Method for constructing dynamic optimization scheduling model of power generation and scouring and silting of reservoir with much sediment

Abstract

The scheme belongs to the field of reservoir scouring and silting dispatching, and particularly relates to a method for constructing a power generation-silting dynamic optimization dispatching model of a multi-sediment reservoir. The method for constructing the power generation-silting dynamic optimization dispatching model of the multi-sediment reservoir comprises the following steps: S10: collecting reservoir topographic scanning data, hydrological monitoring data and ecological space data; the ecological space data comprises an egg production site elevation threshold distribution map and a benthic habitat bottom type zoning map; the data are superimposed into a three-dimensional geographical ecological model through the collection positions of the data; S20: based on historical scouring and silting operation records, extracting scouring and silting equipment dispatching parameters and corresponding geographical ecological models before and after scouring and silting, and establishing a scouring and silting reduction model; and calculating the silting reduction amount according to the geographical ecological models before and after scouring and silting. The scheme solves the problem of extensive damage to reservoir ecology in the prior art scouring and silting dispatching, has dynamic prediction, intelligent dispatching and closed-loop management capabilities, and significantly improves the precision, efficiency and benefit of scouring and silting.

Description

Technical Field

[0001] This scheme belongs to the field of reservoir scour and sedimentation scheduling, specifically involving the construction method of dynamic optimization scheduling model for power generation and scour and sedimentation in reservoirs with high sediment content. Background Technology

[0002] In reservoir operation, scour and sedimentation calculation is a core aspect of understanding the reservoir's operational status. Chinese patent CN113421341A discloses a method for calculating scour and sedimentation based on a regular grid DEM. This method creates a regular grid DEM from multi-period measured topographic data, divides the grid cells diagonally into triangular prisms, and calculates the scour and sedimentation volume of each volume cell according to four models: pure scour, pure sedimentation, semi-scour, and semi-sedimentation. Finally, the total scour and sedimentation volume is obtained by summing these calculations. Compared to traditional regular grid mosaicking methods, cross-sectional methods, and triangular prism methods based on irregular triangular meshes, this method solves the problems of mismatched DEM nodes and large calculation errors in non-regular terrain calculations by accurately matching regular grid nodes and performing categorized volume calculations, thus achieving accurate calculation of scour and sedimentation volume.

[0003] However, this technology suffers from a deficiency in regulating sedimentation and scouring in ecologically sensitive reservoirs: it prioritizes engineering calculations over ecological constraints. Ecologically sensitive reservoirs typically contain ecologically sensitive areas such as fish spawning grounds and feeding grounds, and their bottom sediments are often muddy, making them less tolerant of water disturbances and sediment damage. Although existing methods can accurately calculate sedimentation and scouring volumes using regular grid DEMs, the lack of an ecologically sensitive area identification mechanism prevents dynamic adjustment of sedimentation and scouring operations (i.e., adjusting the scheduling parameters of dredging equipment) based on elevation thresholds of sensitive areas (such as suitable water depth conditions for spawning grounds). In practical applications, sedimentation and scouring operations based solely on technical parameters often lead to abnormally high suspended sediment concentrations, damaging fish spawning substrates and causing a decrease in fish egg hatching rates. Simultaneously, because topographic data is not correlated with bottom sediment type distribution, it is difficult to predict the risk of excessive turbidity in muddy bottom areas caused by high-intensity sedimentation and scouring, resulting in a reduction in benthic habitat area. These problems collectively reflect the issue of traditional technologies causing ecological damage through extensive scheduling when dealing with sedimentation and scouring in ecological reservoirs. Summary of the Invention

[0004] The purpose of this solution is to provide a method for constructing a dynamic optimization scheduling model for power generation and siltation in reservoirs with high sediment content, in order to solve the problem of the extensive and ecologically damaging siltation scheduling of existing technologies.

[0005] To achieve the above objectives, this solution provides a method for constructing a dynamic optimization scheduling model for power generation and sedimentation in multi-sediment reservoirs, comprising the following steps:

[0006] S10: Collect reservoir topographic scanning data, hydrological monitoring data, and ecological spatial data; the ecological spatial data includes a distribution map of spawning ground elevation thresholds and a zoning map of benthic organism habitat substrate types; and overlay the data into a three-dimensional geographic ecological model based on the collection locations of each data point.

[0007] S20: Based on historical scouring and silting operation records, extract the scheduling parameters of scouring and silting equipment and the corresponding geo-ecological models before and after scouring and silting, and establish a scouring and silting reduction model; calculate the silting reduction amount according to the geo-ecological models before and after scouring and silting, and conduct correlation analysis on the weight coefficients of silting reduction amount, scheduling parameters, and hydrological monitoring data and ecological spatial data before and after scouring and silting.

[0008] S30: Obtain the expected sediment reduction, combine the expected sediment reduction with the current geoecological model to obtain the spawning ground elevation threshold distribution, benthic habitat substrate type zoning and hydrological data constraints as safety constraints, determine the adjustment strategy of each weight coefficient in the scour and sediment reduction model according to the safety constraints and the expected sediment reduction, and adjust the scour and sediment reduction model.

[0009] S40: Input the collected current geo-ecological model and the expected siltation reduction into the scour-siltation reduction model to obtain the predicted scheduling parameters.

[0010] The principle and technical effects of this scheme are as follows: First, by collecting reservoir topographic scanning data, hydrological monitoring data, and ecological spatial data, including spawning ground elevation thresholds and benthic substrate type distribution, this scheme constructs a three-dimensional geographic ecological model using spatial overlay technology. This model precisely binds scour and sedimentation calculation units to attributes of ecologically sensitive areas, such as marking sludge removal depth limits for spawning grounds and turbidity disturbance thresholds in benthic areas, forming a foundational data layer that combines topographic accuracy with ecological constraints. This approach transforms scour and sedimentation calculations from extensive regional estimations to refined quantification of grid units, effectively avoiding destructive disturbances to the substrate of fish egg hatching areas and benthic habitats caused by dredging operations. Simultaneously, it achieves spatial coordinate unification and cross-domain correlation of topographic, hydrological, and ecological data, supporting dynamic data updates to reflect the real-time status of the reservoir and providing standardized and visualized underlying data support for subsequent model construction.

[0011] Secondly, this scheme relies on historical records of siltation and sedimentation operations to extract the changes in scheduling parameters and the geographic ecological model before and after siltation and sedimentation. By analyzing and establishing a weighted correlation model of sediment reduction, scheduling parameters, and eco-hydrological data, it quantifies the impact of different siltation and sedimentation intensities on ecological indicators such as water turbidity and fish egg hatching rate. This process transforms expert scheduling experience into calculable model parameters, reducing human decision-making bias and enabling the model to predict the ecological impact of different siltation and sedimentation schemes. Simultaneously, by dynamically adjusting weight coefficients, it automatically adapts to the ecological characteristics of different reservoirs, such as optimizing scheduling emphasis for endangered fish distribution areas or benthic organism-rich areas, without requiring remodeling for a single reservoir, significantly enhancing the model's adaptability and engineering versatility.

[0012] Furthermore, this scheme combines the siltation reduction targets required for power generation with the current geographic and ecological model to generate safety constraints covering spawning ground elevation, benthic substrate types, and hydrological data. These constraints include limiting dredging depth in sensitive areas and controlling the upper limit of water turbidity. The model weight coefficients are adjusted iteratively through algorithms to ensure ecological safety while meeting power generation efficiency requirements. This approach can generate multiple scheduling schemes that balance different priorities, allowing managers to choose according to actual needs and avoiding the extreme, either-or problems of traditional technologies. Simultaneously, by dividing areas according to siltation risk and ecological sensitivity, differentiated scheduling strategies can be implemented, improving the utilization efficiency of dredging equipment while reducing energy consumption from ineffective operations, thus achieving a dynamic balance between power generation efficiency and ecological protection.

[0013] Finally, this solution inputs real-time collected geographic and ecological data with siltation reduction targets into the optimized model. Through calculation, it generates dynamic scheduling parameters that include spatial scour and siltation intensity zoning and time window control. These parameters are then integrated with the reservoir's automated control system, forming a complete closed loop of "data acquisition—model calculation—scheduling execution—effect feedback." This approach supports real-time response to sudden water conditions or ecological changes, rapidly generating emergency scheduling plans; records data throughout the entire process to form a traceable scheduling log, providing long-term training data for continuous model optimization; and outputs visualized decision-making data to assist managers in scientifically evaluating the ecological and economic benefits of scheduling plans. Furthermore, it automatically provides restoration suggestions based on post-scour and siltation ecological monitoring results, constructing a full-chain management system covering scheduling, execution, feedback, and restoration.

[0014] In summary, this solution addresses the problem of existing technologies' extensive and destructive reservoir ecosystems in scouring and sedimentation management. It possesses dynamic prediction, intelligent scheduling, and closed-loop management capabilities, significantly improving the accuracy, efficiency, and effectiveness of scouring and sedimentation.

[0015] Furthermore, in step S10, when constructing the geographic ecological model, the topographic scanning data, hydrological monitoring data, and ecological spatial data are uniformly converted to the same geographic coordinate system according to the collection location to ensure the consistency of various types of data in spatial location. Then, a three-dimensional topographic grid model is constructed based on the topographic scanning data, and each grid cell is assigned a unique spatial coordinate identifier. According to the collection location of the hydrological monitoring data, various hydrological parameters of the hydrological monitoring data are mapped to the corresponding topographic grid cells. Then, according to the distribution characteristics of the ecological spatial data, the spawning ground elevation threshold and the substrate type of benthic habitat are associated with the corresponding topographic grid. Spatial interpolation processing is performed on the superimposed multidimensional data to fill the data collection blind spots and form a continuous and complete three-dimensional geographic ecological model. Finally, the spatial topological relationship of the constructed three-dimensional geographic ecological model is verified to ensure the logical consistency and correlation of topographic, hydrological, and ecological data in spatial location.

[0016] This solution achieves significant technical results through operations such as unifying coordinates of multi-source data, constructing terrain grids, mapping and associating data, interpolation completion, and topology verification. Taking a high-sediment reservoir in the Yellow River as an example, in previous scheduling, due to inconsistent and missing coordinate systems and correlations between topographic, hydrological, and ecological data, scouring and sedimentation operations often damaged fish spawning grounds and were inefficient. However, by adopting the technology of this solution, data misalignment and superposition are avoided by unifying the coordinate system, hydrological parameters such as flow and sediment concentration are accurately mapped based on the terrain grid, scouring and sedimentation operations are constrained by associating spawning ground elevation thresholds with benthic substrate types, interpolation fills data blind spots to improve model integrity, and topology verification ensures data logical consistency. Ultimately, this improves the success rate of fish spawning in the reservoir area while increasing scouring and sedimentation efficiency, achieving synergistic optimization of engineering benefits and ecological protection.

[0017] Furthermore, in step S20, when establishing the scour and sediment reduction model, dredging flow, operation duration, and equipment scheduling power are extracted from historical scour and sedimentation records, along with changes in topographic elevation, hydrological parameter fluctuations, and ecological spatial distribution before and after scour and sedimentation. Based on the geographic ecological model, the sediment reduction amount for each operation area is calculated, including changes in volume and thickness. Then, the correlation between sediment reduction amount and scheduling parameters is analyzed to identify key influencing parameters. Next, a correlation analysis is performed on sediment reduction amount, key scheduling parameters, and hydrological and ecological data to determine the influence relationships and weighting coefficients among the parameters. Subsequently, a scour and sediment reduction model is constructed with scheduling parameters as input and sediment reduction amount as output, and historical data is used for training and validation. Finally, sensitivity analysis is used to evaluate the impact of parameter changes on the output of the scour and sediment reduction model, and the model structure and parameter settings are optimized.

[0018] Taking a reservoir in the upper reaches of the Yangtze River as an example, traditional scour and sedimentation scheduling often leads to "over-dredging" or "excessive sedimentation" due to a lack of data correlation analysis. This solution extracts parameters such as historical dredging flow and operation duration and correlates them with topographic, hydrological, and ecological change data. Through correlation analysis, it identifies the key relationship chain of "flow-sediment concentration-benthic organism damage rate," and constructs a scour and sedimentation reduction model that can accurately predict the sedimentation reduction effect and ecological impact of different scheduling schemes. In practical applications, the optimized scheduling scheme of the scour and sedimentation reduction model reduces the average annual sedimentation in the reservoir area, lowers the rate of benthic organism habitat destruction, and reduces the energy consumption of dredging equipment, achieving a triple benefit improvement in sedimentation control, ecological protection, and cost control. Furthermore, by mining historical scour and sedimentation data under different hydrological conditions, the scour and sedimentation reduction model can automatically adjust parameter weights to adapt to seasonal sedimentation changes, improve scour and sedimentation efficiency during the flood season, and ensure ecological flow during the non-flood season. Sensitivity analysis provides managers with visualized decision-making basis, promoting the scientific nature of scheduling decisions. Moreover, this solution demonstrates good adaptability in different types of reservoirs, effectively reducing the migration cost of scheduling technology.

[0019] Furthermore, when determining the influence relationships between various parameters, a dataset is first constructed. Where Y k X is the amount of sediment reduction. ik Z is a key scheduling parameter. jk Use hydrological monitoring data or ecological data; then calculate the Pearson correlation coefficient between the parameters. Filter | r ab The relevant parameter pairs are set to 0.5; a multiple linear regression model Y is established, and the weight coefficients of Y are estimated by the least squares method. Then, the variance inflation factor of each parameter is calculated, and the parameters are optimized based on the variance inflation factor. Next, the mean square error (MSE) is calculated by ten-fold cross-validation, and the weight coefficients are iteratively adjusted until the MSE converges. Finally, the final weight coefficients are determined, and the standardized weights are calculated based on the weight coefficients to quantify the relative importance of each parameter to the amount of sediment reduction.

[0020] Furthermore, when determining the weighting coefficients among the parameters, a multiple linear regression model is established. The weighting coefficient β is estimated using the least squares method. i γ j Then calculate the variance inflation factor for each parameter. If VIF i If the value is greater than 10, parameter optimization is performed; then, ten-fold cross-validation is used to calculate the mean squared error. Iteratively adjust the weight coefficients until the MSE converges; finally, determine the final weight coefficients. And calculate the standardized weights Quantify the relative importance of each parameter to the amount of sediment reduction.

[0021] This solution focuses on parameter correlation and weight coefficient analysis. Through constructing a multi-dimensional dataset, screening Pearson correlation coefficients, modeling multiple linear regression, optimizing variance inflation factors, and iterative 10-fold cross-validation, it deeply explores the parameter interaction mechanism of the reservoir sludge-deposition system. In an application at a reservoir on the Yellow River, the dataset integrates siltation reduction, scheduling parameters, and hydrological and ecological data. Correlation analysis identifies key correlations such as "dredging flow - siltation volume." Multiple regression combined with least squares method scientifically estimates weights, variance inflation factors eliminate redundant parameters (such as interference from equipment power marginal benefits), and 10-fold cross-validation ensures model stability. The final standardized weights accurately quantify the impact of parameters on siltation reduction (e.g., dredging flow weight reaches 0.4), providing refined parameter support for scheduling strategies. Dynamically optimized parameters adapt to complex operating conditions such as high sediment content and ecological sensitivity during the flood season, ensuring both power generation efficiency and enhancing the long-term maintenance capacity of the reservoir, promoting the transformation of multi-sediment reservoirs from experience-based decision-making to intelligent and precise scheduling.

[0022] Furthermore, the scheduling parameters include the working time and power of the siltation and sedimentation equipment. When the siltation and sedimentation reduction model obtains the scheduling parameters, it also includes the step of adjusting the scheduling parameters: First, based on the type of benthic habitat substrate, sensitive areas are divided in the geographic ecological model. Different upper limits of working power are set according to the fragile substrate and dense organisms in different sensitive areas. The actual working power of the siltation and sedimentation equipment is adjusted according to the upper limit of working power. Then, the siltation and sedimentation volume in each area is calculated based on the expected sedimentation reduction. The working time of the siltation and sedimentation equipment is adjusted in combination with the actual working power of the siltation and sedimentation equipment. Then, the adjusted working time and working power are combined and substituted into the siltation and sedimentation reduction model to simulate and derive the estimated sedimentation reduction volume. The scheduling parameters are dynamically corrected according to the estimated sedimentation reduction volume.

[0023] Taking a reservoir with high sediment content as an example, the reservoir area contains both spawning grounds for migratory fish and regular water storage areas, highlighting the conflict between ecological protection and the need for sediment flushing. This solution, based on a geographic ecological model, finely divides the reservoir area into highly sensitive spawning grounds, moderately sensitive migratory channels, and low-sensitivity regular operation areas. The working power and duration of the sediment flushing equipment are adjusted differently for each area based on its ecological characteristics and sediment flushing needs. In highly sensitive areas, a low-power, long-duration operation strategy is adopted to minimize disturbance to the bottom sediment and protect the spawning and hatching environment for fish. In moderately sensitive areas, the power is appropriately increased and the operation time shortened to ensure unobstructed migratory channels. In low-sensitivity areas, high-power, short-duration operations are used to efficiently reduce sedimentation. Subsequently, the adjusted scheduling parameters are substituted into the sediment flushing reduction model to derive the estimated sediment reduction amount. After comparing this with the target value, the scheduling scheme is precisely optimized through multiple simulations and dynamic fine-tuning of the parameters. Ultimately, this plan not only ensured the survival and reproduction of fish in different areas and maintained the stability of benthic communities, but also efficiently completed the task of reducing siltation, successfully achieving a harmonious balance between the benefits of reservoir engineering and ecological protection.

[0024] Furthermore, when constructing the geographic ecological model, height gauges are rationally deployed in different areas of the reservoir as positioning benchmarks for topographic and ecological information. UAVs equipped with LiDAR devices are used to scan the entire reservoir area from multiple angles along a predetermined route to acquire high-resolution image data of the water surface, reservoir banks, and shallow underwater layers. Image recognition technology is used to process the data, automatically identifying riverbed morphology, shallow beach locations, and vegetation distribution. Fish images are captured and their outlines are marked using contour recognition algorithms. At the same time, the relative positional relationship between the ecological topography and the height gauges is compared to construct a three-dimensional ecological model. This model is then spatially registered, overlaid, and fused with the stored reservoir topographic data to generate a geographic ecological model that integrates topographic and ecological information. Based on the fish outline characteristics and the stored knowledge base, fish species are identified, the number of different fish species is counted, and a spatial distribution heat map is drawn. Seasonal, climatic, and weather information, as well as fish migration, foraging, and reproductive habits, are obtained to analyze fish school behavior patterns. The sensitivity division of each area is dynamically adjusted based on the fish school behavior patterns.

[0025] This solution utilizes height gauges deployed in key areas of the reservoir and LiDAR technology from unmanned aerial vehicles (UAVs) to acquire high-precision image data of the entire area. This allows for precise identification of ecological and topographical information such as riverbed morphology and shallow locations, while also capturing images of the distribution of various rare fish species. Image recognition and 3D modeling are then used to generate a geographic ecological model integrating topographic and ecological information. This model is dynamically divided into different zones (such as highly sensitive areas for fish reproduction and sensitive areas in migration channels) based on fish habits and seasonal climate characteristics. During flushing and siltation scheduling, low-intensity, short-duration operational strategies are implemented for highly sensitive areas to avoid the fish breeding season; for other areas, efficient flushing and siltation schemes are employed to improve fish spawning success rates and maintain the integrity of benthic communities. Furthermore, efficient flushing and siltation operations are conducted in non-sensitive areas defined by the model, combined with precise scheduling parameters, to increase the annual siltation reduction of the reservoir while reducing the failure rate of power generation equipment due to siltation. In addition, the mechanism for dynamically adjusting the division of sensitive areas in this plan enables the reservoir to quickly adapt to changes in fish behavior under different operating conditions such as flood season and dry season, avoiding ecological damage caused by seasonal climate fluctuations, and further realizing the dynamic balance and sustainable development of sediment-laden reservoir scouring and siltation scheduling and ecological protection.

[0026] Furthermore, when acquiring seasonal, climatic, and weather data, based on historical hydrological data and seasonal variation patterns, the periodic characteristics of reservoir inflow and sediment deposition rate are analyzed. Combined with climate forecasts within a preset climate observation period, the impact of weather on water levels is predicted, and a flushing and siltation operation plan is formulated within a preset planning period. Real-time weather forecasts are acquired, and storage capacity is reserved based on the forecast content, and the scheduling parameters of the flushing and siltation equipment are adjusted. Then, the reservoir water storage is coupled with the migration, foraging, and reproductive habits of fish for analysis, and the flushing and siltation operation plan is adjusted based on the analysis results. Finally, the time, area, and intensity of the flushing and siltation operation plan are dynamically adjusted, and the flushing and siltation operation plan is adjusted according to the migration and reproductive habits of fish.

[0027] Taking a reservoir with high sediment content as an example, the reservoir serves functions such as power generation, flood control, and ecological water replenishment, and the reservoir area is an important habitat for various rare fish species. This plan, through analysis of nearly ten years of historical hydrological data and combined with medium- and long-term climate forecasts, developed a seasonal dredging and sedimentation operation plan in advance. Dredging of key areas was completed before the rainy season, effectively improving the reservoir's flood control and storage capacity. Simultaneously, water levels and dredging and sedimentation equipment parameters were dynamically adjusted based on real-time weather forecasts. Storage space was reserved in advance during rainstorm warnings to avoid sediment backflow caused by sudden rises in water levels. After analyzing the reservoir's water storage capacity in conjunction with fish migration and breeding habits, operations in sensitive areas were suspended during the fish breeding season to improve the fish's breeding success rate. During the migration period, reasonable control of water levels and dredging and sedimentation intensity ensured unobstructed fish migration channels. Dynamic adjustments to the dredging and sedimentation plan further increased the annual sediment reduction of the reservoir and reduced downtime of power generation equipment due to siltation, successfully achieving synergistic effects between dredging and sedimentation scheduling and ecological protection in the high-sediment reservoir.

[0028] Furthermore, based on geographical ecological models and fish habits, migration channels are identified and water temperature sensors are deployed to collect water temperature data within these channels. When formulating a dredging and sedimentation plan, specific operational periods are designated within the migration channels. Fish migration behavior is identified based on spatial distribution heat maps and fish behavior within the channels. This fish migration behavior is correlated with the activation of water temperature sensors and water temperature monitoring. Water temperature thresholds are set, and the dredging and sedimentation equipment is activated or deactivated based on the relationship between water temperature and these thresholds. A dynamic feedback mechanism is established to analyze the correlation between fish migration behavior, fish population size, fish species, and water temperature. Water temperature thresholds are adjusted based on these correlations, and the timing and intensity of the dredging and sedimentation equipment are dynamically adjusted according to these thresholds.

[0029] By deploying water temperature sensors and LiDAR technology in fish migration channels, the relationship between fish migration trajectories and water temperature changes is accurately captured. The dynamic threshold setting mechanism of this solution intelligently adjusts the water temperature limits for flushing operations based on the size and species of the fish population. When large-scale migration of endangered fish is detected, the water temperature threshold for initiating operations is automatically lowered, minimizing interference with the fish population. In actual operation, this solution allows flushing operations to avoid peak fish migration periods. After the equipment pauses operations due to water temperature control, the success rate of fish passing through the migration channels is increased, effectively ensuring normal fish migration and reproduction. Simultaneously, based on the dynamic adjustment of water temperature and fish behavior, flushing operations are carried out efficiently during non-sensitive periods, improving the overall operational efficiency of the flushing equipment.

[0030] Furthermore, a database of correspondences between seasons, climate, weather, and fish physiological characteristics is established, and suitable water temperature ranges for fish under different conditions are set. Real-time collected water temperatures are matched with seasonal characteristics, and water temperature thresholds are set based on fish physiological characteristics. Then, the water temperature thresholds are adjusted according to preset climate observation time predictions. Next, the water temperature thresholds are adjusted in real-time according to weather forecasts. Finally, by monitoring fish behavior in migration channels and the siltation operation of siltation equipment, a threshold adjustment feedback mechanism is established. This mechanism is used to compare actual collected fish behavior changes with predicted fish behavior changes based on the siltation operation of the siltation equipment and water temperature, and to correct the water temperature threshold adjustment strategy based on the comparison results.

[0031] By establishing a database of correspondences between seasons, climate, weather, and fish physiological characteristics, the suitable water temperature range for fish breeding season was identified. In spring, the water temperature threshold is automatically lowered to ensure that flushing operations avoid the water temperatures sensitive to fish breeding, thus improving egg survival rates. Based on medium- and long-term climate forecasts, the upward trend of water temperature during periods of high temperature and drought is anticipated in advance, and the threshold is tightened beforehand to prevent flushing operations from exacerbating the temperature rise, effectively reducing fish mortality due to high-temperature stress. Simultaneously, before the arrival of cold waves, the threshold is appropriately relaxed based on weather forecasts, seizing the window of opportunity before the water temperature drops to complete necessary flushing work, improving the annual flushing plan completion rate. Furthermore, this scheme, combined with weather forecasts, adjusts the threshold in time before heavy rain, anticipating a sudden drop in water temperature due to rainfall. By suspending flushing operations in advance, stress responses in fish populations caused by drastic water temperature changes are prevented, improving the success rate of fish migration. By predicting water temperature change trends through climate and weather forecasts and dynamically adjusting the water temperature threshold based on the fish's adaptive temperature under different seasons, climates, and weather conditions, not only is the negative impact of water temperature fluctuations on fish survival and reproduction reduced, but flushing operations are also precisely matched to fish activity patterns. With the water temperature prediction and dynamic regulation mechanism, the stability of the planktonic and benthic community structure in the reservoir is improved, providing a better habitat and foraging environment for fish. At the same time, the silt flushing equipment has reduced the number of ineffective start-ups and shutdowns due to the optimization of water temperature control, reducing equipment wear and tear and significantly lowering maintenance costs. Attached Figure Description

[0032] Figure 1 This is a flowchart illustrating the method for constructing a dynamic optimization scheduling model for power generation and siltation in multi-sediment reservoirs, as described in this invention.

[0033] Figure 2 This is a flowchart illustrating the adjustment of scheduling parameters when the siltation reduction model is solved in an embodiment of the present invention. Detailed Implementation

[0034] The following will describe the concept and technical effects of the present invention clearly and completely with reference to embodiments, so as to fully understand the purpose, features and effects of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are all within the scope of protection of the present invention.

[0035] like Figure 1 As shown, the method for constructing a dynamic optimization scheduling model for power generation and siltation in reservoirs with high sediment content includes the following steps:

[0036] S10: Collect reservoir topographic scanning data, hydrological monitoring data, and ecological spatial data; the ecological spatial data includes a distribution map of spawning ground elevation thresholds and a zoning map of benthic organism habitat substrate types; and overlay the data into a three-dimensional geographic ecological model based on the collection locations of each data point.

[0037] S20: Based on historical scouring and silting operation records, extract the scheduling parameters of scouring and silting equipment and the corresponding geo-ecological models before and after scouring and silting, and establish a scouring and silting reduction model; calculate the silting reduction amount according to the geo-ecological models before and after scouring and silting, and conduct correlation analysis on the weight coefficients of silting reduction amount, scheduling parameters, and hydrological monitoring data and ecological spatial data before and after scouring and silting.

[0038] S30: Obtain the expected sediment reduction, combine the expected sediment reduction with the current geoecological model to obtain the spawning ground elevation threshold distribution, benthic habitat substrate type zoning and hydrological data constraints as safety constraints, determine the adjustment strategy of each weight coefficient in the scour and sediment reduction model according to the safety constraints and the expected sediment reduction, and adjust the scour and sediment reduction model.

[0039] S40: Input the collected current geo-ecological model and the expected siltation reduction into the scour-siltation reduction model to obtain the predicted scheduling parameters.

[0040] In step S10, when constructing the geographic ecological model, the topographic scanning data, hydrological monitoring data, and ecological spatial data are uniformly converted to the same geographic coordinate system according to the collection location to ensure the consistency of various data in spatial location. Then, a three-dimensional topographic grid model is constructed based on the topographic scanning data, and each grid cell is assigned a unique spatial coordinate identifier. According to the collection location of the hydrological monitoring data, various hydrological parameters of the hydrological monitoring data are mapped to the corresponding topographic grid cells. Then, according to the distribution characteristics of the ecological spatial data, the spawning ground elevation threshold and the substrate type of benthic habitat are associated with the corresponding topographic grid. Spatial interpolation processing is performed on the superimposed multidimensional data to fill the data collection blind spots and form a continuous and complete three-dimensional geographic ecological model. Finally, the spatial topological relationship of the constructed three-dimensional geographic ecological model is verified to ensure the logical consistency and correlation of topographic, hydrological, and ecological data in spatial location.

[0041] Specifically, different data sources (such as topographic mapping, hydrological monitoring equipment, and ecological surveys) may use different coordinate systems (such as WGS84 and Beijing 54). These need to be unified to the same benchmark (such as CGCS2000) through coordinate transformation algorithms to avoid data misalignment and overlay due to coordinate differences. If the original data uses a local coordinate system (x... local ,y local It needs to be transformed into the target coordinate system (x) through affine transformation or projection transformation. global ,y global The specific formula is shown in formula (1):

[0042]

[0043] Where a, b, c, d, e, and f are transformation parameters, which can be solved by using the coordinates of known control points.

[0044] Digital elevation model (DEM) technology discretizes terrain scan data (such as elevation points and contour lines) into regular or irregular grids (such as TIN triangular meshes and rectangular grids). Each grid corresponds to a spatial location and elevation value, forming a three-dimensional digital representation of the terrain. If a regular rectangular grid is used, the spatial coordinates of the grid cell can be calculated using the row and column numbers (i,j) and the grid resolution r, as shown in formula (2) below:

[0045]

[0046] Where (x0,y0) are the starting coordinates of the grid, and h(i,j) are the elevation values ​​of the corresponding grid.

[0047] Discrete hydrological monitoring point data are assigned to a topographic grid using spatial interpolation or spatial correlation methods. For example, if the monitoring point is within the grid, the value is directly assigned; if it is located at the grid boundary, the parameter values ​​within the grid can be estimated using inverse distance weighted (IDW) or Kriging interpolation. For the grid point P to be estimated, its hydrological parameter value Z... P From the surrounding n monitoring points Z i The weighted average yields:

[0048]

[0049] Where, d i To monitor the distance from point i to point P, p is the weighting exponent (usually taken as 2).

[0050] Ecological data (such as spawning ground extent and substrate type zoning) is typically stored as isometric or vector data, and its attributes need to be assigned to the corresponding terrain grid through spatial overlay analysis. For example, if a grid is located within a spawning ground polygon, then that grid is assigned the "spawning ground" attribute and an elevation threshold limit.

[0051] Interpolation algorithms (such as bilinear interpolation and spline interpolation) are used to estimate values ​​for missing regions based on known data points, ensuring the spatial continuity of the model. For example, in areas without hydrological monitoring points, the flow value for that area can be obtained by interpolating flow data from surrounding points.

[0052] Spatial topology analysis is used to examine the logical relationships between data (such as whether grids overlap or whether the boundaries of ecologically sensitive areas match the terrain). For example, it verifies whether the elevation of the spawning ground area meets its threshold range to avoid logical contradictions.

[0053] In this embodiment of the scheme, the scheduling optimization project for reservoirs in mountainous areas with high sediment load faces the problem of conflict between power generation demand and ecological protection. Severe soil erosion in the upstream basin and sediment accumulation in the reservoir area lead to decreased power generation efficiency, while insufficient ecological flow in the downstream river affects fish survival.

[0054] When constructing the geographic ecological model, firstly, high-precision topographic scanning data was acquired using LiDAR technology from unmanned aerial vehicles (UAVs). Flow and sediment concentration data were collected from hydrological monitoring stations, combined with data on fish spawning ground distribution and benthic substrate types provided by the fisheries department. Since the topographic data uses the CGCS2000 coordinate system while the hydrological data uses the WGS84 coordinate system, coordinate transformation was performed to unify all data to the CGCS2000 coordinate system. Next, a 10m × 10m rectangular grid model was constructed based on the topographic scanning data, assigning spatial coordinates and elevation values ​​to each grid cell. Using an inverse distance weighting method, discrete hydrological monitoring data were mapped to corresponding grid cells; for example, flow data from a monitoring point was allocated to surrounding grid cells. Simultaneously, ecological data such as fish spawning ground extent and substrate type were spatially overlaid with the topographic grid, marking elevation thresholds for sensitive areas.

[0055] To address the issue of missing data in certain areas, the project team employed Kriging interpolation to fill in the gaps, ensuring the continuity and completeness of the model data. Finally, spatial topology analysis was used to check whether the elevations of the spawning grounds met the constraints, verifying the logical consistency of the data. The ultimately constructed three-dimensional geographic ecological model provides accurate foundational data for subsequent siltation and sedimentation management, satisfying the siltation reduction targets required for reservoir power generation while effectively protecting fish spawning grounds and benthic habitats, achieving a balance between engineering and ecological benefits.

[0056] In step S20, when establishing the scour and sediment reduction model, dredging flow, operation duration, and equipment scheduling power are extracted from historical scour and sedimentation records, along with changes in topographic elevation, hydrological parameter fluctuations, and ecological spatial distribution before and after scour and sedimentation. Based on the geographic ecological model, the sediment reduction amount for each operation area is calculated, including changes in volume and thickness. Then, the correlation between sediment reduction amount and scheduling parameters is analyzed to identify key influencing parameters. Next, a correlation analysis is conducted on sediment reduction amount, key scheduling parameters, and hydrological and ecological data to determine the influence relationships and weighting coefficients among the parameters. Subsequently, a scour and sediment reduction model is constructed with scheduling parameters as input and sediment reduction amount as output, and historical data is used for training and validation. Finally, sensitivity analysis is used to evaluate the impact of parameter changes on the output of the scour and sediment reduction model, and the model structure and parameter settings are optimized.

[0057] Specifically, using a geographic ecological model, the sediment reduction volume and thickness within a unit grid are calculated based on the elevation difference before and after sedimentation. Let the elevation of a grid before sedimentation be h1, the elevation after sedimentation be h2, and the grid area be S. Then, the sediment reduction volume V = S × (h1 - h2), and the sediment reduction thickness Δh = h1 - h2. Correlation analysis is used to quantify the influence of each scheduling parameter on sediment reduction, identifying key parameters with significant impact. The Pearson correlation coefficient is used to measure the correlation between the parameters of the sediment reduction model and the sediment reduction amount. A multivariate sediment reduction model is constructed using multiple regression analysis or machine learning algorithms to quantify the contribution ratio of each factor to the results. By changing individual parameter values, the magnitude of changes in the model output is observed to determine the model's sensitivity to each parameter, thereby allowing for targeted optimization of the sediment reduction model.

[0058] More specifically, when performing correlation analysis on weighted coefficients, the following steps are included:

[0059] S201: Extract the siltation reduction data Y and key scheduling parameter data X. i (i = 1, 2, ..., m) and hydrological and ecological data Z j (j = 1, 2, ..., n), grouped and categorized according to time series or work area to construct a dataset. Where N is the number of data samples;

[0060] S202: Calculate the Pearson correlation coefficient r between the parameters, as shown in the following formula (4):

[0061]

[0062] Where a and b represent different parameters, The mean of the corresponding parameters; filter out |r ab For parameter pairs with a value greater than 0.5, a significant correlation is determined.

[0063] S203: Construct a multiple linear regression model, as shown in the following formula (5):

[0064]

[0065] Where, β i γ j ∈ represents the weight coefficient to be estimated, and ∈ represents the random error term.

[0066] S204: The model parameters are estimated using the least squares method, and the objective function is shown in the following formula (6):

[0067]

[0068] The initial weight coefficients are obtained by solving this optimization problem.

[0069] S205: Calculate the variance inflation factor (VIF) of each parameter. The calculation formula is shown in formula (7) below:

[0070]

[0071] in, Let VIF be the coefficient of determination for a regression model with the i-th parameter as the dependent variable and the remaining parameters as independent variables; if VIF i If the value is greater than 10, then parameter elimination or transformation processing will be performed.

[0072] S206: The model prediction accuracy is evaluated through cross-validation. The ten-fold cross-validation method is used to divide the dataset D into 10 subsets. Nine subsets are used as the training set and one subset is used as the test set. The mean squared error (MSE) is calculated using the formula shown in formula (8) below:

[0073]

[0074] Among them, D s For the s-th test set, N s Its sample size, These are the model's predicted values.

[0075] S207: Based on the cross-validation results, iteratively adjust the weight coefficients until the MSE converges or reaches a preset threshold; finally, determine the stable weight coefficients. The standardized weights of each parameter are calculated as shown in the following formula (9):

[0076]

[0077] in, σ Y These represent the standard deviation of the corresponding parameters and the standardized weights w. i w j This reflects the relative importance of each parameter to the amount of sediment reduction.

[0078] In this embodiment of the scheme, the reservoir suffers from severe soil erosion in the watershed, with an average annual siltation volume of 8 million cubic meters. Simultaneously, the reservoir serves multiple functions, including flood control, power generation, and ecological water replenishment. Traditional scheduling methods struggle to balance dredging and siltation with ecological protection. To establish the dredging and siltation reduction model, nearly 10 years of data on reservoir dredging and siltation operations, hydrological monitoring, and ecological surveys were collected to construct a dataset including dredging flow, equipment power, operation duration, topographic elevation, and fish habitat area. Using a geographic ecological model, the reservoir area was divided into 50m × 50m grid cells. The siltation reduction volume and thickness were calculated by combining the elevation difference before and after dredging and siltation (e.g., a grid cell with an elevation of 325m before dredging and 324.5m after dredging and siltation) with the grid area. Pearson correlation coefficient analysis revealed a correlation coefficient of 0.85 between dredging flow and siltation reduction, and a correlation coefficient of 0.72 between equipment power and benthic disturbance, identifying these as key parameters. After constructing a multiple linear regression model, the weight coefficients were solved using the least squares method. Multicollinearity was detected between "flow velocity" and "sediment concentration" using variance inflation factor detection, leading to parameter optimization and elimination. The model parameters were iteratively adjusted using ten-fold cross-validation. When the mean squared error converged to a preset threshold, the standardized weights of each parameter were calculated (e.g., dredging flow weight 0.42, equipment power weight 0.28). Sensitivity analysis showed that for every 10% increase in equipment power, the marginal benefit of silt reduction decreased by 5%. Based on this, the operational strategy for high-sensitivity areas (fish spawning grounds) was adjusted to "low power, long duration".

[0079] Among them, such as Figure 2 As shown, the scheduling parameters include the working time and power of the siltation and sedimentation equipment. When the siltation and sedimentation reduction model obtains the scheduling parameters, it also includes the step of adjusting the scheduling parameters: First, based on the type of benthic habitat substrate, sensitive areas are divided in the geographic ecological model. Different upper limits of working power are set according to the fragile substrate and dense organisms in different sensitive areas. The actual working power of the siltation and sedimentation equipment is adjusted according to the upper limit of working power. Then, the siltation and sedimentation volume of each area is calculated based on the expected sedimentation reduction. The working time of the siltation and sedimentation equipment is adjusted in combination with the actual working power of the siltation and sedimentation equipment. Then, the adjusted working time and working power are combined and substituted into the siltation and sedimentation reduction model to simulate and derive the estimated sedimentation reduction volume. The scheduling parameters are dynamically corrected according to the estimated sedimentation reduction volume.

[0080] In this embodiment of the scheme, the reservoir suffers from severe siltation due to its rich benthic organisms and wide distribution of fish spawning grounds. When adjusting the scheduling parameters, the reservoir area is first divided into high, medium, and low sensitivity zones based on the benthic habitat sediment data from the geographic ecological model. For the silty clay sediment and dense benthic organisms in the high-sensitivity zone, equipment power is strictly limited, while the power threshold is appropriately increased in the gravelly sediment sediment of the low-sensitivity zone. After calculating the scouring and silting volume for each area based on the annual siltation reduction target, a low-power, long-duration (8 hours / day) strategy is adopted for the high-sensitivity zone, while a high-power (70%), short-duration (5 hours / day) strategy is used for the low-sensitivity zone. Through simulation and verification using the scouring and siltation reduction model, and after optimization and adjustment, the reservoir not only achieved the siltation reduction target and stabilized the benthic community, but also reduced the damaged area of ​​fish spawning grounds by 75%, achieving a win-win situation for both engineering benefits and ecological protection.

[0081] Specifically, when constructing the geographic ecological model, height gauges are rationally deployed in different areas of the reservoir as positioning benchmarks for topographic and ecological information. UAVs equipped with LiDAR devices are used to scan the entire reservoir area from multiple angles along a predetermined route to acquire high-resolution image data of the water surface, reservoir banks, and shallow underwater layers. Image recognition technology is used to process the data, automatically identifying riverbed morphology, shallow beach locations, and vegetation distribution. Fish images are captured and their outlines are marked using contour recognition algorithms. At the same time, the relative positional relationship between the ecological topography and the height gauges is compared to construct a three-dimensional ecological model. This model is then spatially registered, overlaid, and fused with the stored reservoir topographic data to generate a geographic ecological model that integrates topographic and ecological information. Fish species are identified based on fish outline characteristics combined with a stored knowledge base. The number of different fish species is counted, and a spatial distribution heat map is drawn. Seasonal, climatic, and weather information, as well as fish migration, foraging, and reproductive habits, are obtained to analyze fish school behavior patterns. The sensitivity classification of each area is dynamically adjusted based on the fish school behavior patterns.

[0082] In this embodiment of the scheme, when dynamically adjusting the scouring and sedimentation intensity and regional sensitivity based on fish behavior patterns, the core spawning area of ​​the fish school (such as a shallow beach with a water depth of 1.2-2 meters and a fine sandy bottom) is first identified based on a geographic ecological model and a fish habit database, and designated as a highly sensitive area. During the fish egg incubation period (such as in spring, March-April), a strict "zero scouring and sedimentation" policy is implemented in this area, prohibiting any equipment operation. The attachment status of the fish eggs is tracked in real time through sonar monitoring and image recognition technology (such as carp eggs often adhering to the surface of aquatic plants or sand grains), ensuring that the incubation environment is not affected by water flow disturbance. This method can improve the survival rate of fish eggs during this stage.

[0083] Once the fry hatch and enter the juvenile stage (e.g., body length <3cm), the previously highly sensitive area should be adjusted to a moderately sensitive area. Only intermittent operation of the flushing equipment at 10%-15% of its rated power is permitted. This low-intensity disturbance at this stage prevents the juveniles from losing their habitat due to strong currents and, by gently stirring the riverbed, releases a small amount of organic matter (such as humus and benthic diatoms), providing initial food for filter-feeding juveniles and increasing their growth rate during this phase.

[0084] Once the fish reach adulthood (body length ≥10cm, such as during the summer months of June to August), the regional sensitivity level is lowered to a low-sensitivity zone, and the flushing and siltation intensity is moderately increased (e.g., power 30%-50%). This flushing and siltation operation breaks up the sediment layer at the bottom of the river, allowing deposited nutrients such as nitrogen and phosphorus to be fully released and diffused into the upper and middle layers of the water. This triggers a peak in the reproduction of algae (such as diatoms and green algae) and zooplankton (cladocerans and copepods), forming a closed food chain loop of "nutrients → microorganisms → algae → fish". In this embodiment, the chlorophyll a content in the regulated regional water body during the adult stage can increase by 2-3 times, increasing the feeding frequency of fish and improving the monthly weight gain of individual fish compared to the natural state. Furthermore, the increased reservoir capacity due to flushing and siltation can improve the reservoir's power generation efficiency, achieving a win-win situation of ecological goals of "protecting the reproductive base and promoting growth" and economic benefits.

[0085] This dynamic sensitive zone delineation mechanism, which is deeply linked to the fish reproductive cycle, not only ensures the stable development of fish eggs during the spawning period, but also activates the productivity of the water area through siltation and flushing operations during the growth period, ultimately forming a positive cycle of "ecological protection - resource utilization - economic benefits".

[0086] More specifically, fish species identification can employ deep learning classification models (such as ResNet) to learn fish morphological characteristics through training samples; heatmaps, based on kernel density estimation (KDE) algorithms, visually display fish distribution density; and fish behavior analysis utilizes association rule mining (such as the Apriori algorithm) to discover potential connections between environmental factors and fish activity. The kernel density estimation formula is shown in formula (10) below:

[0087]

[0088] Where n is the number of samples, h is the bandwidth, K(·) is the kernel function (such as Gaussian kernel), and x i Let be the location sample of the i-th fish species. Through the above steps, this scheme achieves full automation from data collection to ecological analysis, providing accurate ecological data support for reservoir sludge and sedimentation scheduling, and effectively balancing engineering needs and ecological protection.

[0089] The process involves several steps. First, when acquiring seasonal, climatic, and weather data, historical hydrological data and seasonal variation patterns are used to analyze the periodic characteristics of reservoir inflow and sediment deposition rates. Combined with climate forecasts within a pre-defined observation period, the impact of weather on water levels is predicted, and a pre-defined flushing and siltation operation plan is formulated. Second, real-time weather forecasts are acquired, and storage capacity is reserved based on the forecast information, while the scheduling parameters of the flushing and siltation equipment are adjusted. Third, the reservoir's water storage is coupled with the migration, foraging, and reproductive habits of fish, and the flushing and siltation operation plan is adjusted based on the analysis results. Finally, the timing, area, and intensity of the flushing and siltation operation plan are dynamically adjusted according to the migration and reproductive habits of fish.

[0090] In this embodiment of the scheme, time series analysis methods (such as the ARIMA model) are used to explore the seasonal patterns of hydrological data; climate prediction models (such as ensemble forecasting systems) are used to predict climate events such as precipitation and drought. Time series decomposition model: Y t =T t +S t +R t , where Y t For raw hydrological data, T t S is the trend term. t For seasonal terms, R t For random terms, periodic features are extracted through decomposition.

[0091] Real-time forecast information is obtained using meteorological data API interfaces. Based on reservoir capacity curves and flood evolution models, the regulation water level is calculated. Scheduling parameters are optimized using equipment efficiency models (such as power-dredging volume relationship curves). Reservoir regulation water level calculation: Total safe predicted inflow V. 调蓄 =V 总 -V 安全 -V 预测来水 , of which total V 总 For total storage capacity, safety V 安全 For safe reservoir capacity, predict the inflow V 预测来水 To forecast the amount of water coming in.

[0092] A fish behavior database was established, and the correlation between water level, water temperature, and fish activity was analyzed using association rule mining (such as the Apriori algorithm). Spatial analysis techniques (such as buffer analysis) were employed to delineate ecologically sensitive areas. Association rule support was calculated as: the number of transactions containing the sum and the total number of transactions. Used to quantify the correlation between fish behavior and water level changes.

[0093] Based on geographical ecological models and fish habits, migration channels are identified and water temperature sensors are deployed to collect water temperature data within these channels. When formulating a dredging plan, specific operating periods are designated within the migration channels. Fish migration behavior is identified based on spatial distribution heat maps and fish behavior within the channels. This fish migration behavior is correlated with the activation of water temperature sensors and water temperature monitoring. Water temperature thresholds are set, and dredging equipment is activated or deactivated based on the relationship between water temperature and these thresholds. A dynamic feedback mechanism is established to analyze the correlation between fish migration behavior, fish population size, fish species, and water temperature. Water temperature thresholds are adjusted based on these correlations, and the timing and intensity of dredging equipment operations are dynamically adjusted according to these thresholds.

[0094] Specifically, a database of correspondences between seasons, climate, weather, and fish physiological characteristics is established, and suitable water temperature ranges for fish under different conditions are set. Real-time collected water temperatures are matched with seasonal characteristics, and water temperature thresholds are set based on fish physiological characteristics. Then, the water temperature thresholds are adjusted based on preset climate observation time predictions. Next, the water temperature thresholds are adjusted in real-time according to weather forecasts. Finally, by monitoring fish behavior in migration channels and the siltation operation of siltation equipment, a threshold adjustment feedback mechanism is established. This mechanism is used to compare actual collected changes in fish behavior with predicted changes based on the siltation operation of the siltation equipment and water temperature, and to correct the water temperature threshold adjustment strategy based on the comparison results.

[0095] In this embodiment of the scheme, the reservoir is not only an important source of water for power generation but also a migration channel for various rare fish species. Traditional dredging and siltation operations often disturb the fish's habitat. This embodiment first accurately identifies two main fish migration channels within the reservoir based on a geographical ecological model and fish behavior studies. Water temperature sensors are deployed at key nodes along these channels to monitor water temperature changes in real time. Simultaneously, LiDAR equipment and image recognition technology are used to generate a thermal map of the fish's spatial distribution, accurately tracking the fish's migration dynamics. When formulating the dredging and siltation plan, specific operation periods are designated along the migration channels. When fish migration is detected and the water temperature approaches a preset threshold, the dredging and siltation equipment is automatically suspended. For example, during a silverfish migration, the equipment was stopped in time, allowing the silverfish to pass smoothly through the migration channel and avoiding fish casualties caused by operational interference.

[0096] Based on this, a database of the correspondence between seasons, climate, weather, and fish physiological characteristics was established, allowing for flexible adjustment of water temperature thresholds according to different seasons. During the spring breeding season, the water temperature threshold was lowered to ensure that silt removal operations did not affect egg hatching. During the high-temperature summer period, the threshold was tightened in advance based on climate forecasts to prevent operations from exacerbating water temperature rises and causing stress to the fish. Based on weather forecasts, a potential sudden drop in water temperature was anticipated before heavy rains, allowing for advance adjustment of thresholds and optimization of silt removal operation schedules. By monitoring fish behavior and silt removal operations, the established threshold adjustment feedback mechanism continuously optimizes the strategy. For example, if abnormal fish aggregation was observed after a silt removal operation, the threshold was promptly corrected, resulting in a significant reduction in fish stress responses during subsequent operations.

[0097] The above descriptions are merely embodiments of the present invention, and common knowledge regarding specific structures and characteristics is not elaborated upon here. It should be noted that those skilled in the art can make various modifications and improvements without departing from the structure of the present invention, and these should also be considered within the scope of protection of the present invention. These modifications and improvements will not affect the effectiveness of the present invention or the practicality of the patent. The scope of protection claimed in this application should be determined by the content of its claims, and the specific embodiments described in the specification can be used to interpret the content of the claims.

Claims

1. A method for constructing a dynamic optimization scheduling model for power generation and scouring / deposition in reservoirs with high sediment loads, characterized in that... Includes the following steps: S10: Collect reservoir topographic scanning data, hydrological monitoring data, and ecological spatial data; the ecological spatial data includes a distribution map of spawning ground elevation thresholds and a zoning map of benthic organism habitat substrate types; and overlay the data into a three-dimensional geographic ecological model based on the collection locations of each data point. S20: Based on historical scouring and silting operation records, extract the scheduling parameters of scouring and silting equipment and the corresponding geo-ecological models before and after scouring and silting, and establish a scouring and silting reduction model; calculate the silting reduction amount according to the geo-ecological models before and after scouring and silting, and conduct correlation analysis on the weight coefficients of silting reduction amount, scheduling parameters, and hydrological monitoring data and ecological spatial data before and after scouring and silting. S30: Obtain the expected sediment reduction, combine the expected sediment reduction with the current geoecological model to obtain the spawning ground elevation threshold distribution, benthic habitat substrate type zoning and hydrological data constraints as safety constraints, determine the adjustment strategy of each weight coefficient in the scour and sediment reduction model according to the safety constraints and the expected sediment reduction, and adjust the scour and sediment reduction model. S40: Input the collected current geographic ecological model and the expected siltation reduction into the alluvial-siltation reduction model to obtain the predicted scheduling parameters; In step S20, when establishing the scour and sediment reduction model, dredging flow, operation duration, and equipment scheduling power are extracted from historical scour and sedimentation records, along with changes in topographic elevation, hydrological parameter fluctuations, and ecological spatial distribution before and after scour and sedimentation. Based on the geographic ecological model, the sediment reduction amount for each operation area is calculated, including changes in volume and thickness. Then, the correlation between sediment reduction amount and scheduling parameters is analyzed to identify key influencing parameters. Next, a correlation analysis is performed on sediment reduction amount, key scheduling parameters, and hydrological and ecological data to determine the influence relationships and weighting coefficients among the parameters. Subsequently, a scour and sediment reduction model is constructed with scheduling parameters as input and sediment reduction amount as output, and historical data is used for training and validation. Finally, sensitivity analysis is used to evaluate the impact of parameter changes on the output of the scour and sediment reduction model, and the model structure and parameter settings are optimized. When determining the influence relationships between parameters, first construct a dataset. ,in To reduce the amount of siltation, As a key scheduling parameter, Use hydrological monitoring data or ecological data; then calculate the Pearson correlation coefficient between the parameters. ,filter Relevant parameter pairs; When determining the weighting coefficients among the parameters, a multiple linear regression model is established. The weighting coefficients are estimated using the least squares method. , Then calculate the variance inflation factor for each parameter. ,like Then, parameter optimization is performed; followed by calculation of the mean square error using ten-fold cross-validation. Iteratively adjust the weight coefficients until the MSE converges; finally, determine the final weight coefficients. , And calculate the standardized weights. Quantify the relative importance of each parameter to the amount of sediment reduction; The scheduling parameters include the working time and power of the flushing and siltation equipment.

2. The method for constructing a dynamic optimization scheduling model for power generation and siltation in multi-sediment reservoirs according to claim 1, characterized in that: In step S10, when constructing the geographic ecological model, the topographic scanning data, hydrological monitoring data, and ecological spatial data are uniformly converted to the same geographic coordinate system according to the collection location to ensure the consistency of various data in spatial location. Then, a three-dimensional topographic grid model is constructed based on the topographic scanning data, and each grid cell is assigned a unique spatial coordinate identifier. According to the collection location of the hydrological monitoring data, various hydrological parameters of the hydrological monitoring data are mapped to the corresponding topographic grid cells. Then, according to the distribution characteristics of the ecological spatial data, the spawning ground elevation threshold and the substrate type of benthic organism habitat are associated with the corresponding topographic grid. Spatial interpolation processing is performed on the superimposed multidimensional data to fill the data collection blind spots and form a continuous and complete three-dimensional geographic ecological model. Finally, the spatial topological relationship of the constructed three-dimensional geographic ecological model is verified to ensure the logical consistency and correlation of topographic, hydrological, and ecological data in spatial location.

3. The method for constructing a dynamic optimization scheduling model for power generation and siltation in multi-sediment reservoirs according to claim 2, characterized in that: When the siltation reduction model obtains the scheduling parameters, it also includes the step of adjusting the scheduling parameters: First, based on the type of benthic habitat substrate, sensitive areas are divided in the geographic ecological model. Different upper limits of working power are set according to the fragile substrate and dense organisms in different sensitive areas. The actual working power of the siltation reduction equipment is adjusted according to the upper limit of working power. Then, the siltation volume of each area is calculated based on the expected siltation reduction. The working time of the siltation reduction equipment is adjusted in combination with the actual working power of the siltation reduction equipment. Then, the adjusted working time and working power are combined and substituted into the siltation reduction model to simulate and derive the estimated siltation reduction volume. The scheduling parameters are dynamically corrected according to the estimated siltation reduction volume.

4. The method for constructing a dynamic optimization scheduling model for power generation and siltation in multi-sediment reservoirs according to claim 3, characterized in that: When constructing the geographic ecological model, height gauges are strategically deployed in different areas of the reservoir as benchmarks for topographic and ecological information positioning. UAVs equipped with LiDAR devices are used to scan the entire reservoir area from multiple angles along a predetermined flight path, acquiring high-resolution image data of the water surface, banks, and shallow underwater layers. Image recognition technology is used to process the data, automatically identifying riverbed morphology, shallow water locations, and vegetation distribution. Fish images are captured and their outlines are marked using contour recognition algorithms. Simultaneously, the relative positional relationship between the ecological topography and the height gauges is compared to construct a 3D ecological model. This model is then spatially registered, overlaid, and fused with stored reservoir topographic data to generate a geographic ecological model integrating topographic and ecological information. Fish species are identified based on their outline characteristics and a stored knowledge base. The number of different fish species is counted, and a spatial distribution heatmap is drawn. Seasonal, climatic, and weather data, as well as fish migration, foraging, and reproductive habits, are acquired to analyze fish school behavior patterns. The sensitivity classification of each area is dynamically adjusted based on these patterns.

5. The method for constructing a dynamic optimization scheduling model for power generation and siltation in multi-sediment reservoirs according to claim 4, characterized in that: When acquiring seasonal, climatic, and weather data, based on historical hydrological data and seasonal variation patterns, the periodic characteristics of reservoir inflow and sediment deposition rate are analyzed. Combined with climate forecasts within a preset climate observation period, the impact of weather on water levels is predicted, and a flushing and siltation operation plan is formulated within a preset planning period. Real-time weather forecasts are acquired, and storage capacity is reserved based on the forecast content, and the scheduling parameters of flushing and siltation equipment are adjusted. Then, the reservoir storage capacity is coupled with the migration, foraging, and reproductive habits of fish for analysis, and the flushing and siltation operation plan is adjusted based on the analysis results. Finally, the timing, area, and intensity of the flushing and siltation operation plan are dynamically adjusted, and the plan is adjusted according to the migration and reproductive habits of fish.

6. The method for constructing a dynamic optimization scheduling model for power generation and siltation in multi-sediment reservoirs according to claim 5, characterized in that: Based on geographical ecological models and fish habits, migration channels are identified and water temperature sensors are deployed to collect water temperature within these channels. When formulating a dredging plan, specific operating periods are designated within the migration channels. Fish migration behavior is identified based on the spatial distribution heat map and fish behavior within the migration channels. This fish migration behavior is then correlated with the activation of water temperature sensors and water temperature monitoring. Water temperature thresholds are set, and the dredging equipment is activated or deactivated based on the relationship between water temperature and the water temperature thresholds. Establish a dynamic feedback mechanism to analyze the correlation between fish migration behavior, fish population, fish species, and water temperature. Adjust the water temperature threshold based on the correlation, and dynamically adjust the operation sequence and intensity of the flushing equipment based on the water temperature threshold.

7. The method for constructing a dynamic optimization scheduling model for power generation and siltation in multi-sediment reservoirs according to claim 6, characterized in that: A database of correspondences between seasons, climate, weather, and fish physiological characteristics is established, and suitable water temperature ranges for fish under different conditions are set. Real-time collected water temperatures are matched with seasonal characteristics, and water temperature thresholds are set based on fish physiological characteristics. Then, the water temperature thresholds are adjusted based on preset climate observation time predictions. Next, the water temperature thresholds are adjusted in real-time based on weather forecasts. Finally, by monitoring fish behavior in migration channels and the siltation operation of siltation equipment, a threshold adjustment feedback mechanism is established. This mechanism is used to compare actual collected fish behavior changes with predicted fish behavior changes based on the siltation operation of the siltation equipment and water temperature, and to correct the water temperature threshold adjustment strategy based on the comparison results.

Images