A reservoir flood control multi-objective optimization scheduling method, device and program product

Abstract

The present application relates to the technical field of reservoir flood control scheduling and water resources system intelligent optimization, and particularly relates to a reservoir flood control multi-objective optimization scheduling method, device and program product, to constitute a candidate outflow sequence by outflow of each period in the scheduling period, and to deduce the candidate water level process according to the water balance relationship and the reservoir capacity-water level relationship; to construct a time sequence cumulative sensitivity matrix of the water level to the outflow correction amount, and to solve a quadratic programming with the minimum correction amount two norm as the target, to carry out low disturbance projection repair on the candidate outflow sequence; to recheck the real water level after the repair, and to feed back the residual constraint violation amount to the augmented Lagrange constraint evaluation and Pareto external archive maintenance process; when the gate control is adopted, the repaired outflow is converted into the gate opening hole number and the execution constraint is checked. The present application can improve the feasible solution proportion of the candidate scheduling scheme in the strong constraint flood control scheduling, and enhance the matching of the output scheduling scheme and the engineering operation constraint.

Description

Technical Field

[0001] This invention relates to the field of reservoir flood control scheduling and intelligent optimization technology of water resources systems, specifically to a multi-objective optimization scheduling method, equipment and program products for reservoir flood control, and particularly to a multi-objective optimization scheduling method for reservoir flood control based on the water level-outflow time series cumulative sensitivity matrix, minimum disturbance quadratic programming projection repair and residual constraint verification evaluation. Background Technology

[0002] Reservoir flood control scheduling is a multi-objective, strongly constrained, nonlinear, discrete-continuous hybrid decision-making problem. In engineering practice, the scheduling scheme usually needs to simultaneously meet constraints such as the upper limit of flood control level, necessary water level control, outflow capacity, downstream safe discharge, ecological flow, and final water level, while also taking into account objectives such as flood peak reduction, control of the highest water level in front of the dam, and smooth opening and closing of gates.

[0003] Existing methods for optimizing reservoir or reservoir group scheduling often employ swarm intelligence or evolutionary algorithms such as particle swarm optimization, genetic algorithms, gray wolf algorithms, and whale optimization algorithms to search for scheduling schemes. Infeasible candidate solutions are addressed through penalty functions, boundary truncation, random resampling, constraint corridor correction, or feasible region pre-identification. While these methods can improve the proportion of feasible solutions to some extent, they still present the following problems in strongly constrained flood control scheduling scenarios: First, water level constraints are formed by the accumulation of water balance over time periods; changes in outflow in a single time period affect the water level in that period and subsequent periods, making it difficult for ordinary penalty functions to provide a clear direction for correction. Second, boundary truncation or random resampling can easily disrupt the search direction of the original candidate solutions, causing the population to fluctuate near the feasible region boundary. Third, even after discretizing the number of gate openings, continuous outflow schemes may still exhibit water level verification deviations or frequent gate opening and closing.

[0004] Therefore, there is a need for a multi-objective optimization scheduling method for reservoir flood control that can utilize the time-series cumulative structure formed by water balance and reservoir capacity-water level relationship to perform low-disturbance feasible region projection repair on candidate outflow sequences, and to provide feedback evaluation on residual constraints after verification of actual water levels. Summary of the Invention

[0005] To address the aforementioned technical problems, this invention provides a multi-objective optimization scheduling method, equipment, and program product for reservoir flood control. By constructing a time-series cumulative sensitivity matrix of water level correction to outflow, the strongly coupled water level constraint is transformed into a linearized repair constraint for candidate outflow sequences. A quadratic programming problem is solved with the objective of minimizing the L2 norm of the correction, achieving low-disturbance projection repair of the candidate outflow sequences. After projection repair, the results are verified using actual water balance and reservoir capacity-water level relationships. The residual constraints from this verification are then fed back to the augmented Lagrange constraint evaluation and Pareto external archive maintenance processes, thereby outputting a multi-objective feasible scheduling scheme that meets both flood control safety and engineering operation requirements.

[0006] This invention is achieved through the following technical solution:

[0007] A multi-objective optimization scheduling method for reservoir flood control includes:

[0008] Acquire reservoir scheduling data, which includes inflow sequence, scheduling step size, initial water level, reservoir capacity-water level relationship, water level constraint, outflow capacity constraint, ecological flow constraint, and final water level constraint;

[0009] The candidate outflow sequence is constructed by the outflow of the reservoir in each period during the scheduling period, and a multi-objective optimization model for reservoir flood control is constructed. The multi-objective optimization model for reservoir flood control takes the flood peak reduction index and the highest water level in front of the dam index as scheduling objectives, and further takes the number of gate opening and closing as the scheduling objective when gate control is adopted.

[0010] The population of the swarm intelligence optimization algorithm is initialized to obtain multiple candidate outflow sequences. For any candidate outflow sequence, the reservoir capacity and water level of each time period corresponding to the candidate outflow sequence are deduced based on the water balance relationship and the reservoir capacity-water level relationship to obtain the candidate water level process.

[0011] Based on the candidate water level process, a constraint sensitivity matrix for the water level correction to the outflow is constructed. The constraint sensitivity matrix is ​​used to characterize the temporal cumulative impact of the outflow correction in the k-th time period on the water level in the t-th time period. When k>t, the corresponding matrix element is zero, and when k≤t, the corresponding matrix element is determined by the scheduling step size and the slope of the reservoir capacity-water level relationship at the corresponding water level.

[0012] A linearized water level constraint is established for the candidate outflow sequence based on the constraint sensitivity matrix, and a quadratic programming problem is solved with the objective of minimizing the L2 norm of the outflow correction. The candidate outflow sequence is then projected and repaired into the repaired outflow sequence.

[0013] The repaired outflow sequence is resubstituted into the water balance relationship and the reservoir capacity-water level relationship to verify the actual water level, and the verified constraint residual is obtained.

[0014] Based on the verified constraint residuals, an augmented Lagrange constraint evaluation value is constructed, and the candidate outflow flow sequences after projection repair are compared and selected based on the augmented Lagrange constraint evaluation value.

[0015] The candidate outflow sequence after projection repair and constraint evaluation is sorted by non-dominated sorting. Non-dominated solutions that meet the constraints or whose constraint residuals do not exceed the preset allowable threshold are maintained in the Pareto external archive. The population of the swarm intelligent optimization algorithm is updated from the Pareto external archive by selecting the guiding solution.

[0016] The process iteratively executes the steps of inferring candidate water levels and updating the population of the swarm intelligence optimization algorithm until the termination condition is met. Then, it outputs the set of non-dominated feasible scheduling schemes in the Pareto external archive, or selects the final scheduling scheme from the set of non-dominated feasible scheduling schemes.

[0017] Optionally, the water level constraints include upper limits and lower limits for water levels in each time period;

[0018] The outbound capacity constraints include the minimum and maximum outbound flow rates allowed for each time period;

[0019] The final water level constraint is either a target value constraint for the final water level or an allowable range constraint for the final water level.

[0020] The constraint sensitivity matrix is ​​a lower triangular matrix with time-series cumulative characteristics. Its matrix elements are the partial derivatives of the water level correction amount of the corresponding time period with respect to the outflow of the corresponding time period. The partial derivatives are obtained by analytically differentiating the water balance relationship and the reservoir capacity-water level relationship.

[0021] Optionally, the elements of the constraint sensitivity matrix For the first Water level during the period of time The sensitivity of the outbound flow correction amount for a given period is determined by the following formula: ,in, For the first Water level during a certain period For the first Outbound flow during different time periods For the first Storage capacity during specific time periods;

[0022] ,in, The reservoir capacity-water level curve at water level The derivative at point, For scheduling step size;

[0023] The elements of the constraint sensitivity matrix are: .

[0024] Optionally, the secondary planning uses the correction amount of the candidate outflow sequence. Let the decision variable be the minimum L2 norm of the correction variable, and let the objective function be: ,in, For the first Time period outbound flow correction amount This represents the total number of scheduling periods;

[0025] The constraints of the quadratic programming include:

[0026] Water level constraints, for any given time period ,have ;

[0027] Outbound capacity constraints, for any given time period ,have ;

[0028] Ecological flow constraints, for any given time period ,have ;

[0029] Final water level constraint, ;

[0030] in, Candidate outbound flow sequences, The first one calculated under the candidate outflow sequence Water level during a certain period and The first The lower and upper limits of water level for a given period of time. and The first Minimum and maximum outbound flow rates allowed for a given time period. For the first Lower limit of ecological flow during a given time period. and These are the lower and upper limits of the allowable final water level range, respectively. These are the elements of the constraint sensitivity matrix;

[0031] When the final water level constraint adopts the target value of the final water level season ;

[0032] The repaired outbound flow sequence is as follows: ;

[0033] When the quadratic programming has no feasible solution, non-negative slack variables are configured for each type of linearization constraint, transforming the original constraints that were required to be fully satisfied into soft constraints that allow for finite violations. The quadratic programming with slack variables is solved with the objective of minimizing the sum of the L2 norm of the outflow correction and the weighted penalties of the slack variables. The penalty weights corresponding to the upper limit of the flood control level constraint and the outflow capacity constraint are greater than the penalty weights corresponding to the ecological flow constraint, the final water level constraint, and the lower limit of the water level constraint. After solving, the restored outflow sequence is resubmitted into the water balance relationship and the reservoir capacity-water level relationship for verification, and the remaining constraint residuals after verification are included in the subsequent augmented Lagrange constraint evaluation value.

[0034] Optionally, the verified constraint residue includes the water level constraint residue generated after the verification of the actual water level, the outflow capacity constraint residue, the ecological flow constraint residue, the final water level constraint residue, and the discretized residue generated by converting the outflow flow into the number of gate openings when using gate control.

[0035] The verified constraint residuals are written as inequality constraints. And construct the augmented Lagrangian constraint evaluation value: ,in, The candidate scheduling schemes are after projection restoration and verification of actual water levels. For the first Each constraint corresponds to a dual variable. The penalty coefficient is... To constrain the total number, This is a multi-objective ranking evaluation value used for comparing candidate scheduling schemes. It is a positive part function;

[0036] The dual variables and penalty coefficients are adaptively updated according to the changes in constraint violation during the iteration process.

[0037] Optionally, when the repaired outflow sequence needs to be executed by the gate, the repaired outflow sequence is converted into a gate opening number sequence according to the gate-flow relationship;

[0038] No. The actual outflow rate of the gate during a given time period is determined by the following formula: ,in, For the first Actual outflow from the gate during the specified time period For the first Number of gate openings during the time period For the first Water level during a certain period The gate current function;

[0039] Outbound flow after repair Select the corresponding number of gate opening holes from the set of openable holes. ,make and The deviation meets the preset conditions;

[0040] When the gate is a gate with equal orifice and the same type, the number of opening orifices of the gate is determined by the following formula: ,in, For a single-hole gate at the water level The current carrying capacity below, This represents the maximum number of openings that can be made on the gate. Indicates rounding up;

[0041] After completing the conversion of the number of gate openings, the actual outflow for each period is recalculated based on the gate-flow relationship, and the water level for each period is checked by substituting the water balance relationship and the reservoir capacity-water level relationship. If the check result violates the constraints, the opening sequence with the smaller constraint violation and fewer gate opening and closing times is selected from the adjacent opening number combination.

[0042] Optionally, the gate opening and closing frequency index includes a gate opening and closing orifice frequency index, wherein the gate opening and closing orifice frequency index is: ,in, The number of times the gates are opened and closed during the scheduling period. This represents the absolute value operation;

[0043] The gate opening and closing frequency index also includes the gate adjustment period index, which is: ,in, This refers to the number of time periods during which the number of gate openings changes within the scheduling period. For indicator functions;

[0044] The reservoir flood control multi-objective optimization model also includes gate change rate constraints: ,in, This represents the maximum allowable change in the number of gate openings between adjacent time periods.

[0045] An electronic device includes a processor, a memory, and a computer program stored in the memory and capable of running on the processor. When the processor executes the computer program, it implements the multi-objective optimization scheduling method for reservoir flood control as described above.

[0046] A computer program product includes a computer program / instructions that, when executed by a processor, implement a reservoir flood control multi-objective optimization scheduling method as described above.

[0047] Compared with existing multi-objective reservoir scheduling methods that mainly rely on penalty functions, boundary truncation, random resampling, constraint corridor correction, or feasible region pre-identification, the present invention has at least the following advantages:

[0048] This invention constructs a time-series cumulative sensitivity matrix of water level to outflow correction based on water balance relationship and reservoir capacity-water level relationship, transforming the cross-time period coupled water level constraint into a linearized repair constraint of candidate outflow sequence, so that infeasible candidate solutions have a clear repair direction;

[0049] This invention aims to solve a quadratic programming problem by minimizing the L2 norm of the outflow correction. While satisfying constraints on water level, outflow capacity, ecological flow, and final water level, it tries to retain the original candidate solution search direction generated by the swarm intelligence algorithm, thereby improving the proportion of feasible solutions and search stability under strong constraints.

[0050] This invention re-executes the actual water balance and reservoir capacity-water level relationship verification after projection repair, and feeds back the linearization error, relaxation residue and gate discretization deviation to the augmented Lagrangian constraint evaluation, so that the final output scheme is more in line with the actual engineering execution requirements.

[0051] This invention preserves a multi-objective trade-off scheme between flood peak reduction, dam front maximum water level control, and gate opening and closing smoothness through Pareto external archives, making it easier for dispatchers to select the final dispatch scheme based on flood control safety and operation management preferences. Attached Figure Description

[0052] The accompanying drawings illustrate exemplary embodiments of the present invention and, together with the description thereof, serve to explain the principles of the invention. These drawings are included to provide a further understanding of the invention and are incorporated in and constitute a part of this specification, but do not constitute a limitation on the embodiments of the present invention.

[0053] Figure 1 A flowchart illustrating the overall process of the reservoir flood control multi-objective optimization scheduling method provided in this embodiment of the invention;

[0054] Figure 2 This is a schematic diagram of the constraint sensitivity feasible region projection repair process provided in an embodiment of the present invention. Detailed Implementation

[0055] 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 and embodiments. It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the scope of the invention.

[0056] It should also be noted that, for ease of description, only the parts relevant to the present invention are shown in the accompanying drawings.

[0057] Where there is no conflict, the embodiments and features described in the present invention can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0058] Example 1

[0059] This embodiment provides a multi-objective optimization scheduling method for reservoir flood control, the overall process of which is as follows: Figure 1 As shown, it includes steps S1 to S8.

[0060] Step S1: Obtain reservoir scheduling data.

[0061] Reservoir scheduling data includes:

[0062] Inbound flow sequence ,in For the first Inbound flow during specific time periods For discrete time period indexes, This represents the total number of scheduling periods;

[0063] Scheduling step size This indicates the time interval between adjacent periods; when the inflow... and outbound flow With m 3 When / s is used, the scheduling step size is... Calculated in seconds; when the original scheduling step size is expressed in hours or days, it is first converted to seconds. The unit is consistent with the storage capacity unit.

[0064] initial water level This indicates the water level in front of the dam at the start of the dispatching process;

[0065] Reservoir capacity-water level relationship or its inverse function ,in For storage capacity, Water level;

[0066] Water level constraints and , respectively representing the first The lower and upper limits of the water level allowed for a given time period;

[0067] Outbound capacity constraints and , respectively representing the first Minimum and maximum outbound flow allowed during the time period;

[0068] Ecological flow constraints , indicating the first Lower limit of ecological outflow during specific time periods;

[0069] The final water level constraint can be the target value of the final water level. It can also be the allowable range of the final water level. In a specific example, let This means that the water level in front of the dam must return to the starting water level when the dispatching is completed.

[0070] If gate control is used, the maximum number of gate openings should also be obtained. With the gate's overcurrent function.

[0071] Step S2: Construct a multi-objective optimization model for reservoir flood control.

[0072] Let the candidate outflow sequence be ,in, For the first Outbound flow rate over a given time period. The swarm intelligence optimization algorithm uses the candidate outbound flow rate sequences as candidate search objects.

[0073] The objectives of the reservoir flood control multi-objective optimization model include at least the flood peak reduction index and the highest water level in front of the dam index; when the scheduling plan needs to be implemented through the gates, it also includes the gate opening and closing frequency index.

[0074] In a specific example, the flood peak reduction indicator is taken as... The highest water level index in front of the dam is taken as follows: The number of times the gate is opened and closed is taken as the index. ,in, The number of gate openings was calculated from the sequence of gate openings obtained by converting the repaired outflow sequence.

[0075] Step S3: Initialize the population for the swarm intelligence optimization algorithm to obtain multiple candidate outflow flow sequences. ,in For individual indexes, , Population size.

[0076] Swarm intelligence optimization algorithms refer to optimization algorithms based on population iterative search, including but not limited to particle swarm optimization, gray wolf optimization, whale optimization, or genetic algorithms. This step uses a swarm intelligence optimization algorithm to generate and iteratively update candidate outbound flow sequences within the outbound flow decision space.

[0077] Step S4: Calculate the water level evolution for any candidate outflow sequence.

[0078] Based on the water balance relationship, the reservoir capacity is accumulated periodically by the difference between inflow and outflow in each time period. .in, For the first Storage capacity during different time periods For the first Storage capacity during different time periods For the first Inbound flow during specific time periods For the first Outflow rate for each time period; then, the water level for each time period is obtained from the reservoir capacity-water level relationship. .

[0079] Step S5: Construct the constraint sensitivity matrix of water level on outflow correction, and perform feasible region projection repair on candidate outflow sequences.

[0080] Since the water level is accumulated over time by the outflow through water balance, adjusting the outflow in a certain time period will change the water level in that time period and all subsequent time periods. The sensitivity matrix is ​​a tool used to quantitatively characterize this transmission relationship.

[0081] The candidate outflow sequence generated by the swarm intelligence optimization algorithm is repaired by solving a quadratic programming problem with the objective of minimizing the L2 norm of the correction. This allows the candidate outflow sequence to be projected onto the feasible region defined by water level constraints, outflow constraints, ecological flow constraints, and final water level constraints while preserving the original search direction as much as possible.

[0082] Step S6: Perform dual adaptive processing on the constraints based on the augmented Lagrangian function.

[0083] This step primarily evaluates the constraint residuals that still exist after projection repair and verification using actual water balance and reservoir capacity-water level relationships, as well as the constraint deviations caused by discretization of the number of gate openings. Each constraint condition is then expressed as an inequality constraint. ,in For constrained indexes, , Given the total number of constraints; calculate the constraint violation rate and update the dual variable. With penalty coefficient Thus, the augmented Lagrange constraint evaluation value used for population selection comparison is obtained.

[0084] Step S7, Maintain Pareto external files .

[0085] After projection repair in step S5 and constraint evaluation in step S6, non-dominated sorting is performed on the candidate outflow sequence. Non-dominated solutions that meet the constraints or whose constraint residuals do not exceed the preset allowable threshold are added to the external archive. When the size of the external archive exceeds the capacity limit... At that time, based on the crowd distance Alternatively, the grid density can be reduced to remove overly dense solutions in the target space to maintain scheme diversity; and a guiding solution can be selected from the external archive, combined with the augmented Lagrange constraint evaluation value to update the population and obtain the next generation of candidate outflow sequences.

[0086] Step S8: Iterate through steps S4 to S7 until the number of iterations reaches the upper limit or the termination condition such as convergence is met, and output the set of non-dominated feasible scheduling schemes in the external archive; or, based on the set of schemes, select the final scheduling scheme according to decision preference and output its corresponding outbound sequence or gate opening sequence.

[0087] Through the above steps, the candidate outflow sequence is first repaired by sensitivity projection in each generation, and then evaluated by real water level evolution and augmented Lagrange constraints, so that the population search gradually focuses on feasible and non-dominated scheduling schemes.

[0088] Example 2

[0089] This embodiment is based on Embodiment 1, such as... Figure 2As shown, the construction method of the constraint sensitivity matrix in step S5 and the quadratic programming method used for projection repair are further explained.

[0090] The constraint sensitivity matrix is ​​denoted as ,in Indicates the first Water level during the period of time Sensitivity of outbound flow correction amount for different time periods Index for outbound time period, .

[0091] The partial derivative of the water level with respect to the outflow correction is decomposed into the product of the partial derivative of the water level with respect to the reservoir capacity and the partial derivative of the reservoir capacity with respect to the outflow correction: The partial derivative of water level with respect to reservoir capacity is the reciprocal of the slope of the reservoir capacity-water level curve. The reservoir capacity-water level curve at water level derivative at point This can be approximated by taking the slope of the corresponding segment through piecewise linear interpolation of the reservoir capacity-water level curve. This is derived from the discrete water balance equation. It can be seen that the storage capacity is related to the first The partial derivative of the outbound flow rate over a given period is: , that is, the first The outflow from the reservoir during a given period only affects the reservoir capacity during that period and subsequent periods, but has no impact on the reservoir capacity during previous periods. This is consistent with the time-series cumulative characteristics of water levels.

[0092] Combining the above two equations, we obtain the elements of the sensitivity matrix: .

[0093] Candidate outbound flow sequences generated by swarm intelligence optimization algorithm Correction amount based on outbound flow Let be the decision variables, where Indicates the first The correction amount for candidate outbound flow rate during the time period.

[0094] It should be noted that the quadratic programming in this embodiment is not used to resolve the entire multi-objective scheduling problem, but rather serves as a local repair operator after the swarm intelligence optimization algorithm generates candidate outflow sequences each time. This local repair operator uses the correction amount of the candidate outflow sequence as the decision variable and aims to minimize the L2 norm of the correction amount. While preserving the original candidate sequence search direction as much as possible, it projects the candidate sequence into a linearized feasible region defined by water level constraints, outflow capacity constraints, ecological flow constraints, and final water level constraints.

[0095] The goal of projection repair is to bring candidate sequences into the feasible region while minimizing the perturbation of the original candidate sequences. Therefore, the objective function is to minimize the L2 norm of the correction. .in, Representing vectors The 2-norm (Euclidean norm).

[0096] Constraints include the following four categories:

[0097] Water level constraints, for any given time period ,have ;

[0098] in, and They represent the first Lower and upper limits of water level for different time periods Indicating in the candidate outbound sequence The following is the result of calculation in step S4. Water level during a certain period That is, the first result caused by the correction amount being passed through the sensitivity matrix. Water level increase over time period.

[0099] Outbound capacity constraints, for any given time period ,have ;in and The first Minimum and maximum outbound flow allowed during the time period.

[0100] Ecological flow constraints, for any given time period ,have This ensures that the corrected outflow is not lower than the lower limit of the ecological discharge flow for the corresponding period.

[0101] Final water level constraint, ;in, and These are the lower and upper limits of the permissible final water level range, respectively. When engineering scheduling requires the final water level to reach a specified target value... season The above constraints degenerate into equation-based final water level constraints.

[0102] After solving the above quadratic programming problem, let the corrected outbound sequence be: .

[0103] While satisfying constraints on water level, outflow, ecology, and final water level, the correction L2 norm is minimized, thereby improving the proportion of feasible solutions and convergence stability.

[0104] get Then, the reservoir capacity is recalculated time-by-time according to the water balance relationship, and the actual water level sequence is calculated from the reservoir capacity-water level relationship; when the deviation between the actual water level sequence and the linearized projection result exceeds the preset allowable value, it is adjusted accordingly. The sensitivity matrix is ​​reconstructed as a new candidate sequence and projection repair is performed again.

[0105] The reprojection repair can be repeated until the maximum deviation between the actual water level sequence and the linearized water level sequence does not exceed the preset water level tolerance error, or the number of repeated repairs reaches the preset upper limit. After stopping the reprojection repair, if there is still a constraint residual after the actual water level is checked, the constraint residual is handled according to the method in Example 3. The form is included in the augmented Lagrange constraint evaluation value for subsequent optimal comparison.

[0106] When the quadratic programming problem has no feasible solution, non-negative slack variables are introduced into various linearization constraints.

[0107] Specifically, the upper and lower limits of water level, the upper and lower limits of outflow capacity, the ecological flow constraint, and the final water level constraint are each broken down and uniformly written as the first... Linearized inequality constraints And rewritten as ,in, For the first Each constraint corresponds to a slack variable. To constrain the total number.

[0108] The objective function of a quadratic programming problem with slack variables can be written as follows: ,in, To relax the penalty coefficient, For the first The penalty weights corresponding to each constraint. To reflect the priority of flood control scheduling, the upper limit constraint of flood control water level and the constraint of outflow capacity correspond to... The value is greater than the ecological flow constraint, the final water level constraint, and the lower limit water level constraint. This prioritizes reducing the excess of flood control level and excess of reservoir outflow capacity in the optimization solution. After obtaining the corrected outflow sequence, the actual water level is verified according to the water balance relationship and the reservoir capacity-water level relationship. The remaining constraint residues after verification are then handled according to the method described in Example 3. The form is included in the augmented Lagrange constraint evaluation value.

[0109] Example 3

[0110] This embodiment, based on Embodiment 1, explains the constraint evaluation based on the augmented Lagrangian function in step S6.

[0111] This constraint evaluation is mainly used to handle the constraint residuals that still exist after the actual water level verification following the projection repair, as well as the constraint deviations caused by the discretization of the number of gate openings.

[0112] The augmented Lagrange constraint evaluation in this embodiment does not replace the quadratic programming projection repair in Embodiment 2. Instead, it is used to evaluate the residual constraint violations that still exist after the projection repair, and feeds back the residual constraint violations to the individual selection comparison process of the swarm intelligence optimization algorithm.

[0113] The inequality constraints in the reservoir flood control multi-objective optimization model, as well as the inequality constraints derived from the final water level equality constraint, are uniformly written as follows: ,in Constructing augmented Lagrangian constraint evaluation values: .

[0114] in, This is a multi-objective ranking evaluation value used for comparing the best among individuals in swarm intelligence. For the first The dual variables of each constraint, The penalty coefficient is... When the constraint violation rate has not decreased sufficiently, the penalty coefficient is increased; otherwise, the penalty coefficient remains unchanged, and the dual variable is updated based on the constraint violation rate. This allows the verified constraint residuals to be fed back into the population optimization update process.

[0115] Example 4

[0116] Based on Example 1, this embodiment further explains the conversion method between outflow and gate opening number when the repaired outflow sequence needs to be executed by the gate, as well as the gate opening and closing related objectives and constraints.

[0117] For the outbound flow obtained through projection repair in Example 2 The number of gate openings is determined based on the gate-flow relationship. The actual outflow from the gate during the specified time period is: ,in, For the first Actual outflow from the gate during the specified time period For the first Number of gate openings during the time period For the first Water level during a certain period This is the gate's overcurrent function.

[0118] In a specific example, when the gate is a gate of equal orifice type, the number of opening orifices is determined by the following formula: ,in, For a single-hole gate at the water level The current carrying capacity below, This represents the maximum number of openings that can be made on the gate. Indicates rounding up. and These represent taking the minimum and maximum values, respectively.

[0119] After completing the conversion of the number of gate opening holes, according to The actual outflow rate is recalculated, and the actual water level is calculated by substituting the water balance relationship back into the process. If the actual water level or the actual outflow rate violates the constraints, the sequence of openings with the smaller constraint violation and fewer gate openings is selected from the combinations of adjacent openings.

[0120] The gate opening and closing frequency index includes the gate opening and closing orifice frequency index, which is: ,in, The number of times the gates are opened and closed during the scheduling period. This represents absolute value operations.

[0121] To further differentiate the magnitude of changes in the number of orifices and the frequency of adjustments, the number of gate adjustment periods can also be defined: ,in, This indicates the number of time periods during which the number of gate openings changes within the scheduling period. This is an indicator function.

[0122] In one implementation, to further reduce frequent opening and closing, a gate change rate constraint is set in the reservoir flood control multi-objective optimization model: ,in, This represents the absolute value operation. This represents the maximum allowable change in the number of gate openings between adjacent time periods. By limiting the jump in the number of openings between adjacent time periods, the gate operation in the output scheme becomes smoother, reducing equipment wear.

[0123] Example 5

[0124] An electronic device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the aforementioned multi-objective optimization scheduling method for reservoir flood control.

[0125] Memory is used to store software programs and modules. The processor executes various terminal functions and data processing by running the software programs and modules stored in memory. Memory can mainly include a program storage area and a data storage area. The program storage area can store the operating system, at least one executable program required for a given function, etc.

[0126] The storage data area can store data created based on the use of the terminal. Furthermore, the memory can include high-speed random access memory, and may also include non-volatile memory, such as at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0127] A computer program product includes a computer program / instructions that, when executed by a processor, implement any one of the above-mentioned methods for multi-objective optimization scheduling of reservoir flood control.

[0128] Computer program products include computer programs or instruction sets used to perform specific tasks or achieve specific functions. These programs or instructions are designed to be executed by a processor to implement a series of predefined steps or operations. The program product may be stored in various forms of computer storage media, such as memory, hard disks, solid-state drives, optical discs, or other forms of digital storage devices. It may exist in the form of compiled binary code or in the form of scripts or bytecode that can be executed by an interpreter. Through carefully designed algorithms and logical instructions, the program product enables the processor to process data in a specific order and manner, performing various functions such as data analysis, user interaction, and device control.

[0129] In the description of this specification, the references to terms such as "one embodiment / mode," "some embodiments / modes," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment / mode or example is included in at least one embodiment / mode or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment / mode or example. Moreover, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments / modes or examples. Furthermore, without contradiction, those skilled in the art can combine and integrate the different embodiments / modes or examples described in this specification, as well as the features of different embodiments / modes or examples.

[0130] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this application, "multiple" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0131] Those skilled in the art should understand that the above embodiments are merely for illustrating the present invention and are not intended to limit the scope of the invention. Those skilled in the art can make other changes or modifications based on the above invention, and these changes or modifications still fall within the scope of the present invention.

Claims

1. A multi-objective optimization scheduling method for reservoir flood control, characterized in that, include: Acquire reservoir scheduling data, which includes inflow sequence, scheduling step size, initial water level, reservoir capacity-water level relationship, water level constraint, outflow capacity constraint, ecological flow constraint, and final water level constraint; The candidate outflow sequence is constructed by the outflow of the reservoir in each period during the scheduling period, and a multi-objective optimization model for reservoir flood control is constructed. The multi-objective optimization model for reservoir flood control takes the flood peak reduction index and the highest water level in front of the dam index as scheduling objectives, and further takes the number of gate opening and closing as the scheduling objective when gate control is adopted. The population of the swarm intelligence optimization algorithm is initialized to obtain multiple candidate outflow sequences. For any candidate outflow sequence, the reservoir capacity and water level of each time period corresponding to the candidate outflow sequence are deduced based on the water balance relationship and the reservoir capacity-water level relationship to obtain the candidate water level process. Based on the candidate water level process, a constraint sensitivity matrix for the water level correction to the outflow is constructed. The constraint sensitivity matrix is ​​used to characterize the temporal cumulative impact of the outflow correction in the k-th time period on the water level in the t-th time period. When k>t, the corresponding matrix element is zero, and when k≤t, the corresponding matrix element is determined by the scheduling step size and the slope of the reservoir capacity-water level relationship at the corresponding water level. A linearized water level constraint is established for the candidate outflow sequence based on the constraint sensitivity matrix, and a quadratic programming problem is solved with the objective of minimizing the L2 norm of the outflow correction. The candidate outflow sequence is then projected and repaired into the repaired outflow sequence. The repaired outflow sequence is resubstituted into the water balance relationship and the reservoir capacity-water level relationship to verify the actual water level, and the verified constraint residual is obtained. Based on the verified constraint residuals, an augmented Lagrange constraint evaluation value is constructed, and the candidate outflow flow sequences after projection repair are compared and selected based on the augmented Lagrange constraint evaluation value. The candidate outflow sequence after projection repair and constraint evaluation is sorted by non-dominated sorting. Non-dominated solutions that meet the constraints or whose constraint residuals do not exceed the preset allowable threshold are maintained in the Pareto external archive. The population of the swarm intelligent optimization algorithm is updated from the Pareto external archive by selecting the guiding solution. The process iteratively executes the steps of inferring candidate water levels and updating the population of the swarm intelligence optimization algorithm until the termination condition is met. Then, it outputs the set of non-dominated feasible scheduling schemes in the Pareto external archive, or selects the final scheduling scheme from the set of non-dominated feasible scheduling schemes.

2. The multi-objective optimization scheduling method for reservoir flood control according to claim 1, characterized in that, The water level constraints include the upper limit of water level and the lower limit of water level for each time period; The outbound capacity constraints include the minimum and maximum outbound flow rates allowed for each time period; The final water level constraint is either a target value constraint for the final water level or an allowable range constraint for the final water level. The constraint sensitivity matrix is ​​a lower triangular matrix with time-series cumulative characteristics. Its matrix elements are the partial derivatives of the water level correction amount of the corresponding time period with respect to the outflow of the corresponding time period. The partial derivatives are obtained by analytically differentiating the water balance relationship and the reservoir capacity-water level relationship.

3. The multi-objective optimization scheduling method for reservoir flood control according to claim 2, characterized in that, The elements of the constraint sensitivity matrix For the first Water level during the period of time The sensitivity of the outbound flow correction amount for a given period is determined by the following formula: ,in, For the first Water level during a certain period For the first Outbound flow during different time periods For the first Storage capacity during specific time periods; ,in, The reservoir capacity-water level curve at water level The derivative at point, For scheduling step size; The elements of the constraint sensitivity matrix are: .

4. The multi-objective optimization scheduling method for reservoir flood control according to claim 1, characterized in that, The quadratic programming uses the correction amount of the candidate outflow sequence. Let the decision variable be the minimum L2 norm of the correction variable, and let the objective function be: ,in, For the first Time period outbound flow correction This represents the total number of scheduling periods; The constraints of the quadratic programming include: Water level constraints, for any given time period ,have ; Outbound capacity constraints, for any given time period ,have ; Ecological flow constraints, for any given time period ,have ; Final water level constraint, ; in, Candidate outbound flow sequences, The first one calculated under the candidate outflow sequence Water level during a certain period and The first The lower and upper limits of water level for a given period of time. and The first Minimum and maximum outbound flow rates allowed for a given time period. For the first Lower limit of ecological flow during a given time period. and These are the lower and upper limits of the allowable final water level range, respectively. These are the elements of the constraint sensitivity matrix; When the final water level constraint adopts the target value of the final water level season ; The repaired outbound flow sequence is as follows: ; When the quadratic programming has no feasible solution, non-negative slack variables are configured for each type of linearization constraint, transforming the original constraints that were required to be fully satisfied into soft constraints that allow for finite violations. The quadratic programming with slack variables is solved with the objective of minimizing the sum of the L2 norm of the outflow correction and the weighted penalties of the slack variables. The penalty weights corresponding to the upper limit of the flood control level constraint and the outflow capacity constraint are greater than the penalty weights corresponding to the ecological flow constraint, the final water level constraint, and the lower limit of the water level constraint. After solving, the restored outflow sequence is resubstituted into the water balance relationship and the reservoir capacity-water level relationship for verification, and the remaining constraint residues after verification are included in the subsequent augmented Lagrange constraint evaluation value.

5. The multi-objective optimization scheduling method for reservoir flood control according to claim 1, characterized in that, The verified constraint residue includes the water level constraint residue generated after verifying the actual water level, the outflow capacity constraint residue, the ecological flow constraint residue, the final water level constraint residue, and the discretized residue generated by converting the outflow flow into the number of gate openings when using gate control. The verified constraint residuals are written as inequality constraints. And construct the augmented Lagrangian constraint evaluation value: ,in, The candidate scheduling schemes are after projection restoration and verification of actual water levels. For the first The dual variables corresponding to each constraint The penalty coefficient is... To constrain the total number, This is a multi-objective ranking evaluation value used for comparing candidate scheduling schemes. It is a positive part function; The dual variables and penalty coefficients are adaptively updated according to the changes in constraint violation during the iteration process.

6. The multi-objective optimization scheduling method for reservoir flood control according to claim 1, characterized in that, When the repaired outflow sequence needs to be executed by the gate, the repaired outflow sequence is converted into a sequence of gate opening numbers according to the gate-flow relationship. No. The actual outflow rate of the gate during a given time period is determined by the following formula: ,in, For the first Actual outflow from the gate during the specified time period For the first Number of gate openings during the time period For the first Water level during a certain period The gate current function; Outbound flow after repair Select the corresponding number of gate opening holes from the set of openable holes. ,make and The deviation meets the preset conditions; When the gate is a gate with equal orifice and the same type, the number of opening orifices of the gate is determined by the following formula: ,in, For a single-hole gate at the water level The current carrying capacity below, This represents the maximum number of openings that can be made on the gate. Indicates rounding up; After completing the conversion of the number of gate openings, the actual outflow for each period is recalculated based on the gate-flow relationship, and the water level for each period is checked by substituting the water balance relationship and the reservoir capacity-water level relationship. If the check result violates the constraints, the opening sequence with the smaller constraint violation and fewer gate opening and closing times is selected from the adjacent opening number combination.

7. The multi-objective optimization scheduling method for reservoir flood control according to claim 6, characterized in that, The gate opening and closing frequency index includes the gate opening and closing orifice frequency index, which is as follows: ,in, The number of times the gates are opened and closed during the scheduling period. This represents the absolute value operation; The gate opening and closing frequency index also includes the gate adjustment period index, which is: ,in, This refers to the number of time periods during which the number of gate openings changes within the scheduling period. For indicator functions; The reservoir flood control multi-objective optimization model also includes gate change rate constraints: ,in, This represents the maximum allowable change in the number of gate openings between adjacent time periods.

8. An electronic device comprising a processor, a memory, and a computer program stored in the memory and capable of running on the processor, characterized in that, When the processor executes the computer program, it implements the reservoir flood control multi-objective optimization scheduling method as described in any one of claims 1-7.

9. A computer program product comprising a computer program / instructions, characterized in that, When the computer program / instruction is executed by the processor, it implements a multi-objective optimization scheduling method for reservoir flood control as described in any one of claims 1-7.

Images