Emergency dispatching method for downstream cascade hydropower stations based on artificial intervention in dam-break flood process of barrier lake

Abstract

The application discloses an emergency dispatching method for downstream cascade power stations based on artificial intervention dam-break lake flood process, comprising: obtaining cascade reservoir basic physical data; constructing artificial intervention scenario set, coupling diversion trench excavation parameters and breach evolution process, generating multiple groups of post-intervention dam-break flood process lines as reservoir input flood scenarios; establishing a multi-objective collaborative optimization function system containing flood peak reduction, water level safety control and discharge flow stability targets, and coordinating each target through a dynamic weighting mechanism; using a DPSA-POA coupled optimization framework to solve the cascade joint dispatching problem; based on the multi-scenario optimization results, refining the emergency dispatching rule table; multi-index comprehensive evaluation and visual output of the dispatching scheme. Effectively improve the emergency response capability of cascade power stations under conditions such as dam-break lake dam-break and other emergencies, realize efficient transformation of optimized dispatching results into executable dispatching rules, and provide reliable technical support for emergency rescue decision-making.

Description

Technical Field

[0001] This invention belongs to the field of joint flood control optimization scheduling technology of cascade reservoirs in water conservancy projects, specifically involving an emergency scheduling method for downstream cascade power stations based on artificial intervention in the flood process of landslide dam break. Background Technology

[0002] With global climate change and the increasing frequency of extreme weather events, floods have become one of the major threats to socio-economic and ecological security. According to the UN Water Resources Report (2023), the frequency of global flood events has increased by 30% in the past 20 years, particularly in Asian river basins such as the Yangtze and Yellow Rivers, where the flood situation is becoming increasingly severe. As a major water-producing country, China possesses the world's largest number of reservoirs (approximately 98,000), among which cascade reservoir systems (such as the Three Gorges cascade, the Yalong River cascade, and the Jinsha River cascade) play a crucial role in flood control and disaster reduction.

[0003] Landslide-dammed lakes are a special type of flood disaster caused by geological hazards such as earthquakes and landslides. Practice has shown that cascade reservoir groups play an irreplaceable and crucial role in emergency response to landslide-dammed lake breach floods.

[0004] In the scenario of a landslide-dammed lake bursting its banks and causing a flood, the scheduling context is particularly complex. Landslide-dammed lakes form due to sudden events such as earthquakes and landslides, and the resulting burst floods are characterized by high peak values, short durations, and sudden onset, posing a serious threat to downstream cascade hydropower stations. Inducing the orderly release of water from landslide-dammed lakes through artificial intervention measures (such as excavating diversion channels) is a crucial means of disaster mitigation and risk reduction in emergency response to landslide-dammed lakes. Existing research indicates that parameters such as the cross-sectional and longitudinal structural forms of diversion channels, and the scale of excavation, directly affect the breach evolution process and the morphology of the burst flood. However, the breach process under artificial intervention conditions exhibits significant variability and uncertainty, presenting new technical challenges to the emergency scheduling of downstream cascade hydropower stations.

[0005] Currently, several technical solutions have been proposed for emergency scheduling of cascade reservoirs in response to landslide dammed lake breach floods. CN117933670A proposes a design method for emergency scheduling of upstream and downstream reservoirs of a landslide dammed lake. By assessing the maximum storage and maximum discharge of the dammed lake, it implements downstream reservoir emptying and upstream reservoir flow increase scheduling to reduce the breach flood peak. CN120146721B considers a method for establishing a landslide dammed lake breach risk simulation model and proposes an emergency response method for cascade reservoirs oriented towards breach flood mitigation. It divides emergency scenarios based on the dynamics of the risk and the relationship with flood control resources, enabling dynamic adjustment of emergency decision-making schemes. At the algorithmic level, successive optimization algorithms (POA) and dynamic programming successive approximation algorithms (DPSA) have been applied to solve the time and space dimensionality reduction problems of reservoir group optimization scheduling. However, a comprehensive analysis of existing technologies reveals the following shortcomings: (1) Insufficient adaptability to scenarios of flood events caused by artificial intervention. Existing emergency scheduling technologies for landslide dam breaches and cascade reservoirs generally use fixed design floods or single typical flood events as model inputs, failing to fully reflect the changes in flood morphology caused by the combined effects of natural uncertainties and artificial intervention measures during flood evolution. Especially under artificial treatment of landslide dams, the diversion channel excavation method (depth, width, bottom slope, etc.) directly changes the development of the breach and the outflow process. Existing methods are unable to capture this change in a timely and accurate manner. When the flood process exhibits characteristics such as double peaks, multiple peaks, or steep rises followed by slow declines, the scheduling calculation results are prone to systematic deviations, thereby amplifying the risks in downstream water level control and reservoir utilization.

[0006] (2) The problem of multi-objective conflict lacks an effective coordination mechanism. Existing scheduling models mostly adopt a single optimization objective with peak shaving as the core, or handle multi-objective problems through simple weighting, failing to effectively characterize the inherent conflict relationship between objectives such as water level safety, water level stability, and reservoir capacity recovery. In this case, the optimization results often tend to be local optima, which means that the scheduling scheme performs well in a certain indicator, but has obvious deficiencies in terms of equipment operation safety, river stability, or subsequent flood control capacity.

[0007] (3) The efficiency of solving high-dimensional optimization problems is insufficient to meet the timeliness requirements of emergency response. The joint scheduling of cascade hydropower stations involves multiple reservoirs, multiple control sections, and decision variables in multiple time periods, resulting in a large problem scale and high dimensionality. As the number of cascades and the scheduling time periods increase, the computational complexity of traditional scheduling algorithms increases exponentially. Although dimensionality reduction algorithms such as DPSA and POA have been applied, in the case of sudden disasters such as landslide dam failures, the timeliness requirements for emergency rescue are extremely high, and the computational efficiency of existing algorithms still needs to be improved.

[0008] (4) The optimization results lack effective conversion into executable scheduling rules. Existing methods mostly output optimization results in pure numerical form, lacking explanation and rule-based expression of scheduling decision-making logic. Schedulers need to rely on human experience for secondary interpretation and rule extraction. This not only increases the time cost of emergency decision-making, but also increases the risk of human misjudgment, limiting the rapid deployment and engineering application of scientific scheduling schemes under emergency conditions.

[0009] In summary, existing technologies for emergency dispatching of cascade hydropower stations under landslide dam breach flood scenarios still suffer from problems such as insufficient adaptability to human intervention scenarios, difficulty in coordinating multi-objective conflicts, low efficiency in high-dimensional optimization solutions, and difficulty in directly guiding engineering practice with optimization results. There is an urgent need for an emergency dispatching method that can systematically solve the above-mentioned technical defects. Summary of the Invention

[0010] To address the problems of existing technologies, this invention provides an emergency dispatching method for downstream cascade hydropower stations based on artificial intervention in landslide dammed lake outburst flood processes. By introducing multiple flood process inputs into the system, the method enhances the dispatching model's adaptability to flood uncertainties and differences in artificial response plans. Through the construction of a multi-objective collaborative optimization mechanism, it coordinates the conflicting relationships between key objectives such as peak shaving, water level safety, and stable operation, avoiding dispatching imbalances caused by single-objective driving. An efficient optimization solution framework improves computational efficiency and stability under multi-reservoir joint operation conditions. Furthermore, it automatically converts optimization results into executable dispatching rules, thereby enhancing the emergency response capabilities and engineering application reliability of cascade hydropower stations under sudden disasters such as landslide dammed lake outbursts.

[0011] This invention is implemented as follows: an emergency dispatching method for downstream cascade hydropower stations based on artificial intervention in the flood process of a landslide dam breach, comprising the following steps: S1: Obtain the basic physical data of the cascade reservoirs, including water level-reservoir capacity curves, water level-discharge capacity curves, characteristic water levels, and safety control indicators; S2: Construct a set of artificial intervention scenarios, obtain the diversion channel excavation parameters during the artificial treatment of the landslide dam, the diversion channel excavation parameters include the depth, width and bottom slope of the diversion channel, couple the diversion channel excavation parameters with the breach evolution process, simulate and generate multiple sets of post-intervention breach flood process lines, as the inflow flood input scenarios for the cascade system; S3: Establish a multi-objective collaborative optimization objective function and constraint system. The objective function includes the flood peak reduction objective, the water level safety control objective, and the downstream flow stability objective. The weights of each objective are dynamically adjusted at different scheduling stages through a dynamic weighting mechanism. S4: The DPSA-POA coupled optimization framework is used to solve the tiered joint scheduling problem. The outer layer uses a successive approximation dynamic programming method to reduce the dimensionality of the high-dimensional decision space, and the inner layer uses a stepwise optimization algorithm to perform local fine optimization of key scheduling variables, and outputs the optimal scheduling result set under multiple scenarios. S5: Based on the multi-scenario optimization result samples output in step S4, an emergency dispatch rule table is formed by extracting and summarizing features. The emergency dispatch rule table includes single reservoir dispatch rules and cascade coordination rules.

[0012] Furthermore, the coupling of the diversion channel excavation parameters and the breach evolution process in step S2 specifically includes: establishing a quantitative relationship model between the depth, width, and bottom slope of the diversion channel and the breach widening rate and breach downcutting rate; simulating the time-varying process of longitudinal downcutting and lateral widening of the breach based on different diversion channel excavation conditions; and generating breach flow process lines under different intervention intensities.

[0013] Furthermore, the multi-objective collaborative optimization objective function mentioned in step S3 specifically includes: Objective function of the emergency dispatch model for hydropower station groups:

[0014] In the formula: For reservoir In the The outbound flow rate for each scheduling period, where T is the total number of scheduling periods; Minimum utilization rate of total flood control capacity

[0015] In the formula: Let be the flood control capacity utilization rate of the i-th reservoir at time t; n is the total number of cascade reservoirs; [ [] represents the scheduling time range, and T represents the total number of scheduling time periods; The cascade has the largest total power generation.

[0016] In the formula: Let be the power generation of the i-th reservoir at time t.

[0017] Furthermore, the dynamic weighting mechanism described in step S3 is as follows: based on the warning level of the landslide dam breach flood and real-time hydrological information, the weight coefficients of each objective function are dynamically adjusted. When the warning level rises, the weight of the flood peak reduction objective is increased; when the reservoir water level is close to the safe threshold, the weight of the water level safety control objective is increased; and when the outflow variation is too large, the weight of the outflow stability objective is increased.

[0018] Furthermore, the DPSA-POA coupling optimization framework described in step S4 specifically includes: The outer successive approximation dynamic programming method selects the target reservoir to be optimized in each iteration, fixes the scheduling trajectory of the other reservoirs, and dynamically updates the inflow boundary conditions of the target reservoir according to the cascade linkage relationship, thereby achieving dimensionality reduction decomposition of the high-dimensional scheduling problem. The inner-layer stepwise optimization algorithm introduces a sliding time window mechanism in the scheduling process of the selected reservoir to finely optimize the local scheduling decisions in adjacent time periods. In each time window, dynamic programming is called to search in the discrete state space to solve the optimal local scheduling state that satisfies the constraints, and the scheduling trajectory of the reservoir is updated step by step. Once the inner and outer iterations both meet the convergence conditions, the state trajectory of the entire cascade system is updated until the joint optimization calculation of all reservoirs is completed.

[0019] Furthermore, the constraints in step S4 include: Upstream and downstream hydraulic constraints:

[0020] In the formula: For reservoir In the Inbound traffic during each scheduling period; For reservoir The collection of reservoirs directly upstream; Let be the outflow from reservoir k during the j-th scheduling period; For reservoir In the The upstream water flow rate during each scheduling period; Water balance constraints:

[0021] In the formula: and Reservoirs In the With the The warehouse capacity for each scheduling period; Let be the inflow rate of reservoir i during the j-th scheduling period; Let be the outflow from reservoir i during the j-th scheduling period; The total duration corresponding to a scheduling period; Outbound flow constraints:

[0022] In the formula: and Reservoirs In the The upper and lower limits of traffic constraints within a scheduling period; Water level constraints:

[0023]

[0024] In the formula: For reservoir In the Water level during each scheduling period; and Reservoirs In the The upper and lower limits of water level constraints for each scheduling period; and Reservoirs Water levels at the beginning and end of the scheduling process; and Reservoirs The initial and final water levels; Output constraints:

[0025] In the formula: For reservoir In the Output during each scheduling period; and Reservoirs In the Upper and lower limits of output constraints for each scheduling period: And nonnegativity constraints.

[0026] Furthermore, the extraction of the single reservoir scheduling rules in step S5 specifically includes: The flow threshold is calculated using the quantile method, and the flow is divided into small flow, medium flow, large flow and extra-large flow levels; Water level threshold classification is based on flood control limit water level and design flood level; Staged fitting of linear regression model Q out =aQ in + bΔZ + c, the regression coefficients are estimated using the least squares method, and the coefficient of determination R is used as the regression coefficient. 2 Evaluate the goodness of fit; If-conditional rules are constructed based on flow rate thresholds and water level thresholds, and safety discharge constraints and amplitude limit constraints are added. Add project meaning annotations to the generated rules.

[0027] Furthermore, the extraction of the tiered coordination rules in step S5 specifically includes: The impact of upstream reservoir outflow on downstream reservoir inflow is analyzed, the flood propagation relationship between cascade reservoirs is extracted, and a time-delay coupling equation between upstream outflow and downstream inflow is established. Priority ranking of coordination rules is defined based on safety priority, peak shaving priority, and storage capacity recovery priority; Generate cascade coordination condition rules so that when the upstream outflow exceeds the safety threshold, the downstream reservoir increases its water storage capacity according to a preset ratio.

[0028] Furthermore, it also includes step S6: performing a multi-indicator comprehensive evaluation of the scheduling scheme output in step S4, and visualizing the output through a process line comparison chart and a reservoir capacity evolution chart; the multi-indicator comprehensive evaluation includes peak shaving effect evaluation, water level safety margin evaluation, and scheduling stability evaluation; the visual output includes a flood process line comparison chart before and after scheduling, a reservoir capacity evolution chart for each reservoir, and a scheduling decision support report.

[0029] Furthermore, the method is applied to the emergency joint dispatch of downstream cascade power stations in the scenario of landslide dammed lake outburst flood, wherein the cascade power stations include two or more reservoir power stations arranged in series or in combination.

[0030] The advantages and technical effects of this invention are as follows: (1) By constructing a set of artificial intervention scenarios, a high degree of adaptation between scheduling decisions and emergency response plans is achieved. Due to the impact of intervention measures such as artificial diversion channel excavation on the flood caused by the breach of the landslide dam, the flow process will be drastically reshaped. Traditional technologies are mostly based on fixed flood processes, resulting in a disconnect between scheduling plans and actual operating conditions. This invention establishes a coupled model of diversion channel parameters and breach evolution, pre-simulates multiple intervention scenarios and uses them as inputs, enabling downstream cascade power stations to match the optimal scheduling response strategy in real time according to the actual emergency response dynamics on site, effectively solving the problem of response lag in emergency scheduling.

[0031] (2) The DPSA-POA coupled optimization framework enables rapid decision-making in minutes under complex operating conditions. The joint scheduling of cascade power stations is a typical high-dimensional nonlinear problem. Traditional dynamic programming algorithms encounter the curse of dimensionality when dealing with multiple power stations, resulting in long computation time and difficulty in meeting the extremely high timeliness requirements of emergency rescue. This invention uses DPSA to reduce the dimensionality of the cascade system to a single-library optimization, and combines it with POA for fine optimization of local time windows, shortening the computation time from several hours to minutes.

[0032] (3) A multi-objective dynamic weighting mechanism achieves coordinated protection of the safety and stability of the entire basin. When dealing with landslide dammed lake floods, simply reducing the peak flow often overlooks the risk of overtopping caused by excessively high dam water levels, or the risk of downstream river scouring due to sudden changes in discharge. This invention balances peak reduction, water level safety, and flow stability in the objective function through dynamic weighting, which not only ensures the structural physical safety of the cascade dams, but also protects the stability of the downstream banks by limiting the variation of the discharge flow, thus achieving a leap from the safety of a single power station to the safety of the entire basin system.

[0033] (3) The automatic extraction technology of scheduling rules enables the direct transformation of optimization results into engineering practice. Complex optimization algorithms usually only produce discrete numerical sequences, which are difficult for front-line dispatchers to directly operate in real time based on numerical points in emergency situations. This invention summarizes massive optimization samples into an intuitive "water level-flow" mapping rule table through a rule extraction module, transforming profound mathematical models into easy-to-understand and directly executable operating procedures. This greatly lowers the threshold for emergency decision-making and ensures the accurate implementation of scientific scheduling solutions at the grassroots level. Attached Figure Description

[0034] Figure 1 This is an overall implementation flowchart of an embodiment of the present invention; Figure 2 This is a model logic functional structure diagram of an embodiment of the present invention; Figure 3Timing diagram of the core computation module of the DPSA-POA coupling algorithm in this embodiment of the invention; Figure 4 An illustration of the emergency dispatching results for elevator depots in this embodiment of the invention; Figure 5 Analysis diagram of emergency dispatch results for elevator depots in an embodiment of the present invention. Detailed Implementation

[0035] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0036] The present invention provides an emergency dispatching method for downstream cascade hydropower stations based on artificial intervention in the flood process of a landslide dam break, comprising the following: 1. Overall Technical Approach and Implementation Process The emergency dispatching method for downstream cascade hydropower stations based on artificial intervention in landslide dammed lake flood processes provided by this invention is implemented according to the overall process of "data preparation—model construction—multi-objective optimization—efficient solution—rule extraction—result evaluation". Using the basic physical parameters of the reservoir and multiple flood processes under artificial intervention scenarios as inputs, a cascade reservoir system model and a multi-objective collaborative optimization model are constructed. The DPSA-POA coupled optimization algorithm is used to complete emergency dispatching calculations under multiple scenarios, ultimately forming emergency dispatching rules that can be directly used for on-site decision-making. The dispatching effect is then comprehensively evaluated and visualized. See the detailed process below. Figure 1 .

[0037] 2. Basic Principles of Emergency Dispatch In the event of sudden disasters in a river basin (such as landslides blocking rivers, breaches of barrier lakes, and extreme flood events), the coordinated operation of cascade reservoir groups is a crucial link in preventing the spread of disaster chains and reducing downstream losses. To address the problems of delayed response, insufficient coordination, and rigid rules in traditional experience-based scheduling, this invention constructs a scheduling decision-making system of "algorithm solution—rule generation—scenario simulation—scheme evaluation," realizing the transformation from manual experience to scientific and intelligent scheduling.

[0038] The main task of cascade flood control capacity allocation is to determine different schemes for reserving flood control capacity for each reservoir at different stages of the flood season. Through simulation and comparative analysis of these schemes, the aim is to maximize the flexibility and mobility of cascade flood control scheduling without increasing the flood control risk of the cascade reservoirs, ensuring the safety of downstream flood control targets, while also improving the utilization rate of flood resources at the cascade level. A macro-level approach fails to differentiate the specific utilization and allocation of flood control capacity among the reservoirs at each cascade level. Therefore, further subdivision is needed. Simulation scheduling will still consider individual reservoirs as the basic unit, while comprehensively considering the hydraulic connections between reservoirs to ultimately determine the cascade flood control water storage sequence strategy.

[0039] The transition equation for the flood control state of a reservoir still follows the water balance principle, and its mathematical expression is:

[0040] In the formula: , The first The inflow rate of the reservoir at the beginning and end of the time period, in units of ; , The first The outflow rate of the reservoir at the beginning and end of the time period, in units of ; , The first The water storage capacity of the reservoir at the beginning and end of the time period, in units of .

[0041] The determination of joint emergency dispatch models and their combined application mechanisms should consider the following three principles: (1) Flood control safety of the reservoir itself. When a dangerous situation occurs, the highest water level of the reservoir during the flood control calculation process should not exceed the dam crest elevation. The reservoir should first ensure its own safety to the greatest extent and reduce the risk of failure. After the flood control calculation is completed, the tense flood control situation of the reservoir group should be relieved as soon as possible to ensure the absolute safety of the reservoir group.

[0042] (2) Safety of flood control points of reservoirs. Under the premise that the flood control safety of the reservoir itself is guaranteed, the focus of flood control should be placed on flood control points. Specifically, the maximum flow of the flood control point should not exceed the safe discharge, or the duration of exceeding the safe discharge should be as short as possible and the excess flood volume should be as small as possible, so as to minimize the losses of downstream flood control points.

[0043] (3) In cases where the emergency is extremely urgent or multiple emergencies are encountered together and reservoir failure is unavoidable, the main objective of emergency dispatch should be to prolong the time of reservoir failure as much as possible, to buy more time for the transfer of people and property within the protected area, thereby reducing the loss of life and property.

[0044] 3. Data preparation and scenario construction for human intervention In the data preparation phase, the basic physical data of the cascade reservoirs are first obtained, including water level-storage capacity curves, water level-discharge capacity curves, characteristic water levels, and safety control indicators, which are used to describe the basic storage capacity and operational constraints of a single reservoir.

[0045] Meanwhile, to address the impact of differences in the excavation depth, width, and method of the diversion channel during the artificial handling of landslide dammed lakes on the breach evolution and outflow process, a set of artificial intervention scenarios was constructed. By simulating the breach evolution process under different diversion channel operating conditions, multiple sets of representative post-intervention flood hydrographs were generated as inflow flood input scenarios for the cascade system, thus avoiding the problems of single input and insufficient scenario adaptability in traditional methods.

[0046] 4. Construction of Objective Function and Constraint System for Multi-Objective Collaborative Optimization To address multiple objectives related to safety, efficiency, and ecology during river basin emergencies, a multi-objective scheduling model is established, considering upstream peak shaving, downstream flood control, power plant safety, and power supply. The model's objective functions include minimizing downstream flood risk, smoothing the reservoir outflow process, and maximizing power generation. Simultaneously, reservoir capacity constraints, discharge capacity constraints, and time continuity constraints are introduced to ensure the feasibility and physical rationality of the scheduling scheme. The overall logical functional structure of the model is as follows: Figure 2 As shown.

[0047] (1) Objective function ① To ensure the safety of the hydropower station group system during the flood season to the greatest extent possible, the following formula is used as one of the objective functions of the emergency dispatch model for the hydropower station group:

[0048] In the formula: For reservoir In the Outbound flow during each scheduling period; For reservoir In the The inbound flow during each scheduling period.

[0049] ② Minimum utilization rate of total flood control capacity

[0050] In the formula: : Flood control capacity of the i-th reservoir at time t; n: Total number of cascade reservoirs; t0,t T Start and end times of the scheduling period ③ The total power generation of the cascade is the largest.

[0051] In the formula: : Power generation of the i-th reservoir at time t; n: Total number of cascade reservoirs t0,t T Start and end times of the scheduling period (2) Constraints Taking into account the limitations of the hydropower station's own operating conditions and its downstream flood control responsibilities, the established emergency dispatch model needs to meet the following constraints: ① Upstream and downstream hydraulic constraints:

[0052] In the formula: For reservoir In the Inbound traffic during each scheduling period; For reservoir The collection of reservoirs directly upstream; Let be the outflow from reservoir k during the j-th scheduling period; For reservoir In the The upstream water flow rate during each scheduling period.

[0053] ② Water balance constraints

[0054] In the formula: and Reservoirs In the With the The warehouse capacity for each scheduling period; The inflow rate of reservoir i during the j-th scheduling period; Let be the outflow from reservoir i during the j-th scheduling period; This represents the total duration corresponding to a scheduling period.

[0055] ③Outbound flow constraints

[0056] In the formula: and Reservoirs In the The upper and lower limits of the flow constraints within a scheduling period.

[0057] ④ Water level constraints

[0058]

[0059] In the formula: For reservoir In the Water level during each scheduling period; and Reservoirs In the The upper and lower limits of water level constraints for each scheduling period; and Reservoirs Water levels at the beginning and end of the scheduling process; and Reservoirs The initial and final water levels.

[0060] ⑤ Output constraints

[0061] In the formula: For reservoir In the Output during each scheduling period; and Reservoirs In the The upper and lower limits of the output constraints for each scheduling period.

[0062] ⑥ Non-negativity constraint All variables during each scheduling period, such as water level and outflow, are greater than or equal to 0.

[0063] 5. DPSA-POA Coupled Optimization Solution Mechanism In the solution phase, this invention employs the DPSA-POA coupled optimization framework to efficiently solve the tiered joint scheduling problem.

[0064] The outer layer adopts the successive approximation dynamic programming method (DPSA). In each iteration, the target reservoir to be optimized is selected, the scheduling trajectory of the other reservoirs is fixed, and the inflow boundary conditions of the reservoir are dynamically updated according to the cascade linkage relationship, thereby realizing the dimensionality reduction decomposition of the high-dimensional scheduling problem.

[0065] The inner layer employs a progressive optimization algorithm (POA), which introduces a sliding time window mechanism during the scheduling of selected reservoirs to finely optimize local scheduling decisions within adjacent time periods. Within each time window, dynamic programming is invoked to search the discrete state space, solve for the optimal local scheduling state that satisfies the constraints, and gradually update the scheduling trajectory of the reservoir.

[0066] Once the inner and outer iterations both meet the convergence condition, the state trajectory of the entire cascade system is updated until the joint optimization calculation of all reservoirs is completed, and the optimal scheduling result set under multiple scenarios is output.

[0067] Step 1: Input basic reservoir parameters and model function constraints The objective function used in the model is the optimal one for the current stage:

[0068] In the formula: Let t be the objective function for maximizing peak reduction within time interval t. and for and The outflow rate during the specified time period. All other constraints and reservoir parameters are the same as described above.

[0069] Step 2: Input the initial scheduling scheme The discharge process obtained from the scheduling rules defined in Chapter 2 is used as the initial scheduling process sequence input model, with reservoir capacity changes as the basis. Initial storage capacity during the time period is a decision variable. Storage capacity at the end of the period Given this, we can perform iterative optimization calculations.

[0070] Step 3: Solve in stages Based on the initial scheduling plan, optimization is performed time-by-time starting from the initial moment. First, the initial storage capacity value for the first time period and the storage capacity value at the end of the second time period are determined. Then, [the process is repeated]. Adjustments are made to ensure the objective function is optimal during the first and second time periods, at which point the new storage capacity value is obtained. This process of optimization, performed time-by-time, continues until new results are obtained. '、 , ……,See Figure 3 .

[0071] 6. Emergency Dispatch Rule Extraction Methods and Evaluation Visualization The emergency dispatch rule extraction method is a technique that automatically extracts operable rules from optimized dispatch results. It is applicable to cascade reservoir emergency responses under extreme events such as landslide dam breach floods. The core of this method lies in transforming numerical optimization results into thresholds, equations, and piecewise rules, facilitating rapid decision-making by dispatchers and avoiding the inefficiency and subjectivity of manual rule extraction in traditional methods. Based on optimization data from multiple floods, and combined with statistical and regression analysis, the method ensures the universality and robustness of the rules.

[0072] (1) Extraction of rules for single reservoirs For each reservoir, extract independent rules: ① Threshold calculation Flow thresholds are determined using the quantile method, such as Q1=25%, Q2=50%, Q3=75%, Q4=90%, representing small / medium / large / extra-large flow levels. Water level thresholds are categorized as follows: Z1=flood limit level, Z2=flood limit + 0.5*(design - flood limit), Z3=design flood level.

[0073] ② Stage regression equation Staged fitting of linear model Q out =aQ in+ bΔZ + c, using the least squares method. Calculate R² to evaluate the goodness of fit (>0.85 is considered valid). For example, during the flood season, a low coefficient a (0.4-0.6) is preferred, emphasizing water retention; during the receding season, a high coefficient a (0.8-1.0) is preferred, restoring reservoir capacity.

[0074] ③ Segmentation rule generation If-then rules are constructed based on thresholds, such as "If Q in <Q2and Z<Z2then Q out = 0.5 * Q in (Appropriate discharge during low flow and low water level). Boundary condition: Q out ≤ Safety leakage, |ΔQ| ≤ Amplitude limit.

[0075] ④ Explanation of the meaning of the project Add explanations to the rules, such as "prioritize water storage to maximize peak reduction, but force flood discharge when the water level exceeds Z3".

[0076] (2) Extraction of tiered coordination rules ① Coordination rules Analyze the impact of upstream shipments on downstream operations, such as "If upstream Q..." out "To ensure safety, the downstream storage and interception ratio will be increased by 20%."

[0077] ② Priority sorting Define rule priorities: 1. Safety (water level / discharge exceeding limits); 2. Maximize peak reduction; 3. Smooth recovery.

[0078] ③ Coupling equations Extract the propagation model, such as .

[0079] (3) Evaluation and visualization output of scheduling results Finally, the generated scheduling scheme is comprehensively evaluated based on multiple indicators, including peak shaving effect, water level safety margin, and scheduling stability, and visualized using process line comparison charts and reservoir capacity evolution charts. Simultaneously, a scheduling decision support report is generated, providing intuitive and reliable decision-making basis for scheduling command under emergency conditions. Model output visualizations are shown below. Figure 4 , 5 As shown.

[0080] This invention provides a systematic solution to complex emergency dispatching problems by constructing a complete technical chain of "scenario generation—multi-objective optimization—efficient solution—rule extraction." First, addressing the high uncertainty of the breach-break process under artificial intervention, a set of artificial intervention scenarios is constructed. This set comprehensively considers the coupling relationship between engineering parameters such as the depth and width of the diversion channel and the breach evolution process, simulating and generating multiple representative post-intervention breach-break flood process lines as input scenarios for the dispatching model, thereby improving the model's adaptability to different treatment schemes and flood evolution paths. Second, addressing the significant conflicts among multiple objectives in emergency dispatching, a multi-objective collaborative objective function system is established, encompassing peak flow reduction, water level safety control, and downstream flow stability constraints. A dynamic weighting mechanism coordinates and balances these objectives at different dispatching stages, avoiding safety risks or dispatching imbalances caused by a single dominant objective. Building upon this foundation, and addressing the challenges of high dimensionality and computational complexity in tiered joint scheduling, a DPSA-POA coupled optimization framework is constructed. This framework utilizes a successive approximation dynamic programming method to reduce the dimensionality of the high-dimensional decision space, and combines it with a stepwise optimization algorithm to perform local fine-tuning of key scheduling variables, significantly improving computational efficiency while maintaining solution accuracy. Finally, based on multi-scenario optimization results samples, feature extraction and inductive methods are used to refine an emergency scheduling rule table with engineering operability, effectively transforming optimized scheduling results into practical scheduling decisions.

[0081] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for emergency dispatching of downstream cascade hydropower stations based on artificial intervention in the flood process of a landslide dam break, characterized in that, Includes the following steps: S1: Obtain the basic physical data of the cascade reservoirs, including water level-reservoir capacity curves, water level-discharge capacity curves, characteristic water levels, and safety control indicators; S2: Construct a set of artificial intervention scenarios, obtain the diversion channel excavation parameters during the artificial treatment of the landslide dam, the diversion channel excavation parameters include the depth, width and bottom slope of the diversion channel, couple the diversion channel excavation parameters with the breach evolution process, simulate and generate multiple sets of post-intervention breach flood process lines, as the inflow flood input scenarios for the cascade system; S3: Establish a multi-objective collaborative optimization objective function and constraint system. The objective function includes the flood peak reduction objective, the water level safety control objective, and the downstream flow stability objective. The weights of each objective are dynamically adjusted at different scheduling stages through a dynamic weighting mechanism. S4: The DPSA-POA coupled optimization framework is used to solve the tiered joint scheduling problem. The outer layer uses a successive approximation dynamic programming method to reduce the dimensionality of the high-dimensional decision space, and the inner layer uses a stepwise optimization algorithm to perform local fine optimization of key scheduling variables, and outputs the optimal scheduling result set under multiple scenarios. S5: Based on the multi-scenario optimization result samples output in step S4, an emergency dispatch rule table is formed by extracting and summarizing features. The emergency dispatch rule table includes single reservoir dispatch rules and cascade coordination rules.

2. The emergency dispatching method for downstream cascade hydropower stations based on artificial intervention in the flood process of a landslide dam breach, as described in claim 1, is characterized in that... The coupling of the diversion channel excavation parameters and the breach evolution process in step S2 specifically includes: establishing a quantitative relationship model between the depth, width, and bottom slope of the diversion channel and the breach widening rate and breach downcutting rate; simulating the time-varying process of longitudinal downcutting and lateral widening of the breach based on different diversion channel excavation conditions; and generating breach flow process lines under different intervention intensities.

3. The emergency dispatching method for downstream cascade hydropower stations based on artificial intervention in the flood process of a landslide dammed lake breach, as described in claim 1, is characterized in that... The multi-objective collaborative optimization objective function mentioned in step S3 specifically includes: Objective function of the emergency dispatch model for hydropower station groups: In the formula: For reservoir In the The outbound flow rate for each scheduling period, where T is the total number of scheduling periods; Minimum utilization rate of total flood control capacity In the formula: Let be the flood control capacity utilization rate of the i-th reservoir at time t; n is the total number of cascade reservoirs; [ [] represents the scheduling time range, and T represents the total number of scheduling time periods; The cascade has the largest total power generation. In the formula: Let be the power generation of the i-th reservoir at time t.

4. The emergency dispatching method for downstream cascade hydropower stations based on artificial intervention in the flood process of a landslide dam breach, as described in claim 1, is characterized in that... The dynamic weighting mechanism described in step S3 is as follows: based on the warning level of the landslide dam breach flood and real-time hydrological information, the weight coefficients of each objective function are dynamically adjusted. When the warning level rises, the weight of the flood peak reduction objective is increased; when the reservoir water level is close to the safe threshold, the weight of the water level safety control objective is increased; and when the outflow variation is too large, the weight of the outflow stability objective is increased.

5. The emergency dispatching method for downstream cascade hydropower stations based on artificial intervention in the flood process of a landslide dammed lake breach, as described in claim 1, is characterized in that... The DPSA-POA coupling optimization framework described in step S4 specifically includes: The outer successive approximation dynamic programming method selects the target reservoir to be optimized in each iteration, fixes the scheduling trajectory of the other reservoirs, and dynamically updates the inflow boundary conditions of the target reservoir according to the cascade linkage relationship, thereby achieving dimensionality reduction decomposition of the high-dimensional scheduling problem. The inner-layer stepwise optimization algorithm introduces a sliding time window mechanism in the scheduling process of the selected reservoir to finely optimize the local scheduling decisions in adjacent time periods. In each time window, dynamic programming is called to search in the discrete state space to solve the optimal local scheduling state that satisfies the constraints, and the scheduling trajectory of the reservoir is updated step by step. Once the inner and outer iterations both meet the convergence conditions, the state trajectory of the entire cascade system is updated until the joint optimization calculation of all reservoirs is completed.

6. The emergency dispatching method for downstream cascade hydropower stations based on artificial intervention in the flood process of a landslide dam breach, as described in claim 1, is characterized in that... The constraints in step S4 include: Upstream and downstream hydraulic constraints: In the formula: For reservoir In the Inbound traffic during each scheduling period; For reservoir The collection of reservoirs directly upstream; Let be the outflow from reservoir k during the j-th scheduling period; For reservoir In the The upstream water flow rate during each scheduling period; Water balance constraints: In the formula: and Reservoirs In the With the The warehouse capacity for each scheduling period; Let be the inflow rate of reservoir i during the j-th scheduling period; Let be the outflow from reservoir i during the j-th scheduling period; The total duration corresponding to a scheduling period; Outbound flow constraints: In the formula: and Reservoirs In the The upper and lower limits of traffic constraints within a scheduling period; Water level constraints: In the formula: For reservoir In the Water level during each scheduling period; and Reservoirs In the The upper and lower limits of water level constraints for each scheduling period; and Reservoirs Water levels at the beginning and end of the scheduling process; and Reservoirs The initial and final water levels; Output constraints: In the formula: For reservoir In the Output during each scheduling period; and Reservoirs In the Upper and lower limits of output constraints for each scheduling period: And nonnegativity constraints.

7. The emergency dispatching method for downstream cascade hydropower stations based on artificial intervention in the flood process of a landslide dam breach, as described in claim 1, is characterized in that... The extraction of single reservoir scheduling rules in step S5 specifically includes: The flow threshold is calculated using the quantile method, and the flow is divided into small flow, medium flow, large flow and extra-large flow levels; Water level threshold classification is based on flood control limit water level and design flood level; Staged fitting of linear regression model Q out =aQ in + bΔZ + c, the regression coefficients are estimated using the least squares method, and the coefficient of determination R is used as the regression coefficient. 2 Evaluate the goodness of fit; If-conditional rules are constructed based on flow rate thresholds and water level thresholds, and safety discharge constraints and amplitude limit constraints are added. Add project meaning annotations to the generated rules.

8. The emergency dispatching method for downstream cascade hydropower stations based on artificial intervention in the flood process of a landslide dammed lake breach, as described in claim 1, is characterized in that... The extraction of the tiered coordination rules in step S5 specifically includes: The impact of upstream reservoir outflow on downstream reservoir inflow is analyzed, the flood propagation relationship between cascade reservoirs is extracted, and a time-delay coupling equation between upstream outflow and downstream inflow is established. Priority ranking of coordination rules is defined based on safety priority, peak shaving priority, and storage capacity recovery priority; Generate cascade coordination condition rules so that when the upstream outflow exceeds the safety threshold, the downstream reservoir increases its water storage capacity according to a preset ratio.

9. The emergency dispatching method for downstream cascade hydropower stations based on artificial intervention in the flood process of a landslide dammed lake breach, as described in claim 1, is characterized in that... It also includes step S6: performing a multi-indicator comprehensive evaluation of the scheduling scheme output in step S4, and visualizing the output through process line comparison charts and reservoir capacity evolution charts; the multi-indicator comprehensive evaluation includes peak shaving effect evaluation, water level safety margin evaluation, and scheduling stability evaluation; the visual output includes flood process line comparison charts before and after scheduling, reservoir capacity evolution charts for each reservoir, and scheduling decision support reports.

10. The emergency dispatching method for downstream cascade hydropower stations based on artificial intervention in the flood process of a landslide dammed lake breach, as described in any one of claims 1 to 9, is characterized in that... The method is applied to the emergency joint dispatch of downstream cascade power stations in the scenario of landslide dammed lake outburst flood, wherein the cascade power stations include two or more reservoir power stations arranged in series or in combination.

Images