation of the flood detention and retarding area on the water level of a certain hydrological station are randomly generated. The operation state parameters and the sampling ranges of each parameter of the flood detention and retarding area are shown in Table 1. Considering the enabling conditions of the flood detention and retarding area, from the historical measured typical large floods of different magnitudes and flood compositions, the flood processes in the main flood season corresponding to the water level of another hydrological station exceeding the enabling water level are selected as the inflow boundary set to reflect the changes in hydrological conditions. The selected typical large floods include the floods in 1931, 1935, 1954, 1968, 1996, and 1998. The gate-controlled flood diversion strategy of the flood detention and retarding area is described by the enabling time, flood diversion flow rate, and flood diversion duration. The enabling time is discrete throughout the flood process. For example, in the 1935 flood, the enabling time of the flood detention and retarding area is any time point from 6:00 on July 8, 1935 to 6:00 on July 14, 1935. The discrete range of the flood diversion flow rate is from 0 to the maximum flood diversion flow rate of the flood detention and retarding area. The discrete range of the flood diversion duration is from 0 to 10 days. First, the Latin hypercube sampling method is used to perform global sampling between the upper and lower limits of each variable to generate 600 flood diversion operation states. Then, according to the storage volume limit of the flood detention and retarding area, the flood diversion duration of each sample is corrected. Under each operation state of the flood detention and retarding area, the calibrated one-dimensional hydrodynamic model is used to simulate the water level process of the hydrological station after the flood detention and retarding area diverts flood. Under the same hydrological conditions, this hydrodynamic model is used to simulate the water level process of the hydrological station without flood diversion in the flood detention and retarding area. The water level drop before and after flood diversion is compared to analyze the impact of flood diversion on the water level process of the hydrological station. According to the current effective hydrological forecasting level in the lower reaches of the Yangtze River Basin, the prediction period of this model is set to 7 days. According to the real-time scheduling step size, the prediction time step is set to 6 hours, so there are 29 sampling points within the prediction period. Combining each operation state variable and the water level drop of the hydrological station, an operation sample of the flood detention and retarding area is constructed, with a total of 17,400 simulation samples.

[0035] Table 1 Operation state parameters and change ranges of a certain flood detention and retarding area

[0036] Then, based on multiple operation samples of the flood detention and retarding area, through the optimal information set selection wrapper, the key factors affecting the operation effect of the flood detention and retarding area are identified and optimized.

[0037] Specifically, based on the hydrological factors affecting the operation effect of flood storage and detention areas, including upstream inflow characterizing flood magnitude, water level characterizing water body inertia, downstream backwater effect, and water level fluctuation rate characterizing flood stages, as well as the engineering operation strategies of flood storage and detention areas, combined with the ease of use of key factors and the complex spatial correlation relationships between stations, the flow of each upstream control station, the water level of each downstream control station, the fluctuation rate, and the flood diversion flow, flood diversion days, and influence period of the flood storage and detention area before the activation of the flood storage and detention area are used as possible influencing factors to construct an alternative factor set. According to the alternative factor set and the data on the influence of the operation of the flood storage and detention area on the flood discharge water level of this river section, the optimal information set selection wrapper is used to select the key factors that affect the reduction effect of the flood discharge water level from the alternative factor set.

[0038] The optimization objectives in the optimal information set selection wrapper include: the prediction accuracy of the factor set for the target variable f 1( S i ) being the highest, the dimension of the factor set f 2( S i ) being the lowest, the correlation between the factor set and the target variable f 3( S i ) being the largest, and the redundancy of the factor set f 4( S i ) being the lowest. The prediction accuracy of the factor set for the target variable f 1( S i ) being the highest and the dimension of the factor set f 2( S i ) being the lowest are the main optimization objectives, aiming to ensure a high prediction accuracy of the factor set for the target variable while streamlining the number of selected factors. The correlation between the factor set and the target variable f 3( S i ) being the largest and the redundancy of the factor set f 4( S i ) being the lowest are the secondary optimization objectives, which serve to expand the non-dominated solution set and provide more choices for subsequent factor screening based on expert experience.

[0039] In the optimization process of the optimal information set selection wrapper, the decision variables are set as 0-1 variables, representing whether the factors in each alternative factor set X are selected. For each non-dominated solution, all the decision variables with a value of 1 in the factors are the selected factors, which constitute a set and , multiple non-dominated alternative factor solutions are obtained through one optimization by the non-dominated sorting genetic algorithm, forming a non-dominated solution set. Then, according to non-quantitative factors including whether the observed data is easy to obtain and whether the observation is reliable, an appropriate solution is selected from the non-dominated solution set as the key factor that optimally affects the reduction effect of the flood level.

[0040] More specifically, the factor set for the prediction accuracy of the target variable f 1( S i ) is represented by the symmetric uncertainty SU between the predicted value of the target variable and the measured value y. The greater the prediction accuracy, the stronger the predictability of the factor set for the target variable. SU is an important concept in information theory and is calculated from the information entropy H(‧) of the variable. SU equal to 0 represents that the two variables are completely uncorrelated, and SU equal to 1 represents that the two variables are completely correlated. The factor set for the prediction accuracy of the target variable f 1( S i ) with the highest value is expressed as follows: (1), (2), (3), In the formula, S i is the factor set, x i is each factor in the factor set. Formula (3) uses each selected factor S i in the factor set as the input, and uses the extreme learning machine model to predict the target variable prediction , M is the number of hidden layer nodes, X is the set of alternative variables, represents the sigmoid activation function parameter of the randomly generated m-th hidden layer node parameter, represents the output weight to be calibrated.

[0041] The dimension of the factor set f 2( S i ) that is, the number of factors | S i | in the factor set. The fewer the number of factors in the factor set, the easier it is to use. The dimension of the factor set f 2( S i ) with the lowest value is expressed as follows: (4).

[0042] Correlation between Factor Set and Target Variable f 3( S i ) Use each factor in the factor set x i ( x i ∈ S i ) and the target variable y The sum of symmetric uncertainties is expressed. The greater the correlation between the factor set and the target variable, the more the change of the target variable can be reflected. The correlation between the factor set and the target variable f 3( S i ) The maximum expression is as follows: (5), In the formula, S i is the factor set, x i is each factor in the factor set, X is the alternative factor set, , y is the target variable.

[0043] Factor Set Redundancy f 4( S i ) is the sum of pairwise symmetric uncertainties of each factor in the factor set. The smaller the factor set redundancy, the smaller the similarity of each variable in the factor set and the stronger the diversity of the selected factors. The factor set redundancy f 4( S i ) The lowest expression is as follows: (6).

[0044] In this embodiment, the upstream inflow characterizing the flood magnitude, the water level characterizing the water body inertia, the water level rise and fall rate characterizing the flood stage, and the downstream water level characterizing the backwater effect are comprehensively characterized. Considering the complex spatial correlation relationship between stations, an alternative factor set is constructed, as shown in Table 2. Among them, the upstream inflow is the observation value k time intervals earlier than the flood diversion, kis the flood propagation time from this station to a certain hydrological station. Other variables take the observed values at the moment of flood diversion. The water level drop within the forecast period of a certain hydrological station is used as the target variable. The wrapper algorithm of optimal information set selection is adopted to analyze the key influencing factors of the flood diversion effect in a certain flood detention and storage area in the alternative factor set. The non-dominated sorting genetic algorithm is used as the multi-objective evolutionary algorithm for optimization and solution, and the number of target evaluations is set to 5000 in this algorithm. In the extreme learning machine algorithm, the number of hidden layer neurons with the sigmoid function as the activation function is set to 50, and 10-fold cross-validation is carried out. Considering the randomness of the algorithm, under each input factor, the prediction is repeated 5 times, and the average value of each prediction result is used to estimate the prediction accuracy of each model.

[0045] Table 2 Alternative factor sets analyzed by the wrapper algorithm of optimal information set selection

[0046] Weighing the number of optimal factor sets and the prediction accuracy of the factor sets, select the factor sets with a prediction accuracy within 60% of the highest accuracy. Each alternative subset is composed as Figure 2 shown in the "selection matrix". Each row represents a factor set, and the identified key influencing factors are marked with gray rectangles. The factor sets are arranged in ascending order of dimension from top to bottom. For each dimension, the factor sets are arranged in ascending order of prediction accuracy. The marking color changes with the dimension of the subset. The lighter the gray, the smaller the dimension of the subset. In this example, the dimension of the factor sets is from 2 to 5. The prediction accuracy of each factor set is marked on the right, and the highest prediction accuracy is marked in red.

[0047] The vertical bar connecting multiple rows of gray markings reflects the correlation of the prediction factors with the target variable. The flood diversion strategy variables of the flood detention and storage area, namely the flood diversion flow rate and the flood diversion duration, are included in each subset, forming an uninterrupted vertical bar. The time variable is included in each subset with a prediction accuracy higher than 0.6. This shows that these variables cannot be replaced by other input factors, otherwise it will lead to a significant reduction in the prediction accuracy. These variables are strong correlation factors affecting the flood diversion effect of the flood detention and storage area. Factor set 7 has the highest prediction accuracy and is composed of the water level of a certain hydrological station No. 1 downstream, the water level of a certain hydrological station, the flood diversion flow rate, the flood diversion duration, and the time serial number. Compared with factor set 7, the accuracy of factor set 6 decreases by less than 2%. Therefore, the water level of a certain hydrological station, a more critical control station, is used to replace the water level of a certain hydrological station No. 1 downstream. The guarantee of obtaining the observed values of factor set 6 is improved, and factor set 6 is the selected factor set.

[0048] Next, using various flood detention and storage area operation samples and the key factors affecting the effect of reducing the flood level as input factors, and the amplitude of the water level reduction of each important control station during each period within the forecast period as the target factor, train and verify based on the artificial neural network to construct a prediction model for the impact of flood detention and storage area operation.

[0049] Specifically, using a variety of flood storage and detention area operation samples and the key factors affecting the reduction effect of flood levels as input factors, and the reduction amplitude of the water levels at important control stations in each time period within the prediction period as the target factor. To minimize the computational amount, the smallest structure with acceptable errors is adopted. According to the universal approximation theorem, the number of hidden layers and the form of the transfer function are determined. The number of hidden neural networks is determined through repeated experiments. A two-layer feedforward network with the sigmoid function as the activation function of the hidden layer neurons and the linear function as the activation function of the output layer neurons is used to approximate the multi-dimensional mapping problem, and the artificial neural network structure is determined. The expression is as follows: (7), In the formula, is the water level change at the control station caused by the flood diversion of the flood storage and detention area predicted at time t + τ The prediction of the water level change at the control station caused by the flood diversion of the flood storage and detention area, , T is the prediction period, N is the number of hidden layer neurons with the sigmoid function as the activation function, is the input vector at time t , which is composed of the identified key influencing factors, is the artificial neural network parameter, and the number is N×(2 + D)+1, where D is the dimension of the input space, that is, the number of key influencing factors.

[0050] The Levenberg-Marquardt training algorithm is used to optimize the artificial neural network parameters. In each training, a training set, a smaller validation set, and a test set are set. The training set is used to calculate the gradient and update the network; the validation set is used to determine whether the prediction accuracy stops improving and when to stop training; the test set has no influence on the training and independently measures the performance of the trained network. To avoid overfitting and save training time, when the error metric of the validation set is stable or the prediction error is acceptable, the training iteration is terminated, and the artificial neural network structure and the corresponding parameter values are determined, which are used as the prediction model for the impact of flood storage and detention area operation.

[0051] In this embodiment, 17,400 operation samples of a certain flood storage and detention area are randomly divided into training (70%), validation (15%), and test (15%) subsets. When the number of hidden neurons exceeds 10, the marginal benefit of increasing the number of hidden neurons shows a decreasing trend, and the number of hidden neurons is set to 10.

[0052] The correlation coefficients between the predicted values and the target values of the training, validation, and test sets are as Figure 3 shown. The correlation coefficients of each subset are all higher than 0.99, indicating good prediction performance and no overfitting. The cumulative distribution curve of the absolute value of the prediction error is as Figure 4As shown. The results show that the probability that the prediction error is less than 3 cm is 90%.

[0053] Taking the influence process of flood diversion in a flood detention and retarding area on the water level of a hydrological station under typical flood conditions, and further comparing with the MIKE model to verify the prediction accuracy of the prediction model for the influence of the operation of the flood detention and retarding area established by the present invention. The floods in 1954 and 1998 with basin-wide floods and the regional flood in 1935 are selected for inspection. When the water level forecast of a hydrological station reaches 34.4 m, the flood detention and retarding area is activated. To test the accuracy of the prediction results of the prediction model for the influence of the operation of the flood detention and retarding area established under any flood diversion conditions, during the flood process when the water level of another hydrological station exceeds 34.4 m, the flood diversion timing is randomly selected, and the flood diversion flow rate and flood diversion duration are randomly generated to construct a flood diversion scenario. The change process of the water level drop of a hydrological station after flood diversion in a flood detention and retarding area calculated by different models in each scenario is as Figure 5 , Figure 6 and Figure 7 shown, where Figure 5 is: On July 27, 1954, a flood detention and retarding area diverted flood at a flow rate of 5320 m³ / s for 3.15 d, Figure 6 is: On July 25, 1998, a flood detention and retarding area diverted flood at a flow rate of 5480 m³ / s for 3.95 d; Figure 7 is: On July 13, 1935, a flood detention and retarding area diverted flood at a flow rate of 3160 m³ / s for 3.95 d. In the main coordinate axes, the black solid line marks the change process of the water level drop of a hydrological station simulated by the MIKE model in the sample, and the red solid line marks the change process of the water level drop of a hydrological station predicted by the established model. In the secondary coordinate axes, the gray bar chart marks the forecast error in each time period. As shown in the figure, under the three typical scenarios, the change trend of the water level drop process of a hydrological station simulated by the prediction model for the influence of the operation of the flood detention and retarding area is consistent with that of the MIKE model. Within the 7-day prediction period, under the given flood diversion scenario in 1954, the forecast error in each time period does not exceed 3.5 cm, under the given flood diversion scenario in 1998, the forecast error in each time period does not exceed 3 cm, and under the given flood diversion scenario in 1935, the forecast error in each time period does not exceed 1 cm. Generally, the established prediction model for the influence of the operation of the flood detention and retarding area is relatively accurate in predicting the process of the water level drop of a hydrological station after flood diversion in a flood detention and retarding area.

[0054] Using the traditional one-dimensional hydrodynamic MIKE model for 7-day simulation to obtain the influence of the operation of a flood detention and retarding area on the water level of a hydrological station, the calculation time is about 1.5 min, while the calculation time of the established prediction model for the influence of the operation of the flood detention and retarding area is within 1 second, which greatly reduces the calculation cost while maintaining the accuracy.

[0055] Finally, using the prediction model for the influence of the operation of the flood detention and retarding area, under the water regime and engineering conditions of the operation of each flood detention and retarding area, the influence of the flood passing water level of this river section is predicted.

[0056] Example 2

[0057] The prediction system for the influence of the operation of flood storage and detention areas on the flood level of this river section, the architecture diagram is as Figure 8 shown, and it is composed of a flood storage and detention area operation sample generation module, a key factor identification and optimization module, a flood storage and detention area operation influence prediction model construction module, and a flood storage and detention area operation influence on flood level prediction module.

[0058] The flood storage and detention area operation sample generation module is used to generate various flood storage and detention area operation samples according to the historical typical flood sequence or artificial simulation flood sequence of the flood storage and detention area operation and the corresponding data of the influence of the flood storage and detention area operation on the flood level of this river section.

[0059] The key factor identification and optimization module is used to identify and optimize the key factors affecting the operation effect of the flood storage and detention area by selecting a wrapper through the optimal information set based on various flood storage and detention area operation samples.

[0060] The flood storage and detention area operation influence prediction model construction module is used to take various flood storage and detention area operation samples and the key factors affecting the reduction effect of the flood level as input factors, and take the reduction amplitude of the water level of each important control station in the foresight period as the target factor, and train and verify based on the artificial neural network to construct a flood storage and detention area operation influence prediction model.

[0061] The flood storage and detention area operation influence on flood level prediction module is used to use the flood storage and detention area operation influence prediction model to predict the influence of the flood storage and detention area operation on the flood level of this river section under the water regime and engineering conditions of each flood storage and detention area operation.

[0062] The specific implementation methods of each module in this system are the same as those described in Embodiment 1, and will not be elaborated here.

[0063] Embodiment 3

[0064] A computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, it implements the method for predicting the influence of the operation of the flood storage and detention area on the flood level of this river section as described in Embodiment 1 above, and the prediction system for the influence of the operation of the flood storage and detention area on the flood level of this river section as described in Embodiment 2.

[0065] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a system, or a computer program product. Therefore, the present application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code. The solutions in the embodiments of the present application can be implemented in various computer languages. For example, object-oriented programming languages such as Java, C++, Python, and the interpreted scripting language JavaScript, etc.

[0066] The present application is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to the embodiments of the present application. It should be understood that each flow and / or block in the flowchart and / or block diagram, as well as the combination of flows and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, such that the instructions executed by the processor of the computer or other programmable data processing devices generate means for implementing the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 or multiple blocks.

[0067] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, such that the instructions stored in the computer-readable memory generate a manufactured article including instruction means that implement the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 or multiple blocks.

[0068] These computer program instructions can also be loaded onto a computer or other programmable data processing device, such that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process, so that the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 or multiple blocks.

[0069] Although the preferred embodiments of the present application have been described, those skilled in the art can make additional changes and modifications once they learn the basic creative concepts. Therefore, the appended claims are intended to be construed to include the preferred embodiments as well as all changes and modifications falling within the scope of the present application.

[0070] Obviously, those skilled in the art can make various changes and modifications to this application without departing from the spirit and scope of this application. Thus, if these modifications and variations of this application fall within the scope of the claims of this application and their equivalent technologies, this application is also intended to include these changes and modifications.

Claims

1. Prediction method for the influence of the operation of flood storage and detention areas on the flood level of the river reach, characterized in that, Including: Generating a variety of flood detention area operation samples based on the historical typical flood sequences or artificially simulated flood sequences of the flood detention area operation and the corresponding data on the impact of the flood detention area operation on the flood level of this river section; Based on a variety of flood detention area operation samples, identifying and optimizing the key factors affecting the flood detention area operation effect through an optimal information set selection wrapper; Taking a variety of flood detention area operation samples and the key factors affecting the flood detention area operation effect as input factors, and taking the reduction amplitude of the water levels of important control stations in each time period within the prediction period as the target factor, training and validating based on an artificial neural network, and constructing a prediction model for the impact of flood detention area operation; Using the prediction model for the impact of flood detention area operation to predict the impact of flood detention area operation on the flood level of this river section under the water regime and engineering regime conditions of each flood detention area operation.

2. The method for predicting the influence of the operation of the flood detention and retention area on the flood level of the river section according to claim 1, wherein: The flood detention area operation sample includes the flood diversion operation state of the flood detention area and the impact of the flood detention area operation; The flood diversion operation state of the flood detention area is determined by the incoming water condition and the engineering operation strategy; the method for obtaining the incoming water condition is: according to the activation condition of the flood detention area, selecting floods with the water levels of important control stations reaching a certain magnitude from the historical typical flood sequences or artificially simulated flood sequences to represent different incoming water conditions; the method for obtaining the engineering operation strategy is: obtaining variables including the flood diversion timing, flood diversion discharge, and flood diversion duration from the historical typical flood sequences or artificially simulated flood sequences, using the Latin hypercube sampling method, first considering the change range of each variable separately, globally sampling each variable between the upper and lower limits to initially generate the engineering operation strategy of the flood detention area, and then considering the correlation relationship between variables, and correcting the flood diversion duration in the initially generated flood diversion operation state of the flood detention area according to the flood storage volume limit of the flood detention area and the randomly generated flood diversion discharge to obtain the final engineering operation strategy of the flood detention area; The method for obtaining the impact of the flood detention area operation is: using the water levels of key control stations to represent the flood levels of this river section, running a calibrated and verified one-dimensional hydrodynamic model under the flood diversion operation state of each flood detention area to calculate the water level process of the key control stations under the condition of flood diversion in the flood detention area, running the same hydrodynamic model under each hydrological condition to calculate the water level process of the key control stations under the condition of no flood diversion in the flood detention area, and comparing and analyzing the impact of flood diversion in the flood detention area on the flood levels of this river section to obtain the impact of the flood detention area operation.

3. The method for predicting the influence of the operation of the flood detention and retention area on the flood level of the river reach according to claim 2, wherein: The method for identifying and optimizing the key factors affecting the flood detention area operation effect includes: Based on a variety of flood detention area operation samples, constructing an alternative factor set with the flow rates of each upstream control station, the water levels of each downstream control station, the rise and fall rates, and the flood diversion discharge, flood diversion days, and impact periods of the flood detention area in the time period before the activation of the flood detention area as possible influencing factors; According to the alternative factor set and the data on the impact of the flood detention area operation on the flood levels of this river section, using the optimal information set selection wrapper to optimize the key factors affecting the flood detention area operation effect from the alternative factor set.

4. The method for predicting the influence of the operation of the flood detention and retention area on the flood level of the river reach according to claim 3, characterized in that: The optimization objectives in the optimal information set selection wrapper include: the prediction accuracy of the factor set for the target variable f 1( S i ) Highest, factor set dimension f 2( S i ) minimum, factor set and target variable correlation f 3( S i ) Maximum, factor set redundancy f 4( S i ) is the lowest; the factor set has the lowest prediction accuracy for the target variable f 1( S i ) Maximum and factor set dimension f 2( S i ) is the main optimization goal, and the correlation between the factor set and the target variable f 3( S i ) Maximum sum factor set redundancy f 4( S i ) is the lowest secondary optimization goal; in the optimization process of the optimal information set selection wrapper, the decision variables are set to 0 and 1 variables, representing the factors in each candidate factor set X Whether it is selected, the factors whose decision variables are 1 in each non-dominated solution are the selected factors, forming the set , a number of mutually non-dominated candidate factor solutions are obtained through one-time optimization of the non-dominated sorting genetic algorithm to form a non-dominated solution set. Then, according to non-quantitative factors including whether the observation data is easy to obtain and whether the observation is reliable, an appropriate solution is selected from the non-dominated solution set as the optimal key factor affecting the effect of flood level reduction.

5. The method for predicting the influence of the operation of the flood storage and detention area on the flood level of the river reach according to claim 4, wherein: The prediction accuracy of the target variable for the factor set f 1( S i ) is represented by the symmetric uncertainty SU between the predicted value of the target variable and the measured value y. The highest expression for the prediction accuracy of the target variable for the factor set f 1( S i ) is as follows: （1）， （2）， （3）， In the formula, S i is the factor set, x i is each factor in the factor set, X is the alternative factor set, ; H (‧) is the information entropy of the variable; M is the number of hidden layer nodes, represents the sigmoid activation function parameter of the m-th randomly generated hidden layer node, parameter, represents the output weight to be calibrated; The dimension of the factor set f 2( S i ) which is the number of factors in the factor set| S i |, the dimension of the factor set f 2( S i ) The lowest expression is as follows: （4）； The correlation between the factor set and the target variable f 3( S i ) Use each factor in the factor set x i ( x i ∈ S i ) and the target variable y The correlation between the factor set and the target variable is expressed as the sum of symmetric uncertainties. The expression for the maximum of the correlation between the factor set and the target variable f 3( S i ) is as follows: （5）， In the formula, S i is the factor set, x i is each factor in the factor set, X is the alternative factor set, , y is the target variable; The redundancy of the factor set f 4( S i ) is the sum of the pairwise symmetric uncertainties of the factors in the factor set. The expression with the lowest redundancy f 4( S i ) is as follows: （6）。 6. The method for predicting the influence of the operation of the flood storage and detention area on the flood level of the river reach according to claim 1, wherein: The construction process of the prediction model for the impact of flood detention area operation includes: Using a variety of flood storage and detention area operation samples and the key factors affecting the reduction effect of flood levels as input factors, and taking the reduction amplitude of the water levels at important control stations in each period within the prediction period as the target factor, determine the number of hidden layers and the form of the transfer function according to the general approximation theorem, determine the number of hidden neural networks through repeated experiments, and adopt a two-layer feedforward network with the sigmoid function as the activation function of the hidden layer neurons and the linear function as the activation function of the output layer neurons to approximate the multi-dimensional mapping problem, and determine the artificial neural network structure. The expression is as follows: （7）， In the formula, is the time t + τ the water level change of the control station caused by the flood diversion of the flood storage and detention area predicted, , T is the forecast period, N is the number of neurons in the hidden layer with the sigmoid function as the activation function, is the input vector at time t composed of the identified key influencing factors, is the artificial neural network parameter, with the number of N×(2 + D)+1, where D is the dimension of the input space, that is, the number of key influencing factors; Use the Levenberg-Marquardt training algorithm to optimize the parameters of the artificial neural network. In each training, set a training set, a smaller validation set, and a test set. The training set is used to calculate the gradient and update the network; the validation set is used to determine whether the prediction accuracy stops improving and when to stop training; the test set has no influence on training and independently measures the performance of the trained network. When the error metric of the validation set is stable or the prediction error is acceptable, terminate the training iteration, determine the artificial neural network structure and the corresponding parameter values, and use it as the prediction model for the impact of flood storage and detention area operations.

7. A prediction system for the impact of flood storage and detention area operation on the flood stage of this river section, which implements the method described in any one of claims 1-6 above, is characterized in that: It includes a flood storage and detention area operation sample generation module, a key factor identification and optimization module, a flood storage and detention area operation impact prediction model construction module, and a flood storage and detention area operation impact on flood level prediction module; The flood storage and detention area operation sample generation module is used to generate a variety of flood storage and detention area operation samples according to the historical typical flood sequences or artificially simulated flood sequences of flood storage and detention area operations and the corresponding data on the impact of flood storage and detention area operations on the flood levels of the river section. The key factor identification and optimization module is used to identify and optimize the key factors affecting the flood storage and detention area operation effect based on a variety of flood storage and detention area operation samples through the optimal information set selection wrapper. The flood storage and detention area operation impact prediction model construction module is used to take a variety of flood storage and detention area operation samples and the key factors affecting the flood storage and detention area operation effect as input factors, and take the reduction amplitude of the water levels at important control stations in each period within the prediction period as the target factor, and conduct training and verification based on the artificial neural network to construct a flood storage and detention area operation impact prediction model. The flood storage and detention area operation impact on flood level prediction module is used to use the flood storage and detention area operation impact prediction model to predict the impact of flood storage and detention area operations on the flood levels of the river section under the water conditions and engineering conditions of each flood storage and detention area operation.

8. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, it implements the method for predicting the impact of flood storage and detention area operations on the flood levels of the river section as described in any one of claims 1-6.

Images