Pulse flood dispatching method to mitigate the impact of supersaturated total dissolved gas on fish

Abstract

The invention belongs to the field of ecological protection technology, and specifically relates to a pulse flood dispatching method for mitigating the impact of supersaturated total dissolved gas on fish. Firstly, data are collected to establish a reservoir water level-flow balance equation and a downstream river water level-flow balance equation; a power generation prediction model is constructed; the relationship between the saturation of supersaturated TDG generated by dam flood discharge and environmental factors is determined; then a supersaturated TDG generation prediction model for dam flood discharge is constructed based on a deep neural network; furthermore, the tolerance characteristics of fish to supersaturated TDG are determined; combined with the tolerance of fish's own physiological characteristics to supersaturated TDG, the closing / stopping time of dam flood discharge under different TDG saturation conditions is determined; a multi-objective ecological dispatching model with the goals of minimizing the ecological impact of supersaturated TDG and maximizing power generation is constructed; and a solution is obtained; finally, the distribution of supersaturated TDG in the reservoir area downstream of the dam is obtained through optimization, so as to solve the problems of difficult calibration of parameters and slow calculation speed of traditional mechanism models.

Description

Technical Field

[0001] The invention belongs to the technical field of ecological protection, and in particular relates to a pulse flood dispatching method for alleviating the influence of supersaturated total dissolved gas on fish. Background Art

[0002] Total dissolved gas refers to the total amount of all gases dissolved in the water, usually expressed as volume percentage or mass percentage. Gases that can be dissolved in water include nitrogen, oxygen, carbon dioxide, etc. Flood discharge from a reservoir will draw in air and form a water body that is supersaturated with total dissolved gas (TDG) under the action of turbulence and high pressure, causing fish below the dam to suffer from bubble disease or even die. Its adverse effects will last for tens or hundreds of kilometers and are very likely to cause ecological damage. Therefore, mitigating the impact of supersaturated TDG caused by flood discharge from a reservoir on the suitable habitat of fish below the dam is a necessary and urgent task in the construction of ecological water conservancy.

[0003] Most of the traditional ecological dispatching technologies for mitigating the impact of supersaturated TDG on fish during reservoir flood discharge only consider the impact of the saturation of supersaturated TDG on fish, ignoring the tolerance of fish to supersaturated TDG water bodies, that is, within a certain exposure time, the physiological damage of fish will recover after entering the water body without supersaturated TDG, which will lead to the loss of economic benefits of reservoir ecological dispatching and the reduction of ecological elasticity of the river channel below the dam. In addition, when solving the model, traditional methods rely more on experience and assumptions, and it is difficult to avoid the fish lethal zone. Problems such as "local optimality" and "overfitting" of the model often occur. In fact, within a certain period of time, fish have a certain tolerance to the saturation of supersaturated TDG water bodies. When fish exposed to supersaturated TDG water bodies return to fresh water, the physiological damage caused by supersaturated TDG to fish can be gradually restored. Therefore, the tolerance of fish to supersaturated TDG is incorporated into the multi-objective ecological scheduling of reservoirs as a constraint condition. Through intelligent algorithms, the reservoir flood discharge process is optimized to reduce the water ecological impact of reservoir flood discharge. This will enable efficient and high-quality decision-making on reservoir flood discharge management, mitigate the impact of supersaturated TDG caused by flood discharge on fish below the dam, and improve the economic and ecological benefits of the reservoir.

[0004] The combination of pulse flooding ideas and artificial intelligence technology is rarely reported in reservoir and dam water environment management. Only some scholars have proposed to use pulse floods to generate artificial flood peaks to help the four major carps spawn; use neural network methods to predict the evolution of hydrology and water environment; and use optimization algorithms such as genetic algorithms and ant colony algorithms to regulate the hydrological and hydrodynamic processes of reservoirs to ensure the multi-objective realization of the social, economic and ecological benefits of dam operation. Technical research on multi-objective intelligent scheduling methods for reservoirs that combine the physiological characteristics of fish with the tolerance of supersaturated TDG or other pollutants has not been reported. Therefore, the construction of a reservoir pulse flood intelligent scheduling method that mitigates the impact of supersaturated TDG on fish through artificial intelligence has broad application fields and prospects. Therefore, a pulse flood scheduling method that mitigates the impact of supersaturated total dissolved gas on fish is urgently needed. Summary of the invention

[0005] In view of the prior art, the present invention provides a pulse flood scheduling method for alleviating the impact of supersaturated total dissolved gas on fish.

[0006] The technical solution adopted by the present invention is as follows:

[0007] S1: Construct a multi-objective ecological scheduling model. The model is as follows:

[0008]

[0009] The objective function is to maximize power generation and minimize the ecological impact of oversaturated TDG. The constraints include reservoir water level constraint, downstream river water level constraint, flood discharge flow constraint, power generation flow constraint, power generation constraint, oversaturated TDG saturation constraint, and oversaturated TDG retention time constraint.

[0010] S2: solve the multi-objective ecological scheduling model obtained by S1;

[0011] S3: The distribution of supersaturated TDG in the reservoir area downstream of the dam is obtained through the results of S2, and scheduling is carried out.

[0012] Furthermore, the supersaturated TDG residence time constraint in S1 is:

[0013]

[0014]

[0015] Furthermore, the reservoir water level constraint in S1 is:

[0016]

[0017] The water level constraint of the river channel downstream of the dam in S1 is:

[0018]

[0019] The flood discharge constraint in S1 is:

[0020]

[0021] The power generation flow constraint in S1 is:

[0022]

[0023] The power generation constraint in S1 is:

[0024] 0≤P≤power (8);

[0025] The supersaturated TDG saturation constraint in S1 is:

[0026] C min ≤C≤C max (9);

[0027] Furthermore, the solution formula for the reservoir water level and the river water level below the dam described in S1 is:

[0028] H up (t) = -3E-07·(Q s +Q p -Q in ) 2 -0.0004·(Q s +Q p -Q in )+564.8 (10);

[0029] In the formula, Q in is the inflow flow rate, Q s is the flood discharge, Q p is the power generation flow; H up (t) is the reservoir water level;

[0030] H down (t) = 0.0011·(Q s +Q p )+372.7 (11)

[0031] Q r =Q s +Q p (12)

[0032] In the formula, Q r is the flow rate of the river downstream of the dam, which is obtained by summing the dam flood discharge and the power generation flow; H down (t) is the water level of the river downstream of the dam.

[0033] Furthermore, the solution formula for the power generation in S1 is:

[0034] P(t)=min[g·η p Q p (t)(H up (t)-H down (t)) / 10 6 ,power] (13)

[0035] Where power is the installed capacity of the hydropower station; H up is the reservoir water level; H down is the water level of the river downstream of the dam; Q p is the reservoir power flow; g is the gravitational acceleration; η p is the power generation efficiency of the reservoir;

[0036] And the supersaturated TDG generation prediction model, the model is:

[0037] C(t)=Funtion(X1+…+X7, f, σ) (14)

[0038] Where C(t) is the predicted value of the supersaturated TDG generation saturation obtained based on the ESWT-LSTM deep learning method; X is the high / low frequency component feature set of the model input variables, f is the forget gate; σ is the activation function; when the reservoir does not discharge flood water, the supersaturated TDG saturation is equal to 100%.

[0039] Furthermore, the solution steps in S2 are to first set the initial parameters of the NSGA-II algorithm; then randomly generate the initial population with the reservoir water level in the optimization period as the constraint condition; calculate the operation of the dam under the current population conditions, and determine whether the calculation results meet the constraints given in S1. If not, regenerate the initial population until a parent population P containing N that meets the constraints in S1 is generated. t ; Obtain N offspring groups P that satisfy the S1 constraint condition through arithmetic crossover c ; Obtain N offspring populations P that satisfy the S1 constraint condition through random uniform mutation m ; Then calculate the objective function value, perform non-dominated sorting and crowding distance calculation on the group according to the objective function value, and judge whether the termination condition is met, that is, whether the number of iterations is greater than the maximum number of iterations. If it is met, the result is output, otherwise, the arithmetic crossover step is continued.

[0040] Technical Effects

[0041] (1) The time series feature extraction coupled with LSTM deep learning method is used to accurately identify and enhance the characteristic variables affecting the saturation of TDG generated by dam flood discharge, improve the prediction accuracy of the TDG generation prediction model for dam flood discharge, and solve the problems of difficult parameter calibration, slow calculation speed, complex model structure, high calculation cost and difficult identification of key characteristic variables in traditional mechanism models.

[0042] (2) Based on the fish's own physiological characteristics and tolerance to supersaturated TDG, an ecological scheduling model is constructed to generate supersaturated TDG clusters, increase the area of ​​suitable fish habitat, improve the power generation of the dam operation and the resilience of the water ecosystem below the dam, and solve the problems of low economic benefits and weak ecological resilience of the traditional model.

[0043] (3) The NSGA-Ⅱ algorithm is combined with the reservoir scheduling problem of mitigating the ecological impact of oversaturated TDG, and a reservoir ecological scheduling method based on the NSGA-Ⅱ intelligent algorithm is proposed to solve the problems of slow decision-making speed, low decision-making accuracy and few decision-making options of the traditional model, which can provide more decision-making space for dam managers. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 Schematic diagram of the algorithm flow of embodiment 1 of the present invention.

[0045] Figure 2 This is a schematic diagram of correlation analysis of Example 1 of the present invention.

[0046] Figure 3 This is a schematic diagram of the high / low frequency decomposition of the key influencing factor data of Example 1 of the present invention.

[0047] Figure 4 This is a diagram of the multi-objective ecological scheduling optimization results of Example 1 of the present invention.

[0048] Figure 5 This is a rendering result diagram of the supersaturated TDG distribution characteristics after the optimization process of Example 1 of the present invention. DETAILED DESCRIPTION

[0049] The following will be combined with Example 1 of the present invention and the attached Figure 1-5 , the technical solution of the present invention is clearly and completely described. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0050] Example 1

[0051] S1: First, calibrate the correlation between the reservoir water level and flow, take Xiluodu as the research object, collect reservoir water level and reservoir flow data, analyze the reservoir flow balance relationship, and establish the reservoir flow balance equation; the specific formula is as follows:

[0052] Q w =Q s +Q p -Q in (15)

[0053] In the formula, Q w is the tailwater flow rate, Q in is the inflow flow rate, Q s is the flood discharge, Q p is the power generation flow;

[0054] Then, according to the reservoir water level data and reservoir flow data, the corresponding relationship between the reservoir water level data and the reservoir flow data is established through the least squares method to determine the reservoir water level-flow balance equation; the equation is as follows:

[0055] H up (t) = -3E-07·(Q s +Q p -Q in ) 2 -0.0004·(Q s +Q p -Q in )+564.8 (10)

[0056] In the formula, H up (t) is the reservoir water level;

[0057] The correlation between the water level and flow rate of the river downstream of the dam was determined, the water level data of the river downstream of the Xiluodu Dam and the flood discharge and power generation flow data of the dam were collected, the flow balance relationship of the river downstream of the dam was analyzed, and the flow balance equation of the river downstream of the dam was established; the equation is as follows:

[0058] Q r =Q s +Q p (12)

[0059] Where Qs is the flood discharge, Q p is the power generation flow, Qr is the river flow below the dam, which is obtained by summing the dam discharge flow and the power generation flow;

[0060] According to the water level data and flow data of the river channel below the dam, the corresponding relationship between the water level data and flow data of the river channel below the dam is established, that is, the water level-flow balance equation of the river channel below the dam: The equation is as follows:

[0061] H down (t) = 0.0011·(Qs +Q p )+372.7 (11)

[0062] In the formula, H down (t) is the water level of the river downstream of the dam;

[0063] Construct a power generation prediction model. The power generated by the hydropower station's turbine generator set is called output, that is, the water energy per unit time. The formula is as follows:

[0064] P(t)=g·η p Q p (t)·(H up (t)-H down (t)) / 10 6 (13)

[0065] Where g is the acceleration due to gravity; η p is the power generation efficiency of the reservoir; Qp is the power generation flow of the reservoir; H up (t)-H down (t) is the hydraulic head; Hup is the reservoir water level; Hdown is the river water level below the dam;

[0066] The output of a hydropower station is not only restricted by the flow rate and water head, but also by the rated capacity of the turbine generator set. Then the minimum value of the output and the rated capacity of the turbine generator set is taken; the prediction model is as follows:

[0067] P(r)=min[g·η p Q p (t)·(H up (t)-H down (t)) / 10 6 ,power] (13)

[0068] Where power is the installed capacity of the hydropower station; H up is the water level in the reservoir area; Hdown is the water level in the river downstream of the dam; Q p is the reservoir power flow; g is the gravitational acceleration; η p is the power generation efficiency of the reservoir;

[0069] Identification and enhancement of the influencing factors of the saturation characteristics of supersaturated TDG generated by dam flood discharge. First, collect monitoring data such as water temperature, reservoir water level, river water level, inflow flow, outflow flow, power generation flow, flood discharge flow, supersaturated TDG saturation, and normalize the relevant data; through correlation analysis, determine the correlation between the saturation of supersaturated TDG generated by dam flood discharge and environmental factors such as water level, flow, and water temperature, and calculate the correlation coefficient, see Figure 2As shown in the figure; sort the correlation coefficients, select the environmental factors with larger correlation coefficients as the key influencing factors affecting the saturation of supersaturated TDG generation; decompose the key influencing factor data affecting the saturation of supersaturated TDG generation into low-frequency signals representing the overall trend and high-frequency signals capturing short-term fluctuations through the Extended Stationary Wavelet Transform (ESWT) method, so as to reveal the entanglement relationship of multiple factors with different frequencies in the key influencing factors, realize the enhancement of the characteristic values ​​of the key influencing factors, and improve the prediction accuracy of the saturation of supersaturated TDG generation;

[0070] The mathematical expression of the ESWT model is:

[0071]

[0072] In the formula, A it It is an approximation of the low-frequency information set and can capture the overall trend characteristics of the original data; D it is the high-frequency information set, indicating high-frequency fluctuations, i.e., the short-term mutation characteristics of the original data; m is the decomposition scale, i.e., the number of decomposition levels, and m is 3; the decomposition results of the high- and low-frequency characteristics of the key influencing factors are shown in Figure 3 As shown, first, the original data is decomposed into a low-frequency component A1 and a high-frequency component D1; then, the A1 data is further decomposed into a low-frequency component A1-A1 and a high-frequency component A1-D1 through a filter, and the obtained low-frequency component A1-A1 is further decomposed into a low-frequency component A1-A2 and a high-frequency component A1-D2; at the same time, the high-frequency component D1 is decomposed synchronously according to the same method to obtain high-frequency signal decomposition amounts D1-A2 and D1-D2;

[0073] Based on the ESWT-LSTM deep learning method, the present invention constructs a saturation prediction model for dam flood discharge supersaturation TDG generation; wherein the model input layer is X=(x t-1 , x t-2 ,...x t-n ), represents the input feature vector set of n moments before moment t, that is, the high / low frequency component features of the data at the past moments, the data at the tth moment is the label data, the input data is obtained according to the correlation analysis results, including: normalized outflow-inflow, power generation, flood discharge, inflow, outflow, and water level data; the output data represents the predicted value of the TDG saturation of the LSTM model at the tth moment; the model determines the optimal parameter set through multiple experiments, including: training step, number of iterations, learning rate, loss function form, etc.; 80% of the time series data is used as model training samples, and 20% of the time series data is used as verification data to evaluate the accuracy of the established model. The model is:

[0074] C(t)=Funtion(X1…+X7, f, σ) (14)

[0075] Where C(t) is the predicted value of the saturation generated by the supersaturated TDG; X is the high / low frequency component feature set of the model input variables (i.e. Figure 3 The decomposition results of ESWT model data in the middle), including: the high / low frequency component feature set of the difference between outflow and inflow, the high / low frequency component feature set of power generation flow, the high / low frequency component feature set of flood discharge flow, the high / low frequency component feature set of inflow flow, the high / low frequency component feature set of total outflow flow, the high / low frequency component feature set of upstream and downstream water levels, and the high / low frequency component feature set of water level difference; f is the forget gate; σ is the activation function; when the reservoir does not discharge flood water, the supersaturated TDG saturation is equal to 100%;

[0076] Determine the tolerance characteristics of fish to supersaturated TDG. Fish have a certain tolerance to supersaturated TDG water. The tolerance capacity is expressed as LT50, which is the time required for half of the fish to die in a TDG water body with a specific saturation. Because when fish exposed to supersaturated TDG water bodies return to fresh water, the physiological damage caused by supersaturated TDG to fish can be gradually restored. Therefore, considering the tolerance of fish to supersaturated TDG water bodies, incorporating the tolerance time of fish to supersaturated TDG water bodies as a constraint condition into the multi-objective ecological operation of the reservoir will alleviate the impact of flood discharge supersaturated TDG on fish below the dam.

[0077] The closing / stopping time of the dam flood discharge under different TDG saturation conditions is quantified, and the closing / stopping time of the dam flood discharge under different TDG saturation conditions is determined in combination with the fish's own physiological characteristics and tolerance to supersaturated TDG. In this embodiment, the rock carp is taken as an example. When the rock carp is exposed to 120% saturation TDG water for 14 hours and then exposed to fresh water for 7 hours, the activity can be guaranteed. Then, the closing / stopping time of the dam flood discharge under different TDG saturation conditions is determined to reduce the impact of supersaturated TDG on fish. The mathematical expression is as follows:

[0078]

[0079]

[0080] Where, T open is the continuous flood discharge time (h); T close is the time after continuous flood discharge stops (h); C is the saturation of supersaturated TDG; the maximum continuous flood discharge time is LT 50 The value of LT 50 When the saturation of supersaturated TDG is between 110% and 120%, the continuous flood discharge time (T open ) should be less than 14 hours, and the subsequent stop flood discharge time (Tclose ) should be greater than 7 hours to ensure the normal survival of fish under the dam; when the supersaturated TDG saturation is lower than 110%, fish can survive safely for a long time; when it is higher than 120%, it will affect the physiological characteristics of fish, and the higher the saturation and the longer the exposure time, the more likely the fish will get sick and die; the greater the supersaturated TDG saturation, the shorter the flood discharge time should be;

[0081] A multi-objective ecological dispatch model with the goal of minimizing the ecological impact of oversaturated TDG and maximizing power generation is constructed. According to the multi-objective theory, the flood control, benefit and ecological objectives in the reservoir flood discharge dispatch are analyzed, and the maximum power generation and the minimum ecological impact of oversaturated TDG are selected as the main objectives. Other objectives are converted into constraints, and a multi-objective optimization model of reservoir power generation-oversaturated TDG is established. The objective function is:

[0082]

[0083] In the formula, the objective function is represented by F, and the minimum value of F is taken, Minmize(F); the oversaturated TDG saturation (%) predicted by the LSTM deep learning model will be subtracted by 1 to obtain the minimum target of TDG saturation, C(t)-1; the maximum power generation P(t) is divided by the installed capacity of the hydropower station (Power), and then transformed into the minimum target by (1-P(t) / Power), and the TDG target and the power generation target are coupled (a+b=1) through the multi-objective scheduling model weight coefficients a and b to realize the construction of the multi-objective model;

[0084] The constraints include:

[0085] The reservoir water level constraint is expressed as:

[0086]

[0087] The water level constraint of the river channel below the dam is expressed as:

[0088]

[0089] The flood discharge constraint is expressed as:

[0090]

[0091] The power generation flow constraint can be expressed as:

[0092]

[0093] The power generation constraint can be expressed as:

[0094] 0≤P≤power (8);

[0095] The saturation constraint of the supersaturated TDG can be expressed as:

[0096] C min ≤C≤C max (9);

[0097] The supersaturated TDG residence time constraint is:

[0098]

[0099]

[0100] The specific values ​​of the constraint conditions are shown in Table 1;

[0101] Table 1 Model boundary conditions

[0102]

[0103] The following formula is to solve the above constraints;

[0104] The reservoir water level and flow balance equation is as follows:

[0105] H up (t) = -3E-07·(Q s +Q p -Q m ) 2 -0.0004·(Q s +Q p --Q in )+564.8 (10);

[0106] The water level and flow balance equation of the river channel below the dam is:

[0107] H down (t) = 0.0011·(Q s +Q p )+372.7 (11);

[0108] The power generation prediction model is:

[0109] P(t)=min[g·η p Q p (t)·(H up (t)-H down (t)) / 10 6 , power] (13);

[0110] The supersaturated TDG generation prediction model is:

[0111] C(t)=Function(X1+...+X7, f, σ) (14).

[0112] S2: The NSGA-Ⅱ algorithm is used to solve the multi-objective ecological scheduling model. According to the characteristics of the Xiluodu project, two reservoir flood discharge scheduling schemes are designed, namely, a pulse flood scheduling scheme with equal weights for ecology and power generation (scenario 1) and a pulse flood scheduling scheme with power generation priority (scenario 2). The scheme settings are shown in Table 2:

[0113] Table 2 Scheduling scheme settings

[0114]

[0115] The sequence data of water level in the Xiluodu reservoir area during the dispatch period is used as the decision variable of the optimization algorithm, and each individual in the population is represented by Z = [Z1, Z2, ... Z n ],Z n is the water level of the Xiluodu reservoir at time t, and the dam flood discharge scheduling model constructed by S1 with the goal of minimizing the ecological impact of oversaturated TDG and maximizing power generation is solved by the NSGA-Ⅱ algorithm; the specific steps are as follows:

[0116] First, the initial parameters of the NSGA-Ⅱ algorithm are set, the population size is 20, the maximum number of iterations is 300, the crossover probability is 0.7, and the mutation rate is 0.1;

[0117] Initial population generation: Taking the reservoir water level in the optimization period as the constraint condition, randomly generate the initial population, calculate the flood discharge, power generation, TDG saturation, and upstream and downstream water levels of the dam under the current population conditions, and determine whether the calculation results meet the given constraints. If not, regenerate the initial population until a parent population P containing 20 that meets the S1 constraint conditions is generated. t ;

[0118] Through arithmetic crossover, 20 offspring groups P that meet the S1 constraint are obtained. c ;

[0119] Through random uniform mutation, 20 offspring populations P that meet the S1 constraint are obtained. m ;

[0120] Where P t , P c , P m It is an intermediate variable generated during the calculation process;

[0121] The parent population P t and the offspring population (P c ,P m) to obtain 60 mixed populations, calculate the objective function value, perform non-dominated sorting and crowding distance calculation on the population according to the objective function value, select 20 individuals from the mixed population as the parent population, and determine whether the termination condition is met, that is, whether the number of iterations is greater than the maximum number of iterations. The maximum number of iterations in Example 1 is 300; if the condition is met, the result is output, otherwise, the arithmetic crossover step is returned;

[0122] Output the multi-objective ecological dispatching results of the dam flood discharge with the least ecological impact and the largest power generation of the oversaturated TDG downstream of Xiluodu (characterized by flood discharge flow, power generation flow, reservoir water level, river water level, TDG saturation and power generation); when the flood discharge flow, power generation flow, reservoir water level and river water level are within the threshold range of the multi-objective optimization model constraint conditions (Equations 4, 5, 6 and 7), it means that the reservoir can operate safely and meet the basic safety requirements of reservoir dispatching; when the oversaturated TDG retention time, that is, the time when the TDG saturation is higher than 100% and the time when it is equal to 100% are within the flood discharge start / stop time constraint threshold range (Equations 2 and 3), the ecological target requirements of reservoir dispatching are met; when the power generation generated after the model optimization is higher than the power generation without optimization, the power generation target (that is, economic benefit) of reservoir dispatching is met; the results are as follows Figure 4 shown.

[0123] S3: The optimized flow process, water level process and supersaturated TDG generation saturation are loaded into water environment models such as MIKE 21 and EFDC as boundary conditions, and the distribution of supersaturated TDG in the reservoir or river downstream of the dam after optimization is rendered. The results are as follows: Figure 5 As shown;

[0124] Combination Figure 5 The rendering results show that the constructed model can effectively generate a pulse flood discharge mode, shorten the retention time of supersaturated TDG in the river channel below the dam, reduce the ecological impact of dam flood discharge, and increase the total power generation; the supersaturated TDG generated by Scheme 1 (Scenario 1) has a saturation higher than 130% for only 4 hours, and the subsequent stop flood discharge time can ensure that the fish exposed to the supersaturated TDG water body can recover normal physiological functions. The generated power generation is about 12400MW, which is much higher than the power generation generated by the original flood discharge scheme (unoptimized); the supersaturated TDG generated by Scheme 2 (Scenario 2) has a saturation higher than 130% for 16 hours, which is 4 times that of Scheme 1, but the stop flood discharge time can still ensure the normal survival of fish. The power generation generated by Scheme 2 is about 12600MW, which is much higher than the economic benefits generated by the original flood discharge scheme; The results show that through multi-objective ecological scheduling, the generation of pulse flood scheduling schemes can greatly shorten the retention time of high-saturation TDG in the river channel, provide sufficient time for the recovery of fish physiological functions, and generate higher ecological benefits.

Claims

1. A pulse flood dispatching method for mitigating the impact of supersaturated total dissolved gas on fish, characterized in that: The following steps are involved: S1: Construct a multi-objective ecological scheduling model. The model is as follows: In the formula, the objective function is F, power is the installed capacity of the hydropower station, P(t) is the predicted value of power generation, a and b are the weight coefficients of the ecological objective and power generation objective of the multi-objective ecological scheduling model a+b=1, C(t) is the generation saturation of the oversaturated TDG at time t, and NT is the total time step of the optimization model; The objective function is to maximize power generation and minimize the ecological impact of oversaturated TDG. The constraints include reservoir water level constraint, downstream river water level constraint, flood discharge flow constraint, power generation flow constraint, power generation constraint, oversaturated TDG saturation constraint, and oversaturated TDG retention time constraint. S2: solve the multi-objective ecological scheduling model obtained by S1; S3: The flow process of the dam flood discharge and the supersaturated TDG generation saturation of the dam flood discharge are obtained through the results of S2, and ecological scheduling is carried out; The supersaturated TDG residence time constraint in S1 is: Where, T open is the continuous flood discharge time; T close is the stopping time of flood discharge after the continuous flood discharge stops; C is the saturation of supersaturated TDG; The reservoir water level constraint, the downstream river water level constraint, the flood discharge flow constraint, the power generation flow constraint, the power generation constraint, and the oversaturated TDG saturation constraint in S1 are: Reservoir water level constraints: In the formula, is the minimum reservoir water level, H up is the reservoir water level, is the maximum reservoir water level; The water level constraint of the river channel downstream of the dam in S1 is: In the formula, is the minimum water level in the river below the dam, H dowm is the water level of the river downstream of the dam, is the maximum downstream river water level; The flood discharge constraint in S1 is: In the formula, is the minimum flood discharge, Q s is the flood discharge, is the maximum flood discharge; The generation flow constraint in S1 is: In the formula, is the minimum power generation flow, Q p is the power generation flow, is the maximum power generation flow; The power generation constraint in S1 is: 0≤P≤power (8) In the formula, power is the installed capacity of the hydropower station, and P is the power generation; The supersaturated TDG saturation constraint in S1 is: C min ≤C≤C max (9) In the formula, C min is the minimum supersaturated TDG saturation, C is the supersaturated TDG saturation, C max is the maximum supersaturated TDG saturation.

2. A pulse flood dispatching method for mitigating the impact of supersaturated total dissolved gas on fish according to claim 1, characterized in that: The solution formula for the reservoir water level and the river water level below the dam described in S1 is: H up (t)=-3E-07·(Q s +Q p -Q in ) 2 -0.0004·(Q s +Q p -Q in )+564.8 (10) In the formula, Q in is the inflow flow rate, Q s is the flood discharge, Q p is the power generation flow; H up (t) is the reservoir water level; H down (t)=0.0011·(Q s +Q p )+372.7 (11) Q r =Q s +Q p (12) In the formula, Q r is the flow rate of the river downstream of the dam, which is obtained by summing the dam flood discharge and the power generation flow; H down (t) is the water level of the river downstream of the dam.

3. A pulse flood dispatching method for mitigating the impact of supersaturated total dissolved gas on fish according to claim 1, characterized in that: The solution formula for the power generation in S1 is: P(t)=min[g·η p ·Q p (t)·(H up (t)-H down (t)) / 106,power] (13) Where power is the installed capacity of the hydropower station; H up is the reservoir water level; H down is the water level of the river downstream of the dam; Q in is the reservoir inflow; Q s and Q p are the reservoir flood discharge and power generation flow respectively; g is the gravitational acceleration; η p is the power generation efficiency of the reservoir; when the reservoir is not discharging flood water, the saturation of the supersaturated TDG is equal to 100%; And the formula for supersaturated TDG saturation: C(t)=Funtion(X1+...+X7, f, σ) (14) Where C(t) is the predicted value of the saturation generated by the oversaturated TDG; X is the high / low frequency component feature set of the model input variable, f is the forget gate; and σ is the activation function.

4. A pulse flood dispatching method for mitigating the impact of supersaturated total dissolved gas on fish according to claim 1, characterized in that: The solution steps in S2 are to first set the initial parameters of the NSGA-Ⅱ algorithm; Then, the reservoir water level in the optimization period is used as a constraint condition to randomly generate the initial population, and the population size is set to N, with the value of N ranging from 20 to 500; Calculate the data of the dam under the current population conditions and determine whether the calculation results meet the constraints given in S1. If not, regenerate the initial population until a parent population P containing N people that meet the constraints in S1 is generated. t ; Through arithmetic crossover, we get N offspring groups P that satisfy the S1 constraint. c ; Obtain N offspring populations P that satisfy the S1 constraint condition through random uniform mutation m ; Then calculate the objective function value, perform non-dominated sorting and crowding distance calculation on the group according to the objective function value, and judge whether the termination condition is met, that is, whether the number of iterations is greater than the maximum number of iterations. If it is met, the result is output, otherwise, the arithmetic crossover step is continued.

Images