Optimal Emergency Dispatch Flow Calculation Method and Device for River Water Pollution Accidents Considering Uncertainty

Abstract

The present invention provides a method and device for calculating the optimal emergency dispatch flow in the event of river water pollution accidents considering uncertainty. The method includes: Step 1, data collection; Step 2, based on the water pollution load and river hydrological parameters, establish a numerical simulation model of river water hydrodynamics and water quality for the study reach; Step 3, conduct a sensitivity analysis of river hydrological parameters to screen out the sensitive parameters that have a greater impact on the water quality calculation results; Step 4, use the Latin hypercube sampling method to sample the selected sensitivity parameters to obtain N1 groups of combinations of sensitivity parameters, and input them into the established numerical simulation model of river water hydrodynamics and water quality to obtain the output results of N1 groups of water quality indicators; input the N1 groups of sensitivity parameters and N1 groups of water quality indicators into a neural network model for training to obtain a water quality calculation surrogate model for each water quality indicator; Step 5, establish an optimization model for the dispatch flow; Step 6, conduct uncertainty analysis to obtain the impact of parameter uncertainty on the optimal dispatch flow.

Description

Technical Field

[0001] The present invention belongs to the field of environmental hydraulics, and particularly relates to a method and device for calculating the optimal emergency dispatch flow for river water pollution accidents considering uncertainty. Technical Background

[0002] For the river reaches controlled by sluices and dams, the dilution of river water pollution can be carried out by means of water diversion and regulation, so as to make the water quality of the control section meet the standard and relieve the burden of river water self-purification. However, the regional allocation of water resources and the improvement of river water quality are contradictory to each other, and the determination of the optimal dispatch flow is the focus of research.

[0003] However, the traditional calculation of the optimal emergency dispatch flow for river water pollution dilution is only a simple comparison of schemes, not the optimal result, and the uncertainty of river hydrological parameters is not considered in the calculation process, which leads to the inability of the dispatch flow to play the role of diluting river water pollution or the waste of the dispatch flow. Therefore, in the engineering field of river water pollution control, there is an urgent need for a method for calculating the optimal emergency dispatch flow for river water pollution accidents considering uncertainty. Summary of the Invention

[0004] The present invention is made to solve the above problems, and aims to provide a method and device for calculating the optimal emergency dispatch flow for river water pollution accidents considering uncertainty, which can efficiently calculate the optimal dispatch flow for river water pollution diversion and dilution and perform uncertainty analysis, avoid dispatch waste, and effectively reduce the dispatch decision-making risk. To achieve the above object, the present invention adopts the following solutions:

[0005] <Method>

[0006] As Figure 1 shown, the present invention provides a method for calculating the optimal emergency dispatch flow for river water pollution accidents considering uncertainty, which is characterized by including the following steps:

[0007] Step 1, data collection: Determine the water pollution load situation of the research river reach and the value range of river hydrological parameters (such as roughness, longitudinal dispersion coefficient, degradation coefficients of various water quality indexes, etc.);

[0008] Step 2, model establishment: Based on the water pollution load and river hydrological parameters in Step 1, establish a numerical simulation model of river hydrodynamic water quality for the research river reach;

[0009] Step 3, sensitivity analysis: Use the established numerical simulation model of hydrodynamic water quality to conduct sensitivity analysis of river hydrological parameters, and screen out the sensitive parameters that have a greater impact on the water quality calculation results;

[0010] Step 4, establishment of surrogate model: Use the Latin hypercube sampling method to sample the screened sensitivity parameters to obtain N1 combinations of sensitive parameters, and input them into the established river hydrodynamic and water quality numerical simulation model to obtain the output results of N1 groups of water quality indicators; Input the N1 groups of sensitive parameters and N1 groups of water quality indicators into the neural network model for training to obtain the water quality calculation surrogate model for each water quality indicator; N1 is a positive integer; In the process of establishing the surrogate model in Step 4, the schematic diagram of the artificial neural network model used is shown in Figure 2 ;

[0011] Step 5, establishment of optimal scheduling flow model: Replace the river hydrodynamic and water quality numerical simulation model with the surrogate model and embed it into the particle swarm optimization model, and establish the optimal scheduling flow model with the goal of non-exceeding water quality and minimizing the scheduling flow;

[0012] Step 6, uncertainty analysis: Sample the screened sensitivity parameters to obtain N2 combinations of sensitive parameters, and substitute them into the optimal scheduling flow model to calculate and obtain N2 optimal scheduling flow optimization results. N2 is a positive integer. Conduct statistical analysis on the results to obtain the impact of parameter uncertainty on the optimal scheduling flow.

[0013] The method for calculating the optimal emergency scheduling flow for river pollution accidents considering uncertainty provided by the present invention may further have the following characteristics: In Step 1, the water pollution load situation comes from the investigation of river inflow pollution load; The value ranges of roughness coefficient, longitudinal dispersion coefficient in river hydrological parameters and degradation coefficients of each pollutant index are determined according to historical data.

[0014] The method for calculating the optimal emergency scheduling flow for river pollution accidents considering uncertainty provided by the present invention may further have the following characteristics:

[0015] In Step 2, the river hydrodynamic and water quality numerical simulation model of the research reach is as follows:

[0016]

[0017] In the formula, A is the cross-sectional area; Q is the flow rate; Q branch is the tributary inflow; Z is the water level; g is the acceleration due to gravity; S f is the energy slope; C is the concentration value of the pollutant index; E is the longitudinal dispersion coefficient value; K is the degradation coefficient of this pollutant index;

[0018] The upstream hydrodynamic boundary condition calculated by the above equations is given as the scheduling flow rate, which needs to be assumed; The downstream hydrodynamic boundary condition is obtained by interpolation of the water level-flow relationship curve; The upstream water quality boundary is set to the water quality condition of Class III water, and the downstream water quality boundary is set to the free outfall condition.

[0019] The optimal emergency scheduling flow calculation method for river water pollution accidents considering uncertainty provided by the present invention may further have the following characteristics:

[0020] In step 3, sensitivity analysis is performed based on the following formula:

[0021]

[0022] In the formula, I k represents the sensitivity coefficient of the k-th parameter; x k represents the k-th parameter of the model input; Δx k represents the change in the parameter x k ; y i (x k ) is the calculation result of the water quality index corresponding to the parameter x k ; y i (x k +Δx k ) is the calculation result of the water quality index corresponding to the parameter x k +Δx k ; N c is the number of water quality control sections;

[0023] The greater the sensitivity coefficient of the parameter, the greater the impact on the calculation result of the water quality index, and the stronger the uncertainty of the calculation result of the water quality index brought about by the change of this parameter.

[0024] The optimal emergency scheduling flow calculation method for river water pollution accidents considering uncertainty provided by the present invention may further have the following characteristics:

[0025] In step 5, during the establishment of the scheduling flow optimization model, the objective function to ensure that the water quality does not exceed the standard and the scheduling flow is minimized is as follows:

[0026]

[0027] In the formula, (C - C max ) + = max{C - C max , 0}; Q is the optimal scheduling flow to be obtained; is the concentration value of the i-th water quality index at the j-th control section calculated by the surrogate model; C max is the concentration limit value of this water quality index; M is the number of water quality indexes considered; N c is the number of control sections; η and μ are both weight coefficients used to adjust the weights of the two terms in the objective function, and it is necessary to ensure that when the water quality index exceeds the standard, the objective function increases rapidly.

[0028] The optimal emergency scheduling flow calculation method for river water pollution accidents considering uncertainty provided by the present invention may further have the following characteristics:

[0029] In step 6, statistical analysis is carried out on the calculation results of the optimal scheduling flow corresponding to each group of parameters to obtain the mean value, standard deviation and confidence interval;

[0030] The confidence interval is approximately estimated by using the Chebyshev inequality, and the Chebyshev inequality is as follows:

[0031]

[0032] In the formula, P is the probability; X is the random variable; μ is the mathematical expectation of the random variable X, which can be replaced by the sample mean ; σ is the standard deviation of the random variable X, which can be replaced by the sampling average error S x ; the calculation formula of the sampling average error is as follows:

[0033]

[0034] In the formula, S is the sample standard deviation; N is the number of samples;

[0035] For the given variable X, the overall proportion or probability within the interval [μ - kσ, μ + kσ] is 1 - 1 / k 2 , and thus the interval estimation of the optimal scheduling flow is carried out, and then the confidence interval is obtained.

[0036] <Device>

[0037] Furthermore, the present invention also provides an optimal emergency scheduling flow calculation device for river water pollution accidents considering uncertainty, which is characterized by including:

[0038] An information collection unit, which determines the water pollution load situation, river water hydrological parameters and value ranges of the research river section according to the collected data;

[0039] A model initial construction unit, which establishes a river water dynamic water quality numerical simulation model of the research river section based on the water pollution load and river water hydrological parameters;

[0040] A sensitivity analysis unit, which uses the established water dynamic water quality numerical simulation model to carry out sensitivity analysis of river water hydrological parameters and screens out sensitive parameters that have a greater impact on the water quality calculation results;

[0041] An agent model establishment unit, which samples the screened sensitivity parameters by using the Latin hypercube sampling method to obtain N1 groups of combinations of sensitivity parameters, and inputs them into the established river water dynamic water quality numerical simulation model to obtain the output results of N1 groups of water quality indicators; inputs the N1 groups of sensitivity parameters and N1 groups of water quality indicators into a neural network model for training to obtain a water quality calculation agent model for each water quality indicator;

[0042] The optimal model establishment unit replaces the river water dynamic water quality numerical simulation model with a surrogate model and embeds it into the particle swarm optimization model, and establishes an optimal scheduling flow model with the goal function of non-exceeding water quality and minimizing the scheduling flow;

[0043] The uncertainty analysis unit samples the selected sensitivity parameters to obtain N2 combinations of sensitive parameters, brings them into the optimal scheduling flow model, calculates and obtains N2 optimal scheduling flow optimization results, and conducts statistical analysis on the results to obtain the impact of parameter uncertainty on the optimal scheduling flow;

[0044] The control unit is communicatively connected to the information collection unit, the model initial establishment unit, the sensitivity analysis unit, the surrogate model establishment unit, the optimal model establishment unit, and the uncertainty analysis unit, and controls their operations.

[0045] Preferably, the optimal emergency scheduling flow calculation device for river water pollution accidents considering uncertainty provided by the present invention may further include: a scheduling unit, communicatively connected to the water control system of the water conservancy project and the control unit, based on the optimal scheduling flow optimization result obtained by the uncertainty analysis unit and the impact of parameter uncertainty on the optimal scheduling flow, obtains the optimal scheduling flow for the corresponding time period as scheduling information, and sends the scheduling information to the water control system of the water conservancy project for water use scheduling accordingly.

[0046] Preferably, the optimal emergency scheduling flow calculation device for river water pollution accidents considering uncertainty provided by the present invention may further have the following characteristics: an input display unit, communicatively connected to the control unit, for allowing the user to input operation instructions and performing corresponding displays.

[0047] Preferably, the optimal emergency scheduling flow calculation device for river water pollution accidents considering uncertainty provided by the present invention may further have the following characteristics: in the sensitivity analysis unit, sensitivity analysis is performed based on the following formula:

[0048]

[0049] In the formula, I k represents the sensitivity coefficient of the kth parameter; x k represents the kth parameter of the model input; Δx k represents the change in the parameter x k ; y i (x k ) is the calculation result of the water quality index corresponding to the parameter x k ; y i (x k + Δx k ) is the calculation result of the water quality index corresponding to the parameter x k + Δx k ; N cis the number of water quality control sections; the greater the sensitivity coefficient of a parameter, the greater its impact on the calculation result of the water quality index, and the stronger the uncertainty of the calculation result of the water quality index brought about by the change of this parameter;

[0050] During the establishment process of the optimal model in the optimal model establishment unit for the optimal scheduling flow model, to ensure that the water quality does not exceed the standard and the scheduling flow is minimized, the objective function is as follows:

[0051]

[0052] In the formula, (C - C max ) + = max{C - C max , 0}; Q is the optimal scheduling flow to be obtained; is the concentration value of the i-th water quality index at the j-th control section calculated by the surrogate model; C max is the concentration limit value of this water quality index; M is the number of water quality indexes considered; N c is the number of control sections; both η and μ are weight coefficients used to adjust the weights of the two terms in the objective function to ensure that when the water quality index exceeds the standard, the objective function increases rapidly.

[0053] Functions and effects of the invention

[0054] The optimal emergency scheduling flow calculation method and device for river water pollution accidents considering uncertainty provided by the present invention uses a neural network model in machine learning methods to derive a surrogate model of the hydrodynamic water quality model and applies it to water quality calculation. It considers objectives such as non-exceedance of water quality and minimum scheduling flow in river water pollution dilution in a unique form of objective function. First, parameters are screened by sensitivity analysis, and then uncertainty quantification analysis is carried out; it can give the optimal scheduling flow for pollution dilution when river pollution occurs, rather than just the comparison of schemes, achieving the maximum utilization of water volume; it can consider the objectives of non-exceedance of water quality and minimum scheduling flow simultaneously; it can directly give the optimal scheduling flow and the quantitative analysis of the result uncertainty brought about by the uncertainty of river water hydrological parameters; at the same time, the present invention uses a neural network model to obtain a surrogate model for water quality calculation of the research reach, saving 99% of the calculation load and calculation time.

[0055] In summary, the present invention can efficiently and accurately obtain the optimal scheduling flow and the quantitative analysis result of the result uncertainty brought about by the uncertainty of river water hydrological parameters after river water pollution occurs, providing a more scientific and effective basis for reservoir water use scheduling decisions. Description of the drawings

[0056] Figure 1 is the flow chart of the optimal emergency scheduling flow calculation method for river water pollution accidents considering uncertainty involved in the present invention;

[0057] Figure 2 Schematic diagram of the neural network model adopted in the embodiment of the present invention;

[0058] Figure 3 Trend chart of the change of A1TN index involved in the embodiment of the present invention;

[0059] Figure 4 Trend chart of the change of A1TN index involved in the embodiment of the present invention;

[0060] Figures 5 to 8 Comparison chart of the output results of the proxy model and the simulation results of the hydrodynamic water quality model for each month in the dry season involved in the embodiment of the present invention; wherein, Figure 5 corresponds to January, Figure 6 corresponds to February, Figure 7 corresponds to March, Figure 8 corresponds to December;

[0061] Figures 9 to 12 Calculation result chart of the optimal scheduling flow for each month in the dry season under the determined parameters involved in the embodiment of the present invention; wherein, Figure 9 corresponds to January, Figure 10 corresponds to February, Figure 11 corresponds to March, Figure 12 corresponds to December;

[0062] Figures 13 to 16 Analysis result chart of the parameter uncertainty of the optimal scheduling flow for each month in the dry season involved in the embodiment of the present invention; wherein, Figure 13 corresponds to January, Figure 14 corresponds to February, Figure 15 corresponds to March, Figure 16 corresponds to December. Detailed implementation manners

[0063] The following will describe in detail the specific implementation scheme of the optimal emergency scheduling flow calculation method and device for river water pollution accidents considering uncertainty according to the present invention with reference to the accompanying drawings.

[0064] <Embodiment>

[0065] A is a first-class tributary of B, with a total main stream length of 170.4 km. A certain section A1 of A is 33.31 km long, with a basin area of 763 km 2 , and there are 3 conventional monitoring sections, which are a (inlet section), b, and c from upstream to downstream. Except for the inlet section, the other monitoring sections are water quality control sections. The water function zoning of A1 is Class III water body, and the water quality management target is Class III water. However, Figure 3It can be seen that the water quality situation is not optimistic. After 2013, it gradually deteriorated. From Table 1, the limit values of surface water environmental quality standards, it can be seen that the main water quality index exceeding the standard for A1 is TN. In order to make the water quality of the two water quality control sections meet the standards, a study on the calculation of the optimal regulation flow for water pollution dilution in the A1 river was carried out. Through research, it was found that the pollution of A1 exceeded the standard only in December, January, February, and March during the dry season. In other months, the water quality of the control section could meet the standards according to the existing pollution load conditions. Therefore, only the optimal regulation flow for water pollution dilution in the river during the dry season months was calculated.

[0066] Table 1 Limit Values of Surface Water Environmental Quality Standards Unit: mg / L

[0067]

[0068] As Figure 1 shown, the optimal emergency regulation flow calculation method for river water pollution accidents considering uncertainty provided by the present invention was used to calculate and analyze the optimal regulation flow for water pollution dilution in the A1 river, which specifically included the following steps:

[0069] Step 1, data collection. Determine the water pollution load situation of the research river section and the value ranges of river hydrological parameters (roughness coefficient, longitudinal dispersion coefficient, degradation coefficients of various water quality indicators, etc.);

[0070] According to relevant data, it was assumed that the hydrological parameters of the A1 river obeyed the normal distribution, and the mean values and value ranges were shown in Table 2; the degradation coefficients of each water quality indicator in the dry season months of A1 were shown in Table 3; since the regulation flow Q of A1 was a variable to be optimized, considering the size of the optimal regulation flow and combining with the monthly average flow data of the Liuzigang hydrological station of A1 in the dry season, it was assumed that its value range was 2 - 40 m 3 / s. The pollution load data of A1 were shown in Table 4.

[0071] Table 2 Value Ranges of Hydrological Parameters of the A1 River

[0072]

[0073] Table 3 Mean Values and Value Ranges of Degradation Coefficients of Each Water Quality Indicator in the Dry Season

[0074]

[0075] Table 4 Pollution Load Data of A1

[0076]

[0077] It can be seen from Table 4 that the pollution load of A1 was generalized into 6 sewage outfalls. The main pollution load of A1 was divided into the whole year according to the proportion of the flow rate, and the discharge flow rate of each generalized outfall and the discharge concentration of each pollutant component were obtained, as shown in Table 5.

[0078] Table 5 Discharge Concentrations and Flow Rates at Outfalls in Each Month during the Low Flow Period of A1 Unit: mg / L

[0079]

[0080] Step 2, model establishment. Input the water pollution load and river hydrological parameters in Step 1 to establish a numerical simulation model of river hydrodynamic water quality for the study reach;

[0081] Based on the relevant data in Step 1: A1 terrain, hydrological data, parameter value ranges, hydrodynamic water quality boundary conditions, etc., perform finite difference discretization on the one-dimensional hydrodynamic water quality equations, namely Equation (1), to construct a numerical simulation model of hydrodynamic water quality for A1.

[0082] Step 3, sensitivity analysis. Use the established numerical simulation model of river hydrodynamic water quality to conduct sensitivity analysis of river hydrological parameters, and screen out the sensitive parameters that have a greater impact on the water quality calculation results;

[0083] Taking the calculation of COD concentration as an example, the parameters involved in the analysis are the longitudinal dispersion coefficient E, roughness coefficient n, and degradation coefficient K COD (The regulated flow rate Q is a variable to be solved and must be a sensitive parameter, so sensitivity analysis is not performed), and the specific sensitivity analysis results are shown in Figure 4 . It can be seen from the figure that the impact of the longitudinal dispersion coefficient on the COD at the control section is almost negligible. Therefore, the finally screened sensitive parameters are the roughness coefficient n, the degradation coefficient K, and the regulated flow rate Q.

[0084] Step 4, surrogate model establishment. Use the Latin hypercube sampling method to sample the screened sensitivity parameters to obtain N1 groups of combinations of sensitivity parameters, and input them into the established numerical simulation model of river hydrodynamic water quality to obtain the output results of N1 groups of water quality indicators. Input the N1 groups of sensitivity parameters and N1 groups of water quality indicators into the neural network model for training to obtain a water quality calculation surrogate model for each water quality indicator. During the establishment of the surrogate model, the schematic diagram of the artificial neural network model used is shown in Figure 2 .

[0085] For each month during the low flow period, use the Latin hypercube sampling method to sample the screened parameters to obtain 50 groups of combinations of sensitivity parameters (Q, n, K COD , K TN , K TP ), input these 50 groups of parameters into the established hydrodynamic water quality mathematical model for calculation to obtain 50 groups of output results. Use the parameter combinations and output results as training data to train the neural network model. The specific training results of the surrogate models for the water quality calculation modules in each month during the low flow period are shown in Figures 5 to 8As can be seen from the figure, the concentration results output by the trained neural network model are in good fit with the concentration results calculated by the hydrodynamic water quality model established in the early stage, and the correlation coefficients all reach more than 0.995.

[0086] Step 5, establishment of the optimal scheduling flow model. Replace the river hydrodynamic water quality numerical simulation model with a surrogate model, embed it into the particle swarm optimization model, and establish an optimal scheduling flow model with the goal of non-exceeding water quality and minimizing the scheduling flow.

[0087] For the study on the optimal dilution scheduling flow of water pollution in River A1, there are mainly four water quality indicators to be considered, namely COD, NH3-N, TN, and TP, that is, M = 4; the water quality control sections that need to be considered for A1 are mainly the Chengguan section and the Mujiajing section (the section entering the river), that is, N c = 2. To ensure that the objective function increases rapidly when the water quality indicator exceeds the standard, take η = 1 and μ = 1e+10. Then the objective function at this time is as follows:

[0088]

[0089] Given the determined river hydrological parameters, use the particle swarm optimization algorithm to find the minimum value of formula (6). The only variable is the scheduling flow. After optimization, the optimal scheduling flow for each month in the dry season can be obtained. See Figures 9 to 12 .

[0090] Use the MIKE11 model to test the established optimization model. Given the flow rates in January, February, March, and December as 21.69 m 3 / s, 11.23 m 3 / s, 11.49 m 3 / s, 12.02 m 3 / s respectively. The test results are shown in Table 6. As can be seen from the table, each water quality indicator is below the standard limit value, and the TN at the Mujiajing section is close to the indicator limit value. This shows that the calculation results of the established optimization model are correct and reliable.

[0091] Table 6 Water quality status of monitoring sections under the optimal scheduling flow for each month in the dry season

[0092]

[0093] Step 6, uncertainty analysis. Sample the selected sensitivity parameters to obtain N2 combinations of sensitive parameters, and substitute them into the optimal scheduling flow model to obtain N2 optimal scheduling flow optimization results. Conduct statistical analysis on the results to obtain the influence of parameter uncertainty on the optimal scheduling flow.

[0094] The Monte Carlo method is adopted to study the influence of parameter uncertainty on the calculation result of the minimum upstream flow rate. 500 groups of parameter combinations are selected from the sensitive parameters by Latin hypercube sampling, and they are brought into the particle swarm optimization model to calculate the corresponding minimum upstream flow rate respectively. Finally, statistical analysis is carried out on these calculated optimal scheduling flow rates, and the specific results are shown in Figures 13 to 16 . It can be seen from the figure that the change of parameters has a great influence on the minimum upstream flow rate, and the distribution of the results follows a skewed distribution rather than a normal distribution. This means that the optimal scheduling flow rate calculated by using the parameter mean for the corresponding month is less than the optimal scheduling flow rate with the highest occurrence probability, which may lead to the situation that the decision-making scheduling flow rate does not necessarily meet the need of river water pollution dilution in most cases, and further cause the water quality of the downstream control section to fail to meet the standards. This also fully proves the necessity of parameter uncertainty analysis.

[0095] It can be seen from the above figure that: the uncertainty of the calculation results of the optimal scheduling flow rate in January and March is relatively large, while the uncertainty of the calculation results in February and December is relatively small. The average values of the optimal scheduling flow rates for each month in the dry season finally calculated are: 12.27 m 3 / s (December), 22.30 m 3 / s (January), 12.24 m 3 / s (February) and 11.23 m 3 / s (March). Table 7 gives the optimal scheduling flow rates for each month in the dry season under different confidence intervals.

[0096] Table 7 Optimal scheduling flow rates for each month in the dry season under different confidence intervals Unit: m 3 / s

[0097]

[0098]

[0099] In the process of carrying out parameter uncertainty analysis, with the same computer configuration, the calculation method proposed by the present invention calls the subroutine of the water quality calculation alternative model 1,521,075 times in total, and the total time used is 92.941 s; while it takes 7 s to calculate the water quality result of Case A1 once by using MIKE, and it takes about 123 days to operate 1,521,075 times; in comparison, the calculation efficiency of the method proposed by the present invention is greatly improved.

[0100] After testing, the method of the present invention can not only give the optimal scheduling flow for river water pollution diversion and dilution, but also maximize the utilization of water volume while ensuring that the water quality at the control section does not exceed the standard. At the same time, this method uses a neural network model to obtain a surrogate model for calculating the water quality of the study reach, saving 99% of the computational load and computational time, and improving the drawback that the uncertainty of river hydrological parameters cannot be considered in previous research methods. It can provide more reference basis for scheduling decisions after river water pollution occurs, making the calculation results of the dilution scheduling flow of river water pollution more reasonable.

[0101] Furthermore, this embodiment also provides a device for calculating the optimal emergency scheduling flow for river water pollution accidents that can automatically implement the above method. The device includes an information collection unit, a model initial construction unit, a sensitivity analysis unit, a surrogate model establishment unit, an optimal model establishment unit, an uncertainty analysis unit, a scheduling unit, an input display unit, and a control unit.

[0102] According to the description of step 1 above, the information collection unit determines the water pollution load situation, river hydrological parameters, and their value ranges of the study reach based on the collected data. The data collected here can be data imported or input by the user through the input display unit.

[0103] According to the description of step 2 above, the model initial construction unit establishes a hydrodynamic water quality numerical simulation model of the study reach based on the water pollution load and river hydrological parameters.

[0104] According to the description of step 3 above, the sensitivity analysis unit uses the established hydrodynamic water quality numerical simulation model to conduct sensitivity analysis of river hydrological parameters and screen out sensitive parameters that have a greater impact on the water quality calculation results.

[0105] According to the description of step 4 above, the surrogate model establishment unit uses the Latin hypercube sampling method to sample the selected sensitivity parameters, obtains N1 groups of combinations of sensitivity parameters, and inputs them into the established hydrodynamic water quality numerical simulation model to obtain the output results of N1 groups of water quality indicators; inputs the N1 groups of sensitivity parameters and N1 groups of water quality indicators into the neural network model for training to obtain a water quality calculation surrogate model for each water quality indicator.

[0106] According to the description of step 5 above, the optimal model establishment unit replaces the hydrodynamic water quality numerical simulation model with a surrogate model and embeds it into the particle swarm optimization model to establish an optimization model for the scheduling flow with the goal of non-exceeding water quality and minimum scheduling flow.

[0107] According to the description in step 6 above, the uncertainty analysis department samples the selected sensitivity parameters to obtain N2 combinations of sensitive parameters, and substitutes them into the optimal scheduling flow optimization model. By calculating, N2 optimal scheduling flow optimization results are obtained. Through statistical analysis of the results, the impact of parameter uncertainty on the optimal scheduling flow is obtained.

[0108] Based on the optimal scheduling flow optimization results obtained by the uncertainty analysis department and the impact of parameter uncertainty on the optimal scheduling flow, the scheduling department obtains the optimal scheduling flow for the corresponding period as scheduling information and sends the scheduling information to the water control system of the water conservancy project for water use scheduling accordingly.

[0109] The input display department is used to allow users to input operation instructions and perform corresponding displays.

[0110] The control department is communicatively connected to the information collection department, the initial model building department, the sensitivity analysis department, the surrogate model building department, the optimal model building department, the uncertainty analysis department, the scheduling department, and the input display department to control their operations.

[0111] The above embodiments are merely illustrative examples of the technical solutions of the present invention. The optimal emergency scheduling flow calculation method and device for river water pollution accidents considering uncertainty involved in the present invention are not limited solely to the content described in the above embodiments, but are subject to the scope defined by the claims. Any modifications, supplements, or equivalent replacements made by those skilled in the art to the present invention based on this embodiment are within the scope protected by the claims of the present invention.

Claims

1. Optimal emergency dispatch flow calculation method for river water pollution accidents considering uncertainty, characterized in that, It includes the following steps: Step 1, data collection: Determine the water pollution load situation and the value range of river hydrological parameters in the research reach; Step 2, model establishment: Based on the water pollution load and river hydrological parameters in Step 1, establish a numerical simulation model of river hydrodynamic water quality in the research reach; Step 3, sensitivity analysis: Use the established numerical simulation model of hydrodynamic water quality to conduct sensitivity analysis of river hydrological parameters, and screen out the sensitive parameters that have a greater impact on the water quality calculation results; Step 4, surrogate model establishment: Use the Latin hypercube sampling method to sample the screened sensitivity parameters to obtain combinations of groups of sensitive parameters, and input them into the established river water hydrodynamic and water quality numerical simulation model to obtain output results of groups of water quality indicators; groups of sensitive parameters and groups of water quality indicators are input into the neural network model for training to obtain a water quality calculation surrogate model for each water quality indicator; Step 5, establishment of optimal scheduling flow model: Replace the numerical simulation model of river hydrodynamic water quality with a surrogate model and embed it into the particle swarm optimization model, and establish an optimal scheduling flow model with the goal of non-exceedance of water quality and minimum scheduling flow; Step 6, uncertainty analysis: sample the screened sensitivity parameters to obtain combinations of groups of sensitive parameters, and substitute them into the optimal scheduling flow model to calculate and obtain optimal scheduling flow optimization results. Conduct statistical analysis on the results to obtain the impact of parameter uncertainty on the optimal scheduling flow. Among them, in Step 5, during the establishment of the optimal scheduling flow model, the objective function to ensure non-exceedance of water quality and minimum scheduling flow is as follows: (3) In the formula, ; is the optimal scheduling flow required; is the concentration value of the th water quality index calculated by the surrogate model at the th control section; is the concentration limit value of this water quality index; is the number of water quality indexes considered; is the number of control sections; and are both weight coefficients, used to adjust the weights of the two terms in the objective function to ensure that when the water quality index exceeds the standard, the objective function increases rapidly.

2. The optimal emergency scheduling flow calculation method for river water pollution accidents considering uncertainty according to claim 1, characterized in that: Among them, In Step 1, the water pollution load situation comes from the investigation of river inflow pollution load; the value ranges of roughness coefficient, longitudinal dispersion coefficient and degradation coefficients of each pollutant index in river hydrological parameters are determined according to historical data.

3. The optimal emergency scheduling flow calculation method for river water pollution accidents considering uncertainty according to claim 1, characterized in that: Among them, In Step 2, the numerical simulation model of river hydrodynamic water quality in the research reach is as follows: (1) In the formula, is the cross-sectional area; is the flow rate; is the tributary inflow rate; is the water level; is the acceleration due to gravity; is the energy slope; is the concentration value of the pollutant index; is the longitudinal dispersion coefficient value; is the degradation coefficient of the pollutant index; The upstream hydrodynamic boundary condition calculated by the above equations is given as the scheduling flow; the downstream hydrodynamic boundary condition is obtained by interpolation of the water level-flow relationship curve; the upstream water quality boundary is set as the water quality condition of Class III water, and the downstream water quality boundary is set as the free outfall condition.

4. The optimal emergency scheduling flow calculation method for river water pollution accidents considering uncertainty according to claim 1, It is characterized in that: Among them, in Step 3, sensitivity analysis is carried out based on the following formula: (2) In the formula, represents the sensitivity coefficient of the th parameter; represents the th parameter of the model input; represents the change in the parameter ; is the calculation result of the water quality index corresponding to the parameter ; is the calculation result of the water quality index corresponding to the parameter ; is the number of water quality control sections; The greater the sensitivity coefficient of the parameter, the greater the impact on the water quality index calculation result, and the stronger the uncertainty of the water quality index calculation result brought about by the change of this parameter.

5. The optimal emergency scheduling flow calculation method for river water pollution accidents considering uncertainty according to claim 1, characterized in that: Among them, In Step 6, statistical analysis is carried out on the calculation results of the optimal scheduling flow corresponding to each group of parameters to obtain the mean value, standard deviation and confidence interval; The confidence interval is approximately estimated by the Chebyshev inequality, and the Chebyshev inequality is as follows: (4) In the formula, is the probability magnitude; is a random variable; is the mathematical expectation of the random variable and can be replaced by the sample mean ; is the standard deviation of the random variable and can be replaced by the sampling average error . The calculation formula for the sampling average error is as follows: (5) In the formula, is the sample standard deviation; is the number of samples; For a given variable , the overall proportion or probability within the interval is . Based on this, an interval estimate of the optimal scheduling flow is carried out, and then the confidence interval is obtained.

6. Optimal emergency dispatching flow calculation device for river water pollution accidents considering uncertainty, characterized in that It includes: An information collection department, which determines the water pollution load situation, river hydrological parameters and their value ranges in the research reach according to the collected data; A model initial establishment department, which establishes a numerical simulation model of river hydrodynamic water quality in the research reach based on the water pollution load and river hydrological parameters; A sensitivity analysis department, which uses the established numerical simulation model of hydrodynamic water quality to conduct sensitivity analysis of river hydrological parameters and screen out the sensitive parameters that have a greater impact on the water quality calculation results; The proxy model establishment unit uses the Latin hypercube sampling method to sample the screened sensitivity parameters, obtaining combinations of groups of sensitive parameters, and inputs them into the established river water hydrodynamic and water quality numerical simulation model to obtain output results of groups of water quality indicators; inputs groups of sensitive parameters and groups of water quality indicators into the neural network model for training to obtain a water quality calculation proxy model for each water quality indicator; An optimal model establishment department, which replaces the numerical simulation model of river hydrodynamic water quality with a surrogate model and embeds it into the particle swarm optimization model, and establishes an optimal scheduling flow model with the goal of non-exceedance of water quality and minimum scheduling flow; The uncertainty analysis department samples the selected sensitivity parameters to obtain combinations of groups of sensitive parameters, and substitutes them into the optimal scheduling flow model to calculate and obtain optimal scheduling flow optimization results. Statistical analysis is performed on the results to obtain the impact of parameter uncertainty on the optimal scheduling flow; A control unit, which is communicatively connected to the information collection unit, the initial model building unit, the sensitivity analysis unit, the surrogate model building unit, the optimal model building unit, and the uncertainty analysis unit, and controls their operations; Among them, during the establishment of the optimal model for scheduling flow optimization in the optimal model building unit, the objective function to ensure that the water quality does not exceed the standard and the scheduling flow is minimized is as follows: (3) wherein, ; is the optimal scheduling flow rate sought; is the concentration value of the th water quality index calculated by the surrogate model at the th control section; is the concentration limit value of this water quality index; is the number of water quality indexes considered; is the number of control sections; and are both weight coefficients used to adjust the weights of the two terms in the objective function to ensure that the objective function increases rapidly when the water quality index exceeds the standard.

7. The optimal emergency dispatch flow calculation device for river water pollution accidents considering uncertainty according to claim 6, characterized in that It further includes: A scheduling unit, which is communicatively connected to the water control system of the water conservancy project and the control unit. Based on the optimal scheduling flow optimization result obtained by the uncertainty analysis unit and the impact of parameter uncertainty on the optimal scheduling flow, it obtains the optimal scheduling flow for the corresponding period as scheduling information, and sends the scheduling information to the water control system of the water conservancy project for water use scheduling accordingly.

8. The optimal emergency scheduling flow calculation device for river water pollution accidents considering uncertainty according to claim 6, characterized in that, It further includes: An input display unit, which is communicatively connected to the control unit, and is used for the user to input operation instructions and perform corresponding displays.

9. The optimal emergency dispatch flow calculation device for river water pollution accidents considering uncertainty according to claim 6, It is characterized in that: Among them, In the sensitivity analysis unit, sensitivity analysis is performed based on the following formula: (2) In the formula, represents the sensitivity coefficient of the th parameter; represents the th parameter of the model input; represents the change in the parameter ; is the calculation result of the water quality index corresponding to the parameter ; is the calculation result of the water quality index corresponding to the parameter ; is the number of water quality control sections; The larger the sensitivity coefficient of a parameter, the greater the impact on the calculation result of the water quality index, and the stronger the uncertainty of the calculation result of the water quality index brought about by the change of this parameter.

Images