Method and device for evaluating greenhouse effect of reservoir based on physical driving neural network

Abstract

The embodiment of the application discloses a reservoir greenhouse effect evaluation method based on a physical driving neural network, and the method comprises the following steps: collecting historical data of a target reservoir monitoring station; establishing a river reservoir one-dimensional nitrogen element transport model based on the nitrogen element transport process of the target reservoir; determining a nitrification and denitrification coefficient calibration neural network in the nitrogen element transport process of the target reservoir according to the river reservoir one-dimensional nitrogen element transport model and a corresponding to-be-optimized objective function; inputting the historical data into the to-be-optimized objective function to determine neural network optimization parameters, thereby determining nitrification and denitrification optimization coefficients; and evaluating the greenhouse gas emission of the target reservoir based on the real-time data of the nitrogen element of the target reservoir and the nitrification and denitrification optimization coefficients. The embodiment of the application accurately evaluates the greenhouse gas emission of the reservoir by utilizing the one-dimensional nitrogen element transport equation, combining the actual aquatic environment of the reservoir, and increasing the physical constraint of the nitrification and denitrification coefficient calibration process.

Description

TECHNICAL FIELD

[0001] The embodiment of the present application relates to the technical field of reservoir greenhouse effect evaluation, and in particular to a reservoir greenhouse effect evaluation method based on a physically driven neural network and a reservoir greenhouse effect evaluation device based on a physically driven neural network. BACKGROUND

[0002] Global annual "active" nitrogen caused by human activities leads to serious imbalance of global nitrogen cycle, and causes a series of environmental problems such as water body eutrophication, water body acidification, and greenhouse gas emission. As an important nitrogen sink, the nitrogen cycle based on the nitrification and denitrification of nitrogen elements in rivers and reservoirs is of great significance to the nitrogen budget of the entire ecological system, water body nitrogen pollution, and greenhouse gas emission. The nitrification and denitrification coefficient of the nitrogen element transport process in rivers and reservoirs is an important parameter for reflecting the process, which is mainly affected by water temperature, dissolved oxygen, microbial content, flow rate, and solute interaction. The nitrification and denitrification coefficient is generally obtained by laboratory measurement after water sampling or by inversion of the transport equation.

[0003] The traditional method does not consider the complexity and non-uniformity of the nitrogen transport process in the reservoir, does not evaluate the greenhouse gas emission of the reservoir by a physically constrained method, and cannot accurately evaluate the greenhouse gas emission of the reservoir.

[0004] Therefore, how to improve the accuracy of the evaluation of the nitrogen cycle in rivers and reservoirs and accurately evaluate the greenhouse gas emission of the reservoir is a difficult problem to be solved. SUMMARY

[0005] In view of the above problems, a reservoir greenhouse effect evaluation method based on a physically driven neural network and a reservoir greenhouse effect evaluation device based on a physically driven neural network are provided to overcome the above problems or at least partially solve the above problems.

[0006] According to a first aspect of the embodiment of the present application, a reservoir greenhouse effect evaluation method based on a physically driven neural network is provided, and the method comprises:

[0007] Collecting historical data of a target reservoir monitoring site;

[0008] Establishing a one-dimensional nitrogen element transport model of rivers and reservoirs based on a nitrogen element transport process of the target reservoir;

[0009] Determining a nitrification and denitrification coefficient calibration neural network in the nitrogen element transport process of the target reservoir according to the one-dimensional nitrogen element transport model of rivers and reservoirs, and a corresponding to-be-optimized objective function;

[0010] Inputting the historical data into the to-be-optimized objective function, determining neural network optimization parameters, and thereby determining a nitrification and denitrification optimization coefficient;

[0011] predicting the greenhouse gas emission of the target reservoir based on the real-time data of nitrogen elements of the target reservoir and the nitrification-denitrification optimization coefficient, so as to evaluate the greenhouse effect of the target reservoir.

[0012] Further, the river reservoir one-dimensional nitrogen element transport model is established based on the nitrogen element transport process of the target reservoir, comprising:

[0013] establishing the correlation between the nitrogen elements and the greenhouse gases in the target reservoir based on the historical data;

[0014] establishing the river reservoir one-dimensional nitrogen element transport model based on the correlation and the nitrogen element transport process of the target reservoir.

[0015] Further, after the river reservoir one-dimensional nitrogen element transport model is established based on the correlation and the nitrogen element transport process of the target reservoir, the method further comprises:

[0016] establishing a one-dimensional nitrogen element convection-diffusion equation based on the river reservoir one-dimensional nitrogen element transport model;

[0017] determining the boundary conditions of the nitrogen element convection-diffusion equation.

[0018] Further, the one-dimensional nitrogen element convection-diffusion equation is established based on the river reservoir one-dimensional nitrogen element transport model, comprising:

[0019] establishing the one-dimensional nitrogen element convection-diffusion equation based on the river reservoir one-dimensional nitrogen element transport model according to the following formula one;

[0020] Formula one is:

[0021] ；

[0022] wherein: is the current time; is the spatial position; is the flow parameter; is the diffusion coefficient, which describes the diffusion ability of nitrogen elements in the reservoir; is the concentration of nitrate in the reservoir river; R is the nitrification-denitrification coefficient of water body, which represents the nitrification-denitrification rate of water body and reflects the activity of plankton and microorganisms in water body, which is affected by the oxygen content of water body.

[0023] Further, the boundary conditions include upper boundary conditions and lower boundary conditions.

[0024] The determination of the boundary conditions of the nitrogen element convection-diffusion equation comprises:

[0025] determining the upper boundary condition as an inflow nitrogen element concentration of the target reservoir;

[0026] determining the lower boundary condition as an outflow nitrogen element concentration of the target reservoir.

[0027] Further, the neural network for calibrating the nitrification-denitrification coefficient in the nitrogen element transport process of the target reservoir according to the river-reservoir one-dimensional nitrogen element transport model and the corresponding objective function to be optimized comprise:

[0028] According to the river-reservoir one-dimensional nitrogen element transport model, the neural network for calibrating the nitrification-denitrification coefficient in the nitrogen element transport process of the target reservoir is determined according to the following formula two and formula three;

[0029] Formula two is:

[0030] ；

[0031] Wherein: is a residual error of the river-reservoir one-dimensional nitrogen element transport model; is an approximation of a real nitrogen element concentration; is a current time; is a spatial position, is a flow parameter; is a diffusion coefficient; is an approximation of a real nitrification-denitrification coefficient.

[0032] Formula three is:

[0033] ；

[0034] Wherein: is an approximation error of the neural network output to an observed cross-section nitrogen element; is an approximation of a real nitrogen element concentration; is an observed cross-section nitrogen element concentration.

[0035] Further, the neural network for calibrating the nitrification-denitrification coefficient in the nitrogen element transport process of the target reservoir according to the river-reservoir one-dimensional nitrogen element transport model and the corresponding objective function to be optimized comprise:

[0036] The objective function to be optimized is determined according to the following formula four;

[0037] Formula four is:

[0038]

[0039] Wherein: is an objective function to be optimized; is an approximation of the real nitrification-denitrification coefficient is a fitting model parameter; is an approximation of the real nitrogen element concentration is a fitting model parameter; is a residual of the river-reservoir one-dimensional nitrogen transport model; is an approximation error of the neural network output to the observed cross-section nitrogen element; is a square of the corresponding parameter two-norm.

[0040] Further, the method further comprises:

[0041] inputting the historical data into the to-be-optimized objective function;

[0042] solving the to-be-optimized objective function based on a gradient descent method to obtain the neural network optimization parameter;

[0043] determining the nitrification-denitrification optimization coefficient based on the neural network optimization parameter.

[0044] Further, the method further comprises:

[0045] calculating the gradient of each parameter in the to-be-optimized objective function;

[0046] updating the parameter value of the neural network according to the gradient information of the gradient;

[0047] judging whether the parameter value of the neural network satisfies a convergence condition, and if the parameter value of the neural network satisfies the convergence condition, stopping iteration to obtain the neural network optimization parameter.

[0048] Further, the method further comprises:

[0049] drawing a two-dimensional dot matrix based on the real-time nitrogen element data of the target reservoir and the nitrification-denitrification optimization coefficient;

[0050] predicting the greenhouse gas emission amount of the target reservoir based on the two-dimensional dot matrix to evaluate the greenhouse effect of the target reservoir.

[0051] Further, the method further comprises:

[0052] inputting the nitrification-denitrification optimization coefficient into a one-dimensional nitrogen element convection-diffusion equation;

[0053] solving the one-dimensional nitrogen element convection-diffusion equation based on an implicit difference method to obtain the nitrogen element content of the target reservoir after a preset time;

[0054] determining the loss amount of the nitrogen element of the target reservoir based on the nitrogen element content of the target reservoir after the preset time and the nitrification-denitrification optimization coefficient;

[0055] predicting the greenhouse gas emission of the target reservoir based on the loss amount and the two-dimensional dot matrix, and evaluating the greenhouse effect of the target reservoir.

[0056] Further, the historical data of the target reservoir monitoring station is collected, including:

[0057] The historical data of the monitoring station set up on the trunk stream of the target reservoir area is collected to obtain the historical data of the monitoring station set up on the trunk stream of the target reservoir area.

[0058] The historical data at least includes: flow, average flow rate, water temperature, nitrogen element content, dissolved oxygen, microbial content, and greenhouse gas absorption amount.

[0059] According to a second aspect of the present application, a reservoir greenhouse effect evaluation device based on a physically driven neural network is provided, and the device comprises:

[0060] a data collection module for collecting historical data of a target reservoir monitoring station;

[0061] a first model establishment module for establishing a river reservoir one-dimensional nitrogen element transport model based on the nitrogen element transport process of the target reservoir;

[0062] a neural network establishment module for determining a nitrification-denitrification coefficient rate setting neural network in the nitrogen element transport process of the target reservoir according to the river reservoir one-dimensional nitrogen element transport model, and a corresponding to-be-optimized objective function;

[0063] an optimization coefficient determination module for inputting the historical data into the to-be-optimized objective function to determine a neural network optimization parameter, thereby determining a nitrification-denitrification optimization coefficient;

[0064] an evaluation module for predicting the greenhouse gas emission amount of the target reservoir based on the real-time data of the nitrogen element of the target reservoir and the nitrification-denitrification optimization coefficient, thereby evaluating the greenhouse effect of the target reservoir.

[0065] According to a third aspect of the present application, an electronic device is provided, and the electronic device comprises:

[0066] The processor, the memory, and a computer program stored on the memory and capable of running on the processor, when executed by the processor, implement the method for evaluating the greenhouse effect of a reservoir based on a physically driven neural network according to any one of the preceding.

[0067] According to a fourth aspect of the present application, a computer-readable storage medium is provided, and the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method for evaluating the greenhouse effect of a reservoir based on a physically driven neural network according to any one of the preceding is implemented.

[0068] The technical solution provided by the embodiments of the present application can include the following beneficial effects.

[0069] The embodiments of the present application disclose a method for evaluating the greenhouse effect of a reservoir based on a physically driven neural network, and the method comprises the following steps: collecting historical data of a target reservoir monitoring site; establishing a one-dimensional nitrogen element transport model of a river reservoir based on a nitrogen element transport process of the target reservoir, determining a nitrification and denitrification coefficient calibration neural network in the nitrogen element transport process of the target reservoir according to the one-dimensional nitrogen element transport model of the river reservoir, and a corresponding to-be-optimized objective function, inputting the historical data into the to-be-optimized objective function, determining neural network optimization parameters, and determining nitrification and denitrification optimization coefficients; and evaluating the greenhouse gas emission of the target reservoir based on real-time data of nitrogen elements of the target reservoir and the nitrification and denitrification optimization coefficients. The embodiments of the present application increase the physical constraints of the nitrification and denitrification coefficient calibration process by using the one-dimensional nitrogen element transport equation and combining the actual aquatic environment of the reservoir, and then accurately evaluate the greenhouse gas emission of the reservoir. BRIEF DESCRIPTION OF DRAWINGS

[0070] The accompanying drawings, which form a part of the present application, are intended to provide further understanding of the present application, and the illustrative embodiments of the present application and their description serve the purpose of explaining the present application. The accompanying drawings do not constitute an inappropriate limitation on the present application.

[0071] In order to more clearly illustrate the technical solutions of the present application, the accompanying drawings needed to be used in the description of the present application will be briefly introduced as follows. Obviously, the accompanying drawings in the following description are only some embodiments of the present application, and other accompanying drawings can be obtained by those skilled in the art without any creative labor.

[0072] Figure 1 is a step flow chart of a method for evaluating the greenhouse effect of a reservoir based on a physically driven neural network according to an embodiment of the present application;

[0073] Figure 2 is a schematic diagram of a nitrogen element transport process of a reservoir according to an embodiment of the present application;

[0074] Figure 3is a structural block diagram of a reservoir greenhouse effect evaluation device based on a physically driven neural network provided by an embodiment of the present application. DETAILED DESCRIPTION

[0075] It should be apparent that the described embodiments are only some but not all of the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present application.

[0076] Global nitrogen cycle is seriously imbalanced due to the annual increase of "active" nitrogen caused by human activities, which leads to a series of environmental problems such as water body eutrophication, water body acidification, and greenhouse gas emission. As an important nitrogen sink, the nitrogen cycle based on nitrification and denitrification of nitrogen elements in rivers and reservoirs is of great significance to the nitrogen budget of the entire ecological system, water body nitrogen pollution, and greenhouse gas emission. The nitrification and denitrification coefficients in the nitrogen element transport process of rivers and reservoirs are important parameters for reflecting the process, which are mainly affected by water temperature, dissolved oxygen, microbial content, flow rate, and solute interaction. The coefficients are generally obtained by laboratory measurement after water sampling or by inversion of transport equation parameters.

[0077] The traditional method does not consider the complexity and non-uniformity of the nitrogen transport process in the reservoir, does not evaluate the greenhouse gas emission of the reservoir by a physically constrained method, and cannot accurately evaluate the greenhouse gas emission of the reservoir. Therefore, how to improve the accuracy of the evaluation of the nitrogen cycle of rivers and reservoirs and accurately evaluate the greenhouse gas emission of the reservoir is a difficult problem to be solved.

[0078] In view of the problems in the traditional method, the present application provides a reservoir greenhouse effect evaluation method and a reservoir greenhouse effect evaluation device based on a physically driven neural network. The method comprises the following steps: collecting historical data of a target reservoir monitoring station; establishing a one-dimensional nitrogen element transport model of rivers and reservoirs based on the nitrogen element transport process of the target reservoir; determining a nitrification and denitrification coefficient calibration neural network in the nitrogen element transport process of the target reservoir according to the one-dimensional nitrogen element transport model of rivers and reservoirs, and a corresponding optimization target function; inputting the historical data into the optimization target function to determine neural network optimization parameters, thereby determining a nitrification and denitrification optimization coefficient; and evaluating the greenhouse gas emission of the target reservoir based on real-time data of nitrogen elements of the target reservoir and the nitrification and denitrification optimization coefficient. The present application increases the physical constraint of the nitrification and denitrification coefficient calibration process by using the one-dimensional nitrogen element transport equation and combining the actual aquatic environment of the reservoir, thereby accurately evaluating the greenhouse gas emission of the reservoir.

[0079] REFERENCE Figure 1FIG. 1 shows a step flow chart of a method for evaluating the greenhouse effect of a reservoir based on a physically driven neural network according to an embodiment of the present application. The method can specifically include the following steps:

[0080] Referring to Figure 1 FIG. 2 shows a schematic diagram of a method for evaluating the greenhouse effect of a reservoir based on a physically driven neural network according to an embodiment of the present application. The method can specifically include the following steps:

[0081] Step 101: Collecting historical data of a target reservoir monitoring site;

[0082] By obtaining the monitoring stations set up on the trunk stream and branch stream of the target reservoir area, historical data on the trunk stream and branch stream of the target reservoir area where the monitoring stations are located can be collected, which can include data such as the current flow, average flow rate, water temperature, nitrogen content, dissolved oxygen, microbial content, and greenhouse gas absorption of the target reservoir.

[0083] Step 102: Establishing a one-dimensional nitrogen transport model of a river reservoir based on the nitrogen transport process of the target reservoir;

[0084] Referring to Figure 2 FIG. 3 shows a schematic diagram of the nitrogen transport process in a reservoir.

[0085] The nitrogen transport process in a reservoir is a dynamic process involving multiple links in an ecological system. First, nitrogen enters the reservoir through various pathways such as atmospheric deposition, surface runoff, groundwater recharge, and biological input. Nitrogen undergoes complex transformation processes within the riverbed of the target reservoir area, the water surface of the target reservoir, and the riverbed and water surface of the downstream river channel, including the conversion of ammonia nitrogen (NH3 and NH4+) to nitrate by nitrification, the conversion of nitrate to nitrogen gas (N2) under anaerobic conditions by denitrification, and the conversion of inorganic nitrogen to organic nitrogen by assimilation of aquatic plants and algae. These transformation processes are influenced by factors such as oxygen supply, temperature, and pH value. Nitrogen migration in the reservoir is mainly achieved through horizontal and vertical diffusion by water flow. Finally, nitrogen is output from the reservoir through water flow output at the outlet, adsorption by sediments, and migration or decomposition of organisms, which is affected by factors such as reservoir management, human activities, and climate change.

[0086] The management and control of the nitrogen transport process in a reservoir are crucial for protecting water quality and maintaining ecological balance. Through reasonable water resource management, agricultural practice improvement, sewage treatment, and ecological restoration, the input of nitrogen can be effectively reduced and the output of nitrogen can be improved, thereby reducing the nitrogen load of the reservoir.

[0087] Establishing a one-dimensional chlorine transport model of a river reservoir based on a topological structure can simulate the migration and transformation process of nitrogen in rivers and reservoirs.

[0088] Step 103, determining the nitrification-denitrification coefficient rate of the neural network in the nitrogen transport process of the target reservoir according to the one-dimensional nitrogen transport model of the river reservoir, and the corresponding optimization target function;

[0089] Nitrification-denitrification coefficients are parameters used to describe the rates of nitrification and denitrification in water bodies or soils. These coefficients are very important in environmental science and ecology because they affect nitrogen cycling and biogeochemical processes of nitrogen.

[0090] Nitrification refers to the process by which ammonia nitrogen (NH3 and NH4+) is oxidized to nitrite nitrogen (NO2-) and further oxidized to nitrate nitrogen (NO3-) under the action of microorganisms. The nitrification coefficient is usually expressed as the rate of conversion of ammonia nitrogen to nitrate nitrogen per unit time. This coefficient is influenced by factors such as temperature, pH, oxygen concentration, microbial community structure, and activity.

[0091] Denitrification refers to the process by which microorganisms reduce nitrate nitrogen to nitrogen gas (N2) and release it into the atmosphere under anaerobic conditions. The denitrification coefficient is usually expressed as the rate of conversion of nitrate nitrogen to nitrogen gas per unit time. This coefficient is also influenced by factors such as temperature, pH, oxygen concentration, microbial community structure, and activity.

[0092] In practical applications, nitrification-denitrification coefficients can be determined through laboratory culture experiments, field monitoring, and mathematical models. These coefficients are important for assessing nitrogen cycling in water bodies and soils, designing wastewater treatment systems, predicting the migration and transformation of nitrogen pollutants, and developing environmental management strategies.

[0093] For example, in wastewater treatment, understanding nitrification-denitrification coefficients can help engineers design more effective denitrification processes, such as biological nitrogen removal systems, to reduce nitrogen emissions and protect water environments. In agricultural management, understanding these coefficients can help farmers optimize fertilization strategies, reduce nitrogen loss, improve nitrogen utilization efficiency, and reduce negative environmental impacts.

[0094] In this embodiment, the nitrification-denitrification coefficient rate of the neural network in the nitrogen transport process of the target reservoir is determined based on the one-dimensional nitrogen transport model of the river reservoir, and the corresponding optimization target function. Compared with traditional methods, this embodiment requires less data and adds physical constraints, thereby improving the accuracy of the evaluation of the nitrogen cycle of the river reservoir and accurately evaluating the greenhouse gas emissions of the reservoir.

[0095] Step 104, inputting the historical data into the optimization target function to determine the neural network optimization parameters, thereby determining the nitrification-denitrification optimization coefficient;

[0096] Step 105, predicting the greenhouse gas emission of the target reservoir based on the nitrogen element real-time data of the target reservoir and the nitrification-denitrification optimization coefficient, so as to evaluate the greenhouse effect of the target reservoir.

[0097] In the embodiment, the historical data of the target reservoir monitoring station is collected; a river reservoir one-dimensional nitrogen element transport model is established based on the nitrogen element transport process of the target reservoir, a neural network for calibrating the nitrification-denitrification coefficient rate in the nitrogen element transport process of the target reservoir is determined according to the river reservoir one-dimensional nitrogen element transport model and the corresponding to-be-optimized objective function, the historical data is input into the to-be-optimized objective function, the optimization parameters of the neural network are determined, so as to determine the nitrification-denitrification optimization coefficient; and the greenhouse gas emission of the target reservoir is evaluated based on the nitrogen element real-time data of the target reservoir and the nitrification-denitrification optimization coefficient. The embodiment of the present application accurately evaluates the greenhouse gas emission of the reservoir by using the one-dimensional nitrogen element transport equation, combining the actual aquatic environment of the reservoir, increasing the physical constraint of the nitrification-denitrification coefficient calibration process, and then accurately evaluating the greenhouse gas emission of the reservoir.

[0098] In an embodiment of the present application, the historical data of the target reservoir monitoring station is collected, including:

[0099] The historical data of the monitoring station set up on the trunk stream of the target reservoir is collected.

[0100] The historical data at least includes: flow, average flow rate, water temperature, nitrogen element content, dissolved oxygen, microbial content, and greenhouse gas absorption amount.

[0101] In an embodiment of the present application, the river reservoir one-dimensional nitrogen element transport model is established based on the nitrogen element transport process of the target reservoir, including:

[0102] The correlation between the nitrogen element and the greenhouse gas absorption in the target reservoir is established based on the historical data.

[0103] The river reservoir one-dimensional nitrogen element transport model is established based on the correlation and the nitrogen element transport process of the target reservoir.

[0104] In an embodiment of the present application, after the river reservoir one-dimensional nitrogen element transport model is established based on the correlation and the nitrogen element transport process of the target reservoir, it further includes:

[0105] After the river reservoir one-dimensional nitrogen element transport model is established based on the correlation and the nitrogen element transport process of the target reservoir, it further includes:

[0106] The one-dimensional nitrogen element convection-diffusion equation is established based on the river reservoir one-dimensional nitrogen element transport model.

[0107] determine the boundary condition of the nitrogen element convection diffusion equation.

[0108] In this embodiment, existing data can be collected in advance to preliminarily establish the correlation between the nitrogen element and the greenhouse gas absorption of the target reservoir, and then the correlation between the nitrogen element and the greenhouse gas absorption of the target reservoir is determined according to the nitrogen element transport process of the target reservoir. Figure 2 The one-dimensional nitrogen element transport model of the river reservoir is established based on the topological structure of the nitrogen element transport process of the reservoir shown in FIG. 1. The one-dimensional nitrogen element transport model based on the topological structure can help scientists and engineers to understand the nitrogen cycle process, evaluate the source and influence of nitrogen pollution, and design effective pollution control and ecological restoration schemes.

[0109] In this embodiment, the one-dimensional nitrogen element convection diffusion equation is established based on the correlation between the nitrogen element and the greenhouse gas absorption of the target reservoir and the nitrogen element transport process of the target reservoir, and the one-dimensional nitrogen element convection diffusion equation is determined as a step function of space.

[0110] In an embodiment of the present application, the one-dimensional nitrogen element convection diffusion equation is established based on the one-dimensional nitrogen element transport model of the river reservoir, and includes:

[0111] The one-dimensional nitrogen element convection diffusion equation is established based on the one-dimensional nitrogen element transport model of the river reservoir according to the following formula one:

[0112] Formula one is:

[0113] ；

[0114] Wherein: is the current time; is the spatial position; is the flow parameter; is the diffusion coefficient, which describes the diffusion ability of the nitrogen element in the reservoir; is the concentration of nitrate in the river reservoir; R is the nitrification-denitrification coefficient of the water body, which represents the nitrification-denitrification rate of the water body and reflects the activity of plankton and microorganisms in the water body, which is affected by the oxygen content of the water body.

[0115] By establishing the one-dimensional nitrogen element convection diffusion equation, the space is divided into several intervals, and the nitrification-denitrification coefficient in each interval remains constant, so as to more accurately simulate and predict the transport and conversion process of the nitrogen element in the reservoir.

[0116] In an embodiment of the present application, the boundary condition includes an upper boundary condition and a lower boundary condition.

[0117] The determination of the boundary condition of the nitrogen element convection diffusion equation includes:

[0118] determining the upper boundary condition as the inflow nitrogen element concentration of the target reservoir;

[0119] determining the lower boundary condition as the outflow nitrogen element concentration of the target reservoir.

[0120] In the advection-diffusion equation, boundary conditions play a crucial role in defining the behavior of the system at the boundaries, directly affecting the solution of the equation.

[0121] The upper boundary condition is usually used to describe the material input or output at the highest point of the system, for example, if there is a fixed nitrogen element input at the highest point of the reservoir, it can be set as a Dirichlet boundary condition, specifying a fixed concentration value; if there is no material exchange at the highest point, it can be set as a Neumann boundary condition, specifying a zero flux.

[0122] The lower boundary condition is used to describe the material behavior at the lowest point of the system, for example, if the lowest point of the reservoir is completely closed and material cannot pass through, it can be set as a no-flux boundary condition; if there is material exchange at the lowest point, it can be set as a Robin boundary condition, combining concentration value and flux to describe the exchange process.

[0123] Correct selection and setting of the upper and lower boundary conditions can ensure that the solution of the equation accurately reflects the behavior of the system at the spatial boundary, and provides a reliable basis for the simulation and prediction of the system.

[0124] In this embodiment, the upper boundary condition can be a fixed boundary condition - the nitrogen element concentration of the inflow, and the lower boundary condition of the reservoir can be a free boundary condition with a fixed nitrogen element change rate, generally taking 0. For natural rivers, the lower boundary condition is usually also a fixed boundary condition - the nitrogen element concentration of the outflow.

[0125] In one embodiment of the present application, the neural network for calibrating the nitrification-denitrification coefficient in the nitrogen element transport process of the target reservoir according to the one-dimensional nitrogen element transport model of the river reservoir, and the corresponding optimization target function, include:

[0126] According to the one-dimensional nitrogen element transport model of the river reservoir, the neural network for calibrating the nitrification-denitrification coefficient in the nitrogen element transport process of the target reservoir is determined according to the following formula two and formula three;

[0127] Formula two is:

[0128] ；

[0129] Wherein: is the residual of the one-dimensional nitrogen element transport model of the river reservoir; is an approximation of the true nitrogen element concentration; is the current time; is the spatial position, is the flow parameter; is the diffusion coefficient; is an approximation of the true nitrification-denitrification coefficient.

[0130] Formula three is:

[0131] ；

[0132] wherein: is the approximation error of the neural network output to the observed cross-sectional nitrogen element; is an approximation of the true nitrogen element concentration; is the observed cross-sectional nitrogen element concentration.

[0133] According to the one-dimensional nitrogen element transport model of the target reservoir, a neural network for calibrating the nitrification-denitrification coefficient under the process of nitrogen transport in the reservoir is established, and the complex nonlinear relationship of the nitrification-denitrification process is learned from a large amount of historical data through a machine learning algorithm, so as to improve the accuracy and efficiency of parameter estimation.

[0134] Formula two is the residual error of the one-dimensional nitrogen transport model of the target reservoir, and the neural network model corresponding to the physical driving of the nitrification-denitrification coefficient, wherein the output of the network is is an approximation of the true nitrification-denitrification coefficient, and the model parameter to be fitted is denoted as . The output of the network is is an approximation of the true nitrogen element concentration, and the model parameter to be fitted is denoted as .

[0135] Formula three is the approximation error of the neural network output to the observed cross-sectional nitrogen element .

[0136] In an embodiment of the present application, the neural network for calibrating the nitrification-denitrification coefficient in the nitrogen transport process of the target reservoir is determined according to the one-dimensional nitrogen element transport model of the river reservoir, and the corresponding optimization target function comprises:

[0137] The optimization target function is determined according to formula three as follows:

[0138] Formula four is:

[0139] ；

[0140] wherein: is the optimization target function; is an approximation of the true nitrification-denitrification coefficient fitting model parameters; approximation of the real nitrogen element concentration fitting model parameters; residual of the river reservoir one-dimensional nitrogen transport model; approximation error of the neural network output to the observed cross-section nitrogen element; square of the corresponding parameter two-norm.

[0141] Formula four is to establish the corresponding to be optimized objective function, provide a clear quantitative standard for the optimization process, so as to minimize or maximize the objective function by adjusting the model parameters or decision variables, so as to achieve the optimal solution or satisfactory solution.

[0142] In this embodiment, the fitting model parameters and make its output match the actual observation data, thereby improving the accuracy of the model, revealing the system law, optimizing the system performance, and assisting decision support.

[0143] In an embodiment of the present application, the historical data is input into the to-be-optimized objective function, and the neural network optimization parameter is determined, so as to determine the nitrification and denitrification optimization coefficient, comprising:

[0144] The historical data is input into the to-be-optimized objective function;

[0145] The to-be-optimized objective function is solved based on a gradient descent method to obtain the neural network optimization parameter;

[0146] The nitrification and denitrification optimization coefficient is determined based on the neural network optimization parameter.

[0147] In this embodiment, the to-be-optimized objective function is solved by using a gradient descent method, wherein the basic idea of the gradient descent method is to gradually adjust the parameters along the negative gradient direction of the to-be-optimized objective function by iteration, so as to minimize the to-be-optimized objective function.

[0148] In an embodiment of the present application, the to-be-optimized objective function is solved based on a gradient descent method to obtain the neural network optimization parameter, comprising:

[0149] The gradient of each parameter in the to-be-optimized objective function is calculated;

[0150] The parameter value of the neural network is updated according to the gradient information of the gradient;

[0151] It is judged whether the parameter value of the neural network satisfies a convergence condition, and if the parameter value of the neural network satisfies the convergence condition, such as the parameter value update amplitude of the neural network being less than a preset threshold value or reaching a maximum iteration number, iteration is stopped, so that the neural network optimization parameter is obtained.

[0152] The gradient descent method is adopted to solve the to-be-optimized objective function, and the basic idea of the gradient descent method is to gradually adjust the parameters along the negative gradient direction of the to-be-optimized objective function in an iterative manner, and specifically includes the following steps.

[0153] Step 1, initialization of parameters: a set of initial parameter values are randomly selected as the starting point of the optimization process.

[0154] Step 2, calculation of gradient: the gradient (i.e., partial derivative) of the to-be-optimized objective function with respect to each parameter is calculated at the current parameter value, and the gradient represents the direction of the maximum change rate of the function at the current point.

[0155] Step 3, updating of parameters: the parameter value is updated according to the gradient information and the learning rate.

[0156] Step 4, checking of convergence condition: it is judged whether the convergence condition is satisfied, such as the parameter update amplitude being less than a certain threshold value or reaching a maximum iteration number. If the condition is satisfied, iteration is stopped; otherwise, step 2 is returned to continue iteration.

[0157] The flow data of the nitrogen element content in the collected historical data are input into the established to-be-optimized objective function, and the gradient descent method is adopted to solve the objective function until the optimal neural network optimization parameter is found, so that the output of the neural network can meet the control equation and can approach the observation data as much as possible. The calibrated nitrification-denitrification coefficient is , which is the to-be-solved nitrification-denitrification optimization coefficient.

[0158] In an embodiment of the present application, the nitrogen element real-time data of the target reservoir and the nitrification-denitrification optimization coefficient are used to predict the greenhouse gas emission of the target reservoir, so as to evaluate the greenhouse effect of the target reservoir, which includes:

[0159] A two-dimensional dot matrix graph is drawn based on the nitrogen element real-time data of the target reservoir and the nitrification-denitrification optimization coefficient.

[0160] The greenhouse gas emission of the target reservoir is predicted based on the two-dimensional dot matrix graph, and the greenhouse effect of the target reservoir is evaluated.

[0161] In this embodiment, the nitrogen element real-time data observed by the target reservoir monitoring station in real time and the calibrated nitrification-denitrification optimization coefficient can be drawn into a two-dimensional dot matrix graph, wherein the nitrogen element real-time data at least includes nitrogen element content data and greenhouse gas monitoring quantity records.

[0162] After obtaining the two-dimensional point array diagram, the greenhouse gas emission of the target reservoir is predicted based on the two-dimensional point array diagram, and the greenhouse effect of the target reservoir is evaluated.

[0163] In an embodiment of the present application, the method further comprises:

[0164] The nitrification-denitrification optimization coefficient is input into a one-dimensional nitrogen element convection-diffusion equation;

[0165] The one-dimensional nitrogen element convection-diffusion equation is solved based on an implicit difference method to obtain the nitrogen element content of the target reservoir after a preset time;

[0166] The loss amount of nitrogen element of the target reservoir is determined based on the nitrogen element content of the target reservoir after the preset time and the nitrification-denitrification optimization coefficient;

[0167] The greenhouse gas emission of the target reservoir is predicted based on the loss amount and the two-dimensional point array diagram, and the greenhouse effect of the target reservoir is evaluated.

[0168] In the present embodiment, the nitrification-denitrification optimization coefficient can be input into a one-dimensional nitrogen element convection-diffusion equation, and the one-dimensional nitrogen element convection-diffusion equation is solved based on an implicit difference method, which includes the following steps:

[0169] Step S1, establishing a difference equation: for the time derivative, a backward difference format is used:

[0170]

[0171] For the spatial derivative, a central difference format or other appropriate difference format can be used;

[0172] Step S2, substituting the original equation: substituting the difference format into the original partial differential equation to obtain an algebraic equation about ;

[0173] Step S3, establishing a linear equation set: since the backward difference is used, the obtained algebraic equation is usually nonlinear, and through linearization processing, a linear equation set is obtained; this linear equation set can be in a tri-diagonal matrix form, or can be processed through matrix decomposition technology (such as LU decomposition);

[0174] Step S4, solving the linear equation set: using a numerical method (such as TDMA, iterative method, etc.) to solve the linear equation set to obtain the solution of at the moment ;

[0175] Step S5, iterative calculation: from the initial moment ​Start, step-by-step iterative calculation until the preset time is reached .

[0176] The implicit difference method generally has unconditional stability, that is, the stability of the numerical solution can be maintained even for a larger time step, so solving the one-dimensional nitrogen element convection-diffusion equation based on the implicit difference method can make it have better numerical stability when dealing with convection and diffusion problems, thereby improving the accuracy of evaluating the greenhouse gas emissions of the reservoir.

[0177] After solving the one-dimensional nitrogen element convection-diffusion equation based on the implicit difference method, the nitrogen element content of the target reservoir after the preset time can be obtained , based on the data, the loss amount of the nitrogen element of the target reservoir is determined according to the formula based on the loss amount of the nitrogen element of the target reservoir and the two-dimensional dot matrix, the greenhouse gas emissions of the target reservoir in a future period of time are predicted, and the greenhouse effect of the target reservoir is evaluated.

[0178] The embodiment of the application discloses a reservoir greenhouse effect evaluation method based on a physical driving neural network, which comprises the following steps: collecting historical data of a target reservoir monitoring site; establishing a river reservoir one-dimensional nitrogen element transport model based on the nitrogen element transport process of the target reservoir, determining a neural network of the nitrification and denitrification coefficient rate during the nitrogen element transport process of the target reservoir according to the river reservoir one-dimensional nitrogen element transport model, and a corresponding to-be-optimized objective function, inputting the historical data into the to-be-optimized objective function, determining the optimization parameters of the neural network, and determining the nitrification and denitrification optimization coefficient; and evaluating the greenhouse gas emissions of the target reservoir based on the real-time data of the nitrogen element of the target reservoir and the nitrification and denitrification optimization coefficient. The embodiment of the application increases the physical constraint of the nitrification and denitrification coefficient rate process by using the one-dimensional nitrogen element transport equation and combining the actual aquatic environment of the reservoir, and then accurately evaluates the greenhouse gas emissions of the reservoir.

[0179] It should be noted that, for the method embodiment, in order to simply describe, it is expressed as a series of action combinations, but those skilled in the art should know that the embodiment of the application is not limited by the described action sequence, because according to the embodiment of the application, certain steps can be performed in other order or simultaneously. Secondly, those skilled in the art should know that the embodiments described in the specification all belong to preferred embodiments, and the actions involved are not necessarily necessary for the embodiment of the application.

[0180] Referring to Figure 3 , a structure block diagram of a reservoir greenhouse effect evaluation device based on a physical driving neural network provided by the embodiment of the application is shown, which can specifically include the following modules:

[0181] The data collection module 301 is configured to collect historical data of a target reservoir monitoring site.

[0182] The first model establishment module 302 is configured to establish a river reservoir one-dimensional nitrogen element transport model based on a nitrogen element transport process of the target reservoir.

[0183] The neural network establishment module 303 is configured to determine a nitrification-denitrification coefficient calibration neural network in the nitrogen element transport process of the target reservoir according to the river reservoir one-dimensional nitrogen element transport model, and determine a corresponding to-be-optimized objective function.

[0184] The optimization coefficient determination module 304 is configured to input the historical data into the to-be-optimized objective function, determine neural network optimization parameters, and thus determine nitrification-denitrification optimization coefficients.

[0185] The evaluation module 305 is configured to predict greenhouse gas emissions of the target reservoir based on real-time nitrogen element data of the target reservoir and the nitrification-denitrification optimization coefficients, and thus evaluate the greenhouse effect of the target reservoir.

[0186] The embodiment of the present application discloses a reservoir greenhouse effect evaluation method based on a physically driven neural network, which comprises the following steps: collecting historical data of a target reservoir monitoring site; establishing a river reservoir one-dimensional nitrogen element transport model based on a nitrogen element transport process of the target reservoir, determining a nitrification-denitrification coefficient calibration neural network in the nitrogen element transport process of the target reservoir according to the river reservoir one-dimensional nitrogen element transport model, and determining a corresponding to-be-optimized objective function; inputting the historical data into the to-be-optimized objective function, determining neural network optimization parameters, and thus determining nitrification-denitrification optimization coefficients; and evaluating greenhouse gas emissions of the target reservoir based on real-time nitrogen element data of the target reservoir and the nitrification-denitrification optimization coefficients. The embodiment of the present application accurately evaluates the greenhouse gas emissions of the reservoir by using a one-dimensional nitrogen element transport equation, combining the actual aquatic environment of the reservoir, and increasing the physical constraints of the nitrification-denitrification coefficient calibration process.

[0187] For the device embodiment, it is basically similar to the method embodiment, so the description is relatively simple, and the related parts refer to the part of the method embodiment.

[0188] The embodiment of the present application further provides an electronic device, which comprises:

[0189] The electronic device comprises a processor, a memory, and a computer program stored in the memory and capable of running on the processor, wherein the computer program is executed by the processor to implement each process of the reservoir greenhouse effect evaluation method based on the physically driven neural network, and achieve the same technical effects. To avoid repetition, the details are not described here.

[0190] The embodiment of the present application also provides a computer readable storage medium, the computer readable storage medium stores a computer program, the computer program is executed by a processor to realize each process of the above-mentioned reservoir greenhouse effect evaluation method based on a physical driving neural network, and the same technical effects can be achieved. To avoid repetition, it will not be repeated here.

[0191] Each embodiment in the specification is described in a progressive manner, and each embodiment focuses on the difference from other embodiments. The same or similar parts between each embodiment can be referred to each other.

[0192] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, device or computer program product. Therefore, the embodiments of the present application can be in the form of a complete hardware embodiment, a complete software embodiment or an embodiment combining software and hardware aspects. Moreover, the embodiments of the present application can be in the form of a computer program product implemented on one or more computer usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer usable program code.

[0193] The embodiments of the present application are described with reference to flowcharts and / or block diagrams according to the method, terminal device (system) and computer program product of the embodiments of the present application. It should be understood that each flow and / or block in the flowchart and / or block diagram, and the combination of the flows and / or blocks in the flowchart and / or block diagram can be realized by computer program instructions. These computer program instructions can be provided to a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing terminal device to produce a machine, so that the instructions executed by the computer or other programmable data processing terminal device produce a device for realizing the functions specified in the flow Figure 1 The device for realizing the functions specified in one or more flows and / or blocks Figure 1 The device for realizing the functions specified in one or more flows and / or blocks

[0194] These computer program instructions can also be stored in a computer readable storage medium which can guide the computer or other programmable data processing terminal device to work in a specific way, so that the instructions stored in the computer readable storage medium produce a product including instruction devices, which realize the functions specified in the flow Figure 1 The device for realizing the functions specified in one or more flows and / or blocks Figure 1 The device for realizing the functions specified in one or more flows and / or blocks

[0195] These computer program instructions can also be loaded into a computer or other programmable data processing terminal device, so that a series of operation steps are performed on the computer or other programmable terminal device to produce a computer implemented process, so that the instructions executed on the computer or other programmable terminal device provide a device for realizing the functions specified in the flowFigure 1 one or more processes and / or blocks Figure 1 the steps of a function specified in one or more processes or blocks.

[0196] While preferred embodiments of the application have been described, those skilled in the art will appreciate that other modifications than those specifically described can be made within the scope of the present application. Accordingly, all changes coming within the meaning and equivalency range of the appended claims are intended to be embraced therein.

[0197] Finally, it should be noted that the terms "first", "second", and the like, herein do not denote any order, quantity, combination, or importance, but rather are used to distinguish one element from another, and do not imply singular or plural. Moreover, the term "include", "have", or any other variant thereof, are intended to encompass non-exclusive inclusions, such that processes, methods, articles, or apparatuses that comprise a set of elements not expressly listed are nonetheless included. In addition, elements defined by the statement "comprising a" do not exclude the presence of additional identical elements in the process, method, article, or apparatus that includes the stated elements.

[0198] The above describes in detail the water reservoir greenhouse effect evaluation method and the water reservoir greenhouse effect evaluation device based on a physically driven neural network provided by the present application. The principles and implementation manners of the present application are described by using specific examples. The above description of the embodiments is only used to help understand the method of the present application and its core idea. Meanwhile, for those skilled in the art, the specific implementation manners and application ranges can be changed according to the idea of the present application. In summary, the content of the present description should not be understood as a limitation of the present application.

Claims

1. A method for assessing the greenhouse effect of reservoirs based on physics-driven neural networks, characterized in that, The method includes: Collect historical data from monitoring stations at the target reservoir; A one-dimensional nitrogen transport model of the river and reservoir is established based on the nitrogen transport process of the target reservoir. The nitrification and denitrification coefficient calibration neural network for the nitrogen transport process in the target reservoir is determined based on the one-dimensional nitrogen transport model of the river and reservoir, and the corresponding objective function to be optimized is determined according to the following formula; ； in: The objective function to be optimized; This is an approximation of the true nitrification-denitrification coefficient. The fitting model parameters; This is an approximation of the actual nitrogen concentration. The fitting model parameters; The residual of the one-dimensional nitrogen transport model of the river and reservoir is given. The neural network outputs the approximation error of the observed nitrogen element in the cross-section; To take the square of the corresponding parameter's L2 norm; The historical data is input into the objective function to be optimized to determine the neural network optimization parameters, thereby determining the nitrification-denitrification optimization coefficients. This includes: inputting the historical data into the objective function to be optimized; solving the objective function to be optimized based on a gradient descent method to obtain the neural network optimization parameters; and determining the nitrification-denitrification optimization coefficients based on the neural network optimization parameters. The greenhouse gas emissions of the target reservoir are predicted based on real-time nitrogen data and the nitrification-denitrification optimization coefficient, thereby assessing the greenhouse effect of the target reservoir.

2. The method according to claim 1, characterized in that, The establishment of a one-dimensional nitrogen transport model for rivers and reservoirs based on the nitrogen transport process of the target reservoir includes: Based on the historical data, establish the correlation between nitrogen and greenhouse gases in the target reservoir; Based on the aforementioned correlation and the nitrogen transport process of the target reservoir, a one-dimensional nitrogen transport model for rivers and reservoirs is established.

3. The method according to claim 2, characterized in that, After establishing a one-dimensional nitrogen transport model for the river-reservoir based on the aforementioned correlation and the nitrogen transport process of the target reservoir, the model further includes: A one-dimensional nitrogen convection-diffusion equation is established based on the aforementioned one-dimensional nitrogen transport model of river and reservoir. Determine the boundary conditions for the nitrogen convection-diffusion equation.

4. The method according to claim 3, characterized in that, The establishment of a one-dimensional nitrogen convection-diffusion equation based on the one-dimensional nitrogen transport model of the river and reservoir includes: Based on the one-dimensional nitrogen transport model of the river and reservoir, the one-dimensional nitrogen convection-diffusion equation is established according to the following formula: Formula 1 is: ； in: The current moment; Spatial location; These are flow parameters; The diffusion coefficient describes the diffusion capacity of nitrogen in a reservoir. R represents the concentration of nitrate in the reservoir or river; R is the nitrification-denitrification coefficient, which represents the rate of nitrification-denitrification in the water, reflects the activity of plankton and microorganisms in the water, and is affected by the oxygen content of the water.

5. The method according to claim 3, characterized in that, The boundary conditions include upper boundary conditions and lower boundary conditions; The determination of the boundary conditions for the nitrogen convection-diffusion equation includes: The upper boundary condition is defined as the inflow nitrogen concentration of the target reservoir; The lower boundary condition is defined as the outflow nitrogen concentration of the target reservoir.

6. The method according to claim 1, characterized in that, The step of determining the nitrification and denitrification coefficient calibration neural network and the corresponding objective function to be optimized during the nitrogen transport process in the target reservoir based on the one-dimensional nitrogen transport model of the river and reservoir includes: Based on the one-dimensional nitrogen transport model of the river and reservoir, the nitrification and denitrification coefficients during the nitrogen transport process of the target reservoir are determined by the following formulas 2 and 3, using a calibration neural network. Formula 2 is: ； in: The residual of the one-dimensional nitrogen transport model of the river and reservoir is given. This is an approximation of the actual nitrogen concentration; The current moment; For spatial location, These are flow parameters; The diffusion coefficient is denoted as . This is an approximation of the true nitrification-denitrification coefficients; Formula 3 is: ； in: The neural network outputs the approximation error of the observed nitrogen element in the cross-section; This is an approximation of the actual nitrogen concentration; The observed nitrogen concentration at the cross-section.

7. The method according to claim 1, characterized in that, The gradient descent-based method is used to solve the objective function to obtain the neural network optimization parameters, including: Calculate the gradient of each parameter in the objective function to be optimized; The parameter values ​​of the neural network are updated based on the gradient information of the gradient. Determine whether the parameter values ​​of the neural network meet the convergence condition. If the parameter value update magnitude of the neural network is less than a preset threshold or the maximum number of iterations is reached, stop the iteration to obtain the optimized parameters of the neural network.

8. The method according to claim 1, characterized in that, The method of predicting greenhouse gas emissions from the target reservoir based on real-time nitrogen data and the nitrification-denitrification optimization coefficient, thereby assessing the greenhouse effect of the target reservoir, includes: A two-dimensional dot matrix diagram was drawn based on the real-time nitrogen data of the target reservoir and the nitrification-denitrification optimization coefficients. Based on the two-dimensional dot matrix diagram, predict the greenhouse gas emissions of the target reservoir and assess the greenhouse effect of the target reservoir.

9. The method according to claim 8, characterized in that, The method further includes: The nitrification-denitrification optimization coefficients are input into the one-dimensional nitrogen convection-diffusion equation; The one-dimensional nitrogen convection-diffusion equation is solved using an implicit difference method to obtain the nitrogen content of the target reservoir after a preset time. The amount of nitrogen loss in the target reservoir is determined based on the nitrogen content of the target reservoir after the preset time and the nitrification-denitrification optimization coefficient. Based on the loss amount and the two-dimensional dot matrix, the greenhouse gas emissions of the target reservoir are predicted, and the greenhouse effect of the target reservoir is assessed.

10. The method according to claim 1, characterized in that, The historical data collected from the target reservoir monitoring stations includes: Collect information about the target reservoir area and obtain historical data from monitoring stations established on the main stream and tributaries of the target reservoir area; The historical data includes at least: flow rate, average flow velocity, water temperature, nitrogen content, dissolved oxygen, microbial content, and greenhouse gas uptake.

11. A reservoir greenhouse effect assessment device based on a physics-driven neural network, characterized in that, The device includes: The data collection module is used to collect historical data from the target reservoir monitoring stations; The first model building module is used to build a one-dimensional nitrogen transport model of the river and reservoir based on the nitrogen transport process of the target reservoir. The neural network establishment module is used to determine the nitrification and denitrification coefficients during the nitrogen transport process of the target reservoir based on the one-dimensional nitrogen transport model of the river and reservoir, and to calibrate the neural network and the corresponding objective function to be optimized; the objective function to be optimized is determined according to the following formula; ； in: The objective function to be optimized; This is an approximation of the true nitrification-denitrification coefficient. The fitting model parameters; This is an approximation of the actual nitrogen concentration. The fitting model parameters; The residual of the one-dimensional nitrogen transport model of the river and reservoir is given. The neural network outputs the approximation error of the observed nitrogen element in the cross-section; To take the square of the corresponding parameter's L2 norm; The optimization coefficient determination module is used to input the historical data into the objective function to be optimized, determine the neural network optimization parameters, and thus determine the nitrification-denitrification optimization coefficients. This includes: inputting the historical data into the objective function to be optimized; solving the objective function to be optimized based on a gradient descent method to obtain the neural network optimization parameters; and determining the nitrification-denitrification optimization coefficients based on the neural network optimization parameters. An evaluation module is used to predict the greenhouse gas emissions of the target reservoir based on real-time nitrogen data and the nitrification-denitrification optimization coefficient, thereby assessing the greenhouse effect of the target reservoir.

12. An electronic device, characterized in that, include: A processor, a memory, and a computer program stored in the memory and capable of running on the processor, wherein the computer program, when executed by the processor, implements the steps of the reservoir greenhouse effect assessment method based on a physics-driven neural network as described in any one of claims 1-10.

13. A computer-readable storage medium, characterized in that, A computer program is stored on the computer-readable storage medium, which, when executed by a processor, implements the steps of the reservoir greenhouse effect assessment method based on a physics-driven neural network as described in any one of claims 1-10.

Images