A flood control scheduling method for cascade reservoirs
By constructing a flood control knowledge base and using it to generate flood control scheduling schemes for cascade reservoirs, the problems of time-consuming and error-prone traditional manual decision-making have been solved, achieving efficient and intelligent flood control scheduling decisions and reducing the risks in extreme flood events.
Patent Information
- Application Number
- CN202510861282.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2025-09-01
- Publication Date
- 2026-08-25
- Estimated Expiration
- 2045-09-01
AI Technical Summary
Traditional flood control scheduling methods for cascade reservoirs rely on manual decision-making, which is time-consuming and prone to errors, making it difficult to conduct flood control scheduling quickly and effectively, especially under extreme weather events.
By acquiring the characteristic parameters of target floods and case floods, typical floods are matched and a flood control knowledge base is constructed. The flood control knowledge base is then used to generate flood control scheduling plans for each time period, reducing manual intervention and achieving intelligent decision-making.
It significantly shortens the decision-making cycle for flood control scheduling, reduces scheduling risks during extreme flood events, provides reliable intelligent decision support, and avoids subjective biases from human experience.
Smart Images

Figure CN120806444B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of reservoir scheduling technology, and in particular to a flood control scheduling method for cascade reservoirs. Background Technology
[0002] Developing a flood control scheduling plan for cascade reservoirs is a complex and multi-layered process that requires comprehensive consideration of various factors and objectives. First, cascade reservoirs are typically equipped with multiple flood discharge facilities, such as central outlets, surface outlets, spillways, and spillway tunnels, and the opening degrees of different gates vary. Proper operation of these flood discharge facilities is crucial for flood control scheduling. Second, the discharge capacity of gates is usually not linear; it depends not only on their own opening degree but also on the current water level of the reservoir. Third, frequent gate operations not only increase operational difficulty but may also lead to operational errors. To reduce operational complexity and the risk of error, gate operations should be minimized. Furthermore, after the flood control scheduling plan is formulated, the reservoir needs to undergo inspection and confirmation according to regulations before implementing gate operations, and downstream relevant units must be notified in advance. Therefore, gate operation orders must be issued well in advance.
[0003] Traditional flood control scheduling methods rely primarily on manual decision-making, which requires repeated manual calculations, is time-consuming, and prone to errors. This is especially true when extreme weather events cause significant variations in flood peaks, rendering traditional manual calculation methods inadequate. Summary of the Invention
[0004] Therefore, it is necessary to provide a flood control scheduling method for cascade reservoirs to address the aforementioned technical problems.
[0005] This invention provides a flood control scheduling method for cascade reservoirs, comprising: Obtain the characteristic parameters of the target flood and multiple case floods, determine the matching degree between the target flood and each case flood based on the characteristic parameters of the target flood and the characteristic parameters of each case flood, and take the case flood with the highest matching degree as the typical flood; Starting from the first time period, the system iterates through the operational data of cascade reservoirs under typical flood conditions time periods, determining whether the gate opening of the cascade reservoirs changes in adjacent time periods. If so, the system determines the average inflow of the cascade reservoirs in the next time period, and integrates the average inflow, reservoir water level, and gate opening of the cascade reservoirs in the next time period into flood control knowledge, storing it in the flood control knowledge base. If not, the system continues to iterate through the operational data of the cascade reservoirs under typical flood conditions in the next time period. The process involves obtaining the average inflow of the cascade reservoirs under the target flood within a given time period, and retrieving the most matching flood control knowledge from the flood control knowledge base based on this average inflow. Then, based on this matching knowledge, flood control scheduling is implemented for the target flood to obtain the current time period's flood control scheduling plan. The process then determines whether the total number of time periods for which flood control scheduling plans have been obtained is equal to the total number of time periods for a typical flood. If so, all time periods for the target flood are determined, and the flood control scheduling plans for each time period are summarized to obtain the flood control scheduling plans for the cascade reservoirs under the target flood across all time periods. If not, the average inflow of the cascade reservoirs under the target flood in the next time period is determined.
[0006] Optionally, the average inflow, water level, and gate opening of the cascade reservoirs over a future period can be integrated into flood control knowledge and stored in a flood control knowledge base, specifically including: Using the time corresponding to the first flood control knowledge point as the starting data point, the operation data of cascade reservoirs under typical floods are traversed time periods one by one to determine whether the gate opening changes in adjacent time periods. If so, the average inflow of the cascade reservoirs in the next time period is determined based on the typical flood data, and a flood control knowledge point is constructed and stored in the flood control knowledge base based on the average inflow of the cascade reservoirs in the next time period, the current reservoir water level, and the gate opening. If not, the operation data of the cascade reservoirs in the next time period under typical floods is traversed again. The storage structure of the flood control knowledge base is determined based on the following formula: Flood prevention knowledge i The average inflow, water level, and gate opening of the cascade reservoirs in the next time period; in, i The numbers represent the sequence of flood prevention knowledge.
[0007] Optionally, once all time periods of the target flood have been determined, the flood control scheduling plans for each time period are summarized to obtain the flood control scheduling plans for the downstream cascade reservoirs for all time periods of the target flood, specifically including: Based on the average inflow of the cascade reservoirs under the target flood within a certain period, the current flood control knowledge that best matches the average inflow is obtained from the knowledge base; when the cascade reservoirs open their gates, the current flood control knowledge is searched from the time period of the cascade reservoirs opening their gates. The current flood control knowledge includes the gate opening degree corresponding to the water level and the inflow. Determine the water release periods for the cascade reservoirs. The water release periods are the periods when the water level in the cascade reservoirs exceeds the limit and must be released because the reservoirs are only used for power generation and the gates are not opened. Based on the most suitable current flood control knowledge, conduct flood control scheduling for the target flood and determine the flood control scheduling plan for the current period. Determine if the total number of the current time period is equal to the total number of time periods of the target flood. If so, all time periods of the target flood are determined, and the flood control scheduling plans for each time period are summarized to obtain the flood control scheduling plans for the cascade reservoirs below the target flood in all time periods. The flood control scheduling plans include outflow, water level process, power generation flow, flood discharge flow and gate opening. If not, determine the average inflow of the cascade reservoirs below the target flood in the next time period.
[0008] Optionally, based on the best-matching flood control knowledge, flood control scheduling is carried out for the target flood to determine the flood control scheduling plan for the current period, specifically including: If the cascade reservoirs do not discharge floodwater, and the actual water level difference in the reservoirs is less than the set threshold or exceeds the water level matching the knowledge in the knowledge base, the matching gate opening is not all 0, and the current time period is greater than or equal to 4 time periods from the start of the adjustment period or within 3 time periods from the water release period, then the flood discharge will begin, the gate opening will be set to the gate opening recommended by the flood control knowledge, and the future operating status of the cascade reservoirs will be predicted. If the cascade reservoirs do not discharge floodwater, and the flood control knowledge suggests that all gate openings should be 0, then no floodwater should be discharged, the gate openings should be set to 0, and the future operating status of the cascade reservoirs should be predicted. If the cascade reservoirs are releasing floodwater, find the time of the most recent gate opening, and determine whether to maintain the current gate opening based on the time interval between gate openings; To determine whether to stop flood discharge, if the water level of the cascade reservoirs is lower than the water level corresponding to the flood control knowledge and the opening degree of the corresponding gates is all 0, then stop flood discharge, set the gate opening degree to 0, and predict the future operating status of the cascade reservoirs; if the cascade reservoirs are not discharging floodwater and the gate opening degree recommended by the flood control knowledge is all 0, then do not discharge floodwater, set the gate opening degree to 0, and predict the future operating status of the cascade reservoirs. If flood discharge continues, determine whether to change the gate opening. If the difference between the water level and the flood control level is less than the set threshold, and the difference between the outflow and inflow corresponding to the flood control level is less than the set threshold compared to maintaining the original gate opening, then update the gate opening and determine the current operating status of the cascade reservoirs based on the gate opening and output.
[0009] Optionally, the current operating status of the cascade reservoirs can be determined based on the gate opening and output, specifically including: The flood discharge flow rate is determined based on the gate opening and water level. The tailrace level is determined and the power generation flow rate is obtained based on the power output and flood discharge flow rate; Based on the flood discharge, power generation, and inflow, the reservoir capacity is determined using the water balance equation to update the reservoir water level. The current operating status of the cascade reservoirs is determined based on the updated reservoir water levels and gate openings.
[0010] Optionally, if the cascade reservoirs are releasing floodwater, find the most recent gate opening time, and based on the gate opening time interval, determine whether to maintain the current gate opening, including: Traverse the time series, find the first time period with a non-zero opening degree, and record the first gate opening time and gate opening degree; if no non-zero opening period is found, do not maintain the same opening degree, and then determine whether to stop flood discharge; If the current time period is within the initial gate opening interval after the first gate opening, the same opening degree will be maintained, and the future operating status of the cascade reservoir will be predicted. Initialize the most recent gate opening time to the first gate opening time plus the initial gate opening interval; iterate backward from the current time period to find the time point when the gate opening degree changed most recently, and update the most recent gate opening time. Determine the interval between the current time period and the most recent gate opening time. If it is less than the gate opening interval, maintain the same opening degree and predict the future operating status of the cascade reservoirs. If it exceeds the gate opening interval, do not maintain the same opening degree and determine whether to stop flood discharge.
[0011] Optionally, predict the future operating status of the cascade reservoirs, including: The reservoir water level for the next period is determined based on the power generation flow and gate opening of the cascade reservoirs in the current period. If the reservoir water level in the next period does not exceed the difference between the maximum limit water level and the reserved margin, the current operating status of the cascade reservoirs is determined based on the gate opening and output in the current period. If the reservoir water level exceeds the difference between the maximum limit water level and the reserved margin in the next time period, the power generation flow rate is determined according to the method of not discarding water, and the flood discharge flow rate is determined according to the principle of balancing inflow and outflow, so as to obtain the opening degree of the intermediate gate; the calculation time is reversed to the previous gate opening time, the opening degree of the intermediate gate is used as the gate opening degree of the next time period, and the future operating status of the cascade reservoir is determined according to the gate opening degree and output of the next time period.
[0012] The flood control scheduling method for cascade reservoirs provided in this embodiment of the invention has the following advantages compared with the prior art: This invention selects typical floods through a flood characteristic parameter matching mechanism, extracts and stores flood control-related knowledge from these typical floods, and constructs a flood control knowledge base. Based on this knowledge base, it determines the flood control scheduling schemes for cascade reservoirs under the target flood at all time periods. This solves the problem of repeated manual calculations, which are time-consuming and prone to errors, enabling efficient simulations and significantly shortening the flood control scheduling decision-making cycle.
[0013] Furthermore, in scenarios of rapidly changing floods, this invention directly obtains the optimal scheduling parameters by matching with a flood control knowledge base, which can avoid the subjective bias of human experience and reduce the time spent on manual decision-making from several hours to minutes. This not only significantly reduces the scheduling risk in extreme flood events, but also builds an intelligent decision-making system through continuous learning and optimization of the knowledge base, providing reliable technical support for basin flood control. Attached Figure Description
[0014] Figure 1 This is a flowchart of a flood control scheduling method for a cascade reservoir provided in one embodiment; Figure 2 A flowchart illustrating the generation solution of a flood control scheduling method for cascade reservoirs provided in one embodiment; Figure 3 This is a flowchart illustrating the determination of maintaining gate opening in a flood control scheduling method for a cascade reservoir, as provided in one embodiment. Figure 4 This is a schematic diagram of a cascade reservoir, illustrating a flood control scheduling method for a cascade reservoir as provided in one embodiment. Figure 5 A target flood discharge process diagram for a flood control scheduling method for cascade reservoirs provided in one embodiment; Figure 6 This is a diagram illustrating the typical flood-induced operation of reservoir A in a cascade reservoir flood control scheduling method provided in one embodiment. Figure 7 This is a typical flood-induced gate opening diagram of reservoir A in a flood control scheduling method for a cascade reservoir provided in one embodiment. Figure 8 This is a diagram showing the operation of reservoir A under a target flood in a flood control scheduling method for a cascade reservoir provided in one embodiment. Figure 9 A diagram showing the gate opening degree of reservoir A during a target flood in a flood control scheduling method for a cascade reservoir provided in one embodiment; Figure 10 This is a diagram illustrating the typical flood-induced operation of reservoir B in a cascade reservoir flood control scheduling method provided in one embodiment. Figure 11 This is a typical flood-induced diagram of the gate opening of reservoir B in a cascade reservoir flood control scheduling method provided in one embodiment. Figure 12 This is a diagram showing the operation of reservoir B under a target flood in a flood control scheduling method for a cascade reservoir provided in one embodiment. Figure 13 This is a diagram showing the gate opening degree of reservoir B under target floodwater in a flood control scheduling method for a cascade reservoir provided in one embodiment. Detailed Implementation
[0015] 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 merely illustrative and not intended to limit the invention.
[0016] (1) In the study of reservoir optimization scheduling, the scheduling objectives are mainly based on reducing flood peak flow and staggering peak avoidance, with outflow and water level as decision variables, and the flood control strategies of the reservoir are studied. However, the details are rarely refined to the opening and closing and adjustment of gates, so it is often difficult to apply in practice.
[0017] (2) Flood control scheduling of cascade reservoirs is solved using traditional optimization algorithms. The decision variables involve many objects such as reservoirs, gates, opening degree, and time. Dynamic programming is prone to falling into the curse of dimensionality, while intelligent optimization algorithms have the problem of "premature convergence". Therefore, existing research cannot fully consider the available opening degree of the gates.
[0018] (3) In existing studies, the power generation flow rate is mostly taken as the maximum reference flow rate of the turbine. However, in reality, after the power station is generating power at full load, the power generation flow rate is greatly affected by the reservoir water level and the tailwater level. When the reservoir discharges floodwater, the tailwater level changes greatly, which will lead to an increase in the power generation flow rate. When the reservoir is blocked, the water level increases rapidly, which will reduce the power generation flow rate.
[0019] In one embodiment, a flood control scheduling method for cascade reservoirs is provided, the method comprising: S1. Select the typical flood that best matches the target flood from historical cases; Obtain the characteristic parameters of the target flood and multiple case floods. Determine the matching degree between the target flood and each case flood based on the characteristic parameters of the target flood and the characteristic parameters of each case flood, and take the case flood with the highest matching degree as the typical flood.
[0020] S2. Knowledge Extraction Stage: Determine the storage structure of flood control knowledge, create a flood control knowledge base, and automatically extract and store flood control knowledge from typical floods.
[0021] Starting from the first time period, the system iterates through the operational data of cascade reservoirs under typical flood conditions time periods one by one, determining whether the gate opening of the cascade reservoirs changes in adjacent time periods. If so, the system determines the average inflow of the cascade reservoirs in the next time period, and integrates the average inflow, reservoir water level, and gate opening of the cascade reservoirs in the next time period into flood control knowledge, which is then stored in the flood control knowledge base. If not, the system continues to iterate through the operational data of the cascade reservoirs under typical flood conditions in the next time period.
[0022] S3. Knowledge Application Stage: Utilize the relevant knowledge contained in the knowledge base, combine it with the reservoir operation under the target flood, and make decision-making interventions when necessary to intelligently generate reservoir flood control scheduling plans.
[0023] The process begins by obtaining the average inflow of the cascade reservoirs under the target flood over a given period. Based on this average inflow, the most relevant flood control knowledge is retrieved from the flood control knowledge base. Using this most relevant knowledge, flood control scheduling is implemented for the target flood, resulting in the current flood control scheduling plan. The process then checks if the total number of time periods for which flood control scheduling plans have been obtained is equal to the total number of time periods for a typical flood. If so, all time periods for the target flood are determined, and the flood control scheduling plans for each time period are summarized to obtain the flood control scheduling plans for the cascade reservoirs under the target flood across all time periods. If not, the average inflow of the cascade reservoirs under the target flood in the next time period is determined.
[0024] Furthermore, the selection of the most relevant typical case from historical cases in step S1 can be achieved through various methods, including feature similarity-based matching, text similarity-based matching, rule-based matching, machine learning models, and multi-model hybrid methods. This patent uses feature similarity-based matching as an example to illustrate the steps for selecting typical floods, specifically including:
[0025] S11. Select feature information for case matching, such as peak flow, 12-hour flood volume, 1-day (24-hour) flood volume, 2-day (48-hour) flood volume, total flood volume, average output, discharge volume, initial water level, etc. If it is a multi-peak flood, the flood peak and flood volume are taken from the flood with the largest peak flow. S12. Calculate the characteristic parameters of the target flood based on the selected characteristic information; S13. Calculate the feature parameters of each typical case in the knowledge base based on the selected feature information; S14. Calculate the matching degree between the target flood and the floods in each case based on the feature parameters; S15. Sort the matching degree of each case flood and find the case flood with the highest matching degree with the target flood as the typical flood; Furthermore, the automatic extraction and storage of flood control knowledge from typical floods in S2 specifically includes, S21. Determine the storage structure of the flood control knowledge base. The storage structure is as follows:
[0026] Flood prevention knowledge i The average inflow, water level, and gate opening of the cascade reservoirs in the next time period; in, i This indicates the sequence number of the flood prevention knowledge. Both the case flood and the target flood are data records generated at 1-hour intervals.
[0027] S22. Traverse the operational data of cascade reservoirs under typical floods (such as inflow and outflow rates, water levels, gate openings, etc.). The starting data point is the time corresponding to the first data point. Starting from the first time period, iterate forward step by step, checking whether the gate opening changes in adjacent time periods. If it changes, jump to step S23; otherwise, continue traversing the next time period until all time periods of the case flood have been checked.
[0028] The specific process of traversal is as follows: Flood data in flood control is segmented, with time intervals typically one hour. The future time period for calculating the average inflow is generally three hours. Starting from the first time period, each subsequent time period is pushed forward to the next, until the last time period. During this process, the average inflow within the next time period (e.g., three hours) is continuously calculated. If the remaining time period is less than three hours, the average inflow of the remaining time period is used as the average inflow for the next time period.
[0029] S23. If the gate opening changes, record the corresponding time period. Extract flood control knowledge based on the gate operation, mainly including the following steps: S231. Based on the case flood, calculate the average inflow (actual data) of a typical flood over a future period (e.g., the next three hours). S232. Create a new flood control knowledge entry, which includes the average inflow of a typical flood over a future period (e.g., the next three hours), the current reservoir water level, and the gate opening as the content of this flood control knowledge entry. S233. Store the newly created knowledge in the knowledge base for subsequent analysis and application.
[0030] Furthermore, the flood control scheme for generating the target flood based on the flood control knowledge base in S3 specifically includes: S31. Set the output to the average output of the case, calculate the average inbound flow (forecast data) in the next time period (e.g., the next three hours), and use it for subsequent case matching.
[0031] S32. Based on the average inflow, retrieve the most matching flood control knowledge from the knowledge base; if the reservoir needs to open its gates, search for flood control knowledge during the gate opening period (the period when the gate changes from completely closed to open); this flood control knowledge is the recommended gate opening degree under the corresponding water level and inflow.
[0032] S33. Calculate the water release period of the reservoir (the period when the water level of the reservoir exceeds the limit and must be released when the reservoir is only used for power generation and the gates are not opened). Based on the best matching flood control knowledge, carry out flood control scheduling for the target flood and determine the flood control scheduling plan for the current period. S34. Determine if the current number of time periods is equal to the number of time periods of the target flood. If they are equal, it means that all time periods have been calculated, and the calculation ends. Proceed to S35. Otherwise, proceed to S31 to continue calculating the remaining time periods.
[0033] S35. Output Results: Output the calculated results, including outflow, water level process, power generation flow, flood discharge flow, gate opening, etc.
[0034] Furthermore, the specific steps in S33 for determining the flood control scheduling plan for the current period based on the acquired best-matching flood control knowledge include: S331. If the reservoir is not discharging floodwater and meets three conditions: the reservoir water level is close to (water level difference less than 0.2m, 0.3m, 1m, etc.) or exceeds the water level of the matching knowledge in the knowledge base; the matching gate opening is not all 0; and the current time period is greater than or equal to 4 time periods from the start of the adjustment period or within 3 time periods from the water release period. If all three conditions are met, flood discharge will begin, and the gate opening will be set to the gate opening suggested by the flood control knowledge. Proceed to S336; otherwise, proceed to S332.
[0035] S332. If the reservoir is not discharging floodwater and the flood control knowledge suggests that the gate opening is 0, it means that there is no need to discharge floodwater. Set the gate opening to 0 and proceed to S336; otherwise, proceed to S333.
[0036] S333. If the reservoir is discharging floodwater, find the time of the most recent gate opening and determine whether it is necessary to maintain the current gate opening based on the time interval between gate openings.
[0037] S334. Determine whether it is necessary to stop flood discharge. If the reservoir water level is lower than the water level corresponding to the flood control knowledge and the corresponding gate opening is all 0, it means that the reservoir can stop flood discharge. Set the gate opening to 0 and go to S336. If the reservoir is not discharging floodwater and the gate opening recommended by the flood control knowledge is all 0, it means that there is no need to discharge floodwater. Set the gate opening to 0 and go to S336. If it is necessary to continue discharging floodwater, go to S335.
[0038] S335. Determine whether the gate opening needs to be changed. If the water level difference between the water level and the flood control water level is small (water level difference less than 0.2m, 0.3m, 1m, etc.), and the difference between the outflow and inflow of the flood control water level is smaller than that of the original gate opening, then update the gate opening.
[0039] S336. Predict the future operating status of the reservoir and automatically intervene when necessary to ensure the safe operation of the reservoir during flood control.
[0040] S337. Calculate the current operating status of the reservoir based on the gate opening and output. The main methods are: calculate the flood discharge flow based on the gate opening and water level; calculate the tailrace level and obtain the power generation flow based on the output and flood discharge flow; calculate the reservoir capacity according to the water balance equation based on the flood discharge flow, power generation flow, and inflow, and update the reservoir water level.
[0041] Furthermore, the step S333, which involves finding the most recent gate movement time and determining whether to maintain the current gate opening, specifically includes: S3331. Traverse the time series, find the first non-zero opening period, and record the first gate opening time and gate opening degree; S3332. If no non-zero opening period is found, then it is not necessary to maintain the same opening. Proceed to S334. S3333. If the current time period is within the initial gate opening interval after the first gate opening (referring to the time interval during which the gate opening degree needs to remain unchanged after the first gate opening), then the same opening degree needs to be maintained, and proceed to S336. S3334. Initialize the most recent gate opening time to the first gate opening time plus the initial gate opening interval; S3335. Traverse backwards from the current time period to find the time point when the gate opening changed most recently, and update the time of the most recent gate opening. S3336. Calculate the interval between the current time period and the most recent gate opening time. If it is less than the gate opening interval (the minimum time interval after a gate opening that needs to be maintained after the first opening), then the same opening degree needs to be maintained, and proceed to S336. If it exceeds the gate opening interval, then the same opening degree does not need to be maintained, and proceed to S334.
[0042] Furthermore, the method for S336, which involves appropriate automatic intervention based on the future operation of the reservoir, specifically includes: S3361. Based on the current power generation flow and gate opening, calculate the reservoir water level for the next time period. S3362. If the reservoir water level in the next period does not exceed the maximum limit water level by -0.1 meters (leaving a certain margin), then proceed to S337; if the reservoir water level in the next period exceeds the maximum limit water level by -0.1 meters, it indicates that the knowledge extracted at present cannot fully cover the situation, then proceed to S3363. S3363. Calculate the power generation flow rate according to the method of not wasting water; S3364. Calculate the flood discharge flow according to the principle of balancing inflow and outflow to obtain the new gate opening; S3365. Rewind the calculation time to the most recent gate opening time, set the gate opening to the new calculated opening, to ensure that the reservoir operates within the safe water level range, then proceed to S337.
[0043] Taking two adjacent reservoirs (referred to as Reservoir A and Reservoir B) as an example, a specific embodiment of the present invention is provided, including: (1) Solving for reservoir A S1, Selecting a Typical Case. Since Reservoir A has a large capacity and a wide water level range, the threshold for judging approaching water levels in S331 is set to 1m. The time period for calculating the average inflow is set to three hours.
[0044] The inflow rates of reservoirs A and B during the target flood are shown below. The initial water levels of reservoirs A and B are 864.44 m and 717.92 m, respectively. Reservoirs A and B are matched with rated outputs of 200 MW and 150 MW, respectively. During the target flood, the inflow rates of reservoir A and the flow rates in the AB interval are as follows: Figure 5 As shown.
[0045] The target flood exhibits a three-peak pattern, with the flood mainly occurring in the first half. In the second half, the inflow shows a fluctuating, slightly decreasing trend. The main source of the flood is the inflow from Reservoir A, with the flow rates of all three flood peaks exceeding 1000 m³ / s. The flow rate in the AB interval is relatively small, and its variation trend is consistent with that of the inflow from Reservoir A. The main characteristics of the target flood at Reservoir A are shown in Table 1.
[0046] Table 1. Characteristics of Target Flood in Reservoir A Thirty-eight typical cases of Reservoir A and Reservoir B from 2012 to the present were extracted from historical operational data. After calculating the matching degree, the operational situation of Reservoir A in Case 28 was closest to the target flood, with a matching degree of 0.925. Case 28 (from 01:00 on June 16, 2015 to 04:00 on June 26, 2015) was selected as the typical flood. The characteristic parameters of the typical flood are shown in Table 2.
[0047] Table 2. Flood characteristics corresponding to the best-matched typical flood (Reservoir A) The operation of Reservoir A during a typical flood is shown below. In this case, the flood event exhibits a bimodal pattern, with the main peak flow reaching 2000 m³ / s and the secondary peak flow approximately 1500 m³ / s. During this period, the reservoir operates at full load for continuous power generation, while the outflow varies in a stepwise manner with the dynamic adjustment of the gate opening, as shown below. Figure 6 and Figure 7 As shown.
[0048] S2, Knowledge Extraction Stage: Create a flood control knowledge base for Reservoir A, as shown in Table 3. Extract a total of 18 flood control knowledge items from typical floods. The inflow rate is the average inflow rate over the next three hours. In the gate opening, from left to right, they represent the left middle gate, right middle gate, left surface gate, middle surface gate, and right surface gate.
[0049] Table 3 Flood Control Knowledge Base for Reservoir A Entering the S3 knowledge application stage, starting from the first time period, the reservoir scheduling plan is calculated time by time period.
[0050] In time period 1, S31 calculates the average inflow for the next three hours to be 329 m³ / s, and the reservoir water level to be 864.49 m. S32 determines the best-matching knowledge as knowledge 1. Then, in S33, the flood control scheduling plan is determined. Since the reservoir is not discharging water, S331 calculates the water release period as 13, therefore, the 5th time period (4-hour interval) is required before discharging water. Afterward, the reservoir maintains a state of non-discharging operation.
[0051] In time period 5, S31 calculates the average inflow to the reservoir as 663 m³ / s, and the reservoir water level as 864.90 m. S32 yields the best-matching knowledge as knowledge 1. Moving to S33, the flood control scheduling plan is determined. Since the reservoir is not releasing water, the difference between the reservoir water level of 864.90 m in S331 and the corresponding knowledge level of 867.53 m is significant; therefore, no water release is permitted at this time. Calculations indicate that the future reservoir water level will not exceed the water level limit, thus no artificial intervention is required. The reservoir's operational status is calculated based on the gate opening.
[0052] In time period 11, S31 calculates the average inflow to the reservoir as 867 m³ / s, and the reservoir water level as 866.68 m. At this time, the reservoir needs to open its gates. S32 obtains the best matching knowledge as knowledge 1. Proceeding to S33, the flood control scheduling plan is determined. Since the reservoir is not discharging water, in S331, because the reservoir water level of 866.68 m and the corresponding knowledge level of 867.53 m are close, the gate opening is set at [0.0, 1.0, 0.0, 0.0, 0.0]. After calculation, the future reservoir water level will not exceed the water level limit, so no manual intervention is required. Based on the gate opening, the reservoir operation status is calculated.
[0053] In period 12, the most recent gate opening time was period 11. Since the gate had not been opened before, the first gate opening time was in period 11, with an initial gate opening interval of 2. This period is not within the initial gate opening interval range after the first gate opening, so the gate opening degree of the previous period is maintained. Subsequently, in periods 12, 13, and 14, flood discharge is carried out with an opening degree of [0.0, 1.0, 0.0, 0.0, 0.0].
[0054] In time period 14, after calculation by S336, the reservoir water level will exceed the water level limit in the future, requiring artificial intervention; entering S336, after trial calculation, the new gate opening is obtained as [0.0, 6.0, 0.0, 0.0, 0.0]. The most recent gate operation time was time period 13, so return to time period 13 and discharge floodwater according to the new gate opening.
[0055] During periods 13-16, flood discharge will be carried out at an opening degree of [0.0, 6.0, 0.0, 0.0, 0.0].
[0056] In period 17, S31 calculated the average inflow to the reservoir to be 1059 m³ / s, and the reservoir water level to be 867.58 m. The best matching knowledge was knowledge 6, and S333 determined that it was not necessary to maintain the gate opening. Entering S335, maintaining the gate opening corresponded to a discharge / outflow of 775 / 936 m³ / s, while the discharge / outflow corresponding to the flood control knowledge was 903 / 1064 m³ / s. The gate opening recommended by the flood control knowledge better matched the inflow, therefore the gate opening was changed to [0.0, 6.7, 0.0, 0.0, 0.0]. Subsequently, from period 18 to 21, the gate was maintained at an opening of [0.0, 6.7, 0.0, 0.0, 0.0] for flood discharge.
[0057] In time period 22, S31 calculates the average inflow to the reservoir as 843 m³ / s and the reservoir water level as 867.23 m. The best matching knowledge obtained through S32 is knowledge 7. S333 determines that it is not necessary to maintain the gate opening, and proceeds to S335. Maintaining the gate opening corresponds to a discharge flow of 901 / 1062 m³ / s, while the discharge / outflow corresponding to the flood control knowledge is 623 / 784 m³ / s. The gate opening recommended by the flood control knowledge better matches the inflow, so the gate opening is changed to [0.0, 5.0, 0.0, 0.0, 0.0].
[0058] In time period 163, S31 calculates the average inflow to the reservoir as 241 m³ / s and the reservoir level as 865.76 m. S32 determines the best-matching knowledge as knowledge 16. S333 determines that maintaining the gate opening is unnecessary, proceeding to S334. The gate opening corresponding to the flood control knowledge is 0, and the average inflow and reservoir level corresponding to the flood control knowledge are 248 m³ / s and 867.58 m, respectively. However, the actual average inflow and reservoir level are 241 m³ / s and 865.76 m, respectively. Both the inflow and reservoir level are lower than the flood control knowledge, therefore the reservoir stops discharging floodwater.
[0059] After the above steps are completed and the calculations are finished for each time period, the scheduling plan for reservoir A under the target flood is obtained. The operation of reservoir A is as follows: Figure 8 and Figure 9 As shown.
[0060] from Figure 8and Figure 9 It can be seen that the inflow into Reservoir A fluctuated significantly, gradually declining after three peaks. Because Reservoir A consistently maintained full-load power generation, the power generation flow remained relatively stable. The outflow from Reservoir A exhibited a stepped variation, closely related to the gate opening and largely mirroring the trend of the inflow. This ensured that the reservoir maintained a high water level without exceeding its limits, and no "artificial flooding" occurred. Regarding the reservoir water level, initially, because Reservoir A did not release floodwater, the reservoir impounded floodwater, causing a rapid rise in water level. In the middle stage, the gate opening was dynamically adjusted based on the inflow, maintaining a relatively stable water level. In the later stage, as the water level gradually decreased, the reservoir closed its gates, ceasing flood discharge. Therefore, it is evident that the reservoir effectively controlled the water level by precisely adjusting the gate opening, achieving its flood control objectives, and the scheduling results were quite satisfactory.
[0061] (2) Solving for reservoir B Based on the flow process of reservoir A, the target flood process of reservoir B is obtained. Since reservoir B has a relatively small capacity, the threshold for determining the approaching water level in S331 is set to 0.3m. The time period for calculating the average inflow is set to three hours.
[0062] Proceed to S1 and select a typical case. The characteristics of the target flood at Reservoir B are shown in Table 4:
[0063] Table 4 Characteristics of Target Flood in Reservoir B Thirty-eight typical cases of Reservoir A and Reservoir B from 2012 to the present were extracted from historical operational data. After calculating the matching degree, it was found that the operational situation of Reservoir B in Case 28 was closest to the target flood, with a matching degree of 0.86. Case 28 (from 01:00 on June 16, 2015 to 04:00 on June 26, 2015) was selected as the typical flood. The characteristic parameters of the typical flood are shown in Table 5.
[0064] Table 5. Flood characteristics corresponding to the best-matched typical flood (Reservoir B). The optimal matching case for Reservoir B has the same time frame as the case for Reservoir A. Influenced by the upstream and downstream water conservancy connections, the flood process of Reservoir B changes with the outflow from Reservoir A. The flood in this case mainly exhibits a bimodal characteristic: the main flood peak occurs in the early part of the scheduling period, with a flow rate reaching 2000 m³ / s; a second flood peak occurs in the middle period, but with a relatively smaller flow rate of 1000 m³ / s. The operation of Reservoir B during a typical flood is as follows: Figure 10 and Figure 11 As shown.
[0065] S2, Knowledge Extraction Stage: Create a flood control knowledge base for Reservoir B, as shown in Table 6. Extract a total of 19 flood control knowledge points from typical floods. The inflow rate is the average inflow rate over the next three hours. In the gate opening, from left to right, they represent the left surface gate, the middle surface gate, and the right surface gate.
[0066] Table 6 Flood Control Knowledge Base for Reservoir B Entering the S3 knowledge application stage, starting from the first time period, the reservoir scheduling plan is calculated time by time period.
[0067] In time period 1, S31 calculates the average inflow to the reservoir for the next three hours to be 224 m³ / s, and the reservoir water level to be 717.92 m. S32 then yields the best-matching knowledge as knowledge 1. Moving to S33, the flood control scheduling plan is determined. Since the reservoir is not releasing water, S331 calculates the water release period as 8, therefore, water release will only begin after the 4th time period.
[0068] In time period 5, S31 calculates an average inflow of 245 m³ / s and a reservoir water level of 718.19 m. S32 identifies the best-matching knowledge (Knowledge 1) based on the reservoir gate opening. In S33, the flood control scheduling plan is determined. Since the reservoir is not releasing water, the water release period in S331 is 8. Therefore, in time period 5, the reservoir needs to release water. At this point, the reservoir water level of 718.19 m is already higher than the corresponding knowledge level of 717.61 m. Therefore, the reservoir begins releasing water with a gate opening of [0.125, 0.0, 0.0]. Calculations show that the future reservoir water level will not exceed the water level limit, so no manual intervention is needed. Based on the gate opening, the reservoir operation status is calculated.
[0069] In time period 6, enter S333 to determine whether the gate opening needs to be maintained. The last time the gate was opened was in time period 5. Since the gate was not opened before, the first time the gate was opened was in time period 5. The initial gate opening interval was 2. This time period is not within the initial gate opening interval after the first gate opening. Therefore, the gate opening of the previous time period is maintained.
[0070] In time period 9, S31 calculation yields an average inflow of 453 m³ / s and a reservoir water level of 716.48 m. S32 determines the best-matching knowledge as knowledge 11. The first gate opening time is time period 5, which is currently outside the initial gate opening interval after the first opening. Therefore, the most recent gate opening time is initialized to time period 5 (first gate opening time) plus 2 (initial gate opening interval) = time period 7. Since the gate opening degree has not changed before this, the most recent gate opening time remains unchanged. The interval between the current time period and the most recent gate opening time is 2 hours, which is less than or equal to the 2-hour gate opening interval; therefore, the same opening degree needs to be maintained.
[0071] In time period 10, S31 calculates the average inflow to the reservoir as 498 m³ / s and the reservoir water level as 716.27 m. S32 then obtains the best-matching knowledge as knowledge 16. Based on the obtained best-matching flood control knowledge, S33 determines that maintaining the gate opening is unnecessary in the current flood control scheduling plan. Therefore, S335 determines whether to change the gate opening. Since the gate opening recommended by the knowledge is consistent with the opening in the previous time period, the current opening is maintained.
[0072] In time period 11, S1 calculates the average inflow to the reservoir as 764 m³ / s and the reservoir water level as 716.11 m. S32 determines that the best matching knowledge is knowledge 9. S333 determines that it is not necessary to maintain the gate opening. Entering S335, maintaining the gate opening corresponds to a discharge / outflow of 289 / 488 m³ / s, while the flood control knowledge corresponds to a discharge / outflow of 578 / 778 m³ / s. The gate opening recommended by the knowledge better matches the inflow, so the gate opening is changed to [0.125, 0.0, 0.125].
[0073] After the above steps are completed and the calculations are finished for each time period, the scheduling plan for reservoir B under the target flood is obtained. The operation of reservoir B is as follows: Figure 12 and Figure 13 As shown.
[0074] From the reservoir B operation diagram and gate opening diagram ( Figure 12 , 13 As can be seen, the outflow is adjusted in a stepwise manner according to the increase or decrease of the inflow. The gate opening and closing follows the principle of gradual opening and closing. When the gate is opened to release floodwater, a sudden increase in outflow is avoided, leaving sufficient emergency response time for downstream areas. As the inflow decreases, the gate opening also decreases step by step, keeping the reservoir water level at a high level. At the same time, the reservoir output is always kept at full capacity to make the most of water energy, fully demonstrating the refined management and dynamic response capability of flood control scheduling. However, due to the limitations of flood control knowledge, although the inflow gradually decreases in the later stage of scheduling, the gate still maintains a partial opening, resulting in a faster rate of water level drop. If it is necessary to maintain a high water level for a long period of time, the gate can be closed during time period 161. During the flood control scheduling, there are short-term situations of decreasing-increasing-decreasing gate opening during time periods 50 and 85. These can be addressed by manual intervention to balance the needs of flood control, water storage, and gate operation.
[0075] This invention selects the best-matching case flood as a typical flood, extracts and stores flood control-related knowledge from it, thus forming a knowledge base. The extracted knowledge is then applied to actual flood control scheduling, intelligently generating reservoir flood control scheduling plans. It mainly consists of three stages: case selection, knowledge extraction, and knowledge application. In the case selection stage, the system iterates through reservoir inflow and water level data during typical floods, detecting whether gates have been activated. In the knowledge extraction stage, relevant flood control knowledge is extracted and stored. In the knowledge application stage, the system calculates the average inflow and uses knowledge matching to determine whether gate opening needs adjustment, and simultaneously determines whether decision-making intervention is necessary. When existing knowledge is insufficient, automatic intervention and adjustment are performed to ensure that the reservoir water level remains within a safe range.
[0076] The embodiments described above are merely examples of several implementations of the present invention, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention.
Claims
1. A method for flood control scheduling of cascade reservoirs, characterized in that, include: Obtain the characteristic parameters of the target flood and multiple case floods, determine the matching degree between the target flood and each case flood based on the characteristic parameters of the target flood and the characteristic parameters of each case flood, and take the case flood with the highest matching degree as the typical flood; Starting from the first time period, the operation data of cascade reservoirs under typical floods are traversed time period by time to determine whether the gate opening of cascade reservoirs changes in adjacent time periods; If so, determine the average inflow of the cascade reservoirs in the next period, and integrate the average inflow, reservoir water level and gate opening of the cascade reservoirs in the next period into flood control knowledge and store it in the flood control knowledge base; if not, continue to traverse the operation data of the cascade reservoirs in the next period under typical floods. The process of integrating the average inflow, water level, and gate opening of the cascade reservoirs over a future period into flood control knowledge and storing it in a flood control knowledge base specifically includes: Using the time corresponding to the first flood control knowledge point as the starting data point, the operation data of cascade reservoirs under typical floods are traversed time periods one by one to determine whether the gate opening changes in adjacent time periods. If so, the average inflow of the cascade reservoirs in the next time period is determined based on the typical flood data, and a flood control knowledge point is constructed and stored in the flood control knowledge base based on the average inflow of the cascade reservoirs in the next time period, the current reservoir water level, and the gate opening. If not, the operation data of the cascade reservoirs in the next time period under typical floods is traversed again. The storage structure of the flood control knowledge base is determined based on the following formula: Flood prevention knowledge i The average inflow, water level, and gate opening of the cascade reservoirs in the next time period; in, i The numbers representing the sequence of flood prevention knowledge; The process involves obtaining the average inflow of the cascade reservoirs under the target flood within a given time period, and retrieving the most matching flood control knowledge from the flood control knowledge base based on this average inflow. Then, based on this matching knowledge, flood control scheduling is implemented for the target flood to obtain the current time period's flood control scheduling plan. The process then determines whether the total number of time periods for which flood control scheduling plans have been obtained is equal to the total number of time periods for a typical flood. If so, all time periods for the target flood are determined, and the flood control scheduling plans for each time period are summarized to obtain the flood control scheduling plans for the cascade reservoirs under the target flood across all time periods. If not, the average inflow of the cascade reservoirs under the target flood in the next time period is determined.
2. The flood control scheduling method for cascade reservoirs as described in claim 1, characterized in that, Once all time periods of the target flood have been determined, the flood control scheduling plans for each time period are summarized to obtain the flood control scheduling plans for the downstream cascade reservoirs for all time periods of the target flood, specifically including: Based on the average inflow of the cascade reservoirs under the target flood within a certain period, the current flood control knowledge that best matches the average inflow is obtained from the knowledge base; when the cascade reservoirs open their gates, the current flood control knowledge is searched from the time period during which the cascade reservoirs open their gates, and the current flood control knowledge includes the gate opening degree corresponding to the water level and the inflow. The water release period of the cascade reservoirs is determined, which is the period when the water level of the cascade reservoirs exceeds the limit and must be released when the reservoirs are only used for power generation and the gates are not opened; based on the best matching current flood control knowledge, flood control scheduling is carried out for the target flood, and the flood control scheduling plan for the current period is determined; Determine whether the total number of the current time period is equal to the total number of time periods of the target flood. If so, all time periods of the target flood are determined, and the flood control scheduling plans for each time period are summarized to obtain the flood control scheduling plans for the cascade reservoirs below the target flood in all time periods. The flood control scheduling plans include outflow, water level process, power generation flow, flood discharge flow and gate opening. If not, determine the average inflow of the cascade reservoirs below the target flood in the next time period.
3. The flood control scheduling method for cascade reservoirs as described in claim 2, characterized in that, The method of flood control scheduling based on the best-matching flood control knowledge to determine the flood control scheduling plan for the current period specifically includes: If the cascade reservoirs do not discharge floodwater, and the actual water level difference in the reservoirs is less than the set threshold or exceeds the water level matching the knowledge in the knowledge base, the matching gate opening is not all 0, and the current time period is greater than or equal to 4 time periods from the start of the adjustment period or within 3 time periods from the water release period, then the flood discharge will begin, the gate opening will be set to the gate opening recommended by the flood control knowledge, and the future operating status of the cascade reservoirs will be predicted. If the cascade reservoirs do not discharge floodwater, and the flood control knowledge suggests that all gate openings should be 0, then no floodwater should be discharged, the gate openings should be set to 0, and the future operating status of the cascade reservoirs should be predicted. If the cascade reservoirs are releasing floodwater, find the time of the most recent gate opening, and determine whether to maintain the current gate opening based on the time interval between gate openings; To determine whether to stop flood discharge, if the water level of the cascade reservoirs is lower than the water level corresponding to the flood control knowledge and the opening degree of the corresponding gates is all 0, then stop flood discharge, set the gate opening degree to 0, and predict the future operating status of the cascade reservoirs; if the cascade reservoirs are not discharging floodwater and the gate opening degree recommended by the flood control knowledge is all 0, then do not discharge floodwater, set the gate opening degree to 0, and predict the future operating status of the cascade reservoirs. If flood discharge continues, determine whether to change the gate opening. If the difference between the water level and the flood control level is less than the set threshold, and the difference between the outflow and inflow corresponding to the flood control level is less than the set threshold compared to maintaining the original gate opening, then update the gate opening and determine the current operating status of the cascade reservoirs based on the gate opening and output.
4. The flood control scheduling method for cascade reservoirs as described in claim 3, characterized in that, The determination of the current operating status of the cascade reservoirs based on gate opening and output specifically includes: The flood discharge flow rate is determined based on the gate opening and water level. The tailrace level is determined and the power generation flow rate is obtained based on the power output and flood discharge flow rate; Based on the flood discharge, power generation, and inflow, the reservoir capacity is determined using the water balance equation to update the reservoir water level. The current operating status of the cascade reservoirs is determined based on the updated reservoir water levels and gate openings.
5. The flood control scheduling method for cascade reservoirs as described in claim 3, characterized in that, If the cascade reservoirs are releasing floodwater, the process involves finding the most recent gate opening time and, based on the gate opening time interval, determining whether to maintain the current gate opening, including: Traverse the time series, find the first time period with a non-zero opening degree, and record the first gate opening time and gate opening degree; if no non-zero opening period is found, do not maintain the same opening degree, and then determine whether to stop flood discharge; If the current time period is within the initial gate opening interval after the first gate opening, the same opening degree will be maintained, and the future operating status of the cascade reservoir will be predicted. Initialize the most recent gate opening time to the first gate opening time plus the initial gate opening interval; iterate backward from the current time period to find the time point when the gate opening degree changed most recently, and update the most recent gate opening time. Determine the interval between the current time period and the most recent gate opening time. If it is less than the gate opening interval, maintain the same opening degree and predict the future operating status of the cascade reservoirs. If it exceeds the gate opening interval, do not maintain the same opening degree and determine whether to stop flood discharge.
6. The flood control scheduling method for cascade reservoirs as described in claim 5, characterized in that, The prediction of the future operating status of the cascade reservoirs includes: The reservoir water level for the next period is determined based on the power generation flow and gate opening of the cascade reservoirs in the current period. If the reservoir water level in the next period does not exceed the difference between the maximum limit water level and the reserved margin, the current operating status of the cascade reservoirs is determined based on the gate opening and output in the current period. If the reservoir water level exceeds the difference between the maximum limit water level and the reserved margin in the next time period, the power generation flow rate is determined according to the method of not discarding water, and the flood discharge flow rate is determined according to the principle of balancing inflow and outflow, so as to obtain the opening degree of the intermediate gate; the calculation time is reversed to the previous gate opening time, the opening degree of the intermediate gate is used as the gate opening degree of the next time period, and the future operating status of the cascade reservoir is determined according to the gate opening degree and output of the next time period.
Citation Information
Patent Citations
Cascade power station gate control strategy generation and rolling optimization method and system
CN114839943A
Reservoir intelligent flood control optimal scheduling method and system
CN117787599A