A reservoir evaporation control method fusing evaporation law analysis and dispatching scenario simulation

Abstract

This invention discloses a reservoir evaporation control method that integrates evaporation law analysis and scheduling scenario simulation, relating to the field of reservoir control. This method deploys a multi-zone remote sensing microclimate acquisition point system in the reservoir area, using remote sensing points in the main dam area, the intake area, and the tributary shallow bay area as representatives to achieve three-dimensional heterogeneous microclimate coverage. Combined with multiple sensor groups such as total radiation meters, aerosol lidar, and stratified temperature and humidity probes, it can achieve high-frequency, 5-minute time-series sampling of water evaporation driving factors and spatial resolution identification of local disturbance behavior. Compared to existing technologies that rely on intermittent measurements from single-point water surface evaporation pans or meteorological stations, this method significantly improves the completeness of environmental baseline data for evaporation behavior prediction, providing high-resolution input conditions for subsequent heat flux modeling and feedback control, and possessing dual technical advantages of accuracy and real-time performance.

Description

Technical Field

[0001] This invention relates to the field of reservoir control technology, specifically to a reservoir evaporation control method that integrates evaporation law analysis and scheduling scenario simulation. Background Technology

[0002] In large reservoir systems in arid and semi-arid plateau regions, typical meteorological characteristics such as significant diurnal temperature gradients, large fluctuations in radiation energy, and frequent air disturbances are observed year-round. This results in nonlinear, high-frequency disturbances in the water evaporation process, interspersed with complex condensation and replenishment behaviors. As closed hydrological systems, reservoirs are not only driven by thermal factors such as solar radiation and wind disturbances, but also constrained by the coupling of multiple complex factors including microclimate stability and aerosol nucleation potential. Therefore, traditional linear scheduling methods are insufficient to reflect the multi-field coupling mechanism of heat, humidity, and dynamics in such systems. Constructing a closed-loop control method based on evaporation potential prediction, condensation compensation identification, and scheduling response feedback has become a key challenge in the regulation of plateau lakes and water conservation reservoir groups.

[0003] Currently, mainstream reservoir scheduling systems often rely on rigid water allocation or planned regulation based on water level, inflow, and demand load. Their response to evaporation losses is limited to periodically adjusting water balance sheets or using regression compensation based on empirical loss deduction coefficients. They fail to dynamically identify the real-time trends in water body evaporative heat flux, and especially lack modeling and utilization of nighttime condensation energy replenishment behavior. This one-way push deficiency in evaporation regulation leads to a common phenomenon of simultaneous amplification of water release and evaporation during clear, hot days. That is, maximum scheduling is implemented during the period of strongest energy input, while the window of condensation capacity at night is completely not activated. This results in the peak of water body heat output lagging behind the peak of energy input on a daily scale, making it impossible to construct a dynamic water vapor balance mechanism of daytime release and nighttime replenishment. At the same time, existing models mostly use single-parameter prediction models such as wind speed, humidity, and temperature to construct potential evaporation, while failing to effectively collect and model key microphysical variables such as disturbance frequency and nucleation particle size. This prevents the accuracy of the regulation response from penetrating to the underlying logic of the condensation mechanism.

[0004] Due to the lack of a control strategy based on the dynamic complementarity of evaporation and condensation, existing reservoir scheduling systems are prone to heat flux overload and subsequent de-control phenomena. This means that the evaporative heat energy released per unit of water release exceeds the nighttime energy replenished by natural condensation. Over the long term, this leads to secondary effects such as abnormal water level drops, microclimate instability, and increased frequency of water body edge drying. Simultaneously, the scheduling response window is fixed and rigid because it does not consider condensation potential. This causes the water transfer threshold to fail to converge in time during periods of high daytime evaporation, while the system misses the opportunity to store energy during periods of high nighttime replenishment potential due to the lack of a strategic release window, resulting in a double imbalance of "excessive daytime evaporation plus wasted nighttime cooling." Once these systemic deviations accumulate in high-frequency scheduling cycles, they can not only cause mismatches in regional water supply plans but also disrupt the surrounding microclimate structure, triggering feedback anomalies such as water vapor disturbances and unstable aerosol inversion paths, thereby exacerbating the system's thermal disturbance effects. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a reservoir evaporation control method that integrates evaporation law analysis and scheduling scenario simulation, thus solving the problems mentioned in the background technology.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a reservoir evaporation control method integrating evaporation law analysis and scheduling scenario simulation, comprising the following steps:

[0007] S1. Deploy remote sensing microclimate collection points in the reservoir area and use sensor groups to collect evaporation environment data, and then transmit the evaporation environment data to the edge computing terminal.

[0008] S2. Perform time-series data normalization and preprocessing on the evaporation environment data in the edge computing terminal to obtain a dimensionless data set. Based on the dimensionless data set, establish the evaporation heat flux qevp and the condensation compensation heat flux qcnd.

[0009] S3. Calculate the diurnal water vapor energy difference factor ΔEdn based on the evaporation heat flux qevp and the condensation compensation heat flux qcnd. Then, conduct a preliminary comparative evaluation based on the diurnal water vapor energy difference factor ΔEdn to determine the condensation compensation potential.

[0010] S4. Based on the preliminary comparative evaluation results, the adjustment response mechanism is triggered. Based on the diurnal water vapor energy difference factor ΔEdn, the responsive scheduling window function Amod is output, and the water release capacity is adjusted based on the responsive scheduling window function Amod.

[0011] S5. The responsive scheduling window function Amod is correlated with the evaporative heat flux qevp to calculate the behavior energy consumption matching index. Based on the behavior energy consumption matching index Uctrl, a secondary comparison evaluation is performed, and the corresponding scheduling strategy is executed.

[0012] Preferably, S1 includes S11 and S12;

[0013] S11. Remote sensing microclimate collection points are set up in the reservoir area, including remote sensing points in the main dam area, remote sensing points in the intake area, and remote sensing points in the shallow bay area of ​​the tributaries; and a sensor group is configured at each remote sensing microclimate collection point, with the sampling frequency of the sensor group set to 5 minutes each time, to collect the evaporation environment data of each remote sensing microclimate collection point in real time.

[0014] The remote sensing points for the main dam area are set up in the main dam area of ​​the reservoir.

[0015] The remote sensing points for the water inlet area are located in the water inlet area of ​​the reservoir;

[0016] The remote sensing points for the tributary shallow bay area are set in the tributary shallow bay area of ​​the reservoir;

[0017] The sensor group includes a total radiation meter, a stratified temperature and humidity probe, an aerosol lidar, a water vapor density sensor, and a temperature and humidity sensor.

[0018] The evaporation environment data includes thermal radiation intensity parameter Rsol, nighttime convection gradient factor Xconv, upper-air aerosol condensation potential energy factor Rgsa, and water vapor disturbance frequency Vwv.

[0019] S12. The evaporation environment data collected by each remote sensing microclimate collection point is transmitted to the edge computing terminal via the LoRa communication protocol through the communication module deployed in the sensor group, and the evaporation environment data is synchronized through GNSS timestamp.

[0020] Preferably, S2 includes S21;

[0021] S21. Receive evaporation environment data in real time at the edge computing terminal, and perform time-series alignment of all parameters using a sliding window mechanism to obtain the structured data matrix Xsync = {x ij}n×m, where x ij This represents the evaporation environment data at the i-th time point and the j-th time point, where n represents the number of time points and m represents the number of types of evaporation environment parameters.

[0022] Preferably, S2 further includes S22;

[0023] S22. Preprocess the structured data matrix Xsync. The preprocessing involves using the Z-score standardization method to transform the evaporation environment data of each column in the structured data matrix Xsync into a standard dimensionless parameter with a standard deviation of 1 and a mean of 0, thereby obtaining a dimensionless data set.

[0024] Preferably, S2 further includes S23;

[0025] S23. Combine the solar radiation intensity parameter Rsol and the water vapor disturbance frequency parameter Vwv in the dimensionless dataset through the evaporation driving factor fitting model to construct the evaporation heat flux parameter qevp, which is used to characterize the evaporation latent heat release capacity.

[0026] The nighttime convection gradient factor Xconv, the upper-air aerosol condensation potential energy factor Rgsa, and the water vapor disturbance frequency parameter Vwv from the dimensionless dataset are combined and processed through a condensation nucleation potential modeling process to construct the condensation replenishment heat flux parameter qcnd, which is used to characterize the nighttime water vapor condensation behavior.

[0027] Preferably, S3 includes S31;

[0028] S31. Based on the time-span integration operation of the evaporation heat flux parameter qevp and the condensation heat flux parameter qcnd, the heat released per unit area during the evaporation process and the heat absorbed per unit area during the condensation process are integrated and compared within their respective time periods to obtain the diurnal water vapor energy difference factor ΔEdn. ​​The diurnal water vapor energy difference factor ΔEdn is calculated by subtracting the integral value of the condensation heat flux in the nighttime time period tn from the integral value of the evaporation heat flux in the daytime time period td.

[0029] Preferably, S3 further includes S32;

[0030] S32. A preset condensation compensation potential reference threshold Eth is used as the boundary based on the maximum recoverable energy under the optimal condensation capacity condition in historical monitoring data.

[0031] A preliminary comparative assessment was conducted by comparing the condensation compensation potential reference threshold Eth with the diurnal water vapor energy difference factor ΔEdn to determine the nighttime condensation replenishment potential and evaporation loss condensation compensation potential in the reservoir water system. The specific assessment content is as follows:

[0032] When the diurnal water vapor energy difference factor ΔEdn is greater than or equal to the condensation compensation potential reference threshold Eth, it indicates insufficient condensation and abnormal evaporation, and the regulation response mechanism is immediately triggered.

[0033] When the diurnal water vapor energy difference factor ΔEdn is less than the condensation compensation potential reference threshold Eth, it indicates that the diurnal energy is balanced and condensation is sufficient, and no intervention is needed.

[0034] Preferably, S4 includes S41;

[0035] S41. After preliminary comparative evaluation of the trigger regulation response mechanism, based on the numerical results of the diurnal water vapor energy difference factor ΔEdn and combined with the principle of thermal balance regulation, a type of scheduling window function mapping model is constructed. This type of scheduling window function mapping model uses the diurnal water vapor energy difference factor ΔEdn as the input variable and the regulation time window width as the output target to dynamically generate a responsive scheduling window function Amod. The responsive scheduling window function Amod is the regulation capacity function. The responsive scheduling window function Amod is passed as the control input parameter to the variable water release capacity regulation module for preliminary actual water level adjustment.

[0036] Preferably, S5 includes S51;

[0037] S51. After the initial actual water level adjustment, the variable water release capacity control module of the reservoir scheduling system extracts the energy consumption per unit area at time t and the actual water release flow at time t corresponding to the actual scheduling behavior, qactual(t). Within the set observation period T, the ratio of the evaporation heat flux parameter qevp(t) at time t to the energy consumption per unit area at time t corresponding to the actual scheduling behavior is calculated to form an instantaneous matching index. Combined with the weight of the ratio of the responsive scheduling window function Amod(t) at the current time point t to the actual water release flow at time t, a combined kernel function is formed. Then, the combined kernel function is processed by integral averaging within the time interval [0,T] to output the behavior energy consumption matching degree index Uctrl.

[0038] Preferably, S5 further includes S52;

[0039] S52. Based on the behavioral energy consumption matching index Uctrl, a secondary comparative evaluation is performed to determine the initial control status of the actual water level adjustment. Based on the results of the secondary comparative evaluation, the corresponding scheduling strategy is executed. The specific evaluation content is as follows:

[0040] When the behavior energy consumption matching index Uctrl < 0.85, it indicates that the control state is ideal, the behavior is reasonable, and the current scheduling strategy should be maintained.

[0041] When 0.85 ≤ behavior energy consumption matching degree index Uctrl < 1, it indicates that the control state is critically deviating and there is a slight deviation. At this time, the tightening strategy is executed, and the responsive scheduling window function Amod(t) at time t is iteratively regenerated until the control state is ideal and the iteration stops.

[0042] When the behavior energy consumption matching index Uctrl≥1, the expression control state is severely over-consumed and the strategy is out of control. At this time, forced compression is started, compressing the responsive scheduling window function Amod(t) at time t to no more than 35%, freezing 12:00-16:00, and opening window permissions.

[0043] This invention provides a reservoir evaporation control method that integrates evaporation law analysis and scheduling scenario simulation. It has the following beneficial effects:

[0044] (1) This method deploys a multi-zone remote sensing microclimate acquisition point system in the reservoir area, using remote sensing points in the main dam area, the intake area, and the tributary shallow bay area as representatives to achieve three-dimensional heterogeneous coverage of the microclimate. Combined with multiple types of sensor groups such as total radiation meters, aerosol lidar, and stratified temperature and humidity probes, it can achieve high-frequency, time-series sampling every 5 minutes and spatial resolution identification of local disturbance behavior of water evaporation driving factors. Compared with the intermittent measurement methods of existing technologies that rely on single-point water surface evaporation pans or meteorological stations, this scheme has the advantages of high throughput, low hysteresis, wide regional distribution, and diverse parameter configurations. It significantly improves the integrity of the environmental basis data for evaporation behavior prediction and provides high-resolution input conditions for subsequent heat flux modeling and feedback control, with dual technical advantages of accuracy and real-time performance.

[0045] (2) This method introduces an evaporation heat flux parameter qevp based on the solar radiation intensity parameter Rsol and the water vapor disturbance frequency parameter Vwv. Simultaneously, it combines the nighttime convection gradient factor Xconv and the upper-air aerosol condensation potential energy factor Rgsa to construct a condensation replenishment heat flux parameter qcnd. Through time-series integral analysis, a diurnal water vapor energy difference factor ΔEdn is generated. For the first time, this method incorporates the latent heat released by evaporation and the latent heat absorbed by condensation into a unified heat energy conservation calculation model, overcoming the problem of existing technologies that only consider energy dissipation from the evaporation side. This method establishes a novel modeling framework of "heat, humidity, and disturbance coupling drive + diurnal asymmetric mechanism analysis + disturbance suppression term deduction," which not only improves the scientific accuracy of water body energy loss estimation but also provides physically based decision mediator variables for dynamic scheduling, filling the modeling gap in current control models that lack quantification of condensation contribution.

[0046] (3) This method constructs a responsive scheduling window function Amod and a behavior energy consumption matching index Uctrl to realize a two-stage comparative evaluation mechanism for scheduling behavior: First, by using a scheduling window function mapping model, the diurnal water vapor energy difference factor ΔEdn is transformed into a dynamically controllable scheduling capacity input, forming a closed-loop response path of "prediction → water release capacity convergence"; Second, by using the behavior energy consumption matching index Uctrl to comprehensively characterize "scheduling behavior energy consumption matching" and "behavior strategy capacity matching", a self-iteration and saturation protection mechanism for fine-tuning strategies is realized. Compared with the rigid threshold water release scheduling mode in the existing technology, this method can automatically compress the high evaporation period window or freeze the release authority during key periods according to the dynamic changes in water vapor heat flux, thereby realizing the forward-looking containment of abnormal evaporation scenarios and the strategic release of condensation replenishment potential. It has strong dynamic adaptability and control precision advantages, significantly improving the reservoir's regulation stability and water-saving capacity under extreme meteorological conditions. Attached Figure Description

[0047] Figure 1 This is a schematic diagram illustrating the steps of a reservoir evaporation control method that integrates evaporation law analysis and scheduling scenario simulation according to the present invention.

[0048] Figure 2 Freezing control chart during high evaporation periods;

[0049] Figure 3 Comparison of daytime evaporative heat flux qevp and nighttime condensation replenishment heat flux qcnd integrals. Detailed Implementation

[0050] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0051] Example 1

[0052] This invention provides a reservoir evaporation control method that integrates evaporation law analysis and scheduling scenario simulation. Please refer to [link / reference]. Figure 1 This includes the following steps:

[0053] S1. Deploy remote sensing microclimate collection points in the reservoir area and use sensor groups to collect evaporation environment data, and then transmit the evaporation environment data to the edge computing terminal.

[0054] S2. Perform time-series data normalization and preprocessing on the evaporation environment data in the edge computing terminal to obtain a dimensionless data set. Based on the dimensionless data set, establish the evaporation heat flux qevp and the condensation compensation heat flux qcnd.

[0055] S3. Calculate the diurnal water vapor energy difference factor ΔEdn based on the evaporation heat flux qevp and the condensation compensation heat flux qcnd. Then, conduct a preliminary comparative evaluation based on the diurnal water vapor energy difference factor ΔEdn to determine the condensation compensation potential.

[0056] S4. Based on the preliminary comparative evaluation results, the adjustment response mechanism is triggered. Based on the diurnal water vapor energy difference factor ΔEdn, the responsive scheduling window function Amod is output, and the water release capacity is adjusted based on the responsive scheduling window function Amod.

[0057] S5. The responsive scheduling window function Amod is correlated with the evaporative heat flux qevp to calculate the behavior energy consumption matching index. Based on the behavior energy consumption matching index Uctrl, a secondary comparison evaluation is performed, and the corresponding scheduling strategy is executed.

[0058] In this embodiment, the method, through the multi-point remote sensing microclimate acquisition layout in step S1 and the collaborative work of the high-frequency sensor group, can spatially cover key hydrological heterogeneous areas such as the main dam, the intake, and tributary shallow bays, and temporally achieve minute-level evaporation environment data sampling; then, in step S2, dimensionless processing and parameter fusion modeling are completed through the edge computing terminal to construct the evaporation heat flux parameter qevp and the condensation replenishment heat flux parameter qcnd, which can dynamically respond to the micro-disturbance environment; subsequently, in step S3, the diurnal water vapor energy difference factor ΔEdn is generated by quantitatively assessing the diurnal energy difference, and the condensation potential and The dynamic balance of evaporation loss enables the identification of evaporation anomalies from the perspective of "energy conservation." In step S4, the responsive scheduling window function Amod is generated based on ΔEdn. ​​By controlling the compression or relaxation of scheduling capabilities with energy offset, a mapping bridge is constructed between daytime and nighttime energy offsets and actual water level regulation. Finally, in step S5, by constructing the behavior energy consumption matching index Uctrl, a dynamic matching assessment between water release behavior and evaporation demand is achieved. The scheduling strategy is iteratively optimized through the Uctrl secondary discrimination mechanism, establishing a complete control logic chain from identification and regulation to closed-loop rebalancing. This method not only breaks through the traditional "one-way passive suppression" approach in evaporation control, but also introduces nighttime condensation potential as an active scheduling factor for the first time. It also achieves fine-grained quantitative modeling of water energy flow in the heat flux space, effectively improving the early warning capability and regulation accuracy for extreme evaporation behavior. Ultimately, this method achieves three significant improvements: in prediction accuracy, the regional adaptability of evaporation assessment is significantly enhanced through a microclimate distributed sensing network and perturbation-sensitive modeling; in regulation capability, the responsive scheduling window function Amod dynamically compresses the scheduling window, enabling forward-looking constraints on abnormal energy consumption behavior; and in strategy closed-loop, a feedback mechanism driven by the behavior-energy consumption matching index Uctrl achieves precise alignment between scheduling behavior and energy demand. In summary, the technical solution presented in this application not only solves the core problems of existing reservoir evaporation regulation models, such as response lag, lack of quantitative modeling of condensation potential, and disconnect between scheduling behavior and energy consumption, but also achieves significant improvements in system adaptability, water-saving scheduling, and model physical interpretability, demonstrating significant technological advancement value and potential for widespread application.

[0059] Example 2

[0060] Please see Figure 1 Specifically: S1 includes S11 and S12;

[0061] S11. Remote sensing microclimate collection points are set up in the reservoir area, including remote sensing points in the main dam area, remote sensing points in the intake area, and remote sensing points in the shallow bay area of ​​the tributaries; and a sensor group is configured at each remote sensing microclimate collection point, with the sampling frequency of the sensor group set to 5 minutes each time, to collect the evaporation environment data of each remote sensing microclimate collection point in real time.

[0062] Remote sensing points in the main dam area are set up in the main dam area of ​​the reservoir, where the water body heat capacity response is the strongest, the wind field is stable, and the light reflection on the water surface is the most concentrated, reflecting the main trend of overall evaporation of the reservoir.

[0063] Remote sensing points in the intake area are set up in the reservoir's intake area to detect common water disturbances and disturbances to the atmosphere structure caused by water diversion paths, and to identify the impact of water diversion activities on local evaporation disturbances.

[0064] Remote sensing points in the tributary shallow bay area are set up in the tributary shallow bay area of ​​the reservoir. The water is shallow and has low thermal inertia, making it easy for condensation core zones to form, thus capturing microscale nighttime condensation events.

[0065] The sensor group includes a total radiation meter, a stratified temperature and humidity probe, an aerosol lidar, a water vapor density sensor, and a temperature and humidity sensor;

[0066] Evaporation environment data include thermal radiation intensity parameter Rsol, nighttime convection gradient factor Xconv, upper-air aerosol condensation potential energy factor Rgsa, and water vapor disturbance frequency Vwv;

[0067] The thermal radiation intensity parameter Rsol is calculated as the energy flux per unit area by measuring the temperature rise of the radiant heat on the thermistor element of the total radiation meter installed at the unobstructed main dam area remote sensing point, including direct and diffuse light.

[0068] The nighttime convection gradient factor Xconv is obtained by using stratified temperature and humidity probes at three heights (0.5m, 2m, and 4m above the water surface) to collect temperature and relative humidity in real time and calculate the gradient. The larger the gradient, the stronger the nighttime air convection and the easier it is to produce condensation. The superposition of convective heat transfer and humidity migration describes the possibility of cold air sinking and humidified air rising.

[0069] The upper-level aerosol condensation potential energy factor Rgsa was determined by using aerosol lidar to observe the concentrations of aerosols in the atmosphere (PM0.3-PM2.5) and their average particle size. Based on the equation meteorological physics, the concentrations of atmospheric aerosols (PM0.3-PM2.5) and average particle size are fitted to obtain the expression, which characterizes the ability of fine particulate matter in the atmosphere, such as salt particles and dust, to provide condensation nuclei at night and is a triggering factor for non-explicit condensation.

[0070] The water vapor disturbance frequency Vwv is obtained by continuously recording the water vapor content in the air using a water vapor density sensor. The number of second derivative changes is analyzed within a fixed time window, and a disturbance judgment threshold is set. The number of times the disturbance judgment threshold is exceeded is considered a sudden change in water vapor concentration, and frequency analysis is performed. Frequent disturbances can disrupt the balance of the water vapor layer and induce abnormal evaporation; they can also cause the condensation layer to be broken down.

[0071] S12. The evaporation environment data collected by each remote sensing microclimate collection point is transmitted to the edge computing terminal via the LoRa communication protocol through the communication module deployed in the sensor group, and the evaporation environment data is synchronized through GNSS timestamp to ensure the temporal consistency and regional resolution effectiveness of the evaporation environment data.

[0072] In this embodiment, the method achieves precise coverage of water body evaporation heterogeneity, disturbance, and condensation-sensitive areas by scientifically deploying remote sensing microclimate acquisition points in the main dam area, the intake area, and the shallow bay areas of tributaries in S11. The functional differences in the acquisition points in each area, such as capturing the overall evaporation trend of the reservoir in the main dam area, identifying the impact of water diversion disturbances at the intake, and detecting nighttime condensation formation conditions in the shallow bay areas of tributaries, ensure a comprehensive understanding of the spatial heterogeneity of evaporation behavior. Combined with a multi-source sensor group consisting of a total radiation meter, stratified temperature and humidity probes, aerosol lidar, and water vapor density sensor, the system can acquire key evaporation control factors such as thermal radiation intensity parameter Rsol, nighttime convection gradient factor Xconv, upper-level aerosol condensation potential energy factor Rgsa, and water vapor disturbance frequency Vwv at high frequency, realizing a quantitative characterization of the core driving mechanism of the evaporation process. In S12, by uploading data using the LoRa communication protocol through the communication modules configured at each acquisition point, and combining this with GNSS timestamps for cross-point data timing alignment, the time drift and spatial distortion problems under multi-point concurrent acquisition are effectively solved, ensuring the spatiotemporal consistency and processing reliability of evaporation environment data. This implementation scheme achieves the following: improved environmental perception coverage: by deploying a structured acquisition point network, taking into account the spatial diversity of reservoir evaporation behavior and microclimate interference characteristics; improved physical targeting of environmental parameter perception: by selectively using sensor parameters with physical interpretability, such as Rgsa and Xconv, multi-dimensional capture of condensation conditions and evaporation intensity is achieved; enhanced data transmission stability and timing consistency: through LoRa low-power wide-area network and GNSS synchronization mechanism, continuous, reliable, and low-latency transmission capabilities are ensured over a large area of ​​water. Overall, step S1 constructs a solid data foundation and environmental perception front-end in the entire reservoir evaporation control system, providing highly reliable and spatially resolved environmental information support for subsequent flux modeling, scheduling simulation, and feedback evaluation, greatly improving the responsiveness, accuracy, and predictability of reservoir fine-tuning.

[0073] Example 3

[0074] Please see Figure 1 Specifically: S2 includes S21;

[0075] S21. Receive evaporation environment data in real time at the edge computing terminal, and perform time-series alignment of all parameters using a sliding window mechanism to obtain the structured data matrix Xsync = {x ij}n×m, where x ij This represents the evaporation environment data at time point i, n represents the number of time points, i.e. the length of the time series obtained by continuous sampling. For example, if data is collected every 5 minutes, there will be 288 time points in 24 hours. m represents the number of types of evaporation environment parameters, i.e. the evaporation environment data collected at each time point, such as the four core parameters collected in this method.

[0076] S2 also includes S22;

[0077] S22. Preprocess the structured data matrix Xsync. The preprocessing method is to use the Z-score standardization method to transform the evaporation environment data of each column in the structured data matrix Xsync into a standard dimensionless parameter with a standard deviation of 1 and a mean of 0, thereby obtaining a dimensionless data set.

[0078] S2 also includes S23;

[0079] S23. Combine the solar radiation intensity parameter Rsol and the water vapor disturbance frequency parameter Vwv in the dimensionless dataset through the evaporation driving factor fitting model to construct the evaporation heat flux parameter qevp to characterize the evaporation latent heat release capacity. The evaporation driving factor fitting model is based on the variable simplification of the classic Penman evaporation model. The total solar radiation intensity parameter Rsol is used to replace the net radiation index to optimize the physical interpretation under clear sky conditions. At the same time, the water vapor disturbance frequency parameter Vwv is used to replace the traditional wind speed function to optimize the response sensitivity to evaporation behavior in the disturbed environment.

[0080] The net radiation factor in the traditional model is replaced by the solar radiation intensity parameter Rsol. The reason is that in cloudless or partly cloudy weather, the actual net radiation of the earth's surface is mainly determined by solar radiation. Therefore, the two are highly correlated, and the replacement simplifies the requirements for measurement equipment.

[0081] The wind speed function is replaced by the water vapor disturbance frequency parameter Vwv to reflect the reinforcing effect of frequent disturbances on evaporation behavior in areas such as plateaus, wind gaps, or canyons. The disturbance frequency, as a micro-disturbance excitation index, has a higher physical interpretation capability.

[0082] The nighttime convection gradient factor Xconv, the upper-level aerosol condensation potential energy factor Rgsa, and the water vapor disturbance frequency parameter Vwv from the dimensionless dataset are combined using a condensation nucleation potential modeling process to construct the condensation return heat flux parameter qcnd, which characterizes nighttime water vapor condensation behavior. The condensation nucleation potential modeling process is based on... Based on the theory and principles of environmental disturbance, this paper considers the condensation nucleation capacity and atmospheric vertical stability as positive contributing factors, and introduces disturbance frequency as a nucleation inhibition term to form a comprehensive modeling method for nighttime latent heat release behavior. By considering the temperature and humidity gradients in the vertical direction at night, the stability of the condensation environment is reflected. At the same time, the potential of fine particles and aerosols in the atmosphere as condensation nuclei at night is combined as triggering factors for the possibility of condensation. Furthermore, the adverse effects of water vapor disturbance on nucleation stability are eliminated, thus constructing a condensation heat flux model that can simultaneously reflect nucleation conditions, condensation driving forces, and disturbance disturbances.

[0083] Throughout the modeling process, all parameters are normalized via edge computing terminals before use to ensure their dimensionlessness, thus ensuring comparability and stability in the modeling process. When each physical quantity is substituted into the model, the unit system remains consistent. The evaporation rate and condensation rate are based on the liquid surface loss / replenishment thickness per unit length per unit time. The flux output is uniformly represented as the energy value per unit area per unit time, ensuring the consistency and interpretability of the model output.

[0084] In this embodiment, the method introduces a sliding window mechanism in the edge computing terminal to perform temporal alignment of the original evaporation environment data, constructing a structured data matrix Xsync. This ensures that each parameter has a consistent temporal semantic relationship at different time points, improving the temporal accuracy of data processing and the robustness of model input. S22 further uses the Z-score normalization method to perform dimensionless processing on each column of parameter data, eliminating parameter scale differences and constructing a dimensionless data set under a unified dimensional system, laying a mathematical foundation for subsequent multi-source parameter fusion modeling. In S23, the solar radiation intensity parameter Rsol and the water vapor disturbance frequency parameter Vwv from the dimensionless data set are innovatively introduced into the classic Penman evaporation model framework to construct an evaporation driving factor fitting model for calculating the evaporation heat flux parameter qevp. This effectively simplifies the dependence on factors that are difficult to measure precisely, such as net radiation and wind speed, improving the physical adaptability of evaporation behavior modeling under clear skies and disturbed environments. Simultaneously, it employs... Based on theory and disturbance principles, a condensation nucleus potential energy model was established by combining the nighttime convection gradient factor Xconv, the upper-air aerosol condensation potential energy factor Rgsa, and the water vapor disturbance frequency parameter Vwv. This model was used to calculate the condensation recovery heat flux parameter qcnd, comprehensively characterizing the nighttime condensation heat release behavior from three perspectives: nucleation, driving force, and disturbance. Both flux models were normalized to ensure that the physical quantity output is the energy value per unit area per unit time (J / m²). 2 The S2 step ( / s) enhances the model's versatility and interpretability across time and location. It achieves high-precision temporal alignment and standardized normalization of multi-source evaporation environment data, ensuring data modeling consistency and dimensional uniformity. A heat flux modeling mechanism with physical foundations and engineering simplification is constructed, significantly improving the availability and prediction accuracy of the two key energy factors, qevp and qcnd. It overcomes the limitations of traditional models' dependence on meteorological stations, enhancing sensitivity and discriminative ability to local microclimate disturbances.

[0085] Example 4

[0086] Please see Figure 1 and Figure 3 Specifically: S3 includes S31;

[0087] S31. The time-span integration operation based on the evaporation heat flux parameter qevp and the condensation heat flux parameter qcnd is constructed based on the basic principles of energy conservation and heat flux accumulation over time in physics. The heat released per unit area during the evaporation process and the heat absorbed per unit area during the condensation process are integrated and compared in their respective time periods to obtain the diurnal water vapor energy difference factor ΔEdn. ​​The diurnal water vapor energy difference factor ΔEdn is calculated by subtracting the integral value of the condensation heat flux in the nighttime time period tn from the integral value of the evaporation heat flux in the daytime time period td.

[0088] The specific calculation formula is: △Edn=(∫tdqevp(t)dt)-(∫tnqcnd(t)dt), where td represents the daytime period, tn represents the nighttime period, qebp(t) represents the evaporation heat flux parameter at time t, qcnd(t) represents the condensation compensation heat flux parameter at time t, and dt represents the time integral function;

[0089] (∫tdqevp(t)dt) represents the cumulative value of the evaporative heat flux per unit area qevp during the daytime period. Its physical meaning is that it represents the total latent heat absorbed by water bodies during the daytime evaporation process, which converts liquid water into water vapor.

[0090] (∫tnqcnd(t)dt) represents the cumulative value of the condensation return heat flux parameter qcnd per unit area during the nighttime phase. Its physical meaning is that it represents the total latent heat released during the nighttime environment when water vapor is condensed into droplets.

[0091] Specifically, the units of the evaporation heat flux parameter qevp and the condensation compensation heat flux parameter qcnd are both joules per square meter per second (J / m²). 2 / s), after time integration, is converted into the cumulative energy per unit area during the day and night (J / m²). 2 To ensure dimensional consistency during calculations, the upper and lower limits of integration are determined based on the local "daytime" and "nighttime" durations as defined in the environmental monitoring system, typically using local sunrise and sunset times for separation.

[0092] In this embodiment, the evaporation heat flux parameter qevp and the condensation compensation heat flux parameter qcnd can be numerically integrated using the sliding time window method. By using heat flux data with a resolution of five minutes or ten minutes, a fixed-width rectangular integral is performed on the daytime and nighttime time periods to obtain the cumulative value of evaporation energy and the cumulative value of condensation energy, and then the value of the diurnal water vapor energy difference factor ΔEdn is calculated.

[0093] S3 also includes S32;

[0094] S32. Preset condensation compensation potential reference threshold Eth. The condensation compensation potential reference threshold Eth is based on the maximum recoverable energy under the optimal condensation capacity condition in historical monitoring data as the boundary. It is used to evaluate whether the replenishment potential of nighttime condensation in the current water system can meet the evaporation loss.

[0095] A preliminary comparative assessment was conducted by comparing the condensation compensation potential reference threshold Eth with the diurnal water vapor energy difference factor ΔEdn to determine the nighttime condensation replenishment potential and evaporation loss condensation compensation potential in the reservoir water system. The specific assessment content is as follows:

[0096] When the diurnal water vapor energy difference factor ΔEdn is greater than or equal to the condensation compensation potential reference threshold Eth, it indicates insufficient condensation and abnormal evaporation, and the regulation response mechanism is immediately triggered.

[0097] When the diurnal water vapor energy difference factor ΔEdn is less than the condensation compensation potential reference threshold Eth, it indicates that the diurnal energy is balanced and condensation is sufficient, and no intervention is needed.

[0098] In this embodiment, in S31, the method employs the principles of heat flux conservation and time integration from physics to integrate the daytime evaporation heat flux parameter qevp and the nighttime condensation replenishment heat flux parameter qcnd over time periods, calculating the difference in total latent heat release per unit area between day and night, and generating a day-night water vapor energy difference factor ΔEdn. ​​This factor not only reflects the balance between evaporation and condensation dynamics under current climate and hydrological conditions but also provides a clear basis for quantitative decision-making in subsequent scheduling and regulation. In particular, the use of sliding time window integration and high-frequency data input, such as a 5-minute resolution, gives the day-night water vapor energy difference factor ΔEdn good timeliness and trend capture capabilities, significantly improving its sensitivity to abnormal evaporation or insufficient condensation. In S32, a dynamic heuristic evaluation system is constructed by constructing a condensation compensation potential reference threshold Eth and comparing it in real time with the day-night water vapor energy difference factor ΔEdn. ​​This reference threshold Eth is derived from the energy recovery capacity under optimal condensation conditions in historical monitoring data, possessing practical rationality in an engineering context. Compared to traditional evaporation models that only provide static calculation results, this scheme can make adjustment decisions based on whether the diurnal water vapor energy difference factor ΔEdn exceeds the condensation compensation potential reference threshold Eth, achieving real-time judgment on whether "condensation is sufficient to compensate for evaporation". When the diurnal water vapor energy difference factor ΔEdn exceeds the threshold, the system can immediately trigger the adjustment response mechanism; while under energy balance conditions, excessive intervention is avoided, reflecting the ideas of differentiated control and energy consumption optimization. Therefore, the purpose and beneficial effects achieved by this implementation method are mainly reflected in the following aspects: based on the principle of energy conservation, a two-way quantitative model for the latent heat behavior of evaporation and condensation is constructed, forming a complete energy budget framework; the dynamic accuracy and real-time performance of the diurnal water vapor energy difference factor ΔEdn calculation are significantly improved through the sliding integral method, enabling the evaluation mechanism to have immediate judgment capabilities; and the introduction of the condensation compensation potential reference threshold Eth as a dynamic trigger threshold is used to construct an active judgment mechanism, avoiding misjudgments in traditional passive scheduling.

[0099] Example 5

[0100] Please see Figure 1 Specifically: S4 includes S41;

[0101] S41. After preliminary comparative evaluation of the trigger regulation response mechanism, based on the numerical results of the diurnal water vapor energy difference factor ΔEdn and combined with the principle of heat balance regulation, a type of scheduling window function mapping model is constructed. This type of scheduling window function mapping model uses the diurnal water vapor energy difference factor ΔEdn as the input variable and the regulation time window width as the output target to dynamically generate a responsive scheduling window function Amod. The responsive scheduling window function Amod is a regulation capacity function that reflects the degree to which the water release window should be expanded. The responsive scheduling window function Amod is then passed as a control input parameter to the variable water release capacity regulation module for preliminary actual water level adjustment.

[0102] The responsive scheduling window function Amod calculates its output through the following type of scheduling window function mapping model;

[0103]

[0104] In the formula, Amod(t) represents the responsive scheduling window function at time t, Qref(t) represents the daily reference water supply at time t, Eref represents the upper limit of historical condensation replenishment energy, used to normalize the day-night water vapor energy difference factor ΔEdn, and α represents the response sensitivity coefficient, which controls the compression intensity and is set by the user according to experience to be between 0.5 and 1.0.

[0105] The design principle of the responsive scheduling window function Amod draws on environmental feedback scheduling theory and proportional weakening feedback mechanism in control engineering. It performs dynamic compression based on the default reference flow. The compression factor is dynamically adjusted according to the ratio of the diurnal water vapor energy difference factor ΔEdn to the reference energy value. The larger the diurnal water vapor energy difference factor ΔEdn is, the stronger the compression. A first-order linear compression function is used, which is convenient for algorithm implementation and has a simple physical interpretation. This algorithm belongs to the nonlinear scaling first-order weakening model and is widely used in energy balance control and flow control strategies.

[0106] One type of scheduling window function mapping model originates from the Logistic suppression nonlinear mapping model in mathematical physics and is widely used in control modeling of ecological, hydrological, and energy systems with characteristics of "lag, response, and saturation".

[0107] △Edn

[0108] Since Eref is dimensionless, the entire bracket is also dimensionless. After multiplying it by the daily reference water supply Qref(t) at time t, the output responsive scheduling window function Amod(t) at time t still maintains m. 3 / s, so the units on both sides of the formula are consistent, which is logically and physically reasonable and has a rigorous dimension.

[0109] In this embodiment, the method, through step S4 and sub-step S41, constructs and outputs a responsive scheduling window function Amod based on the real-time analysis results of the diurnal water vapor energy difference factor ΔEdn, achieving precise linkage between energy difference perception and scheduling behavior control. Specifically, this implementation method, based on the principle of thermal balance regulation, uses the diurnal water vapor energy difference factor ΔEdn as the adjustment input variable, introduces a type of scheduling window function mapping model, and combines it with the historical condensation replenishment energy upper limit value Eref to complete the normalization process, dynamically outputting a window adjustment capability that matches the degree of coupling with the diurnal water vapor energy difference factor ΔEdn. ​​In the implementation of this method, the responsive scheduling window function Amod is not only constructed as a water release capacity control function, but also controls its compression intensity by setting the response sensitivity coefficient α to an empirical range of 0.5 to 1.0, thereby achieving a sensitive response to changes in the diurnal water vapor energy difference factor ΔEdn. In terms of functional form, this model borrows from the Logistic suppression nonlinear mapping mechanism, which is widely used in behavioral modeling with "hysteresis, response, and saturation" characteristics in fields such as ecosystems and hydrological energy control, and has good physical interpretability and dynamic regulation performance. Through this function, the reservoir scheduling system can automatically determine the intensity of the current energy imbalance and dynamically compress the reference water supply Qref(t) to generate the responsive scheduling window function Amod(t) at time t, thereby realizing real-time water level regulation behavior. Therefore, this method establishes a functional bridge from energy difference determination to water body scheduling capacity compression, bridging the technical gap between previous evaporation compensation assessment and water release scheduling control. By employing a first-order linear compression algorithm incorporating the proportional weakening feedback principle, the control response process is both simple and feasible, and possesses good engineering adaptability. While maintaining the original physical dimensions of Qref(t), it achieves dimensional consistency and energy consistency in the output of the responsive scheduling window function Amod(t) at time t, ensuring the physical rigor of the model. Through dynamic mapping response to the diurnal water vapor energy difference factor ΔEdn, a synchronization mechanism between evaporation anomalies and control capacity compression is constructed, significantly improving the water resource system's adaptability to extreme climates and short-term fluctuations. Overall, this step effectively improves the dynamic accuracy and energy efficiency of reservoir scheduling, providing a central control module support for the entire evaporation control strategy system.

[0110] Example 6

[0111] Please see Figure 1 and Figure 2 Specifically: S5 includes S51;

[0112] S51. After the initial actual water level adjustment, the variable water release capacity control module of the reservoir scheduling system extracts the energy consumption per unit area at time t and the actual water release flow at time t corresponding to the actual scheduling behavior. Within the set observation period T, the ratio of the evaporation heat flux parameter qevp(t) at time t to the energy consumption per unit area at time t corresponding to the actual scheduling behavior is calculated to form an instantaneous matching index. Combined with the ratio weight of the responsive scheduling window function Amod(t) at the current time point t and the actual water release flow at time t, a combined kernel function is formed. Then, the combined kernel function is integrated and averaged over the time interval [0,T] to output the behavior energy consumption matching degree index Uctrl, which is used to characterize the dynamic fit between scheduling behavior and energy consumption demand.

[0113] The behavior energy consumption matching index Uctrl is calculated and output using the following algorithm formula;

[0114]

[0115] In the formula, dt represents the time integral function;

[0116] This represents an instantaneous matching indicator, used to analyze whether behavior is excessive or lacking in energy release;

[0117] This represents the ratio of the compressed scheduling window capacity to the actual behavior, used to analyze whether scheduling closely follows the policy capacity.

[0118] Since both indicators are fractions, both sides are dimensionless parameters. The first ratio reflects behavior-energy consumption matching; the second ratio reflects behavior-strategy capacity matching; their product forms a composite matching factor.

[0119] For a specific process example, let qactual(t) = 60.27 J / m 2 / s; Qactual(t) = 40m 3 / s, Amod(t)=48m 3 / s, qevp(t)=85J / m 2 ;

[0120] Single-point integral function value The average integral over all points within one hour yields a behavior energy consumption matching index Uctrl = 0.59.

[0121] S5 also includes S52;

[0122] S52. Based on the behavioral energy consumption matching index Uctrl, a secondary comparative evaluation is performed to determine the initial control status of the actual water level adjustment. Based on the results of the secondary comparative evaluation, the corresponding scheduling strategy is executed. The specific evaluation content is as follows:

[0123] When the behavior energy consumption matching index Uctrl < 0.85, it indicates that the control state is ideal, the behavior is reasonable, and the current scheduling strategy should be maintained.

[0124] When 0.85 ≤ behavior energy consumption matching degree index Uctrl < 1, it indicates that the control state is critically deviating and there is a slight deviation. At this time, the tightening strategy is executed, and the responsive scheduling window function Amod(t) at time t is iteratively regenerated until the control state is ideal and the iteration stops.

[0125] The tightening strategy reduces the upper limit of the release function of the responsive scheduling window function Amod(t) at time t by tightening the response sensitivity coefficient α by 10%.

[0126] The response sensitivity coefficient α determines the response strength of the responsive scheduling window function Amod to the change of the diurnal water vapor energy difference factor ΔEdn, which is reflected as the scaling factor between the diurnal water vapor energy difference factor ΔEdn and the responsive scheduling window function Amod.

[0127] When the response sensitivity coefficient α is large, even slight changes in the diurnal water vapor energy difference factor ΔEdn will be amplified into a large change in the responsive scheduling window function Amod, resulting in excessive adjustment of the system's water release range.

[0128] When the behavior energy consumption matching index Uctrl enters the 0.85–1 range, it means that the system is “approaching overconsumption” or “behavior is following too fast” energy consumption signal, and there has been an excessive energy release or lagging regulation.

[0129] If the response sensitivity coefficient α is not adjusted at this point, the response will still be amplified in the next strategy iteration, which may lead to further deviation and cause "strategy positive feedback to go out of control".

[0130] When the behavior energy consumption matching degree index Uctrl≥1, the expression control state is severely over-consumed and the strategy is out of control. At this time, forced compression is started, compressing the responsive scheduling window function Amod(t) at time t to no more than 35%, and freezing 12:00-16:00, i.e. the high evaporation period, and opening the window.

[0131] In this embodiment, step S5 of the method constructs a closed-loop matching evaluation and dynamic correction mechanism between scheduling behavior and energy consumption demand through sub-steps S51 and S52, realizing full-process linkage control from energy efficiency identification of water release behavior to adaptive optimization of strategy. Specifically, S51 introduces the "behavior energy consumption matching degree index Uctrl" to comprehensively analyze the energy adaptability between the actual energy consumption per unit area (qactual) and the evaporation heat flux (qevp), and simultaneously combines the strategy fit between the responsive scheduling window function (Amod) and the actual water release flow rate (Qactual). A composite matching factor is formed through the two dimensionless ratios, and then the integral average is processed within a set time interval to accurately quantify whether the current scheduling behavior is consistent with the energy demand and strategy objectives of the evaporation environment. Based on this, S52 constructs a targeted secondary comparison evaluation decision logic by performing hierarchical judgment on the value of the behavior energy consumption matching degree index Uctrl. When the value is less than 0.85, it indicates that the scheduling behavior is well controlled and the energy consumption is well matched, and the current strategy is maintained. When it is between 0.85 and 1, it indicates that the control is approaching the critical point, and the system automatically tightens the response sensitivity coefficient α to iteratively compress the scheduling window function Amod to improve the matching degree. When it is greater than or equal to 1, it is considered that the strategy is significantly out of control, and the system will implement a dual intervention of "forced compression + time freeze" to close the window during the high evaporation period, immediately curb excessive behavior, and prevent energy waste and ineffective loss of water resources. This implementation introduces a comprehensive evaluation index, Uctrl, that integrates behavior, energy consumption, and strategy. It establishes a quantitative basis for the behavioral energy efficiency of the scheduling system, addressing the lack of perception of actual energy consumption in traditional water resource scheduling. A differentiated adaptive adjustment mechanism of disturbance-free iteration and strategy freezing is constructed, enabling the system to respond to both minor deviations and severe loss of control, ensuring both flexibility and rigidity in regulation. The behavioral energy consumption matching index Uctrl is highly coupled with actual behavioral parameters, reflecting whether behavior closely follows the strategy capacity boundary, enhancing the system's dynamic identification and rapid correction capabilities for strategy execution deviations. By freezing open window permissions during high evaporation periods, energy waste and evaporation loss during critical periods are effectively suppressed, strengthening the control stability against extreme weather or abnormal water levels.

[0132] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A reservoir evaporation control method integrating evaporation law analysis and scheduling scenario simulation, characterized in that: Includes the following steps: S1. Deploy remote sensing microclimate collection points in the reservoir area and use sensor groups to collect evaporation environment data, and then transmit the evaporation environment data to the edge computing terminal. S2. Perform time-series data normalization and preprocessing on the evaporation environment data in the edge computing terminal to obtain a dimensionless data set. Based on the dimensionless data set, establish the evaporation heat flux qevp and the condensation compensation heat flux qcnd. S2 also includes S23; S23. Combine the solar radiation intensity parameter Rsol and the water vapor disturbance frequency parameter Vwv in the dimensionless dataset and calculate them using the evaporation driving factor fitting model to construct the evaporation heat flux parameter qevp. The nighttime convection gradient factor Xconv, the upper-air aerosol condensation potential energy factor Rgsa, and the water vapor disturbance frequency parameter Vwv in the dimensionless dataset are combined and processed through the condensation nucleation potential energy modeling process to construct the condensation replenishment heat flux parameter qcnd. S3. Calculate the diurnal water vapor energy difference factor ΔEdn based on the evaporation heat flux qevp and the condensation compensation heat flux qcnd. Then, conduct a preliminary comparative evaluation based on the diurnal water vapor energy difference factor ΔEdn to determine the condensation compensation potential. S4. Based on the preliminary comparative evaluation results, trigger the adjustment response mechanism, output the responsive scheduling window function Amod based on the diurnal water vapor energy difference factor ΔEdn, and adjust the water release capacity based on the responsive scheduling window function Amod. S5. The responsive scheduling window function Amod is correlated with the evaporative heat flux qevp to calculate the behavior energy consumption matching degree index. Based on the behavior energy consumption matching degree index Uctrl, a secondary comparison evaluation is performed, and the corresponding scheduling strategy is executed.

2. The reservoir evaporation control method according to claim 1, which integrates evaporation law analysis and scheduling scenario simulation, is characterized in that: S1 includes S11 and S12; S11. Remote sensing microclimate collection points are set up in the reservoir area, including remote sensing points in the main dam area, remote sensing points in the intake area, and remote sensing points in the shallow bay area of ​​the tributaries; and a sensor group is configured at each remote sensing microclimate collection point, with the sampling frequency of the sensor group set to 5 minutes each time, to collect the evaporation environment data of each remote sensing microclimate collection point in real time. The remote sensing points for the main dam area are set up in the main dam area of ​​the reservoir. The remote sensing points for the water inlet area are located in the water inlet area of ​​the reservoir; The remote sensing points for the tributary shallow bay area are set in the tributary shallow bay area of ​​the reservoir; The sensor group includes a total radiation meter, a stratified temperature and humidity probe, an aerosol lidar, a water vapor density sensor, and a temperature and humidity sensor. The evaporation environment data includes thermal radiation intensity parameter Rsol, nighttime convection gradient factor Xconv, upper air aerosol condensation potential energy factor Rgsa, and water vapor disturbance frequency Vwv. S12. The evaporation environment data collected by each remote sensing microclimate collection point is transmitted to the edge computing terminal via the LoRa communication protocol through the communication module deployed in the sensor group, and the evaporation environment data is synchronized through GNSS timestamp.

3. The reservoir evaporation control method according to claim 1, which integrates evaporation law analysis and scheduling scenario simulation, is characterized in that: S2 includes S21; S21. Receive evaporation environment data in real time at the edge computing terminal, and perform time-series alignment of all parameters using a sliding window mechanism to obtain the structured data matrix Xsync={x ij }n×m, where x ij This represents the evaporation environment data at the i-th time point and the j-th time point, where n represents the number of time points and m represents the number of types of evaporation environment parameters.

4. The reservoir evaporation control method according to claim 3, which integrates evaporation law analysis and scheduling scenario simulation, is characterized in that: S2 further includes S22; S22. Preprocess the structured data matrix Xsync. The preprocessing involves using the Z-score standardization method to transform the evaporation environment data of each column in the structured data matrix Xsync into a standard dimensionless parameter with a standard deviation of 1 and a mean of 0, thereby obtaining a dimensionless data set.

5. The reservoir evaporation control method according to claim 1, which integrates evaporation law analysis and scheduling scenario simulation, is characterized in that: S3 includes S31; S31. Based on the time-span integration operation of the evaporation heat flux parameter qevp and the condensation heat flux parameter qcnd, the heat released per unit area during the evaporation process and the heat absorbed per unit area during the condensation process are integrated and compared within their respective time periods to obtain the diurnal water vapor energy difference factor ΔEdn. ​​The diurnal water vapor energy difference factor ΔEdn is calculated by subtracting the integral value of the condensation heat flux in the nighttime time period tn from the integral value of the evaporation heat flux in the daytime time period td.

6. The reservoir evaporation control method according to claim 5, which integrates evaporation law analysis and scheduling scenario simulation, is characterized in that: S3 further includes S32; S32. A preset condensation compensation potential reference threshold Eth is used as the boundary based on the maximum recoverable energy under the optimal condensation capacity condition in historical monitoring data. A preliminary comparative assessment was conducted by comparing the condensation compensation potential reference threshold Eth with the diurnal water vapor energy difference factor ΔEdn to determine the nighttime condensation replenishment potential and evaporation loss condensation compensation potential in the reservoir water system. The specific assessment content is as follows: When the diurnal water vapor energy difference factor ΔEdn is greater than or equal to the condensation compensation potential reference threshold Eth, it indicates insufficient condensation and abnormal evaporation, and the regulation response mechanism is immediately triggered. When the diurnal water vapor energy difference factor ΔEdn is less than the condensation compensation potential reference threshold Eth, it indicates that the diurnal energy is balanced and condensation is sufficient, and no intervention is needed.

7. The reservoir evaporation control method according to claim 6, which integrates evaporation law analysis and scheduling scenario simulation, is characterized in that: S4 includes S41; S41. After preliminary comparative evaluation of the trigger regulation response mechanism, based on the numerical results of the diurnal water vapor energy difference factor ΔEdn and combined with the principle of thermal balance regulation, a type of scheduling window function mapping model is constructed. This type of scheduling window function mapping model uses the diurnal water vapor energy difference factor ΔEdn as the input variable and the regulation time window width as the output target to dynamically generate a responsive scheduling window function Amod. The responsive scheduling window function Amod is the regulation capacity function. The responsive scheduling window function Amod is passed as the control input parameter to the variable water release capacity regulation module for preliminary actual water level adjustment.

8. The reservoir evaporation control method according to claim 7, which integrates evaporation law analysis and scheduling scenario simulation, is characterized in that: S5 includes S51; S51. After the initial actual water level adjustment, the variable water release capacity control module of the reservoir scheduling system extracts the energy consumption per unit area at time t and the actual water release flow rate at time t corresponding to the actual scheduling behavior. Within the set observation period T, the ratio of the evaporation heat flux parameter qevp(t) at time t to the energy consumption per unit area at time t corresponding to the actual scheduling behavior is calculated to form an instantaneous matching index. Combined with the weight of the ratio of the responsive scheduling window function Amod(t) at the current time point t to the actual water release flow rate at time t, a combined kernel function is formed. Then, the combined kernel function is processed by integral averaging within the time interval [0,T] to output the behavior energy consumption matching degree index Uctrl.

9. The reservoir evaporation control method according to claim 8, which integrates evaporation law analysis and scheduling scenario simulation, is characterized in that: S5 also includes S52; S52. Based on the behavioral energy consumption matching index Uctrl, a secondary comparative evaluation is performed to determine the initial control status of the actual water level adjustment. Based on the results of the secondary comparative evaluation, the corresponding scheduling strategy is executed. The specific evaluation content is as follows: When the behavior energy consumption matching index Uctrl < 0.85, it indicates that the control state is ideal, the behavior is reasonable, and the current scheduling strategy should be maintained. When 0.85 ≤ behavior energy consumption matching degree index Uctrl < 1, it indicates that the control state is critically deviating and there is a slight deviation. At this time, the tightening strategy is executed, and the responsive scheduling window function Amod(t) at time t is iteratively regenerated until the control state is ideal and the iteration stops. When the behavior energy consumption matching degree index Uctrl≥1, the expression control state is severely over-consumed and the strategy is out of control. At this time, forced compression is started, compressing the responsive scheduling window function Amod(t) at time t to no more than 35%, freezing 12:00-16:00, and opening window permissions.

Images