A flood routing scheme recommendation method and system based on multi-objective optimization

Abstract

The application provides a flood routing scheduling scheme recommendation method and system based on multi-objective optimization, relates to the technical field of data management, and comprises the following steps: collecting basic situation information of a flood routing scheduling object; establishing a residual storage flood propagation model according to the basic situation information; performing parameter identification on the residual storage flood propagation model based on recent upstream and downstream flow, and determining the parameters of the residual storage flood propagation model; inputting the upstream inflow into the residual storage flood propagation model, and outputting the downstream released outflow; determining a deterministic inflow prediction sequence based on the upstream inflow, and constructing a probability prediction set matrix according to an error disturbance term; reducing the probability prediction set matrix to a scenario tree, and determining a multi-scenario inflow sequence set with branch probabilities; constructing a scheduling control constraint model based on the multi-scenario inflow sequence set; and determining a Pareto candidate scheduling scheme set through a multi-stage stochastic model prediction control algorithm.

Description

Technical Field

[0001] This invention relates to the field of data management technology, and in particular to a method and system for recommending flood evolution scheduling schemes based on multi-objective optimization. Background Technology

[0002] In actual operation, the propagation of floods in river networks exhibits significant lag, peak reduction, and storage / discharge effects, resulting in a high degree of coupling between the flood evolution process and the dispatching decision-making process. Simply relying on static experience or localized information is insufficient to meet the needs of refined flood control dispatching. By introducing multi-objective optimization and probabilistic scenario analysis mechanisms, under the premise of meeting flood control safety constraints, the comprehensive performance of different dispatching schemes in terms of peak reduction and staggering, downstream risk control, operational stability, and reservoir recovery capacity can be systematically evaluated. This provides dispatchers with quantifiable, comparable, and recommendable decision support solutions.

[0003] However, existing flood control technologies are mostly based on deterministic inflow forecasts, generally ignoring the uncertainty of inflow forecast errors amplifying with the increase in forecast period, making it difficult to reflect the multiple scenarios of actual flood evolution. Furthermore, traditional methods often separate flood evolution calculations from the control decision-making process, failing to explicitly incorporate the lag and peak-shaving effects of flood propagation in river sections into the optimization model. This results in insufficient adaptability of control results under complex river network conditions, making it difficult to provide comprehensive, robust, and interpretable control scheme support for practical flood control decisions. Summary of the Invention

[0004] In view of the shortcomings of the prior art, the purpose of this invention is to provide a flood evolution scheduling scheme recommendation method based on multi-objective optimization. This method can solve the technical problem that traditional methods often separate the flood evolution calculation and scheduling decision-making process, and fail to explicitly introduce the lag and peak-shaving effect of flood propagation in river sections into the optimization model. As a result, the scheduling results are not adaptable to complex river network conditions and it is difficult to provide comprehensive, robust and interpretable scheduling scheme support for actual flood control decisions.

[0005] A first aspect of this invention proposes a method for recommending flood evolution scheduling schemes based on multi-objective optimization, comprising: S1: Collect basic situational information of flood evolution and scheduling objects; S2: Based on basic situational information, establish a residual reservoir capacity flood propagation model; S3: Based on recent upstream and downstream flow rates, perform parameter identification on the residual reservoir capacity flood propagation model to determine the parameters of the residual reservoir capacity flood propagation model; S4: Input the upstream inflow into the residual reservoir flood propagation model after determining the parameters, and output the downstream outflow; S5: Based on the upstream inflow, determine the deterministic inflow prediction sequence, and add the error disturbance term to the deterministic inflow prediction sequence to construct the probabilistic prediction set matrix; S6: Based on the probability prediction set matrix, construct a scene tree and determine a set of multi-scenario inflow sequences with branch probabilities; S7: Based on a set of inflow sequences under multiple scenarios, and combined with the reservoir water balance equation and the outflow released downstream, a scheduling control constraint model is constructed; S8: Solve the scheduling control constraint model using a multi-stage stochastic model predictive control algorithm to determine the Pareto candidate scheduling scheme set; S9: Screen the Pareto candidate scheduling scheme set to determine the recommended flood evolution scheduling scheme.

[0006] A second aspect of this invention proposes a flood evolution scheduling scheme recommendation system based on multi-objective optimization, comprising: a processor and a memory; The memory stores programs or instructions that can run on the processor, and when the programs or instructions are executed by the processor, they implement the steps of the recommended method for flood evolution scheduling schemes based on multi-objective optimization as described in the first aspect.

[0007] A third aspect of the present invention provides a readable storage medium on which a program or instructions are stored, and when the program or instructions are executed by a processor, the steps of the flood evolution scheduling scheme recommendation method based on multi-objective optimization as described in the first aspect are implemented.

[0008] The beneficial effects of the technical solutions provided in the embodiments of the present invention include at least the following: In this embodiment of the invention, a residual reservoir capacity flood propagation model is introduced to explicitly characterize the flood propagation lag and peak-shaving effect in river sections. Based on parameter identification, the flood evolution calculation results are used as an important input to the scheduling control constraint model, achieving close coupling between the flood evolution process and the scheduling decision-making process. Simultaneously, by introducing an error perturbation that increases with the prediction step size into the deterministic inflow prediction sequence, a probabilistic prediction set matrix is ​​constructed and further dimensionality reduced to form a multi-scenario inflow sequence set with branch probabilities. This allows the scheduling optimization process to systematically reflect the uncertainty of future inflows. Furthermore, a multi-stage stochastic model predictive control algorithm is used to solve the multi-scenario scheduling control constraint model, obtaining a Pareto candidate scheduling scheme set that balances multiple objectives such as flood control safety, flood peak reduction, downstream risk control, and scheduling stability. A recommended scheme is then selected through comprehensive evaluation, effectively improving the scientific rigor, robustness, and engineering applicability of the flood scheduling scheme and reducing the risk of scheduling decision-making errors under extreme inflow conditions. Attached Figure Description

[0009] The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Throughout the drawings, the same reference numerals denote the same parts. Obviously, the drawings described below are merely some embodiments of the present invention. Those skilled in the art can obtain other drawings based on these drawings without any creative effort.

[0010] Figure 1 This is a flowchart illustrating a method for recommending flood evolution scheduling schemes based on multi-objective optimization, as provided in an embodiment of the present invention.

[0011] Figure 2 This is a schematic diagram of the structure of a flood evolution scheduling scheme recommendation system based on multi-objective optimization provided in an embodiment of the present invention. Detailed Implementation

[0012] To enable those skilled in the art to better understand the technical solutions in the embodiments of the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. It should be understood that these descriptions are merely exemplary and are not intended to limit the scope of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0013] The following description, in conjunction with the accompanying drawings, details the method for recommending flood evolution scheduling schemes based on multi-objective optimization provided by the embodiments of the present invention through specific implementations and application scenarios.

[0014] Reference manual attached Figure 1 The diagram illustrates a flowchart of a flood evolution scheduling scheme recommendation method based on multi-objective optimization provided by an embodiment of the present invention.

[0015] This invention provides a method for recommending flood evolution scheduling schemes based on multi-objective optimization, which may include the following steps: S1: Collect basic status information of the flood evolution and scheduling objects.

[0016] Among them, the basic situation information of the flood evolution scheduling object refers to the basic data set used to characterize the overall operating status and external constraints of the current flood scheduling system, and to reflect the real operating status and structural characteristics of the flood scheduling object at the current moment.

[0017] In one possible implementation, the basic situational information includes: real-time water level and capacity of the reservoir, operational status information of gates and flood discharge facilities, downstream control section flow restriction threshold, and the topology of river network nodes-river segments-reservoirs-flood diversion projects. Those skilled in the art can set the magnitude of the downstream control section flow restriction threshold according to actual needs; this invention does not impose any limitations on this.

[0018] It should be noted that by collecting basic situational information on flood evolution and scheduling objects at the beginning of the scheduling calculation, the actual operational status of the reservoir-river network-control section at the current moment can be comprehensively and accurately grasped. This provides a unified and reliable data foundation for subsequent flood propagation modeling, parameter identification, and scheduling optimization. This step helps to avoid model bias caused by missing information or inconsistent state assumptions, ensuring that the residual reservoir capacity flood propagation model and scheduling control model can truly reflect actual engineering conditions. This improves the accuracy and reliability of flood evolution simulation results and scheduling decision results, and enhances the adaptability of the entire scheduling scheme to complex water systems and multi-project joint operation scenarios.

[0019] In one possible implementation, the process after S1 and before S2 includes: Based on basic situational information and combined with network flow theory, a directed graph of the river network is constructed.

[0020] Network flow theory, based on directed graphs, describes the transmission of matter or energy within a network through nodes and directed edges. It is used to characterize the distribution and conservation of flow within a network structure. River network directed graphs, on the other hand, are graph structures built using reservoirs, river channels, flood diversion projects, etc., as nodes, river segments as directed edges, and the direction of water flow as the directed relationship. They are used to describe the transmission paths and topological relationships of floods throughout the river network system.

[0021] It should be noted that by combining network flow theory and constructing a directed graph of the river network based on fundamental situational information, the complex river network structure and its water transport relationships can be transformed into a formalized and computable graph model, clearly expressing the propagation paths, convergence, and diversion relationships of floods between different river segments and nodes. This approach facilitates the application of water conservation and scheduling constraints within a unified framework, enhancing the structural consistency and scalability of flood evolution modeling. Simultaneously, it provides a clear topological foundation for the subsequent construction of residual reservoir flood propagation models, the generation of multi-scenario scheduling constraints, and the implementation of optimization algorithms, thereby improving the systematic nature and engineering applicability of overall scheduling analysis and optimization calculations.

[0022] S2: Based on basic situational information, establish a flood propagation model for residual reservoir capacity.

[0023] The residual reservoir capacity flood propagation model is an equivalent dynamic model used to describe the lag, peak reduction, and storage and discharge characteristics of floods during their propagation within a river segment. This model takes the upstream inflow of the river segment as input, introduces the residual reservoir capacity state variable to characterize the river segment's temporary storage capacity for floods, and combines parameters such as propagation delay and residual coefficient to simulate the propagation process of floods within the river segment and the outflow released downstream, thereby reflecting the regulation and delay role of the river segment in the evolution of floods.

[0024] In one possible implementation, S2 specifically includes: S201: Determine the residual reservoir capacity state and parameter set of the directed edges in the directed graph of the river network, wherein the parameter set includes: river segment propagation delay, residual coefficient and initial residual reservoir capacity.

[0025] Among them, residual reservoir capacity refers to the equivalent water volume in a river segment that is not immediately released but temporarily retained and participates in the subsequent flood evolution at a certain scheduling moment, used to characterize the river segment's flood storage capacity. River segment propagation delay refers to the number of time steps required for floodwater to propagate from the upstream section to the downstream section, used to reflect the lag characteristics of flood propagation. The residual coefficient is a parameter representing the proportion of water entering the river segment that is retained as residual reservoir capacity, used to characterize the river segment's peak shaving and regulation intensity.

[0026] S202: Based on the remaining storage capacity status and parameter set, calculate the remaining storage capacity status update equation:

[0027] in, express t Residual storage capacity at time +1 express t Residual storage capacity at any given time I t-TT express t - TT The inflow rate entering the river segment at any given time, taking propagation delay into account. TT The subsequent inflow, O t express t The outflow of a river section at a given time refers to the amount of water released from that section of the river to the downstream area.

[0028] Specifically, the propagation time of the river section TT This indicates that the incoming water from upstream will not immediately affect the current river section, but will only have an effect on the river section after TT time steps.

[0029] Specifically, the formula is used to describe the evolution of residual reservoir capacity of floods in a river segment between adjacent scheduling times. By introducing propagation delay and outflow release terms, it achieves equivalent modeling of flood lag and peak shaving effects.

[0030] S203: Based on the residual reservoir capacity state update equation, establish a linear relationship between the outflow of the river section and the release of residual reservoir capacity:

[0031] in, The residual coefficient represents the proportion of water entering the river section that is temporarily retained as residual storage capacity. Those skilled in the art can set the value of the residual coefficient according to actual needs, but this invention does not limit it.

[0032] S204: Based on the propagation delay of the river section, the outflow of the upstream node at the corresponding time is mapped to the delayed time of the storage and discharge node of the river section to form a time connection relationship that characterizes the lag characteristics of flood propagation in the river section.

[0033] Among them, the storage and release node refers to the equivalent functional node introduced in the directed graph of the river network, which is used to realize the storage and release distribution of the inflow of the river section in the time dimension.

[0034] S205: Based on the time connection relationship, storage and discharge nodes are introduced between the upstream and downstream nodes of each river section. At the storage and discharge nodes, according to the linear release relationship, the inflow into the river section is allocated as the outflow for downstream transmission and the residual reservoir capacity for the next scheduling time.

[0035] Of this, part is used as the "release flow" corresponding to the outflow of the river section, and part is used as the "residual reservoir capacity" corresponding to the next moment.

[0036] Specifically, at the storage and release node, the inflow into the river section is not directly equivalent to the downstream outflow. Instead, it is calculated by combining the residual reservoir capacity, residual coefficient, and propagation delay relationship. Through this calculation process, the downstream outflow of the river section at the corresponding time is obtained, and the residual reservoir capacity at the next scheduling time is updated simultaneously. The downstream outflow and the residual reservoir capacity at the next scheduling time are determined according to the following formula:

[0037] in, express t + TT Residual storage capacity at any given time express( t + TT The remaining storage capacity at time )+1, O t+TT express t + TT The outflow of a river section at a given time refers to the amount of water released from that section of the river to the downstream area.

[0038] Specifically, the formula is used to describe the process by which upstream inflow, after experiencing a propagation delay in the river section, is stored and released at the storage and release node according to the residual coefficient.

[0039] S206: After completing the construction of the relationship between storage and discharge nodes and flow distribution, apply network flow conservation constraints to any node in the directed graph of the river network, except for source and sink nodes, to establish a residual capacity flood propagation model that satisfies the water conservation condition:

[0040] in, V This represents the set of nodes in a directed graph of a river network. S This represents the source node, i.e., the source of the inflow. T This represents the sink node, i.e., the final outflow boundary. V \ S , T This represents all nodes except the source and sink nodes. n Represents any intermediate node. Represents a node n The set of outgoing arcs, i.e., from the node n The set of all directed edges from which a point originates. e Represents a directed edge. Represents a node n The set of incoming arcs, Indicates that in a directed edge e The amount of water transmitted upstream.

[0041] Among them, network flow conservation constraints refer to the water conservation conditions applied to non-source and non-sink nodes in the directed graph of a river network based on network flow theory, which are used to ensure that the inflow and outflow at any node remain consistent. Source nodes and sink nodes refer to the nodes from which floods originate and the nodes from which floods ultimately exit or control the boundary, respectively.

[0042] Conservation constraints are used to ensure that, in the directed graph of the river network, all nodes except source and sink nodes, the inflow and outflow of water are equal, thereby ensuring the consistency of water volume and physical rationality in the flood evolution calculation process.

[0043] It should be noted that by introducing parameters such as residual reservoir capacity, propagation delay, and residual coefficients, the flood propagation process in the river section is modeled step by step. Furthermore, by setting storage and discharge nodes and network flow conservation constraints in the directed graph of the river network, the complex processes of flood propagation lag, peak shaving, and storage and discharge can be transformed into a recursive and constrained computational model while maintaining clear physical meaning. This modeling approach not only ensures the consistency and physical rationality of water volume during flood propagation in the river network but also facilitates coupling with subsequent scheduling optimization models. This allows flood evolution results to directly participate in the construction of scheduling constraints and objective functions, thereby improving the stability, scalability, and engineering applicability of flood evolution simulation in multi-objective optimization scheduling.

[0044] S3: Based on recent upstream and downstream flow rates, perform parameter identification on the residual reservoir capacity flood propagation model to determine the parameters of the residual reservoir capacity flood propagation model.

[0045] Among them, the recent upstream and downstream flow refers to the time series data of inflow and outflow obtained by actual measurement or monitoring at the upstream and downstream sections of the river during the current flood process or its immediate vicinity, which is used to reflect the actual flood propagation characteristics.

[0046] Specifically, the residual reservoir capacity flood propagation model is used to describe the propagation lag and peak reduction process of floods within a river segment. It takes the upstream inflow as input, characterizes the storage and retention effect of the river segment through the residual reservoir capacity state, and outputs the outflow released downstream from the river segment. The outflow process is further used as input to the scheduling and control constraint model to limit the safety conditions and scheduling objectives of the downstream control section, thereby realizing the coupling of the flood evolution process and the scheduling decision process.

[0047] It should be noted that by identifying the parameters of the residual reservoir flood propagation model based on recent upstream and downstream measured flows, the model parameters can adaptively reflect the current flood conditions and the actual river operation, avoiding systematic errors introduced by using empirical or fixed parameters. This step helps improve the consistency between flood propagation simulation results and actual observations, enabling the model to accurately characterize the lag and peak-shaving characteristics of river sections under different flood processes. This provides a reliable and dynamically updated model foundation for subsequent flood evolution calculations and the construction of scheduling control constraints, improving the accuracy and robustness of the overall scheduling scheme under complex inflow conditions.

[0048] In one possible implementation, S3 specifically includes: S301: Determine the identification time domain length based on the recent upstream inflow sequence and the measured downstream outflow sequence.

[0049] Among them, the upstream section inflow sequence and the downstream section measured outflow sequence refer to the flow observation time series obtained at the upstream and downstream sections of the river section within the time window, respectively.

[0050] The identification time domain length refers to the length of the time window used for parameter identification, that is, selecting a recent continuous flood event data as the analysis period for model identification. Specifically, the length is T.

[0051] S302: Determine the initial parameter combination for the residual reservoir capacity flood propagation model.

[0052] It should be noted that those skilled in the art can set the initial parameter combination according to actual needs, and this invention does not limit it.

[0053] S303: Based on the identified time-domain length, construct the first objective function for parameter identification:

[0054] in, Minimize Indicates minimization. Horizon Indicates the length of the identification time domain. express t The measured flow rate at the downstream section at a given time (representing the actual observed flow rate obtained by hydrological stations or monitoring equipment at the downstream control section or river outlet). express t At any given combination of parameters Below, the outflow calculated by the residual reservoir capacity flood propagation model, i.e. t The model simulates the flow rate at any given time.

[0055] The objective function is a mathematical function used to measure the degree of deviation between the simulated flow and the measured flow. In this step, parameter optimization is achieved by minimizing the sum of the squared errors of the two.

[0056] S304: Based on the RSM dynamic equations and network structure, construct the first constraint condition.

[0057] The constraints include: network flow conservation constraints, residual capacity state recursion constraints, parameter value range constraints, and first and last residual capacity closure constraints.

[0058] Among them, the RSM dynamic equation refers to the state update equation describing the evolution of residual reservoir capacity over time using the residual reservoir capacity flood propagation model.

[0059] Among them, the network flow conservation constraint refers to the constraint condition that ensures that all intermediate nodes in the directed graph of the river network satisfy the water conservation relationship. The first and last residual reservoir capacity closure constraint refers to the constraint that ensures that the residual reservoir capacity state at the beginning and end of the identification time domain is consistent, in order to avoid non-physical drift of the residual reservoir capacity state during the parameter identification process.

[0060] S305: Using a genetic algorithm, combined with the first objective function and the first constraint condition, a global search and iterative update of the initial parameter combination is performed until the genetic algorithm converges, thus determining the parameters of the residual reservoir capacity flood propagation model.

[0061] Among them, the genetic algorithm is a global optimization algorithm based on the population evolution mechanism. It achieves iterative search of the parameter space through selection, crossover and mutation operations. The genetic algorithm is a mature existing technology, and will not be described in detail here.

[0062] It should be noted that by aligning the measured flow sequences of upstream and downstream within a unified identification time domain, and constructing a parameter identification model with the goal of minimizing outflow error under the constraints of network flow conservation and residual reservoir capacity dynamics, the mathematical and physical consistency of the parameter identification results can be effectively guaranteed. At the same time, using a genetic algorithm to perform a global search on parameters such as propagation delay, residual coefficient, and initial residual reservoir capacity helps to avoid getting trapped in local optima, so that the identified model parameters can truly reflect the propagation and peak-shaving characteristics of the river segment under the current flood process, thereby improving the accuracy and stability of the residual reservoir capacity flood propagation model in subsequent flood evolution calculations and scheduling optimization applications.

[0063] S4: Input the upstream inflow into the residual reservoir flood propagation model after determining the parameters, and output the downstream outflow.

[0064] The upstream inflow refers to the time series of water entering the system at the upstream section of a river or reservoir during the flood's evolution, used to characterize the external flood inflow process. The downstream outflow refers to the time series of water released downstream from the downstream section of the river or storage / discharge node after the flood has undergone propagation, retention, and peak reduction effects, used to characterize the actual impact of the flood's evolution on the downstream area.

[0065] It should be noted that by inputting the upstream inflow into the residual reservoir flood propagation model with completed parameter identification, the lag and peak-shaving effects generated during flood propagation within the river section can be realistically simulated while maintaining computational efficiency. This allows the downstream outflow to reflect not only the intensity of the inflow but also the comprehensive impact of the river section's regulation and storage characteristics. This step enables the flood evolution results to serve as a quantitative input directly for the construction of subsequent scheduling and control constraints, effectively avoiding the bias caused by simply equating downstream outflow with upstream inflow, thereby improving the adaptability of scheduling decisions to the flood propagation process and the engineering rationality.

[0066] S5: Based on the upstream inflow, determine the deterministic inflow prediction sequence and add the error disturbance term to the deterministic inflow prediction sequence to construct the probability prediction set matrix.

[0067] Among them, the deterministic inflow forecast sequence refers to the inflow forecast result of a single path within the future forecast period obtained based on hydrological forecasting models or empirical extrapolation methods, without considering forecast errors and uncertainties. The error disturbance term refers to the random or parameterized disturbance quantity used to characterize the inflow forecast error and uncertainty. The probabilistic forecast set matrix refers to the multiple possible future inflow forecast trajectories generated by introducing error disturbances on the deterministic inflow forecast sequence, organized in matrix form, to describe the probability distribution characteristics of future water inflow evolution.

[0068] It should be noted that by introducing an error disturbance term into the deterministic inflow forecast sequence and constructing a probabilistic forecast set matrix, the uncertainties that may exist in future water inflow processes can be systematically characterized. This allows scheduling optimization to no longer rely on a single forecast result, thereby significantly reducing the impact of forecast bias on scheduling decisions. This step helps to preserve the main trend characteristics of water inflow forecasts, while reflecting extreme water inflow scenarios and different evolution paths through multi-member forecasts. This provides a sufficient information foundation for subsequent scenario tree construction and multi-stage stochastic optimization, improving the robustness and safety of flood control schemes under complex and uncertain water inflow conditions.

[0069] Specifically, the magnitude of the error disturbance term gradually increases with the increase of the prediction step size to reflect the higher uncertainty of the long-term prediction compared to the short-term prediction, thereby generating multiple possible inflow prediction trajectories and constructing a probability prediction set matrix.

[0070] In one possible implementation, S5 specifically includes: S501: Based on upstream inflow and combined with empirical extrapolation, construct a deterministic inflow forecast sequence for the future forecast period.

[0071] Among them, the empirical extrapolation method refers to the method of extending and predicting future inflows based on recent measured trends in inflow, using statistical or empirical rules. The future forecast period refers to the forward-looking time range considered in scheduling decisions.

[0072] Specifically, the deterministic inflow prediction sequence is as follows: ,in, q DSF This represents a deterministic inflow forecast sequence. q k Indicates the first k Deterministic inflow forecast within a prediction step, k =1,2,..., N , N Indicates the length of the future forecast period.

[0073] S502: Based on the deterministic inflow prediction sequence, a stochastic disturbance model is constructed to describe the uncertainty of the prediction by introducing an error disturbance term.

[0074] Specifically, the uncertainty prediction sequence is as follows: ,in, Indicates the first k Uncertainty inflow prediction within a prediction step, Indicates the first k The relative prediction error within each prediction step, i.e. the relative error perturbation term, can be set by those skilled in the art according to actual needs, and this invention does not limit it.

[0075] S503: By combining the statistical characteristics of the error disturbance term, the random disturbance model is parameterized to construct an error magnitude function that increases with the future forecast period:

[0076] in, Indicates the first k The error magnitude function within the prediction step, i.e., the first prediction step... k The uncertainty magnitude of the inflow forecast error within a prediction step. This represents the standard deviation of the initial prediction error. The error growth factor can be set by those skilled in the art according to actual needs, and this invention does not limit it.

[0077] Among them, the error magnitude function refers to the functional form used to describe the change of prediction error uncertainty with the prediction step size.

[0078] This formula is used to describe the characteristic that the error in river inflow prediction gradually increases with the forecast period, so as to reflect the distribution law that the uncertainty of long-term prediction is higher than that of short-term prediction.

[0079] S504: Determine the error distribution based on the error amplitude function.

[0080] Specifically, the error distribution assumes that the relative prediction error follows a normal distribution with a mean of zero and a standard deviation that is a function of the error magnitude.

[0081] Here, error distribution refers to the probabilistic assumption about the random nature of prediction errors.

[0082] S505: Construct a probability prediction set matrix based on the error distribution.

[0083] Specifically, the expression for the probability prediction set matrix is: ,in, Q PSF Represents the probability prediction set matrix.q k,m Indicates the first m The predicted trajectory is in the first k The inflow rate for each predicted step length, k =1,2,..., N , N Indicates the length of the future forecast period. m =1,2,..., M , M This represents the number of predicted members, i.e., the number of predicted trajectories generated by random perturbation for the same deterministic inflow prediction sequence.

[0084] Each column represents a complete inflow forecast trajectory, and each row represents the multi-member forecast results at the same forecast step size within the forecast period.

[0085] It should be noted that by first constructing a deterministic inflow forecast sequence and then introducing an error perturbation model that increases with the forecast period, the objective law of gradually increasing uncertainty in long-term forecasts can be reasonably reflected while maintaining the stability of the main trend of inflow forecasts. This method can generate a multi-member, multi-trajectory probability forecast set matrix, providing a sufficient and structured data foundation for subsequent scenario tree construction. This allows the scheduling optimization process to fully consider different inflow evolution scenarios, thereby improving the robustness, risk control capability, and engineering practical value of flood control schemes under uncertain conditions.

[0086] S6: Based on the probability prediction set matrix, construct the scenario tree and determine the set of multi-scenario inflow sequences with branch probabilities.

[0087] Here, a scenario tree is a tree-like structure model used to represent the gradual unfolding of uncertainty over time. Each node corresponds to a representative inflow state under different prediction step sizes, and branches characterize the differentiation of the inflow evolution path. Branch probability refers to the probability weight of each branch occurring in the scenario tree. A multi-scenario inflow sequence set refers to a set of inflow flow scenario sequences with occurrence probabilities, formed by the paths from the root node to the leaf node of the scenario tree, used to describe multiple possible evolutionary scenarios of future inflow.

[0088] In one possible implementation, S6 specifically includes: S601: Based on the row matrix in the probability prediction set matrix, group the prediction members to determine multiple candidate branches.

[0089] Candidate branches refer to representative groups formed by aggregating prediction members with similar values ​​within the same prediction step.

[0090] In this context, each prediction member represents a future inflow prediction sequence.

[0091] Specifically, the predicted members are grouped by clustering methods, and predicted members with similar values ​​are merged into the same candidate branch to reduce the number of predicted members and maintain the main distribution characteristics.

[0092] S602: Calculate the statistical representative value of the predicted members within each candidate branch.

[0093] Specifically, the formula for calculating the statistical representative value is as follows: ,in, Indicates the prediction step size k Next b The set of predicted members contained in each candidate branch, Indicates the prediction step size k Next b The statistical representative value of each candidate branch, q k,m Indicates the first m The predicted trajectory is in the first k The predicted inflow rate for each step.

[0094] Among them, the statistical representative value refers to the statistical quantity used to characterize the overall characteristics of a candidate branch, which is usually the average value of the inflow of the predicted members within that branch at the corresponding prediction step size.

[0095] S603: Based on the prediction time order, connect the candidate branches at adjacent prediction steps to determine the tree structure that gradually unfolds over time, i.e., the scene tree.

[0096] Specifically, clustering branches at adjacent prediction steps are connected so that each prediction step's branch node is derived only from the branch node of the previous prediction step, thus forming a tree structure that gradually unfolds over time to represent the gradual differentiation process of uncertainty over time.

[0097] S604: Calculate the probability of occurrence of the corresponding scene tree branch based on the number of predicted members in each candidate branch.

[0098] Specifically, the formula for calculating the probability of occurrence is as follows: ,in, Indicates the prediction step size k Next b The probability of each candidate branch occurring. M This indicates the predicted number of members.

[0099] S605: Take the path from the root node to the leaf node along the scene tree as the river flow scenario sequence, and take the product of the occurrence probabilities of each path as the scenario occurrence probability of the river flow scenario sequence.

[0100] The probability of a scenario occurring refers to the joint probability of the entire scenario path from the root node to the leaf node.

[0101] Specifically, the probability of a scenario occurring is calculated as follows: ,in, p j Indicates the first j The probability of a scenario occurring along a given path. Indicates the prediction step size k The first scene in the next scene tree b j ( k The probability of ) candidate branches appearing, b j ( k ) indicates the first j The scenario path in the prediction step size k Candidate branch index at that time.

[0102] S606: Determine the set of multi-scenario inflow sequences based on the probability of scenario occurrence.

[0103] It should be noted that by clustering the predicted members in the probability prediction set matrix and constructing a scenario tree structure that expands over time using statistical representative values, the main distribution characteristics of inflow uncertainty can be preserved while effectively compressing the prediction scale, giving multi-scenario information a clear temporal evolution logic. This method not only significantly reduces the computational complexity of multi-stage stochastic optimization, but also, by explicitly introducing branch probabilities and scenario occurrence probabilities, enables subsequent scheduling models to evaluate the impact of different inflow scenarios in a probabilistically weighted manner, thereby improving the robustness and risk perception capability of flood control decisions under complex and uncertain inflow conditions.

[0104] S7: Based on a set of inflow sequences under multiple scenarios, and combined with the reservoir water balance equation and the outflow released downstream, a scheduling control constraint model is constructed.

[0105] Among them, the reservoir water balance equation refers to the basic constraint equation used to describe the relationship between reservoir inflow, outflow, and capacity changes between adjacent scheduling times. The scheduling control constraint model refers to the mathematical model used in the scheduling optimization process to uniformly describe system state updates, physical boundary conditions, and operational constraints, and to limit the feasible range of scheduling decision variables and system state.

[0106] Specifically, the inputs to the scheduling control constraint model mainly include a set of multi-scenario inflow sequences and their corresponding scenario occurrence probabilities, downstream outflow calculated by the residual reservoir flood propagation model, initial reservoir capacity and water level status, scheduling time step, and physical operational constraint parameters of the reservoir and flood discharge facilities (such as upper and lower water level limits, discharge capacity curves, etc.). The output of the model is the optimal scheduling control decision sequence and its state evolution results for different scenarios in each predicted scheduling period, specifically including the reservoir discharge flow or gate opening, reservoir capacity and water level change trajectory at each time point, and forming a Pareto candidate scheduling scheme set that can be used for screening and recommendation under the meaning of multi-objective optimization.

[0107] In one possible implementation, S7 specifically includes: S701: Use the set of multi-scenario inflow sequences as a perturbation sequence, and express the probability of scenario occurrence as the weight of each scenario.

[0108] The disturbance sequence refers to the time series variable introduced as an external uncertain input during the scheduling optimization process. In this method, it mainly refers to the multi-scenario inflow sequence.

[0109] Each inflow scenario corresponds to a possible future water inflow evolution path.

[0110] S702: Determine the second objective function based on the perturbation sequence and the weights of each scenario:

[0111] in, Indicates the foreseeable future Internally, simultaneously for state variables x and control variables u Optimize min Indicates minimization. p j Indicates the first j The probability of each scenario occurring. J ( ) represents the stage cost function. E ( ) represents the terminal cost function. Indicates the first j Under the given scenario path, the system predicts the step size. k State variables at time, Indicates the first j Under the given scenario path, in the prediction step size k The scheduling decisions made at that time (such as reservoir discharge, gate opening, etc.) Indicates the first j Prediction step size under multiple scenario paths k The disturbance amount at that time, i.e., the uncertain external input, Indicates the firstj Under the given scenario path, the system predicts the step size. N The state variables at time (in flood control problems, they usually correspond to reservoir capacity, reservoir water level, and the state of residual reservoir capacity at the end of the river section). Indicates the first j Under the given scenario path, in the prediction step size N The scheduling decisions made at that time Indicates the first j Prediction step size under multiple scenario paths k The amount of disturbance at that time.

[0112] It should be noted that the disturbance mainly refers to the multi-scenario inflow sequence.

[0113] Specifically, the system refers to the water conservancy dispatching system affected by dispatch control and flood evolution, which is a dynamic water volume evolution system composed of reservoirs, river sections and their associated regulation and storage units.

[0114] S703: Based on the reservoir water balance equation, construct the system state update constraint framework between adjacent scheduling times:

[0115] in, s t express t The remaining storage capacity at any given time, i.e., the storage capacity status. s t-1 express t The remaining storage capacity at time -1 Indicates the scheduling time step. express t Inbound flow at any time express t real-time scheduling of traffic release express t Other outbound traffic at any given time.

[0116] S704: Rewrite the update constraint skeleton into residual equality constraints for optimization:

[0117] in, r t express t The residual of the reservoir water balance equation at time t.

[0118] Among them, the residual equation constraint transforms the water balance relationship into an equation form suitable for optimization solutions.

[0119] S705: Based on the reservoir capacity-water level relationship, the reservoir capacity status (residual reservoir capacity) is mapped to the forebay water level.

[0120] Specifically, ,in, fb t express t The water level in the forebay at any given time, that is, the water level height corresponding to the forebay (in front of the dam) of the reservoir. f sl ( ) represents the reservoir capacity-water level mapping function.

[0121] S706: Construct the second constraint condition based on the forebay water level and residual equation constraint.

[0122] Specifically, the constraints include: (physical boundary constraints of water level) setting upper and lower limits for water level in each time period to ensure that it is not lower than the minimum operating water level and does not exceed the maximum safe / operating water level.

[0123] (Flood discharge (spillway / gate) flow boundary and capacity curve constraints): During the scheduling process, upper and lower limits are set for the discharge flow of the reservoir at each scheduling time. The scheduled discharge flow shall not be less than the minimum allowable discharge flow at any scheduling time, and shall not be greater than the maximum allowable discharge flow at the corresponding time. The maximum allowable discharge flow is determined by the current forebay water level through a preset flood discharge capacity curve function.

[0124] S707: Based on the second objective function and the second constraint, and combined with the probability of scenario occurrence, construct a scheduling control constraint model.

[0125] It should be noted that by incorporating multi-scenario inflow sequences as perturbation sequences and using the probability of scenario occurrence as weights into the objective function, and simultaneously constructing constraints such as water balance, reservoir capacity-water level mapping, and discharge capacity within a unified framework, the scheduling optimization process can simultaneously consider the uncertainty of inflow and the physical operational characteristics of water conservancy projects. This modeling approach ensures that scheduling decisions under different inflow scenarios are subject to consistent and strict constraints, effectively avoiding the generation of infeasible or high-risk scheduling schemes. This provides a rigorous and feasible model foundation for multi-stage stochastic optimization solutions, significantly improving the safety, stability, and engineering feasibility of flood control scheduling schemes in complex and uncertain environments.

[0126] S8: Solve the scheduling control constraint model using a multi-stage stochastic model predictive control algorithm to determine the Pareto candidate scheduling scheme set.

[0127] Among them, the multi-stage stochastic model predictive control algorithm refers to a control algorithm that optimizes the system state and control variables in stages within the rolling prediction time domain by combining multi-scenario uncertainty information. It achieves dynamic decision-making for uncertain systems by re-predicting and re-optimizing at each scheduling stage. The Pareto candidate scheduling scheme set refers to a set of non-dominant scheduling schemes obtained by solving under multi-objective optimization conditions, in which each scheme is superior to other schemes in at least one scheduling objective, and there is no situation where it is completely superior to another scheme in all objectives.

[0128] In one possible implementation, S8 specifically includes: S801: Based on the scenario tree, the flood evolution scheduling problem is represented as a multi-stage stochastic decision structure.

[0129] Among them, the multi-stage stochastic decision structure refers to the optimization structure that represents the scheduling problem as making decisions for different stochastic scenarios in multiple time stages.

[0130] S802: Based on the multi-stage stochastic decision-making structure, the optimal control sequence and state trajectory for various scenarios are determined through probability weighting within the foreseeable future period.

[0131] Specifically, in the scheduling time domain Based on the water level of the reservoir's forebay fb and scheduling and releasing traffic Q s To determine the relevant variables for decision-making, a comprehensive scheduling objective function is constructed. J ( fb , Qs By weighting and superimposing multiple scheduling sub-objectives, a unified optimization objective is formed. By minimizing the comprehensive scheduling objective function, a coordinated trade-off is achieved among various scheduling objectives, including flood control safety, downstream risk control, scheduling stability, and operational recovery, thereby determining the optimal scheduling scheme within the entire predictive scheduling time domain.

[0132] The optimal control sequence refers to the sequence of optimal scheduling control variables corresponding to each scheduling stage within the predictive scheduling time domain. The state trajectory refers to the path of the evolution of system state variables (such as reservoir water level and capacity) over time.

[0133] S803: Based on the optimal control sequence and state trajectory, perform multi-stage stochastic MPC solution on the scheduling control constraint model.

[0134] Among them, stochastic MPC solution refers to the control process of repeatedly solving the optimization problem in the rolling time domain by combining stochastic scenario information. Stochastic MPC solution is a mature existing technology, and will not be described in detail here.

[0135] S804: Based on the solution results, execute the preliminary scheduling control decision corresponding to the root node of the scenario tree.

[0136] S805: Reconstruct the scene tree based on the system state and inflow prediction after execution.

[0137] S806: Based on the reconstructed scene tree, repeat steps S801 to S805 until multiple sets of scheduling control solutions corresponding to different objective trade-offs are obtained.

[0138] S807: Determine the Pareto candidate scheduling scheme set based on multiple sets of scheduling control solutions.

[0139] Specifically, the values ​​of multiple sets of scheduling control solutions on each scheduling objective index are compared, and the schemes that are simultaneously superior to other schemes on all objectives are eliminated. The scheduling control solutions that do not dominate each other are retained to form a Pareto candidate scheduling scheme set.

[0140] It should be noted that by representing the flood evolution scheduling problem as a multi-stage stochastic decision structure and combining scenario trees with a probability weighting mechanism for rolling optimization, scheduling decisions can be dynamically adjusted at each stage based on the latest state and forecast information, thereby effectively addressing uncertainties in inflow and changes in system state. Furthermore, by repeatedly constructing scenario trees and obtaining multiple sets of scheduling control solutions under different objective trade-offs, and then selecting candidate schemes based on the Pareto criterion, it is helpful to comprehensively characterize the conflicts and balances among various scheduling objectives. This provides a scientific, rich, and engineering-interpretable decision-making basis for the final scheme recommendation, significantly improving the flexibility, robustness, and intelligence of flood scheduling schemes.

[0141] S9: Screen the Pareto candidate scheduling scheme set to determine the recommended flood evolution scheduling scheme.

[0142] The recommended scheme refers to the final scheduling scheme that is selected from the Pareto candidate scheduling scheme set through unified evaluation criteria and decision-making rules, and has the best overall performance and is applicable to the current flood scheduling scenario.

[0143] It should be noted that further screening of the Pareto candidate scheduling scheme set allows for the introduction of engineering experience and decision-making preferences while fully preserving the diversity of multi-objective optimization results. This enables a quantitative comparison of the comprehensive performance of different schemes, avoiding one-sided decisions based solely on a single objective. This step helps transform complex multi-objective optimization results into intuitive and actionable scheduling recommendations, allowing dispatchers to quickly obtain optimal or near-optimal recommended schemes while prioritizing flood control safety. This significantly improves the practicality, interpretability, and efficiency of flood control scheduling decisions.

[0144] Specifically, based on the evaluation results of each candidate scheme in the Pareto candidate scheduling scheme set on multiple scheduling objective indicators, a comprehensive evaluation index system is constructed, and each candidate scheme is uniformly quantitatively scored. The multiple scheduling objective indicators include at least the degree of downstream control section flow exceeding limits, duration of exceeding limits, flood peak reduction effect, reservoir terminal water level deviation, and scheduling operation smoothness. After dimensionless normalization of each scheduling objective indicator, the comprehensive evaluation score of each candidate scheme is calculated according to the preset objective importance weights. Under the premise of satisfying the priority conditions of flood control safety constraints, the scheduling scheme with the best comprehensive evaluation score is selected as the recommended scheme output.

[0145] In this embodiment of the invention, a residual reservoir capacity flood propagation model is introduced to explicitly characterize the flood propagation lag and peak-shaving effect in river sections. Based on parameter identification, the flood evolution calculation results are used as an important input to the scheduling control constraint model, achieving close coupling between the flood evolution process and the scheduling decision-making process. Simultaneously, by introducing an error perturbation that increases with the prediction step size into the deterministic inflow prediction sequence, a probabilistic prediction set matrix is ​​constructed and further dimensionality reduced to form a multi-scenario inflow sequence set with branch probabilities. This allows the scheduling optimization process to systematically reflect the uncertainty of future inflows. Furthermore, a multi-stage stochastic model predictive control algorithm is used to solve the multi-scenario scheduling control constraint model, obtaining a Pareto candidate scheduling scheme set that balances multiple objectives such as flood control safety, flood peak reduction, downstream risk control, and scheduling stability. A recommended scheme is then selected through comprehensive evaluation, effectively improving the scientific rigor, robustness, and engineering applicability of the flood scheduling scheme and reducing the risk of scheduling decision-making errors under extreme inflow conditions.

[0146] Reference manual attached Figure 2 The diagram shows a structural schematic of a flood evolution scheduling scheme recommendation system based on multi-objective optimization provided by an embodiment of the present invention.

[0147] This invention provides a flood evolution scheduling scheme recommendation system 20 based on multi-objective optimization, comprising: a processor 201 and a memory 202; The memory 202 stores programs or instructions that can run on the processor 201. When the program or instructions are executed by the processor 201, they implement the steps of the above-described method for recommending flood evolution scheduling schemes based on multi-objective optimization and achieve the same technical effect. To avoid repetition, the present invention will not elaborate further.

[0148] It should be understood that the processor 201 in this embodiment of the invention may be a Central Processing Unit (CPU), or it may be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor.

[0149] It should also be understood that the memory 202 in the embodiments of the present invention can be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory. The non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. The volatile memory can be random access memory (RAM), which is used as an external cache. By way of example, but not limitation, many forms of random access memory are available, such as static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link dynamic random access memory (SLDRAM), and direct memory bus RAM (DR RAM).

[0150] The above embodiments can be implemented, in whole or in part, by software, hardware (such as circuits), firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of the present invention are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more sets of available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. A semiconductor medium can be a solid-state drive.

[0151] It should be understood that, in various embodiments of the present invention, the order of the above-mentioned processes does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0152] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0153] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the devices, apparatuses, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0154] In the several embodiments provided by this invention, it should be understood that the disclosed devices, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another device, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0155] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0156] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0157] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0158] This invention provides a readable storage medium comprising: storing a program or instructions on the readable storage medium, wherein when the program or instructions are executed by a processor, the program or instructions implement the steps of the above-described method for recommending flood evolution scheduling schemes based on multi-objective optimization, and can achieve the same technical effect. To avoid repetition, this invention will not elaborate further.

[0159] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the embodiments of the present invention, and are not intended to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the protection scope of the present invention.

Claims

1. A method for recommending flood evolution scheduling schemes based on multi-objective optimization, characterized in that... ,include: S1: Collect basic situational information of flood evolution and scheduling objects; S2: Based on the aforementioned basic situation information, establish a residual reservoir capacity flood propagation model; S3: Based on recent upstream and downstream flow rates, perform parameter identification on the residual reservoir capacity flood propagation model to determine the parameters of the residual reservoir capacity flood propagation model; S4: Input the upstream inflow into the residual reservoir flood propagation model after determining the parameters, and output the downstream outflow; S5: Based on the upstream inflow, determine a deterministic inflow prediction sequence, and add an error disturbance term to the deterministic inflow prediction sequence to construct a probability prediction set matrix; S6: Based on the probability prediction set matrix, construct a scene tree and determine a set of multi-scenario inflow sequences with branch probabilities; S7: Based on the multi-scenario inflow sequence set, combined with the reservoir water balance equation and the downstream outflow, a scheduling control constraint model is constructed; S8: Solve the scheduling control constraint model using a multi-stage stochastic model predictive control algorithm to determine the Pareto candidate scheduling scheme set; S9: Screen the Pareto candidate scheduling scheme set to determine the recommended flood evolution scheduling scheme.

2. The method for recommending flood evolution scheduling schemes based on multi-objective optimization according to claim 1, characterized in that... The basic situation information includes: the real-time water level and capacity of the reservoir, the operating status of the gates and flood discharge facilities, the flow restriction threshold of the downstream control section, and the topology of the river network node-river section-reservoir-flood diversion project.

3. The method for recommending flood evolution scheduling schemes based on multi-objective optimization according to claim 1, characterized in that... The process includes the following steps after S1 and before S2: Based on the aforementioned basic situational information and combined with network flow theory, a directed graph of the river network is constructed.

4. The method for recommending flood evolution scheduling schemes based on multi-objective optimization according to claim 3, characterized in that... S2 specifically includes: S201: Determine the residual reservoir capacity state and parameter set of the directed edges in the directed graph of the river network, wherein the parameter set includes: river segment propagation delay, residual coefficient, and initial residual reservoir capacity; S202: Calculate the residual storage capacity state update equation based on the residual storage capacity state and the parameter set; S203: Based on the residual reservoir capacity state update equation, establish a linear release relationship between the outflow of the river section and the residual reservoir capacity; S204: Based on the propagation delay of the river section, the outflow of the upstream node at the corresponding time is mapped to the delay time of the river section's storage and discharge node to form a time connection relationship characterizing the lag characteristics of flood propagation in the river section; S205: Based on the aforementioned time connection relationship, storage and discharge nodes are introduced between the upstream and downstream nodes of each river segment, and at the storage and discharge nodes, according to the aforementioned linear release relationship, the inflow into the river segment is allocated as the outflow for downstream transmission and the residual reservoir capacity for the next scheduling time; S206: After completing the construction of the storage and discharge nodes and flow distribution relationship, apply network flow conservation constraints to any node in the directed river network graph except for the source node and sink node, so as to establish the residual reservoir flood propagation model that satisfies the water conservation condition.

5. The method for recommending flood evolution scheduling schemes based on multi-objective optimization according to claim 1, characterized in that... S3 specifically includes: S301: Determine the identification time domain length based on recent upstream cross-section inflow sequences and measured downstream cross-section outflow sequences; S302: Determine the initial parameter combination for the residual reservoir capacity flood propagation model; S303: Based on the identified time-domain length, construct the first objective function for parameter identification; S304: Based on the RSM dynamic equations and network structure, construct the first constraint condition; S305: Using a genetic algorithm, combined with the first objective function and the first constraint, a global search and iterative update of the initial parameter combination is performed until the genetic algorithm converges, thereby determining the parameters of the residual reservoir capacity flood propagation model.

6. The method for recommending flood evolution scheduling schemes based on multi-objective optimization according to claim 1, characterized in that... S5 specifically includes: S501: Based on the upstream inflow, and combined with empirical extrapolation, construct a deterministic inflow prediction sequence for the future forecast period; S502: Based on the deterministic inflow prediction sequence, a stochastic perturbation model is constructed to describe the prediction uncertainty by introducing the error perturbation term; S503: Based on the statistical characteristics of the error disturbance term, the random disturbance model is parameterized to construct an error magnitude function that increases with the future forecast period; S504: Determine the error distribution based on the error amplitude function; S505: Based on the error distribution, construct the probability prediction set matrix.

7. The method for recommending flood evolution scheduling schemes based on multi-objective optimization according to claim 1, characterized in that... S6 specifically includes: S601: Based on the row matrix in the probability prediction set matrix, group the prediction members to determine multiple candidate branches; S602: Calculate the statistical representative value of the predicted member within each of the candidate branches; S603: Based on the prediction time order, connect the candidate branches at adjacent prediction steps to determine the tree structure that gradually unfolds over time, i.e., the scene tree; S604: Calculate the probability of occurrence of the corresponding scene tree branch based on the number of predicted members in each candidate branch; S605: The path along the scene tree from the root node to the leaf node is taken as the river flow scenario sequence, and the product of the occurrence probabilities on each of the paths is taken as the scenario occurrence probability of the river flow scenario sequence; S606: Determine the set of multi-scenario inflow sequences based on the probability of occurrence of the scenario.

8. The method for recommending flood evolution scheduling schemes based on multi-objective optimization according to claim 1, characterized in that... S7 specifically includes: S701: The set of multi-scenario inflow sequences is used as a perturbation sequence, and the probability of scenario occurrence is expressed as the weight of each scenario; S702: Determine the second objective function based on the perturbation sequence and the weights of each scenario; S703: Based on the reservoir water balance equation, construct the system state update constraint framework between adjacent scheduling times; S704: Rewrite the updated constraint skeleton into residual equality constraints for optimization; S705: Based on the reservoir capacity-water level relationship, the reservoir capacity status is mapped to the forebay water level; S706: Based on the forebay water level and the residual equation constraint, construct the second constraint condition; S707: Based on the second objective function and the second constraint condition, and combined with the probability of scenario occurrence, construct the scheduling control constraint model.

9. The method for recommending flood evolution scheduling schemes based on multi-objective optimization according to claim 1, characterized in that... S8 specifically includes: S801: Based on the scenario tree, the flood evolution scheduling problem is represented as a multi-stage stochastic decision structure; S802: Based on the multi-stage stochastic decision-making structure, the optimal control sequence and state trajectory for various scenarios are determined through probability weighting within the foreseeable future period; S803: Based on the optimal control sequence and the state trajectory, perform multi-stage stochastic MPC solution on the scheduling control constraint model; S804: Based on the solution results, execute the preliminary scheduling and control decision corresponding to the root node of the scenario tree; S805: Reconstruct the scene tree based on the system state and inflow prediction after execution; S806: Based on the reconstructed scene tree, repeat steps S801 to S805 until multiple sets of scheduling control solutions corresponding to different objective trade-offs are obtained; S807: Determine the Pareto candidate scheduling scheme set based on the multiple sets of scheduling control solutions.

10. A flood evolution scheduling scheme recommendation system based on multi-objective optimization, characterized in that... This includes: processor and memory; The memory stores programs or instructions that can run on the processor, and when the processor executes the programs or instructions, it implements the steps of the flood evolution scheduling scheme recommendation method based on multi-objective optimization as described in any one of claims 1 to 9.

Images