Maskinggen river flood forecasting method based on flood similarity analysis and electronic equipment

Abstract

The invention discloses a Muskinggen river flood forecasting method based on flood similarity analysis, and belongs to the technical field of hydrological forecasting. The method comprises the following steps: constructing a historical flood database; obtaining an upstream section forecast inflow process of the target flood and actual outflow at a downstream section forecast starting moment; selecting a multi-dimensional flood characteristic index, determining a weight by adopting a hybrid cross weighting method, calculating a comprehensive similarity between a target flood and each historical flood by utilizing an unequal weight distance coefficient method, and screening similar historical floods meeting conditions according to the comprehensive similarity; taking the minimum mean square error of the downstream observation and calculated flow of the historical flood as a target, dynamically calibrating the parameter flow calculation coefficients C, C and C of the Muskinggen model by adopting an improved particle swarm optimization algorithm, and further solving a storage capacity constant K and a flow proportion coefficient x; and finally, performing downstream flood process forecasting of the target flood by adopting the optimized parameters. According to the method, forecasting errors caused by fixed parameters of a traditional method are overcome, and the flood forecasting precision and reliability are remarkably improved.

Description

Technical Field

[0001] This invention relates to the field of river flood forecasting and water conservancy information utilization, and in particular to the Muskingen River flood forecasting method based on flood similarity analysis, and electronic equipment for implementing the method. Background Technology

[0002] River flood simulation is a core component of a basin flood forecasting and flood control decision support system. Its purpose is to accurately and quickly predict the flood process at downstream sections based on known flood processes at upstream sections, providing a scientific basis for key decisions such as reservoir operation, dike defense, and personnel evacuation.

[0003] Among numerous flood forecasting methods, the Muskingan method has become one of the most widely used river flood forecasting methods both domestically and internationally due to its clear physical concepts, simple calculations, low data requirements, and good accuracy under certain conditions. The main parameters of the Muskingan model, the storage constant K and the flow-to-weight ratio x, both have physical meanings and are related to the characteristics of the flood; therefore, the model parameters differ for each flood. Currently, the conventional practice in Muskingan river flood forecasting is to directly use the model parameters estimated from historical floods for current river flood forecasting, employing uniform model parameters for all floods. This approach is clearly unreasonable. Summary of the Invention

[0004] To address the issues of computational complexity and low prediction accuracy caused by using uniform model parameters for all floods in existing river flood forecasting methods, this proposal specifically suggests a Muskingan river flood forecasting method based on flood similarity analysis.

[0005] The technical solution is a flood forecasting method for the Muskingan River channel based on flood similarity analysis, comprising the following steps:

[0006] S101: Construct a historical database of historical floods, which stores upstream cross-sectional flood process data and corresponding downstream cross-sectional flood process data of multiple historical floods;

[0007] S102: Obtain the target flood The upstream cross-section forecast flood process data and the downstream cross-section forecast start time actual outflow data;

[0008] S103: Construct a flood similarity index system based on the multi-dimensional characteristics of floods;

[0009] S104: Extract target flood The flood multidimensional characteristic values ​​are standardized, and the weights of each characteristic index are determined by a hybrid cross-weighting method.

[0010] S105: Based on the standardized characteristic indicators and determined weights, the target flood is calculated using the non-equivalent distance coefficient method. Overall similarity to various historical floods;

[0011] S106: Based on the comprehensive similarity value, select floods from the historical flood database that are similar to the target flood. The most similar historical flood ;

[0012] S107: Based on historical floods The optimization objective is to minimize the mean square error between the observed flow at the downstream section and the flow at the downstream section calculated based on the flood process at the upstream section. An improved particle swarm optimization algorithm is used to optimize and estimate the flow calculation coefficients C0, C1, and C2 in the Muskingen flow calculation formula.

[0013] S108: Based on the flow calculation coefficients C0, C1, and C2 obtained from the optimization estimation, determine the parameter storage constant K and flow proportion coefficient x of the Muskingen model;

[0014] S109: Substitute the parameters K and x into the Muskingen channel flood calculation model to calculate the target flood. Evolutionary calculations are performed to predict the target flood. The downstream section of the flood process.

[0015] In the above or some embodiments, the multi-dimensional flood characteristics mentioned in step S103 include peak flow, total flood volume, flood duration, and peak shape coefficient.

[0016] In the above or some embodiments, after standardizing each flood characteristic index, the coefficient of variation of the flood characteristic index is calculated based on the standard deviation and the average value of the flood characteristic index; the weight of each flood characteristic index is calculated using a hybrid cross-weighting method based on subjective perception and objective conditions. .

[0017] In the above or some embodiments, the comprehensive similarity in step S105 is: Obtain the target flood after standardization. The characteristic index values ​​X i,f_t and historical floods The characteristic index values ​​X i,fh , where i = 1, 2, ..., n, and n is the total number of flood characteristic indicators; With historical floods Non-equitable Euclidean distance between Calculate the target flood With historical floods Overall similarity between , 1- When the similarity exceeds If the threshold is exceeded, the historical flood is determined to be a similar flood.

[0018] In the above or some embodiments, the improved particle swarm optimization algorithm described in step S107 is improved by employing a linear decreasing strategy for each step. and Perform iterative calculations, using a linear incremental strategy. Perform iterative calculations.

[0019] In the above or some embodiments, the optimization objective function in step S107 is: Where MSE is the mean squared error, N is the number of data points, and Q is the mean squared error. o (t) represents historical floods The flow rate Q is observed at the downstream section at time t. c (t) represents the downstream cross-sectional flow rate at time t, calculated using the Muskingen formula and the upstream cross-sectional flow rate.

[0020] In the above or some embodiments, the improved particle swarm optimization algorithm described in step S107, wherein the particle velocity and position updates and parameter adaptive adjustments are performed according to the following steps:

[0021] Initialize the particle swarm, with the position vector X of each particle e. e This represents a candidate solution for a set of flow calculation coefficients (C0, C1, C2), and its velocity vector V is initialized. e ;

[0022] Define the fitness function as historical floods The mean square error (MSE) between the observed flow rate and the calculated flow rate at the downstream cross section is calculated, and the fitness value of each particle is calculated.

[0023]

[0024]

[0025]

[0026] In each iteration t, the inertia weight ω, individual cognitive factor c1, and social cognitive factor c2 are dynamically adjusted according to the following formula:

[0027] Where T is the preset maximum number of iterations. , and The range of values ​​is ;

[0028] Update the velocity V of each particle e according to the following formula. e (t+1) and position X e (t+1)

[0029] ,

[0030] Where Pe(t) is the individual historical best position of particle e, Pg(t) is the global historical best position of the particle swarm, and rand1 and rand2 are uniform random numbers in the interval [0,1].

[0031] Based on the updated particle positions, recalculate the fitness value of each particle, and update the individual historical best position Pe(t+1) and the global historical best position Pg(t+1).

[0032] Repeat the steps until the maximum number of iterations T is reached or the preset convergence accuracy is met. Finally, output the global historical best position Pg as the optimized estimate of the flow calculation coefficients (C0, C1, C2).

[0033] To address the issues of computational complexity and low prediction accuracy caused by using uniform model parameters for all floods in existing river flood forecasting methods, this solution proposes an electronic device for implementing the Muskingen River Flood Forecasting Method based on flood similarity analysis.

[0034] The method includes a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor, when executing the program, implements the steps of the method as described in any one of claims 1 to 5.

[0035] Compared with existing technologies, the Muskingan River flood forecasting method based on flood similarity analysis and improved particle swarm optimization provided by this invention has the following significant advantages and beneficial effects:

[0036] Significantly Improved Forecast Accuracy: This invention overcomes the inherent limitation of fixed parameters in the traditional Muskingan method by introducing flood similarity analysis. By assigning mixed cross-weights to the non-equilibrium distance coefficient method, it can intelligently select historical floods from historical flood reservoirs that are most similar in characteristics to the current target flood, and dynamically calibrate the model parameters accordingly. This method makes the parameter combination (K,x) no longer globally optimal, but rather a "local optimum" for the current flood pattern, greatly enhancing the model's adaptability to specific floods and improving forecast accuracy, effectively reducing forecast errors for abnormal floods and floods with significant characteristic differences.

[0037] Enhanced Efficiency and Reliability of Parameter Optimization: This invention employs an improved Particle Swarm Optimization (PSO) algorithm for parameter calibration, addressing the issues of standard PSO algorithms easily getting trapped in local optima and exhibiting slow convergence speed in the later stages of iteration. By adopting a linear decreasing strategy for the inertia weight ω and the individual cognitive factor c1, and a linear increasing strategy for the social cognitive factor c2, a dynamic balance between the algorithm's global exploration capability and local exploitation capability is achieved. This strategy enables the algorithm to extensively search the solution space in the early stages of iteration and quickly converge to a high-precision solution in the later stages, thus finding the optimal parameters that minimize simulation errors more efficiently and reliably, avoiding the blindness and instability of traditional trial-and-error methods or simple optimization algorithms.

[0038] The method exhibits enhanced adaptability and robustness: The overall scheme of this invention possesses a high degree of adaptability. Its core lies in "adapting to different floods"—different target floods are automatically matched with different similar historical floods, and different optimal parameters are used for calculation accordingly. This design enables the invention to better handle complex flood events of different sources, seasons, and magnitudes, demonstrating stronger robustness (stability) to changes in river characteristics and the diversity of flood processes, and reducing the risk of forecast failure due to the use of unsuitable parameters.

[0039] Finally, this invention achieves high accuracy without imposing a significant computational burden. The improved PSO algorithm enhances convergence speed, and the similarity analysis stage quickly identifies the most relevant calibration data, avoiding the need for optimization calculations on the entire historical database each time. The entire method has low data requirements, requiring only conventional hydrological observation data, and is computationally fast, meeting the timeliness requirements of real-time flood forecasting and possessing significant engineering application value.

[0040] In summary, this invention effectively solves the problem of fixed parameters in the traditional Musking method by organically integrating intelligent matching of flood similarity with an improved intelligent optimization algorithm. It has achieved significant progress in forecast accuracy, reliability, adaptability, and efficiency, providing a more advanced and reliable technical means for river flood forecasting. Attached Figure Description

[0041] Figure 1 This is a flowchart of the method of the present invention.

[0042] Figure 2 This is a schematic diagram illustrating the principle of the model of the present invention. Detailed Implementation

[0043] Embodiments of the present invention are described in detail below, examples of which are illustrated in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain the present invention, and should not be construed as limiting the present invention.

[0044] In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0045] Example:

[0046] The present invention provides a flood forecasting method for the Muskingan River channel based on flood similarity analysis, the specific implementation steps of which are as follows:

[0047] S101: Constructing a historical flood database

[0048] Collect and organize complete flood event data for m historical events (e.g., 30 events) along the river section, including the flood hydrograph I(t) at upstream station B and the synchronous flood hydrograph Q(t) at downstream station C. Store all data in a database.

[0049] S102: Acquire target flood data

[0050] Real-time acquisition of the target flood that needs to be forecasted The flow process data (I1, I2, ..., Iᵢ) of upstream station B from the start of the rise to the present.

[0051] S103: Extract and standardize flood characteristic indicators

[0052] From historical databases of every flood and target flood From this, four flood characteristic indicators are extracted: peak flow rate Q. m The total flood volume W, flood duration T, and peak shape coefficient α (α = peak flow / average flow) constitute a 4-dimensional flood characteristic vector.

[0053] The maximum standardization method is used to standardize all feature indices. For any feature index i, the standardization formula is:

[0054]

[0055] in, This is the original value of the i-th feature of the j-th flood. It is the standardized value, where i takes all values ​​between 1 and n.

[0056] The steps for determining the weighting coefficients of flood characteristic indicators are as follows:

[0057] To reasonably reflect the importance of different characteristic indicators in flood similarity discrimination, the coefficient of variation-G1 hybrid cross-weighting method is used to determine the weight of each indicator. Weights of each indicator The determination process is as follows:

[0058] First, experts will be appointed based on their professional expertise to assess the peak flow rate Q. m The importance of the following indicators is ranked: total flood volume W, flood duration T, and peak shape coefficient α. For example, peak flow Q m The total flood volume W, flood duration T, and peak shape coefficient α.

[0059] Calculate the coefficient of variation weights of rating indicators :

[0060] ,( )

[0061]

[0062] In the formula, For the first The standard deviation of each flood characteristic indicator For the first The coefficient of variation of each flood characteristic index For the first The average value of each flood characteristic index.

[0063] Using the coefficient of variation to determine the characteristic indicators of adjacent floods and The ratio of importance :

[0064]

[0065] Step 4: Based on the characteristic indicators of adjacent floods and The ratio of importance We can obtain the hybrid cross weight of the i-th index under the criterion layer for that criterion layer based on the coefficient of variation-G1 method:

[0066]

[0067] Will Substitution ,( The weights of each evaluation indicator can be obtained. The weighting of the (i-1)th evaluation index in the criterion layer for this criterion layer is a hybrid cross-weighting method based on the coefficient of variation-G1 method. To calculate the importance ratio; The weight is the weight of the i-th indicator in the criterion layer for the mixed cross-weighting of the criterion layer.

[0068] S104: Calculate weights and flood similarity

[0069] Based on the determined weight coefficients of the four characteristic indicators The weights and flood similarity are calculated. In this embodiment, the weights are assumed to be: =[0.4,0.3,0.2,0.1].

[0070] Calculate the target flood using the non-equitable distance coefficient method. With every historical flood Euclidean distance between and similarity The calculation formula is as follows:

[0071] ;

[0072] ;

[0073] 1- ;

[0074] Historical floods with similarity values ​​exceeding a set threshold were calculated. The threshold setting can be based on empirical verification and summarization, with 0.7-0.9 being the preferred choice; of course, it is also possible to filter out values ​​related to the target flood. Similarity The highest historical flood .

[0075] S105: Calibration of Musking Parameters Using an Improved PSO Algorithm

[0076] Based on the most similar historical floods selected The upstream and downstream data were used as the basis for calibration. The particle swarm size was set to N=30, and the maximum number of iterations was set to T=500.

[0077] S105.1 Initialize the particle swarm: the position vector X of each particle e =[C0,C1,C2], randomly initialized within a reasonable range. Velocity vector V e Similarly, it is initialized randomly.

[0078] S105.2 Define the fitness function: The fitness function is the historical flood. The mean square error (MSE) between the observed flow rate at downstream station C and the flow rate calculated using the Muskingen formula Q2=C0·I2+C1·I1+C2·Q1.

[0079] S105.3 Iterative Optimization: In each iteration t:

[0080] a. Dynamically adjust parameters: Calculate the parameters for the current iteration using the following formula:

[0081] ω(t)=0.9-(0.9-0.4)(t / 500)

[0082] c1(t) = 2 - (2 - 0)(t / 500)

[0083] c2(t)=0+(2-0)(t / 500)

[0084] b. Update particle velocity and position: For each particle e, update according to the following formula:

[0085]

[0086]

[0087] Where Pe(t) is the individual historical best position of particle e, Pg(t) is the global historical best position of the particle swarm, and rand1 and rand2 are uniform random numbers in the interval [0,1].

[0088] c. Evaluation and Update: Calculate the fitness value (MSE) for each new location, and update the individual optimal location Pe(t) and the global optimal location Pg(t).

[0089] S105.4 Termination and Output: After iteration, the global optimal position Pg(t) = [C0, C1, C2] is output. This is based on the following formula:

[0090]

[0091]

[0092] Solving the simultaneous equations yields the final Muskingen model parameters K and x.

[0093] S106: Conduct flood forecasting

[0094] Substitute the optimal parameters K and x obtained from step S105 into the Muskingen channel flood calculation model. Utilize the target flood... Given the known flood process data (I1,I2,...,Iᵢ) of upstream station B, the flood process (q1,q2,...,qᵢ,...) of downstream station C is calculated and predicted hour by hour until the flood peak passes, thus completing the forecast of the entire flood process.

[0095] Effect verification:

[0096] The calculation is performed using data from three flood events in a certain river section in 1960, 1961, and 1964 from the literature.

[0097] (1) Validation of the effectiveness of flood similarity analysis method based on non-equivalent distance coefficient method

[0098] Data from three flood events in Chenggou Bay inflow in 1960, 1961, and 1964 were analyzed. First, a flood similarity index system was established using peak discharge, total flood volume, flood duration, and peak shape coefficient. Second, the indexes were ranked by importance: peak discharge > total flood volume > flood duration > peak shape coefficient. Then, the weights of the four flood indexes were calculated, specifically the weights of peak discharge, total flood volume, flood duration, and peak shape coefficient. Finally, the non-equilibrium distance coefficient method was used to calculate the similarity between the floods.

[0099] The similarity between the three floods calculated by this method, and its comparison with hydrological similarity methods and time series similarity search algorithms in the literature, are shown in Table 1 below:

[0100] Due to the characteristics of flood processes, there is no definitive true value for flood similarity; rather, it depends on different flood similarity evaluation methods. However, generally speaking, all reasonable and effective flood similarity evaluation methods should tend to be consistent in their assessment of the similarity between the same floods. As can be seen from the data in the table above, there are only minor differences between our method and the hydrological similarity methods and time series similarity search algorithms in the literature in calculating the similarity of the same flood; they are basically consistent. Therefore, the data demonstrates the rationality and effectiveness of our method. Because our method uses a hybrid cross-weighting method in determining the weights of flood characteristic indicators, combining the advantages of subjective and objective weighting methods, it can reflect both the objective information of actual data and incorporate the experience of decision-makers or users. Therefore, our method has unique advantages in its underlying principles.

[0101] (2) Validation of the improved particle swarm optimization algorithm in model parameter optimization

[0102] An improved particle swarm optimization (PSO) algorithm was developed by incorporating a strategy of "linearly decreasing inertia weights + dynamic adjustment of cognitive factors," thus addressing the insufficient convergence accuracy in the later stages of the conventional PSO algorithm. First, the algorithm parameters were set: particle swarm size, maximum number of iterations, and termination condition (reaching the maximum number of iterations). Parameter optimization was performed using data from the 1961 floods, resulting in optimized model parameters. , , Calculate in reverse , The calibrated parameters were substituted into a simulation of the 1961 flood, and the simulation results of this algorithm were compared with those of other algorithms in the literature. The comparison of various evaluation indicators is shown in Table 2 below:

[0103] As can be seen from the comparison in the table above, the improved particle swarm algorithm achieves the same flood simulation effect as the multi-agent genetic algorithm, and outperforms the conventional particle swarm algorithm, the accelerated genetic algorithm, and the nonlinear programming method.

[0104] (3) Application example of the Muskingan channel flood calculation method based on flood similarity analysis

[0105] Assuming a flood will occur in the same river section, the flood process data from the 1964 flood is used as the flood process forecast for this river section in the next few days. In order to predict the flood control situation in Linqing downstream in advance and make preparations for flood defense, it is necessary to forecast the flood process at the Linqing section in the next few days.

[0106] At this point, the flood data for 1960, 1961, and 1964 are all historical flood data. According to the data in Table 1 above, the similarity between the flood events of this river segment in 1960, 1961, and 1964 and the flood event predicted for this river segment are 0.79, 0.92, and 1.0, respectively. Using the flood data from these three years, the corresponding optimal model parameters were estimated, and the model parameter optimization and calibration results are shown in Table 3.

[0107] The flood process at the Linqing section during this flood was predicted using model parameters from the floods of 1960, 1961, and 1964, respectively. The relevant statistical indicators are shown in Table 4.

[0108]

[0109]

[0110]

[0111] According to the similarity between floods in Table 2, the 1964 flood is most similar to the current flood, followed by the 1961 flood, and the 1960 flood has the lowest similarity. The comparison results in Table 3 show that directly using the model parameters from the 1964 flood to predict the current flood event at the Linqing section is the most effective, with an average absolute error of 10.252 m. 3 / s, with an average relative error of 2.602%; directly using the model parameters from the 1961 flood to predict the current flood process at the Linqing section was the second best, with an average absolute error of 10.494 m. 3 / s, with an average relative error of 2.646%; directly using the model parameters from the 1960 flood to predict the current flood process at the Linqing section yielded the best results, with an average absolute error of 11.026 m. 3The average relative error is 2.731%. This shows that the more similar the flood processes are, the more similar the corresponding model parameters will be. Therefore, using historical floods with the highest similarity to the current flood as sample data for model parameter calibration can effectively and quickly determine model parameters that are more suitable for the current flood, and effectively reduce flood forecast errors.

[0112] Compared with existing technologies, this solution has the following technical advantages:

[0113] (1) Significantly improved forecast accuracy: This invention overcomes the inherent defect of fixed parameters in the traditional Muskingan method by introducing flood similarity analysis. By assigning mixed cross-weights to the non-equilibrium distance coefficient method, it can intelligently select the historical floods most similar to the current target flood characteristics from historical flood reservoirs, and dynamically calibrate the model parameters accordingly. This method makes the parameter combination (K,x) no longer the global optimum, but a "local optimum" for the current flood pattern, which greatly enhances the model's adaptability to specific floods and forecast accuracy, and effectively reduces the forecast error of abnormal floods and floods with large characteristic differences.

[0114] (2) Enhanced Efficiency and Reliability of Parameter Optimization: This invention employs an improved Particle Swarm Optimization (PSO) algorithm for parameter calibration, making key improvements to address the problems of the standard PSO algorithm easily getting trapped in local optima and having slow convergence speed in the later stages of iteration. By adopting a linear decreasing strategy for the inertia weight ω and the individual cognitive factor c1, and a linear increasing strategy for the social cognitive factor c2, a dynamic balance between the algorithm's global exploration capability and local development capability is achieved. This strategy enables the algorithm to extensively search the solution space in the early stages of iteration and to quickly converge to a high-precision solution in the later stages, thereby finding the optimal parameters that minimize simulation errors more efficiently and reliably, avoiding the blindness and instability of traditional trial-and-error methods or simple optimization algorithms.

[0115] (3) The method is more adaptive and robust: The overall scheme of the present invention has a high degree of adaptability. Its core lies in "differentiating according to flood" - different target floods will automatically match different similar historical floods, and different optimal parameters will be used for calculation accordingly. This design enables the present invention to better handle complex flood events of different sources, seasons and magnitudes, and shows stronger robustness (stability) to changes in river characteristics and the diversity of flood processes, reducing the risk of forecast failure due to the use of unsuitable parameters.

[0116] Finally, this invention maintains high accuracy without imposing a significant computational burden. The improved PSO algorithm increases convergence speed, and the similarity analysis stage quickly filters out the most relevant calibration data, avoiding the need for optimization calculations on the entire historical database each time. The entire method has low data requirements, requiring only conventional hydrological observation data, and is computationally fast, meeting the timeliness requirements of real-time flood forecasting and possessing good engineering application value. In summary, this invention effectively solves the problem of parameter fixation in the traditional Muskingan method by organically integrating intelligent flood similarity matching with an improved intelligent optimization algorithm, achieving significant progress in forecast accuracy, reliability, adaptability, and efficiency, and providing a more advanced and reliable technical means for river flood forecasting.

[0117] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention.

Claims

1. A Muskingum river flood forecasting method based on flood similarity analysis, characterized in that, The method comprises the following steps: S101: constructing a historical database of historical floods, which stores upstream cross-section flood process data and corresponding downstream cross-section flood process data of multiple historical floods; S102: Obtain the target flood upstream section forecast flood process data and downstream section forecast starting time actual outflow data; S103: constructing a flood similarity index system based on multi-dimensional characteristics of the floods; S104: extract target flood The multi-dimensional characteristic values of the flood are standardized, and a hybrid cross-assignment method is used to determine the weight of each characteristic index. S105: Based on the standardized feature index and the determined weight, the target flood is calculated by using the non-weighted distance coefficient method The comprehensive similarity between each field and historical flood; S106: According to the comprehensive similarity value size, a historical flood similar to the target flood is screened out from the historical flood database the most similar historical flood ; S107: with historical floods The improved particle swarm optimization algorithm is used to optimize the flow calculation coefficients C0, C1 and C2 in Muskingum flow calculation formula, with the minimum mean square error between the flow at the downstream section and the flow at the downstream section calculated based on the flood process at the upstream section as the optimization target. S108: determining the parameters of the Muskingum model, i.e., the storage constant K and the flow proportionality coefficient x, according to the estimated flow calculation coefficients C0, C1 and C2. S109: Substitute the parameters K and x into the Muskingum channel flood routing model, and perform evolution calculation on the target flood to obtain the downstream section flood process of the target flood .

2. The method of claim 1, wherein, The multi-dimensional flood characteristics in step S103 include peak flow, total flood volume, flood duration and peak shape coefficient.

3. The method according to claim 1 or 2, characterized in that, The variation coefficient of the flood characteristic index is calculated according to the standard deviation of the flood characteristic index and the average value of the flood characteristic index after the standardization of each flood characteristic index; and the weight of each flood characteristic index is calculated by using a hybrid cross weighting method based on subjective cognition and objective conditions .

4. The method of claim 3, wherein, The comprehensive similarity in step S105 is ; obtaining the characteristic index value X i,f_t of the target flood after standardization, and the characteristic index value X i,fh of the historical flood , wherein i=1, 2,..., n, and n is the total number of flood characteristic indexes; calculating the non-weighted Euclidean distance between the target flood and the historical flood ; calculating the comprehensive similarity between the target flood and the historical flood , 1- , when the similarity exceeds a set threshold value, determining that the historical flood is a similar flood.

5. The method of claim 4, wherein, The improved particle swarm optimization algorithm in step S107 is improved in that the linear decreasing strategy is respectively adopted to and iterative calculation is performed, and the linear increasing strategy is adopted to iterative calculation is performed.

6. The method of claim 5, wherein, The optimization objective function in step S107 is: where MSE is the mean square error, N is the number of data points, Q o (t) is the historical flood Q c (t) is the calculated downstream section flow at time t obtained by using the Muskingum formula and upstream section flow calculation.

7. The method of claim 6, wherein, The improved particle swarm optimization algorithm in step S107 is updated in the following steps: Initialize the swarm of particles, each particle e's position vector X e Represent a candidate solution for a set of flow calculation coefficients (C0, C1, C2) and initialize its velocity vector V e ; Define the fitness function as historical floods The mean square error (MSE) between the observed flow rate and the calculated flow rate at the downstream cross section is calculated, and the fitness value of each particle is calculated. In each iteration t, the inertia weight ω, the individual cognitive factor c1, and the social cognitive factor c2 are dynamically adjusted according to the following formulas: Wherein, T is a preset maximum iteration number, The value range of a is , And The value range of a is ; The velocity V of each particle e is updated according to the following formula e (t+1) and position X e (t+1) where Pe(t) is the individual historical optimal position of particle e, Pg(t) is the global historical optimal position of the particle group, and rand1 and rand2 are uniform random numbers in the interval [0, 1]. According to the updated particle position, the fitness value of each particle is recalculated, and the individual historical optimal position Pe(t+1) and the global historical optimal position Pg(t+1) are updated; the steps are repeated until the maximum iteration number T is reached or the preset convergence precision is met, and finally the global historical optimal position Pg is output as the optimized estimation value of the flow calculation coefficients (C0, C1, C2).

8. The method of claim 1, wherein, In step 8, the parameters of the Muskingum model, i.e., the storage constant K and the flow proportionality coefficient x, are determined according to the following formula: is the calculation period.

9. An electronic device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, The processor implements the steps of the method of any one of claims 1 to 5 when executing the program. The processor implements the steps of the method of any one of claims 1 to 5 when executing the program.

Images