Method and system for dynamic automatic monitoring of reservoir water temperature based on multi-point cooperation

Abstract

This invention discloses a method and system for dynamic automatic monitoring of reservoir water temperature based on multi-point collaboration, belonging to the field of reservoir water temperature monitoring technology. The method includes: generating a water release scheduling scheme based on the water release scheduling needs of the target reservoir and obtaining a water temperature monitoring dataset of the target reservoir; performing sensor reliability correction on the water temperature monitoring dataset to obtain a reliable water temperature dataset; obtaining a water temperature stratification model; performing multi-dimensional correlation accident simulation and adjustment to construct a water release regulation space; performing collaborative optimization under multi-modal risk prediction to obtain a water release regulation optimization strategy; and dynamically adjusting the water release regulation optimization strategy in a closed loop. This invention solves the technical problems of inaccurate risk prediction and lack of dynamic adaptation and adjustment mechanisms in existing technologies, leading to insufficient scheduling safety and adaptability. It achieves the technical effect of realizing accurate risk control and dynamic adaptation of reservoir water release scheduling, improving scheduling safety and adaptability.

Description

Technical Field

[0001] This invention relates to the field of reservoir water temperature monitoring technology, specifically to a method and system for dynamic automatic monitoring of reservoir water temperature based on multi-point collaboration. Background Technology

[0002] In the field of reservoir water release scheduling, water temperature changes directly affect the downstream aquatic ecological balance, the safety of industrial and agricultural water use, and the operational stability of water conservancy projects. Accurate monitoring of water temperature distribution and dynamic regulation of water release strategies are core industry requirements. Existing technologies mostly rely on data from single monitoring points or simple hierarchical models, which suffer from problems such as incomplete water temperature monitoring coverage, insufficient data reliability, and low accuracy of hierarchical simulation. Furthermore, they are difficult to comprehensively predict and coordinate the management of multimodal risks such as cold water discharge, warm water impact, and ecological damage. They also lack a closed-loop adjustment mechanism that dynamically adapts to real-time changes in reservoir water temperature, resulting in insufficient safety and ecological adaptability of water release scheduling schemes, and failing to fully meet the refined scheduling needs in complex reservoir environments.

[0003] Existing technologies suffer from inaccurate risk prediction for reservoir water release scheduling and a lack of dynamic adaptation and adjustment mechanisms, resulting in insufficient scheduling safety and adaptability. Summary of the Invention

[0004] This application provides a method and system for dynamic automatic monitoring of reservoir water temperature based on multi-point collaboration, which is used to address the technical problems in the prior art where the prediction of reservoir water release scheduling risks is inaccurate and the lack of dynamic adaptation and adjustment mechanisms leads to insufficient scheduling safety and adaptability.

[0005] In view of the above problems, this application provides a method and system for dynamic automatic monitoring of reservoir water temperature based on multi-point collaboration.

[0006] The first aspect of this application provides a method for dynamic automatic monitoring of reservoir water temperature based on multi-point collaboration, the method comprising: A water release scheduling plan is generated based on the water release scheduling requirements of the target reservoir, and a water temperature monitoring dataset of the target reservoir is obtained based on a multi-point collaborative monitoring network. The water temperature monitoring dataset is then subjected to sensor reliability calibration based on the multi-point collaborative monitoring network to obtain a reliable water temperature dataset. A virtual hierarchical reconstruction based on the reliable water temperature dataset under near-deep water temperature change analysis is performed to obtain a water temperature hierarchical model. Based on the water temperature hierarchical model, the water release scheduling plan is adjusted through multi-dimensional correlation accident simulation to construct a water release regulation space. Based on the water temperature hierarchical model, collaborative optimization under multi-modal risk prediction is performed on the water release regulation space to obtain a water release regulation optimization strategy. Finally, the water release regulation optimization strategy is dynamically adjusted in a closed loop based on the multi-point collaborative monitoring network.

[0007] A second aspect of this application provides a multi-point collaborative automatic monitoring system for reservoir water temperature dynamics, the system comprising: The system includes the following modules: a water temperature monitoring dataset acquisition module, used to generate a water release scheduling plan based on the water release scheduling requirements of the target reservoir, and acquire the water temperature monitoring dataset of the target reservoir based on a multi-point collaborative monitoring network; a reliable water temperature dataset acquisition module, used to perform sensor reliability correction on the water temperature monitoring dataset based on the multi-point collaborative monitoring network to acquire a reliable water temperature dataset; a water temperature stratification model acquisition module, used to perform virtual stratification reconstruction on the reliable water temperature dataset under the analysis of near-deep water temperature changes to acquire a water temperature stratification model; a water release regulation space construction module, used to perform multi-dimensional correlation accident inference and regulation on the water release scheduling plan based on the water temperature stratification model to construct a water release regulation space; a water release regulation optimization strategy acquisition module, used to perform collaborative optimization under multi-modal risk prediction on the water release regulation space based on the water temperature stratification model to acquire a water release regulation optimization strategy; and a dynamic closed-loop adjustment module, used to perform dynamic closed-loop adjustment on the water release regulation optimization strategy based on the multi-point collaborative monitoring network.

[0008] One or more technical solutions provided in this application have at least the following technical effects or advantages: A water release scheduling plan is generated based on the water release scheduling requirements of the target reservoir, and the water temperature monitoring dataset of the target reservoir is obtained. The water temperature monitoring dataset is then subjected to sensor reliability calibration to obtain a reliable water temperature dataset. A virtual hierarchical reconstruction based on the analysis of near-deep water temperature changes is performed to obtain a water temperature hierarchical model. The water release scheduling plan is then adjusted through multi-dimensional correlation accident simulation to construct a water release regulation space. The water release regulation space is then optimized collaboratively under multi-modal risk prediction to obtain a water release regulation optimization strategy. Finally, the water release regulation optimization strategy is dynamically adjusted in a closed loop based on the multi-point collaborative monitoring network. This achieves the technical effect of realizing precise risk control and dynamic adaptation of reservoir water release scheduling, improving scheduling safety and adaptability. Attached Figure Description

[0009] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0010] Figure 1 A schematic diagram of the process for a multi-point collaborative automatic monitoring method for reservoir water temperature dynamics provided in this application embodiment; Figure 2 A schematic diagram of the structure of a reservoir water temperature dynamic automatic monitoring system based on multi-point collaboration provided in this application embodiment.

[0011] Figure labeling: Water temperature monitoring dataset acquisition module 10, reliable water temperature dataset acquisition module 20, water temperature stratification model acquisition module 30, water release regulation space construction module 40, water release regulation optimization strategy acquisition module 50, dynamic closed-loop adjustment module 60. Detailed Implementation

[0012] This application provides a method and system for dynamic automatic monitoring of reservoir water temperature based on multi-point collaboration, which addresses the technical problems in the prior art where inaccurate prediction of reservoir water release scheduling risks and the lack of dynamic adaptation and adjustment mechanisms lead to insufficient scheduling safety and adaptability.

[0013] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0014] Example 1, as Figure 1 As shown, this application provides a method for dynamic automatic monitoring of reservoir water temperature based on multi-point collaboration, the method comprising: Step S100: Generate a water release scheduling plan based on the water release scheduling requirements of the target reservoir, and obtain the water temperature monitoring dataset of the target reservoir based on the multi-point collaborative monitoring network.

[0015] Specifically, the core requirements for water release scheduling of the target reservoir are first comprehensively reviewed, covering key aspects such as the water consumption scale of the water supply recipients, key water consumption periods, minimum and maximum water release limits, downstream ecological protection standards, and flood control and drought relief requirements. Combining the reservoir's hydrological characteristics, historical scheduling data, watershed water resource allocation plans, and relevant industry standards, a suitable water release scheduling plan is generated through multi-factor comprehensive modeling and analysis. Simultaneously, relying on a multi-point collaborative monitoring network composed of multiple water temperature monitoring points in the target reservoir, water temperature data at different locations and depths in the reservoir are collected in real time using the sensors deployed at each monitoring point. The collected scattered data is then summarized, organized, and preliminarily formatted to form a water temperature monitoring dataset covering the entire reservoir area and reflecting the spatiotemporal distribution of water temperature.

[0016] Step S200: Perform sensor reliability correction on the water temperature monitoring dataset according to the multi-point collaborative monitoring network to obtain a reliable water temperature dataset.

[0017] Specifically, relying on a multi-point collaborative monitoring network consisting of multiple water temperature monitoring points in the target reservoir, the system employs multi-node data cross-verification, real-time sensor status diagnosis, and intelligent identification and correction of abnormal data to perform sensor reliability calibration on the acquired water temperature monitoring dataset. By comparing synchronous water temperature monitoring data from different monitoring points during the same period, abnormal deviations caused by equipment failure or environmental interference are eliminated. The system also verifies the power supply stability, signal transmission quality, and calibration status of the sensors at each monitoring point, correcting system errors. Finally, using a normal data distribution model built based on historical data from the monitoring network, data exceeding reasonable thresholds are interpolated or replaced, ultimately forming a reliable water temperature dataset whose accuracy, completeness, and reliability meet the requirements of subsequent analysis.

[0018] Step S300: Perform virtual hierarchical reconstruction on the trusted water temperature dataset under the analysis of near-deep water temperature changes to obtain a water temperature stratification model.

[0019] Specifically, firstly, based on the water depth coordinate information in the reliable water temperature dataset, the K-means clustering algorithm is used to classify the water depth characteristics of the data, dividing it into several continuous water depth intervals with similar water temperature distribution patterns, and outputting multiple corresponding water temperature distribution sequences. Then, the sliding window comparison method is used to calculate the point-by-point differences of the water temperature distribution sequences of adjacent water depth intervals, extracting multiple water temperature change characteristics of neighboring depths, such as water temperature gradient, rate of change, and stability. Subsequently, a twin evaluation index system is constructed, covering the consistency of change trends, numerical deviation threshold, and fluctuation synchronicity, and pairwise matching evaluations are performed on the water temperature change characteristics of each neighboring depth to generate neighboring depth change twin sequences. Finally, based on this twin sequence, a three-dimensional interpolation and hierarchical boundary fitting algorithm is used to virtually reconstruct the reliable water temperature dataset, clarifying the boundary range, core water temperature parameters, and vertical distribution patterns of each water temperature layer, and generating a water temperature stratification model that can accurately reflect the three-dimensional distribution characteristics of reservoir water temperature.

[0020] Step S400: Based on the water temperature stratification model, perform multi-dimensional correlation accident simulation and adjustment on the water release scheduling scheme to construct the water release regulation space.

[0021] Specifically, based on the constructed water temperature stratification model, this study employs accident scenario simulation, characteristic identification, causal tracing, and multi-decision combination methods to conduct multi-dimensional correlation accident simulation and adjustment of water release scheduling schemes. For cold water discharge accidents, the water temperature stratification model is used to simulate the accident process and obtain cold water discharge simulation data. Top-level characteristics of the accident are identified, and a multi-level causal tracing simulation path is established. Based on this path, multiple parameters of the scheduling scheme, such as water release volume, water release duration, and outlet location, are correlated and adjusted to generate the first water release adjustment decision set. Similarly, scenario simulation and parameter adjustment are performed for warm water impact and ecological damage accidents, respectively, to obtain the second and third water release adjustment decision sets. Finally, the three decision sets are comprehensively integrated and optimized to construct a water release adjustment space covering different accident response scenarios and containing multiple adjustment scheme possibilities.

[0022] Step S500: Based on the water temperature stratification model, perform collaborative optimization under multimodal risk prediction on the water release regulation space to obtain the water release regulation optimization strategy.

[0023] Specifically, a collaborative optimization strategy for water release regulation is obtained by using a water temperature stratification model to explore the water release regulation space. First, various regulation schemes within the water release regulation space are virtually executed using the water temperature stratification model, generating multiple virtual water release regulation data. Then, a cold water discharge risk prediction network is trained using a historical water release regulation set as input and a historical cold water discharge risk set as output. The virtual data is input into this network to obtain a cold water discharge risk sequence. Simultaneously, warm water impact risk sequences and ecological damage risk sequences are predicted and obtained separately, and integrated to form a water release risk prediction map. Subsequently, based on water release risk constraints including cold water discharge, warm water impact, and ecological damage risks, a multimodal risk constraint collaborative optimization is performed on the water release regulation space to obtain a water release regulation optimization space. Next, weights are assigned to the three types of multimodal water release risk factors to construct a water release coupled risk model. Finally, this model is used to iteratively optimize the water release regulation optimization space, selecting the scheme with the lowest risk and best suitability, generating a water release regulation optimization strategy.

[0024] Step S600: Dynamically close-loop adjust the water release regulation optimization strategy according to the multi-point collaborative monitoring network.

[0025] Specifically, the water temperature stratification model is first subjected to real-time anomaly detection through data deviation analysis and threshold comparison to capture the abnormal fluctuation characteristics of reservoir water temperature in the spatiotemporal dimensions. Combining the meteorological trends around the reservoir, watershed inflow prediction data, and ecological environment evolution patterns, multi-dimensional trend extrapolation is performed on the identified water temperature anomalies to predict the intensity, duration, and scope of impact of the anomalies, thus obtaining the anomaly trend. Based on this trend, parameters such as the layout of monitoring points, data acquisition frequency, and sensor sensitivity of the multi-point collaborative monitoring network are adaptively and correlatedly adjusted to optimize the allocation of monitoring resources. An optimized collaborative monitoring network is established to cover abnormal areas and respond more efficiently. During the execution of the water release regulation optimization strategy, real-time water temperature data of the reservoir is collected simultaneously through the optimized collaborative monitoring network to form an updated water temperature dataset. The updated water temperature dataset is compared and analyzed with the predicted data of the water temperature stratification model to calculate the deviation value. Based on the magnitude and trend of the deviation, the core parameters in the water release regulation optimization strategy, such as the water release volume, water release time, and water outlet selection, are dynamically corrected and iteratively updated to construct a closed-loop mechanism of monitoring-analysis-regulation-feedback, ensuring that the water release scheduling is always accurately adapted to the actual changes in the reservoir water temperature.

[0026] In one possible implementation, step S300 further includes: Step S310: Classify water depth characteristics based on the trusted water temperature dataset to obtain multiple water depth and water temperature distribution sequences.

[0027] Step S320: Compare the water temperature changes at adjacent water depth locations based on the multiple water depth and water temperature distribution sequences to obtain the water temperature change characteristics of multiple adjacent depths.

[0028] Step S330: Perform twin evaluation based on the multiple adjacent deep water temperature change characteristics to obtain adjacent deep water change twin sequences.

[0029] Step S340: Perform virtual hierarchical reconstruction of the reliable water temperature dataset based on the adjacent depth change twin sequence to generate the water temperature hierarchical model.

[0030] Specifically, for the acquired reliable water temperature dataset, the water depth coordinates and matching water temperature monitoring values ​​corresponding to each data point are first extracted. The data is then preprocessed to remove invalid and redundant data and to synchronize timestamps and calibrate coordinates. Subsequently, the K-means clustering algorithm is used, with water depth as the core clustering dimension. Combining the total water depth span of the reservoir, hydrological environmental characteristics, and water temperature data distribution density, a reasonable number of clusters and iteration termination conditions are preset. By calculating the Euclidean distance between data points, the water depth data is clustered and grouped, dividing several continuous water depth intervals with similar water temperature distribution characteristics. Finally, the water temperature data corresponding to each cluster is arranged in ascending order of water depth, forming multiple water depth and water temperature distribution sequences specific to each water depth interval that can accurately reflect the water temperature variation with water depth within that interval.

[0031] For the multiple water depth and temperature distribution sequences already acquired, the distribution sequence pairs corresponding to adjacent water depth positions are determined according to the continuous sorting relationship of water depth intervals. Using a sliding window comparison combined with difference calculation, the water temperature data in each pair of adjacent sequences are matched and analyzed point by point. The key parameters such as water temperature difference, water temperature change rate, change amplitude and fluctuation stability of adjacent water depth positions in the same time dimension are calculated. Through statistical modeling, the unique water temperature change patterns of each adjacent water depth position are extracted, and abnormal interference factors are eliminated. Finally, multiple adjacent water depth temperature change characteristics that can accurately characterize the water temperature difference features and correlation patterns of different adjacent water depth intervals are obtained.

[0032] First, a multi-dimensional twin evaluation index system is constructed, including trend matching degree, numerical fluctuation synchronization rate, mutation threshold matching degree, and stable interval overlap rate. The Analytic Hierarchy Process (AHP) is used to determine the weight of each index. Then, multiple adjacent deep water temperature change characteristics are paired in pairs according to water depth order. The similarity of each pair of characteristics under each evaluation index is calculated. The cosine similarity algorithm is used to calculate the trend matching degree, and the mean square error algorithm is used to calculate the numerical deviation. The weights are combined to obtain a weighted sum to obtain the comprehensive similarity score. A similarity threshold is set, such as 85%, and characteristic combinations that meet the threshold requirement are selected. These successfully matched characteristic combinations are sequentially connected in the order of their corresponding adjacent water depth intervals. At the same time, the correlation strength parameter and matching confidence of the characteristics are supplemented, and finally, an adjacent deep water temperature change twin sequence that can accurately reflect the correlation law and matching relationship of adjacent deep water temperature changes is formed.

[0033] Based on neighboring depth variation twin sequences and combined with the physical characteristics of the vertical distribution of reservoir water temperature, an improved adaptive density peak clustering (ADPClust) algorithm is used to virtually reconstruct a reliable water temperature dataset into layers. First, the characteristic matching strength and association reliability in the neighboring depth variation twin sequences are used as weighting factors and incorporated into the density calculation of data points, strengthening the guiding role of adjacent water depth temperature variation patterns on clustering. Then, by adaptively determining the cluster center and cutoff distance, abrupt boundaries and stable regions of water temperature distribution are identified, dividing similar and continuous water depth intervals into the same water temperature layer. Simultaneously, the radial basis function (RBF) interpolation algorithm is used to complete and smooth the water temperature data within each layer, accurately fitting the water temperature distribution surface of each layer. Finally, core parameters such as boundary water depth, average water temperature, water temperature gradient, and distribution standard deviation of each layer are integrated to generate a water temperature layering model that can three-dimensionally present the vertical layering structure of reservoir water temperature and dynamically reflect the water temperature characteristics of each layer.

[0034] In one possible implementation, step S400 further includes: Step S410: Based on the water temperature stratification model, perform cold water discharge accident simulation and adjustment on the water release scheduling scheme to obtain the first water release adjustment decision set.

[0035] Step S420: Based on the water temperature stratification model, perform temperature water shock accident simulation and adjustment on the water release scheduling scheme to obtain the second water release adjustment decision set.

[0036] Step S430: Based on the water temperature stratification model, perform ecological damage accident simulation and adjustment on the water release scheduling scheme to obtain the third water release regulation decision set.

[0037] Step S440: Combine the decisions based on the first water release regulation decision set, the second water release regulation decision set, and the third water release regulation decision set to generate the water release regulation space.

[0038] Specifically, firstly, based on a water temperature stratification model, using MIKE3 or EFDC three-dimensional hydrodynamic-water temperature coupled simulation software, core parameters such as water release volume, outlet location, and scheduling duration from the water release scheduling plan are input to simulate the occurrence process of cold water discharge accidents under different operating conditions, obtaining cold water discharge simulation data covering the cold water temperature distribution, diffusion range, impact duration, and degree of disturbance to downstream water temperature; then, feature extraction is performed on the simulation data using a random forest algorithm to identify top-level characteristics of the discharge accident, such as the intensity of the accident's impact, diffusion rate, and recovery difficulty; subsequently, based on a Bayesian network model, with the top-level characteristics as target nodes and the water temperature stratification structure... Using structural parameters, water release parameters, and reservoir topography as parent nodes, a multi-level causal tracing network is constructed to clarify the correlation paths and influence weights of various factors with cold water release accidents, and to establish a complete cold water release accident simulation path. Finally, based on this simulation path, a multi-objective optimization algorithm, such as NSGA-II, is adopted to reduce the risk of cold water release, ensure water supply demand, and control scheduling costs as optimization objectives. The algorithm performs multi-dimensional correlation adjustment and combination iteration on water release threshold, water release time allocation, multi-outlet switching logic, and water temperature control parameters in the water release scheduling scheme, generating multiple sets of adjustment schemes that meet different risk prevention and control levels, and integrating them to form the first water release regulation decision set.

[0039] Based on a water temperature stratification model, this study first accurately extracts water temperature distribution data for different water layers in the reservoir, clarifying the thickness, distribution range, and core temperature parameters of the warm water layer. Combined with key information from the water release schedule, such as the release volume, outlet elevation, and scheduling cycle, a simulated scenario of a warm water impact accident is constructed. Numerical simulation technology is then used to simulate the concentrated release of warm water under different operating conditions, obtaining simulation data on the warm water impact, including the diffusion rate, affected water area, downstream water temperature rise, and duration. Based on the simulation data, a feature recognition algorithm is employed to extract the impact intensity level, ecological disturbance degree, and other parameters. The top-level characteristics of warm water shock accidents are restored to their difficulty level. Further, multi-level causal tracing of these top-level characteristics is conducted to clarify the intrinsic relationship between factors such as the location of the warm water layer, the gradient of water release flow, and the rhythm of water release, and the occurrence and impact of the accident. A complete simulation path for warm water shock accidents is established. Finally, based on this simulation path, multi-dimensional correlation adjustments and combination optimizations are performed on the dynamic allocation ratio of water release volume, the timing of water outlet switching, and water temperature pre-regulation parameters in the water release scheduling scheme. Multiple sets of adjustment schemes that can effectively reduce the risk of warm water shock and adapt to different scenario requirements are generated and integrated to form a second water release regulation decision set.

[0040] Based on the aforementioned water temperature stratification model, an ecological damage accident simulation was conducted on the water release scheduling scheme. Combining water temperature stratification data with downstream ecologically sensitive indicators, such as suitable water temperature ranges for fish and temperature thresholds for aquatic plant growth, the simulation explored the impact of reservoir discharge on the downstream ecosystem under different discharge parameters. This yielded ecological damage simulation data encompassing information such as habitat temperature disturbances, changes in ecological community stability, and species survival risk levels. Based on this simulation data, accident outcome characteristics were identified, extracting top-level characteristics of the ecological damage accident, including the scope of impact, degree of damage, and recovery period length. Multi-level causal tracing was performed on these top-level characteristics based on the simulation data to clarify the correlation logic and influence weights between factors such as water temperature stratification imbalance, abnormal discharge water temperature, and flow fluctuation amplitude and the ecological damage accident, establishing an ecological damage accident projection path. Based on this projection path, the water release scheduling scheme was adjusted using multi-parameter correlation. Core parameters such as dynamic allocation of discharge volume, selection of discharge outlets, and control of suitable water temperature discharge periods were combined and optimized to generate multiple adjustment schemes that effectively reduce the risk of ecological damage, which were then integrated to form a third water release adjustment decision set.

[0041] First, the first, second, and third water release regulation decision sets are preprocessed. Conflicts are eliminated through rule verification, such as contradictory water release parameters, overlapping and incompatible water release periods, and invalid regulation schemes that do not comply with basic reservoir scheduling constraints, such as exceeding reservoir capacity or violating ecological baseflow limits. Duplicate or highly similar schemes are also deduplicated and merged. Next, a multi-dimensional decision combination evaluation system is constructed, covering core evaluation indicators such as adaptability to accident risk prevention and control, feasibility of scheduling execution, rationality of cost control, and adaptability to ecological protection. A weighted summation method is used to cross-combine and screen effective schemes from the three decision sets. Finally, all evaluated combination schemes are integrated to form a diversified regulation scheme set encompassing different risk response scenarios, parameter configurations, and scheduling priorities. This set comprehensively covers the individual and collaborative prevention and control needs of three types of accidents: cold water discharge, warm water impact, and ecological damage. Ultimately, a water release regulation space is generated that provides sufficient and effective selection space for subsequent multimodal risk prediction and collaborative optimization.

[0042] In one possible implementation, step S410 further includes: Step S411: Simulate a cold water discharge accident based on the water temperature stratification model to obtain cold water discharge simulation data.

[0043] Step S412: Based on the cold water discharge simulation data, identify the characteristics of the accident results and determine the top-level characteristics of the discharge accident.

[0044] Step S413: Based on the cold water discharge simulation data, perform multi-level causal tracing of the top-level characteristics of the discharge accident and establish a cold water discharge accident projection path.

[0045] Step S414: Based on the simulated path of the cold water discharge accident, perform multi-parameter correlation adjustment on the water release scheduling scheme to generate the first water release adjustment decision set.

[0046] Specifically, based on the core data such as the water temperature distribution, thickness, and temperature gradient of each water layer in the reservoir accurately depicted by the water temperature stratification model, and combined with key parameters such as the water release volume, outlet location and elevation, scheduling duration, and release rhythm specified in the water release scheduling plan, a cold water discharge accident simulation scenario is constructed using the MIKE3 three-dimensional hydrodynamic-water temperature coupled simulation software. By inputting reservoir topographic data and boundary conditions, such as watershed inflow, meteorological data, and key parameters of the bottom cold water layer, the dynamic process of concentrated discharge of the bottom low-temperature water body under different operating conditions is simulated. The diffusion path, temperature decay law, and mixing characteristics of cold water in the reservoir area and downstream river channel are accurately reproduced. Finally, cold water discharge simulation data covering multiple dimensions of information such as the time-series change of cold water discharge flow, temperature curve of the discharged water body, scope and area of ​​the affected water area, water temperature response data of each downstream section, and duration of the accident are obtained.

[0047] First, the simulated cold water discharge data was preprocessed. The Z-score standardization method was used to unify the data dimensions, and box plots were used to remove outliers caused by simulation errors and boundary condition fluctuations, ensuring data reliability. Then, an accident outcome characteristic identification system was constructed, covering four dimensions: temperature impact, spatial diffusion, duration, and ecological impact. The temperature impact dimension includes the magnitude of the water temperature drop, extreme low temperatures, and the difference between the temperature and the downstream suitable water temperature. The spatial diffusion dimension includes the range of low temperature impact, diffusion rate, and covered water area. The duration dimension includes the duration of low temperature and recovery period. The ecological impact dimension includes… This includes the degree of damage to the suitable water temperature range for aquatic organisms and the level of habitat disturbance. Based on this system, a random forest algorithm is used for feature extraction. Preprocessed simulated data is used as the algorithm input. Multiple decision trees are constructed to vote and filter data features, and the importance score of each feature is calculated. Core features with an importance score higher than a preset threshold, such as 0.7, are selected. Finally, the selected core features are classified and integrated to clarify the quantitative indicators and representational significance of each feature. Ultimately, the top-level characteristics of cold water discharge accidents that can comprehensively and accurately reflect the severity, scope of impact, and core hazards of cold water discharge accidents are determined.

[0048] Using simulated cold water discharge data as the core data source, a Bayesian network algorithm is employed to conduct multi-level causal tracing. First, the top-level characteristics of the discharge accident are identified, such as the magnitude of the sudden drop in water temperature and the range of low-temperature impact, serving as the target nodes of the network. Then, key influencing factors are selected from the simulated data as parent nodes, including parameters related to water temperature stratification, such as cold water layer thickness, water temperature gradient, and cold water layer distribution depth; water release scheduling parameters, such as release volume, outlet elevation, scheduling duration, and release rhythm; and reservoir environmental parameters, such as reservoir flow velocity, water level, and topographic features. Through statistical analysis of the simulated data, the relationship between each parent node and the target node is calculated. A conditional probability table between nodes is used to quantify the influence of different factors on the top-level characteristics. At the same time, a multi-level node association structure is constructed based on data correlation. The upper-level nodes are the core influencing factors, such as the thickness of the cold water layer and the discharge volume, while the lower-level nodes are the derived influencing factors, such as the cold water diffusion rate and the mixing efficiency with the surrounding water bodies. Then, through the inference mechanism of Bayesian network, the transmission path of each factor from the bottom to the top-level characteristics is clarified, the role mechanism and priority of key influencing nodes are identified, and weakly correlated or irrelevant factors are eliminated. Finally, a logically coherent, hierarchically clear, quantifiable and traceable cold water discharge accident simulation path is formed.

[0049] Based on the established simulation path of cold water discharge accidents, the correlation weights and influence logic between core parameters such as water release volume, outlet elevation, scheduling duration, and water release rhythm and the risk of cold water discharge accidents are clarified. The NSGA-II multi-objective optimization algorithm is used for multi-parameter correlation adjustment. First, optimization objectives are set. The primary objective is to reduce the risk of cold water discharge accidents below a preset threshold. The secondary objectives are to ensure the normal water supply demand downstream and control scheduling energy consumption costs. Then, the reservoir capacity limit, water release facility carrying capacity, and ecological base flow baseline are used as constraints. The key parameters identified in the simulation path are used as algorithm input variables. Through operations such as population initialization, crossover mutation, non-dominated sorting, and congestion calculation, multiple rounds of combined iterations are performed on each parameter to generate multiple sets of adjustment schemes that meet different optimization priorities. Subsequently, the feasibility of the generated schemes is verified and the effect is evaluated. Schemes that exceed the constraints or have poor risk control effects are eliminated. Finally, the selected effective schemes are integrated to form the first water release regulation decision set covering different parameter configurations and adapting to various operating conditions.

[0050] In one possible implementation, step S500 further includes: Step S510: Perform multimodal risk prediction on the water release regulation space based on the water temperature stratification model, and establish a water release risk prediction map.

[0051] Step S520: Based on the water release risk prediction map, perform multimodal risk constraint collaborative optimization on the water release regulation space according to the water release risk constraints to obtain the water release regulation optimization space.

[0052] Step S530: Based on the multimodal water release risk factors, weights are allocated to establish a water release coupled risk model. The multimodal water release risk factors include cold water discharge risk, warm water impact risk, and ecological damage risk.

[0053] Step S540: Perform iterative optimization of the water release coupling risk in the water release regulation optimization space according to the water release coupling risk model to generate the water release regulation optimization strategy.

[0054] Specifically, based on the core data such as the water temperature distribution, thickness, and gradient of each water layer in the reservoir accurately provided by the water temperature stratification model, all regulation schemes within the water release regulation space are virtually executed. The simulation examines the dynamic changes in water temperature, diffusion paths, and comprehensive impacts on downstream areas under different schemes, obtaining multiple virtual water release regulation data covering multi-dimensional information such as water temperature time-series curves, diffusion range parameters, and ecological response data. Subsequently, a cold water discharge risk prediction network is constructed, using historical water release regulation data as input and corresponding cold water discharge risk records as output. The network is trained using a CNN-LSTM hybrid deep learning model, extracting spatial features from the virtual data through convolutional layers and capturing temporal variation patterns through LSTM layers. After completing network training, multiple virtual data sets of water release regulation are input into the network, outputting a cold water release risk sequence. Simultaneously, the Gradient Boosting Tree (XGBoost) algorithm is used to predict warm water impact risk, using features such as warm water diffusion rate and temperature superposition amplitude in the virtual data to train the model and predict the warm water impact risk sequence. Combining downstream ecologically sensitive indicators, such as the suitable water temperature range for species, the virtual data is quantitatively analyzed using the ecological risk index method to generate an ecological damage risk sequence. Finally, the three types of risk sequences are structured and visualized according to risk type, probability of occurrence, impact level, and time dimension to establish a comprehensive water release risk prediction map covering multimodal risk information.

[0055] Based on the quantified cold water discharge risk sequence, warm water impact risk sequence, and ecological damage risk sequence in the water release risk prediction map, multi-dimensional water release risk constraints are defined. The cold water discharge risk constraint is that the low temperature impact range does not exceed the downstream sensitive water area and the water temperature drop is ≤3℃. The warm water impact risk constraint is that the downstream water temperature rise is ≤2℃ and the impact duration is ≤72 hours. The ecological damage risk constraint is that the habitat disturbance level is ≤II and the species survival risk index is ≤0.3. Subsequently, the multi-objective constraint optimization algorithm NSGA-III is used for collaborative optimization. Each regulation scheme within the water release regulation space is used as the algorithm input, with the core optimization objective being to simultaneously satisfy the three types of risk constraints, using the reservoir as a case study. Constraints such as reservoir capacity limits, maximum capacity of water release facilities, and ecological baseflow baseline are used to conduct multiple rounds of screening and iteration on all schemes through operations such as population initialization, non-dominated sorting based on reference points, crossover mutation, and crowding maintenance. During the iteration process, the three types of risk data of each scheme in the risk prediction map are retrieved one by one and accurately compared with the preset constraints. Schemes with any risk index exceeding the constraint range are eliminated, while schemes that meet the constraints and meet the standards for scheduling feasibility and cost rationality are retained. When the iteration reaches the preset number of times or the screening results converge, all effective schemes that have passed the multi-dimensional constraint verification are integrated to finally form a water release regulation optimization space that focuses on low risk and high adaptability.

[0056] To address three multimodal water release risk factors—cold water discharge risk, warm water impact risk, and ecological damage risk—an analytic hierarchy process (AHP) was employed for weight allocation. First, a hierarchical structure was constructed, comprising the three risk factors. The target layer represented the multimodal water release risk weight allocation, while the criterion layer comprised the three risk factors. Subsequently, experts in water conservancy engineering, ecological environment, and hydrological scheduling were invited to pairwise score the importance of each risk factor in the criterion layer. Using a 1-9 scale, where 1 indicates equal importance, 3 indicates slightly more important, 5 indicates significantly more important, 7 indicates strongly more important, and 9 indicates extremely more important, the inverse is used for the opposite. This constructed a judgment matrix. Finally, the largest eigenvalue and its corresponding eigenvector of the judgment matrix were calculated, and the eigenvectors were analyzed. The initial weights are obtained through normalization and then a consistency test is performed. The test index is CR = CI / RI, where CI is the consistency index and RI is the average random consistency index. When CR < 0.1, the judgment matrix meets the consistency requirements; otherwise, the expert scores need to be adjusted until they meet the requirements. Finally, the weight coefficients of each risk factor are determined, such as cold water discharge risk weight w1 = 0.4, warm water impact risk weight w2 = 0.3, and ecological damage risk weight w3 = 0.3. Based on this, a water release coupled risk model is established, setting the water release coupled risk value = cold water discharge risk value × w1 + warm water impact risk value × w2 + ecological damage risk value × w3. The three types of single risk factors are quantitatively integrated through this linear weighting function to achieve accurate assessment and quantitative characterization of the comprehensive risk of the water release regulation scheme.

[0057] Each regulation scheme within the water release regulation optimization space is taken as the object to be evaluated. First, the risk values ​​of cold water discharge, warm water impact, and ecological damage corresponding to each scheme are extracted. The data comes from the water release risk prediction map. Substituted into the established water release coupling risk model, the quantitative value of the water release coupling risk for each scheme is calculated. Then, an improved particle swarm optimization (PSO) algorithm is used for iterative optimization, with the optimal scheme having the minimum water release coupling risk value as the core optimization objective. Each regulation scheme is mapped to a particle in the particle swarm, and the positions of the particle swarm, the parameter combinations and velocities of the corresponding schemes, the corresponding parameter adjustment ranges, and algorithm parameters such as the number of iterations, inertia weight, and learning factor are set. In each iteration, the position and velocity of particles are updated by individual extreme values, the historical best coupling risk value of a single particle, the global extreme value, and the historical best coupling risk value of the entire particle swarm. At the same time, the feasibility of the updated particles is verified by combining the actual constraints of reservoir scheduling, such as the upper limit of water release, the operation restrictions of the water outlet, and the requirements of ecological base flow, and invalid schemes that exceed the constraints are eliminated. As the iteration progresses, schemes with lower coupling risk values ​​are continuously screened out. When the iteration reaches the preset number or the global extreme value remains stable for several consecutive rounds, such as no better value appears for 10 consecutive rounds, the algorithm converges. At this time, the adjustment scheme mapped by the particle corresponding to the global extreme value is the optimal scheme, which is determined as the water release regulation optimization strategy.

[0058] In one possible implementation, step S510 further includes: Step S511: Virtually execute the water release regulation space according to the water temperature stratification model to obtain multiple virtual water release regulation data.

[0059] Step S512: Using the historical data of water release regulation as input information and the historical data of cold water discharge risk as output information, train the cold water discharge risk prediction network.

[0060] Step S513: Input the multiple virtual data of water release regulation into the cold water discharge risk prediction network to obtain the cold water discharge risk sequence.

[0061] Step S514: Based on the multiple virtual data of water release regulation, predict the risk of warm water impact and obtain the warm water impact risk sequence.

[0062] Step S515: Based on the multiple virtual data of water release regulation, predict the ecological damage risk and obtain the ecological damage risk sequence.

[0063] Step S516: Organize the cold water discharge risk sequence, the warm water impact risk sequence, and the ecological damage risk sequence to generate the water release risk prediction map.

[0064] Specifically, based on core data such as the precise output of the water temperature stratification model, including the water temperature distribution, thickness, temperature gradient, and spatial relationships of different water layers in the reservoir, a three-dimensional hydrodynamic-temperature coupled simulation algorithm, such as the MIKE3 algorithm, is used to virtually execute all regulation schemes within the water release regulation space. Key parameters of each regulation scheme, such as the water release volume, outlet location and elevation, scheduling duration, and water release rhythm, are used as algorithm inputs. Combined with reservoir topographic data and boundary conditions, such as watershed inflow, meteorological data, and core water temperature stratification data, the algorithm analyzes the water flow equations and water... Numerical solutions to the temperature transport equations are used to simulate the dynamic temperature changes of the released water body under different scenarios, the diffusion path in the reservoir area and downstream river channels, the mixing process with surrounding water bodies, and the comprehensive impact on downstream water areas. Finally, for each regulation scheme, virtual data of water release regulation is output, which includes time-series curves of water release flow, time-series data of temperature of the released water body, quantitative parameters of diffusion range such as influence radius, coverage area, water temperature response data of key downstream sections, and temperature disturbance feedback in ecologically sensitive areas. The virtual data corresponding to all schemes are integrated to form multiple virtual data of water release regulation.

[0065] First, historical data on water release scheduling of the target reservoir were collected, including parameters such as release volume, outlet location and elevation, scheduling duration, and release rhythm. This data was then integrated to form a structured historical data set of water release scheduling. Simultaneously, actual records of cold water release risks under corresponding historical scheduling scenarios were compiled, covering indicators such as the impact range of low temperatures, the magnitude of sudden water temperature drops, the severity of accidents, and downstream ecological feedback. After quantification, the accident levels were classified into 1-5 levels to construct a historical data set of cold water release risks. Subsequently, a CNN-LSTM hybrid deep learning network was built as the cold water release risk prediction network. The CNN layer was used to extract spatial features from the historical data set of water release scheduling, such as the spatial correlation between outlet distribution and water temperature stratification. The LSTM layer is used to capture temporal features, such as the time patterns of changes in water release volume and dynamic evolution of water temperature. The historical data of water release regulation is divided into a training set and a validation set in a 7:3 ratio. The training set is used for iterative optimization of network parameters, and the validation set is used for real-time evaluation of prediction accuracy. During training, the mean squared error (MSE) is used as the loss function, and the loss value is minimized through the Adam optimization algorithm. The weights of the convolutional kernels, the state parameters of the LSTM units, and the weights of the fully connected layers are iteratively adjusted. After each iteration, the model performance is verified using the validation set. When the MSE of the validation set is lower than a preset threshold (e.g., 0.03) for 5 consecutive iterations and the model shows no overfitting, training is stopped, and the construction of the cold water release risk prediction network is completed.

[0066] First, multiple virtual data sets for water release regulation are preprocessed. The Z-score standardization algorithm is used to unify the dimensions of indicators in different dimensions of the data, such as water flow rate, downstream water temperature, and diffusion range parameters. Outliers caused by boundary condition fluctuations or calculation errors during the simulation are removed using the box plot method to ensure the consistency and reliability of the input data. Then, the preprocessed virtual data sets for water release regulation are sequentially input into a pre-trained CNN-LSTM cold water discharge risk prediction network according to the schemes within the water release regulation space. The network extracts spatial features from the data through convolutional layers, such as the spatial correlation between the outlet location and the cold water layer distribution, and the spatial morphology of the diffusion range. Then, the LSTM layer captures temporal features, such as the dynamic trend of water temperature change and the temporal fluctuation pattern of flow rate. The extracted features are fused and mapped through fully connected layers to output a quantified value of cold water discharge risk corresponding to each virtual data set for water release regulation. The quantified value ranges from 0 to 1, with the value closer to 1 indicating a higher risk. Finally, the quantified values ​​of cold water discharge risk corresponding to all virtual data sets are arranged sequentially according to the input order to form a cold water discharge risk sequence that corresponds one-to-one with the schemes in the water release regulation space.

[0067] Core feature parameters related to warm water impact were extracted from multiple virtual data sets of water release regulation, including warm water diffusion rate, temperature difference between the discharged water and downstream water, quantified value of warm water influence range, duration of temperature rise, and peak water temperature at downstream sensitive sections, among other multi-dimensional indicators. Subsequently, a gradient boosting tree (XGBoost) algorithm was used to construct a warm water impact risk prediction model, with the extracted feature parameters used as model input variables. Simultaneously, relevant data on historical warm water impact events in the reservoir were collected, and the risk levels of warm water impacts in historical events were quantified into 0-1 values ​​as labels based on their impact severity. The model was then trained, and the tree structure was iteratively optimized. The learning rate and regularization parameters are adjusted to minimize the mean squared error between the predicted value and the true label until the model's prediction accuracy on the validation set meets the preset requirements, such as accuracy ≥ 90%. After training, the feature parameters corresponding to each virtual water release regulation data are input into the model one by one. The model outputs a quantitative value of the warm water impact risk corresponding to each virtual data by weighted combination and nonlinear mapping of the features. The closer the value is to 1, the higher the risk. Finally, all the quantitative values ​​are arranged in order of the schemes in the water release regulation space to form a warm water impact risk sequence corresponding to each regulation scheme, so as to achieve accurate quantitative characterization of the warm water impact risk of each scheme.

[0068] Core characteristic parameters related to ecological damage were extracted from multiple virtual data sets of water release regulation, covering multi-dimensional indicators such as the magnitude and duration of temperature disturbances in downstream ecologically sensitive areas, the degree of habitat integrity damage, deviations from suitable water temperature ranges for aquatic organisms, and changes in species survival probability. Subsequently, a combined algorithm of Analytic Hierarchy Process (AHP) and Support Vector Machine (SVM) was used to predict ecological damage risk. First, experts in ecological environment and water conservancy scheduling were invited to pairwise score the importance of each characteristic parameter using the AHP algorithm, constructing a judgment matrix and completing a consistency check. The weight coefficients of each parameter were calculated, such as: deviation from suitable water temperature range weight 0.35, degree of habitat damage weight 0.3, change in species survival probability weight 0.2, magnitude of temperature disturbance weight 0.1, and duration of disturbance weight 0.05. Then, the weighted values ​​were... The feature parameters are used as input to the SVM model, with the quantified risk level of historical ecological damage events, in the range of 0-1 (0 representing no risk and 1 representing extremely high risk), as labels. The kernel function parameters of the SVM model are optimized using a grid search method, with the RBF kernel function selected. The penalty coefficient C=10 and the kernel function parameter γ=0.1 are optimized to minimize the model prediction error until the prediction accuracy on the validation set is ≥88%. After training, the weighted feature parameters corresponding to each water release regulation virtual data are input into the SVM model one by one. The model outputs the quantified ecological damage risk value corresponding to each virtual data through nonlinear mapping. Finally, all quantified values ​​are arranged sequentially according to the scheme order within the water release regulation space to form an ecological damage risk sequence corresponding to each regulation scheme, achieving accurate quantitative representation of the ecological damage risk of each scheme.

[0069] First, the core correlation logic of the cold water discharge risk sequence, the warm water impact risk sequence, and the ecological damage risk sequence is clarified. Using the unique identifier of each regulation scheme within the water release regulation space as an index, the three types of risk quantification values ​​corresponding to each scheme are matched one-to-one, constructing a four-dimensional correlation data structure of scheme identifier - cold water discharge risk value - warm water impact risk value - ecological damage risk value. Then, the three types of risk values ​​are standardized and unified to ensure they all fall within the 0-1 quantification range for easy horizontal comparison. Based on risk level classification rules, such as 0-0.3 for low risk, 0.3-0.7 for medium risk, and 0.7-1 for high risk, each risk value is labeled with a corresponding level label. Finally, graph visualization technology is used, with nodes... The system presents various regulation schemes in a visual format, with node size mapping the comprehensive risk level of water release coupling. This risk level is initially calculated by weighting three types of risk values. Node colors distinguish the dominant risk type, such as blue for cold water discharge risk, red for warm water impact risk, and green for ecological damage risk. The thickness of the edges represents the similarity of risk characteristics between different schemes. Finally, the system integrates scheme identifiers, three types of risk quantification values, risk level labels, visualized nodes, and relationships to generate a clear, complete, and intuitive water release risk prediction map. This provides a centralized and visualized presentation of the multimodal risks of all regulation schemes, offering clear data support and decision-making reference for subsequent risk constraint optimization.

[0070] In one possible implementation, step S600 further includes: Step S610: Perform anomaly detection on the water temperature stratification model to obtain the anomaly characteristics of the reservoir water temperature.

[0071] Step S620: Based on the reservoir environment trend, perform trend extrapolation on the abnormal characteristics of the reservoir water temperature to obtain the abnormal water temperature change trend.

[0072] Step S630: Adaptively adjust the multi-point collaborative monitoring network according to the abnormal water temperature change trend to obtain an optimized collaborative monitoring network.

[0073] Step S640: When executing the water release regulation optimization strategy, the updated water temperature dataset of the optimized collaborative monitoring network is acquired simultaneously.

[0074] Step S650: Dynamically correct and iteratively update the water release regulation optimization strategy based on the updated water temperature dataset.

[0075] Specifically, it adopts an isolation forest-based Isolation approach. Forest's anomaly detection algorithm comprehensively monitors the water temperature stratification model. It first extracts core data from the model's output, including real-time water temperature values ​​for each water layer, water temperature gradient distribution, water layer thickness parameters, and time-series temperature variation curves at different depths, constructing a multi-dimensional input feature set. The algorithm then randomly samples the feature set data to build multiple isolated trees. Leveraging the ease with which abnormal water temperature data can be quickly isolated in the feature space, an anomaly score is calculated for each data point, ranging from 0 to 1, with scores closer to 1 indicating a higher degree of anomaly. Combining this with historical water temperature data from the target reservoir and statistically analyzed for normal fluctuation thresholds, such as the normal fluctuation range and reasonable gradient change intervals for each water layer, an anomaly score threshold is set, such as 0.7. Data points with scores higher than the threshold and exceeding the historical normal fluctuation range are then selected. Further analysis is conducted on the water layer location, water temperature deviation magnitude, anomaly occurrence time, duration, and abrupt temperature gradient changes corresponding to these anomalies. Simultaneously, factors other than water temperature anomalies, such as equipment malfunctions and data transmission interference, are excluded. Finally, the algorithm integrates these findings to form a reservoir water temperature anomaly characteristic profile containing information such as anomaly water layer identification, specific values ​​of anomaly parameters, degree of deviation from the normal range, and preliminary characteristics of anomaly evolution.

[0076] First, multi-dimensional trend data of the reservoir environment are collected, covering meteorological trends, including time-series changes in temperature, precipitation, sunshine duration, and wind speed; watershed inflow trends, including dynamic evolution data of inflow volume, inflow temperature, and inflow water quality; and reservoir operation trends, including water level adjustment plans, historical water release frequency and intensity, and operational status of water conservancy facilities. These data are standardized and preprocessed to eliminate dimensional differences. Then, the trend characteristics of each environmental factor are extracted using the sliding window method. Subsequently, a combined algorithm of LSTM network and Monte Carlo simulation is used for trend extrapolation. The reservoir water temperature anomaly characteristics, including the location of the anomalous water layer, the magnitude of water temperature deviation, the duration of the anomaly, and gradient abrupt change characteristics, are combined with the extracted environmental trend characteristics as the input to the LSTM network. The network learns the temporal correlation between historical water temperature anomalies and environmental factors to capture the dynamic evolution logic of water temperature anomalies with environmental changes, and outputs the basic evolution trend of water temperature anomalies. At the same time, it introduces the Monte Carlo simulation algorithm to set up a variety of extreme environmental trend scenarios, such as extreme high temperature and low rainfall, continuous heavy rainfall and surge in water inflow, and simulates the fluctuation of environmental factors under different scenarios through a large number of random samples to calculate the development probability of water temperature anomalies under each scenario. Finally, it integrates the temporal prediction results of the LSTM network and the probability analysis results of the Monte Carlo simulation to form a water temperature anomaly change trend containing information such as anomaly intensity change curve, path of water layer expansion, duration prediction interval, probability of risk occurrence under different scenarios and peak intensity prediction.

[0077] Based on the trend of abnormal water temperature changes, the core monitoring objectives are clearly defined, including the current location of the abnormal water layer, the predicted expansion path, key water temperature indicators such as the magnitude of the anomaly, the rate of gradient change, and the potential impact area. An adaptive weighted correlation algorithm is used to dynamically adjust the multi-point collaborative monitoring network. First, the spatial distance between each existing monitoring point and the core anomaly area and the predicted expansion area is calculated using the Euclidean distance algorithm. Then, the correlation between historical water temperature data and anomaly characteristics at each monitoring point is analyzed using the Pearson correlation coefficient. The spatial distance weight and the data correlation weight are weighted and fused at a ratio of 0.4:0.6 to obtain the comprehensive priority weight for each monitoring point; the higher the weight, the greater the monitoring value. Subsequently, the sampling frequency of the monitoring points is adjusted according to the comprehensive priority weight, prioritizing high-weight monitoring points (i.e., weight ≥ 0.7). The sampling interval was shortened from the usual 1 hour to 15 minutes. For medium-weighted monitoring points (0.3 ≤ weight < 0.7), a 30-minute sampling interval was maintained. For low-weight monitoring points (weight < 0.3), a 1-hour sampling interval was retained. At the same time, a link optimization algorithm was used to reconstruct the data transmission network of the monitoring points, establishing a dedicated transmission channel between high-priority monitoring points and the data center to reduce data transmission latency. For potential abnormal expansion areas not covered by existing monitoring points identified in trend analysis, the spatial distribution characteristics of the areas were analyzed using the K-means clustering algorithm to determine the optimal deployment location and number of temporary monitoring points, triggering deployment instructions to supplement monitoring coverage. Ultimately, an optimized collaborative monitoring network was formed that focuses on abnormal areas, dynamically adapts sampling frequency, achieves efficient data transmission, and provides coverage without blind spots.

[0078] After the execution process of the water release regulation optimization strategy is initiated, the all-time data acquisition mode of the optimized collaborative monitoring network is triggered simultaneously. In this network, each fixed monitoring point continuously collects real-time water temperature data at the corresponding monitoring location according to the adaptively adjusted sampling frequency: high-priority monitoring points every 15 minutes, medium-priority monitoring points every 30 minutes, and low-priority monitoring points every hour. Newly added temporary monitoring points simultaneously track the water temperature dynamics of potential anomaly expansion areas. The raw water temperature data collected by all monitoring points is transmitted to the data processing center in real time through an optimized dedicated transmission channel or priority communication link. The data processing center performs millisecond-level cleaning on the received raw data, quickly removing anomalies caused by equipment failure and signal interference. At the same time, it completes data format standardization and time sequence alignment processing, and integrates to form an updated water temperature dataset containing multi-dimensional information such as the full-time water temperature curve of each monitoring point, dynamic tracking data of key indicators in anomaly areas, real-time snapshots of water temperature distribution across the entire reservoir, and dynamic data on water layer gradient changes. This ensures that the dataset can accurately and in real time reflect the true changes in reservoir water temperature during the strategy execution process.

[0079] The updated water temperature dataset obtained from the optimized collaborative monitoring network is compared with the initial prediction data from the water temperature stratification model in multiple dimensions. Deviation values ​​of key indicators such as real-time water temperature, water temperature gradient changes, and evolution of abnormal areas in each water layer are extracted. Based on set deviation thresholds, such as absolute water temperature deviation ≤ 0.5℃ and gradient change deviation ≤ 0.2℃ / m, it is determined whether correction is needed. If the deviation is within the allowable range, the current water release regulation optimization strategy continues to be implemented. If the deviation exceeds the threshold, combined with the abnormal water temperature change trend, a multi-parameter dynamic optimization algorithm is used to adjust the core regulation parameters in the strategy, such as the water release allocation ratio, etc. The operation mode of the combined water outlet and the segmented setting of the scheduling time were adjusted in a targeted manner. At the same time, the corrected parameters were substituted into the water release coupling risk model and the coupling risk value was recalculated to ensure that the corrected strategy still meets the risk constraints of cold water discharge, warm water impact and ecological damage. Subsequently, the corrected strategy that has been verified is used as the new current execution strategy. The updated water temperature dataset is continuously received for real-time deviation monitoring. A closed-loop process of data comparison-deviation judgment-parameter correction-risk verification is carried out in a loop to realize the dynamic iterative update of the water release regulation optimization strategy and ensure that the strategy is always accurately adapted to the actual changes in the reservoir water temperature.

[0080] In one possible implementation, step S100 further includes: The multi-point collaborative monitoring network includes multiple water temperature monitoring points in the target reservoir.

[0081] Specifically, the multi-point collaborative monitoring network is a three-dimensional monitoring system built to meet the water temperature monitoring needs of the target reservoir. Its core comprises multiple water temperature monitoring points distributed at different horizontal and water depth locations within the reservoir. Horizontally, the monitoring points cover key areas of the reservoir, including the upstream, midstream, and downstream sections, as well as near the shore and the center of the lake. This includes key monitoring areas such as the vicinity of the discharge outlet and ecologically sensitive areas, while also considering special locations prone to water temperature fluctuations, such as the reservoir's edge and tributary confluences, ensuring no blind spots in the horizontal direction. Vertically, monitoring points are strategically placed according to the reservoir's water depth gradient, from surface water to deep water, comprehensively capturing water temperature differences and vertical distribution characteristics at different depths, accurately reflecting the reservoir's water temperature stratification structure. These monitoring points simultaneously collect real-time water temperature data, achieving dynamic perception of the entire reservoir's water temperature across all areas and depths through collaborative operation. This provides comprehensive and accurate basic data support for subsequent water temperature data correction, stratification model construction, and water release strategy optimization.

[0082] In one possible implementation, step S520 further includes: The risk constraints for water release include cold water discharge risk constraints, warm water impact risk constraints, and ecological damage risk constraints.

[0083] Specifically, the aforementioned water release risk constraints are a multi-dimensional constraint system designed to ensure the safety of reservoir water release scheduling and reduce ecological and environmental impacts. The core of these constraints covers three categories: cold water release risk constraints, warm water impact risk constraints, and ecological damage risk constraints. Cold water release risk constraints focus on limiting the negative impacts of releasing low-temperature water bodies, clearly defining the minimum temperature threshold of the released water body, the upper limit of the sudden temperature drop, and the maximum range of low-temperature impact, to prevent a sharp drop in downstream water temperature from harming aquatic life, agricultural irrigation, and industrial water use. Warm water impact risk constraints address the potential risks of high-temperature water discharge, setting the maximum temperature difference between the released water body and downstream natural water bodies, the upper limit of the duration of temperature rise, and the maximum coverage area of ​​warm water diffusion, to prevent excessively high water temperatures from leading to reduced dissolved oxygen and imbalance in the aquatic ecosystem. Ecological damage risk constraints revolve around the comprehensive impact of reservoir water release on the downstream ecological environment, specifying indicators such as habitat disturbance level thresholds, upper limits of deviation from the suitable water temperature range for aquatic organisms, and safe ranges for species survival risk indices, comprehensively avoiding ecosystem damage and biodiversity loss caused by water release. The three types of constraints complement each other and work together to provide clear standards for risk screening of water release regulation space, ensuring that the final water release regulation optimization strategy can not only meet the scheduling needs, but also effectively control various environmental risks.

[0084] Example 2 is based on the same inventive concept as the multi-point collaborative automatic monitoring method for reservoir water temperature in the previous examples, such as... Figure 2 As shown, this application provides a multi-point collaborative automatic monitoring system for reservoir water temperature dynamics. The system and method embodiments in this application are based on the same inventive concept. The system includes: The water temperature monitoring dataset acquisition module 10 is used to generate a water release scheduling plan based on the water release scheduling requirements of the target reservoir, and to acquire the water temperature monitoring dataset of the target reservoir based on a multi-point collaborative monitoring network.

[0085] The reliable water temperature dataset acquisition module 20 is used to perform sensor reliability correction on the water temperature monitoring dataset based on the multi-point collaborative monitoring network to acquire a reliable water temperature dataset.

[0086] The water temperature stratification model acquisition module 30 is used to perform virtual stratification reconstruction on the reliable water temperature dataset under the analysis of near-deep water temperature changes, and to obtain the water temperature stratification model.

[0087] The water release regulation space construction module 40 is used to perform multi-dimensional correlation accident simulation and adjustment of the water release scheduling scheme based on the water temperature stratification model, and to construct the water release regulation space.

[0088] The water release regulation optimization strategy acquisition module 50 is used to perform collaborative optimization under multimodal risk prediction on the water release regulation space based on the water temperature stratification model, and to acquire the water release regulation optimization strategy.

[0089] The dynamic closed-loop adjustment module 60 is used to dynamically adjust the water release regulation optimization strategy according to the multi-point collaborative monitoring network.

[0090] Furthermore, the system also includes: Based on the reliable water temperature dataset, water depth characteristics are classified to obtain multiple water temperature distribution sequences at different depths; based on the multiple water temperature distribution sequences at different depths, water temperature changes at adjacent depths are compared to obtain multiple neighboring depth water temperature change characteristics; based on the multiple neighboring depth water temperature change characteristics, twin evaluation is performed to obtain neighboring depth change twin sequences; based on the neighboring depth change twin sequences, the reliable water temperature dataset is virtually hierarchically reconstructed to generate the water temperature hierarchical model.

[0091] Furthermore, the system also includes: Based on the water temperature stratification model, the water release scheduling scheme is simulated and adjusted for cold water discharge accidents to obtain a first water release adjustment decision set; based on the water temperature stratification model, the water release scheduling scheme is simulated and adjusted for warm water impact accidents to obtain a second water release adjustment decision set; based on the water temperature stratification model, the water release scheduling scheme is simulated and adjusted for ecological damage accidents to obtain a third water release adjustment decision set; and the first water release adjustment decision set, the second water release adjustment decision set, and the third water release adjustment decision set are combined to generate the water release adjustment space.

[0092] Furthermore, the system also includes: The cold water discharge accident simulation is performed on the water release scheduling scheme based on the water temperature stratification model to obtain cold water discharge simulation data; the accident result characteristics are identified based on the cold water discharge simulation data to determine the top-level characteristics of the discharge accident; multi-level causal tracing is performed on the top-level characteristics of the discharge accident based on the cold water discharge simulation data to establish a cold water discharge accident inference path; and multi-parameter correlation adjustment is performed on the water release scheduling scheme based on the cold water discharge accident inference path to generate the first water release adjustment decision set.

[0093] Furthermore, the system also includes: Based on the water temperature stratification model, multimodal risk prediction is performed on the water release regulation space to establish a water release risk prediction map. Based on the water release risk prediction map, multimodal risk constraint collaborative optimization is performed on the water release regulation space according to the water release risk constraints to obtain the water release regulation optimization space. Based on the multimodal water release risk factors, a water release coupled risk model is established by weighting the multimodal water release risk factors, which include cold water discharge risk, warm water impact risk, and ecological damage risk. Based on the water release coupled risk model, water release coupled risk iterative optimization is performed on the water release regulation optimization space to generate the water release regulation optimization strategy.

[0094] Furthermore, the system also includes: The water release regulation space is virtually executed according to the water temperature stratification model to obtain multiple virtual water release regulation data. A cold water discharge risk prediction network is trained using the historical water release regulation data as input and the historical cold water discharge risk data as output. The multiple virtual water release regulation data are input into the cold water discharge risk prediction network to obtain a cold water discharge risk sequence. Warm water impact risk prediction is performed based on the multiple virtual water release regulation data to obtain a warm water impact risk sequence. Ecological damage risk prediction is performed based on the multiple virtual water release regulation data to obtain an ecological damage risk sequence. The cold water discharge risk sequence, the warm water impact risk sequence, and the ecological damage risk sequence are then combined to generate the water release risk prediction map.

[0095] Furthermore, the system also includes: Anomaly detection is performed on the water temperature stratification model to obtain abnormal characteristics of reservoir water temperature; trend extrapolation of the abnormal characteristics of reservoir water temperature is performed based on reservoir environmental trends to obtain abnormal water temperature change trends; adaptive correlation adjustment is performed on the multi-point collaborative monitoring network based on the abnormal water temperature change trends to obtain an optimized collaborative monitoring network; when executing the water release regulation optimization strategy, the updated water temperature dataset of the optimized collaborative monitoring network is obtained simultaneously; the water release regulation optimization strategy is dynamically corrected and iteratively updated based on the updated water temperature dataset.

[0096] Furthermore, the system also includes: The multi-point collaborative monitoring network includes multiple water temperature monitoring points in the target reservoir.

[0097] Furthermore, the system also includes: The risk constraints for water release include cold water discharge risk constraints, warm water impact risk constraints, and ecological damage risk constraints.

[0098] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, the above description focuses on specific embodiments of this specification. Additionally, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some implementations, multitasking and parallel processing are possible or may be advantageous.

[0099] The above description is only a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

[0100] This specification and accompanying drawings are merely illustrative examples of this application and are intended to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from its scope. Therefore, if such modifications and variations fall within the scope of this application and its equivalents, this application intends to include such modifications and variations.

Claims

1. A method for dynamic automatic monitoring of reservoir water temperature based on multi-point collaboration, characterized in that, The method includes: A water release scheduling plan is generated based on the water release scheduling requirements of the target reservoir, and the water temperature monitoring dataset of the target reservoir is obtained based on the multi-point collaborative monitoring network. Based on the multi-point collaborative monitoring network, the water temperature monitoring dataset is subjected to sensor reliability correction to obtain a reliable water temperature dataset. The reliable water temperature dataset is subjected to virtual hierarchical reconstruction under the analysis of near-deep water temperature changes to obtain a water temperature stratification model; Based on the water temperature stratification model, the water release scheduling scheme is subjected to multi-dimensional correlation accident simulation and adjustment to construct a water release adjustment space. Based on the water temperature stratification model, a collaborative optimization under multimodal risk prediction is performed on the water release regulation space to obtain a water release regulation optimization strategy. The water release regulation optimization strategy is dynamically adjusted in a closed loop based on the multi-point collaborative monitoring network.

2. The method for dynamic automatic monitoring of reservoir water temperature based on multi-point collaboration as described in claim 1, characterized in that, The reliable water temperature dataset is subjected to virtual hierarchical reconstruction based on the analysis of near-deep water temperature changes to obtain a water temperature stratification model, including: Based on the reliable water temperature dataset, water depth characteristics are classified to obtain multiple water depth and water temperature distribution sequences. Based on the multiple water depth and water temperature distribution sequences, the water temperature changes at adjacent water depth locations are compared to obtain the water temperature change characteristics of multiple adjacent depths. Based on the aforementioned multiple adjacent deep water temperature change characteristics, a twin evaluation is performed to obtain adjacent deep water change twin sequences; The reliable water temperature dataset is virtually reconstructed using the adjacent depth variation twin sequence to generate the water temperature stratification model.

3. The method for dynamic automatic monitoring of reservoir water temperature based on multi-point collaboration as described in claim 1, characterized in that, Based on the water temperature stratification model, the water release scheduling scheme is subjected to multi-dimensional correlation accident simulation and adjustment to construct a water release adjustment space, including: Based on the water temperature stratification model, the water release scheduling scheme is simulated and adjusted for cold water discharge accidents to obtain the first water release adjustment decision set. Based on the water temperature stratification model, the water release scheduling scheme is simulated and adjusted for temperature water shock accidents to obtain a second water release adjustment decision set. Based on the water temperature stratification model, the water release scheduling scheme is simulated and adjusted to simulate ecological damage accidents, and a third water release adjustment decision set is obtained. The water release regulation space is generated by combining decisions based on the first water release regulation decision set, the second water release regulation decision set, and the third water release regulation decision set.

4. The method for dynamic automatic monitoring of reservoir water temperature based on multi-point collaboration as described in claim 3, characterized in that, Based on the water temperature stratification model, the water release scheduling scheme is simulated and adjusted for cold water discharge accidents to obtain a first water release adjustment decision set, including: Based on the water temperature stratification model, a cold water discharge accident simulation was performed on the water release scheduling scheme to obtain cold water discharge simulation data. Based on the simulated cold water discharge data, the characteristics of the accident results are identified to determine the top-level characteristics of the discharge accident. Based on the cold water discharge simulation data, multi-level causal tracing is performed on the top-level characteristics of the discharge accident to establish a cold water discharge accident simulation path; Based on the simulated path of the cold water discharge accident, the water release scheduling scheme is adjusted using multiple parameters to generate the first water release adjustment decision set.

5. The method for dynamic automatic monitoring of reservoir water temperature based on multi-point collaboration as described in claim 1, characterized in that, Based on the water temperature stratification model, a collaborative optimization strategy for water release regulation is obtained by performing multimodal risk prediction on the water release regulation space, including: Based on the water temperature stratification model, a multimodal risk prediction of the water release regulation space is performed, and a water release risk prediction map is established. Based on the aforementioned water release risk prediction map, the water release regulation space is optimized through multimodal risk constraint collaborative search according to the water release risk constraints, thereby obtaining the water release regulation optimization space. A water release coupled risk model is established by weighting multimodal water release risk factors, including cold water discharge risk, warm water impact risk, and ecological damage risk. Based on the water release coupling risk model, the water release regulation optimization space is iteratively optimized to generate the water release regulation optimization strategy.

6. The method for dynamic automatic monitoring of reservoir water temperature based on multi-point collaboration as described in claim 5, characterized in that, Based on the aforementioned water temperature stratification model, a multimodal risk prediction is performed on the water release regulation space, and a water release risk prediction map is established, including: The water release regulation space is virtually executed according to the water temperature stratification model to obtain multiple virtual water release regulation data. Using the historical data of water release regulation as input information and the historical data of cold water discharge risk as output information, a cold water discharge risk prediction network is trained. The multiple virtual data of water release regulation are input into the cold water discharge risk prediction network to obtain the cold water discharge risk sequence; Based on the multiple virtual data of water release regulation, the risk of warm water shock is predicted, and a warm water shock risk sequence is obtained. Based on the multiple virtual data of water release regulation, ecological damage risk is predicted, and an ecological damage risk sequence is obtained. The risk sequences of cold water discharge, warm water impact, and ecological damage are compiled to generate the water release risk prediction map.

7. The method for dynamic automatic monitoring of reservoir water temperature based on multi-point collaboration as described in claim 1, characterized in that, The water release regulation optimization strategy is dynamically adjusted in a closed loop according to the multi-point collaborative monitoring network, including: Anomaly detection is performed on the water temperature stratification model to obtain the anomaly characteristics of the reservoir water temperature; Based on the reservoir environment trend, the abnormal characteristics of the reservoir water temperature are extrapolated to obtain the abnormal water temperature change trend. Based on the abnormal water temperature change trend, the multi-point collaborative monitoring network is adaptively correlated and adjusted to obtain an optimized collaborative monitoring network. When executing the water release regulation optimization strategy, the updated water temperature dataset of the optimized collaborative monitoring network is acquired simultaneously. The water release regulation optimization strategy is dynamically corrected and iteratively updated based on the updated water temperature dataset.

8. The method for dynamic automatic monitoring of reservoir water temperature based on multi-point collaboration as described in claim 1, characterized in that, The multi-point collaborative monitoring network includes multiple water temperature monitoring points in the target reservoir.

9. The method for dynamic automatic monitoring of reservoir water temperature based on multi-point collaboration as described in claim 5, characterized in that, The risk constraints for water release include cold water discharge risk constraints, warm water impact risk constraints, and ecological damage risk constraints.

10. A reservoir water temperature dynamic automatic monitoring system based on multi-point collaboration, characterized in that, The system is used to implement the multi-point collaborative dynamic automatic monitoring method for reservoir water temperature according to any one of claims 1-9, and the system includes: The water temperature monitoring dataset acquisition module is used to generate a water release scheduling plan based on the water release scheduling requirements of the target reservoir, and to acquire the water temperature monitoring dataset of the target reservoir based on a multi-point collaborative monitoring network. The reliable water temperature dataset acquisition module is used to perform sensor reliability correction on the water temperature monitoring dataset based on the multi-point collaborative monitoring network to acquire a reliable water temperature dataset. The water temperature stratification model acquisition module is used to perform virtual stratification reconstruction on the reliable water temperature dataset under the analysis of near-deep water temperature changes, and to obtain the water temperature stratification model. The water release regulation space construction module is used to perform multi-dimensional correlation accident simulation and adjustment on the water release scheduling scheme according to the water temperature stratification model, and construct the water release regulation space. The water release regulation optimization strategy acquisition module is used to perform collaborative optimization under multimodal risk prediction on the water release regulation space based on the water temperature stratification model, and to acquire the water release regulation optimization strategy. The dynamic closed-loop adjustment module is used to dynamically adjust the water release regulation optimization strategy according to the multi-point collaborative monitoring network.

Images