A method and system for monitoring water levels in water conservancy projects based on artificial intelligence

Abstract

This application provides an artificial intelligence-based method and system for monitoring water levels in water conservancy projects. The method involves acquiring meteorological parameters, soil moisture parameters, and water level measurement parameters generated by magnetic levitation displacement sensing from a group of reservoirs across a basin, analyzing the dynamic changes of these parameters, and calculating the correlation strength coefficient reflecting the hydraulic interaction between reservoirs. This coefficient is then jointly encoded with the water level measurement parameters to construct a dynamic correlation map across reservoirs. Real-time water level signals from each reservoir are simultaneously collected and matched with a pre-set fluctuation feature library to identify abnormal water level intervals related to water flow path nodes. By combining the mapping relationship between the dynamic correlation map and the abnormal fluctuation intervals, the risk level of water level imbalance is determined, and control commands, including flood discharge path optimization and water storage capacity adjustment, are generated accordingly. The technical solution provided in this application can improve the efficiency and accuracy of water level monitoring in water conservancy projects.

Description

Technical Field

[0001] This application relates to the field of water level monitoring technology for water conservancy projects, and in particular to a water level monitoring method and system for water conservancy projects based on artificial intelligence. Background Technology

[0002] Coordinated water level regulation across river basin reservoirs is crucial for effective water resource management, especially in the face of climate change, flood warnings, and drought management. To ensure coordinated operation among reservoirs and prevent upstream and downstream areas from facing floods or droughts due to improper operation of a single reservoir, a highly intelligent monitoring system is needed.

[0003] Currently, existing solutions include water level monitoring based on Internet of Things (IoT) technology. This solution deploys a series of high-precision sensors in various reservoirs to collect water information in real time and transmits the data to a central processing platform using wireless communication networks. The central platform then uses big data analytics algorithms to process the collected information, thereby achieving dynamic monitoring and analysis of the overall operational status of the reservoir group. This method not only improves the accuracy of data collection but also greatly enhances the speed and efficiency of information processing, providing strong support for making scientific and rational reservoir scheduling decisions.

[0004] While IoT-based water level monitoring solutions excel in enhancing data processing capabilities and decision support, they also have certain limitations. First, the reliance on numerous sensors and a complex network architecture results in high initial construction costs and relatively high maintenance difficulty. Second, in response to extreme weather or natural disasters, data loss may occur due to sensor damage or network outages, impacting the overall system's stability and reliability. Finally, although big data analytics provides powerful data processing capabilities, in certain situations, such as inappropriate model parameter settings or poor input data quality, the analysis results may deviate from reality, affecting the accuracy of decision-making. These issues limit the widespread adoption and effectiveness of this solution in practical applications. Summary of the Invention

[0005] This application provides a water level monitoring method and system for water conservancy projects based on artificial intelligence, in order to solve the problems of low efficiency and poor accuracy in water level monitoring of water conservancy projects in the prior art.

[0006] Firstly, this application provides an artificial intelligence-based method for monitoring water levels in hydraulic engineering projects, including:

[0007] Meteorological parameters, soil moisture parameters, and water level measurement parameters of a cross-basin reservoir group are obtained. The water level measurement parameters are generated by magnetic levitation displacement sensing and have a linear mapping relationship with water level changes.

[0008] Based on the dynamic variation patterns of the meteorological parameters and soil moisture parameters, a correlation strength coefficient reflecting the hydraulic interaction between reservoirs is obtained. The correlation strength coefficient is then jointly encoded with the water level measurement parameters to generate a dynamic correlation map across reservoirs.

[0009] Real-time water level signals from each reservoir are collected synchronously, and the real-time water level signals are spatiotemporally matched with a preset water level fluctuation feature library to extract abnormal water level fluctuation ranges associated with water flow path nodes in the cross-reservoir dynamic association map.

[0010] The risk level of water level imbalance of the cross-basin reservoir group is determined by the mapping relationship between the cross-reservoir dynamic correlation map and the abnormal water level fluctuation range.

[0011] Based on the water level imbalance risk level, control instructions are generated, which include flood discharge path optimization schemes and water storage capacity adjustment ratios.

[0012] Optionally, the step of jointly encoding the correlation strength coefficient and the water level measurement parameters to generate a cross-reservoir dynamic correlation map includes:

[0013] The weight ratio of meteorological parameters and soil moisture parameters in the correlation strength coefficient is dynamically allocated based on the instantaneous rate of change of meteorological parameters and the cumulative amount of soil moisture parameters.

[0014] The temporal fluctuation amplitude of the water level measurement parameters is superimposed with the weight ratio to generate a hydraulic influence value characterizing the influence of the external environment on the reservoir node, and a hydraulic transmission path between reservoir nodes is constructed based on the difference between the hydraulic influence values ​​of adjacent reservoir nodes.

[0015] Based on the synchronous change trend of the water level measurement parameters in the hydraulic transmission path, the connection strength of the hydraulic transmission path is dynamically corrected to obtain the corrected connection strength.

[0016] The geographical coordinates, hydraulic transmission paths, and corrected connection strengths of the reservoir nodes are bound together to generate a dynamic cross-reservoir association map.

[0017] Optionally, the step of superimposing the temporal fluctuation amplitude of the water level measurement parameters with the weight ratio to generate a hydraulic influence value characterizing the influence of the external environment on the reservoir nodes, and constructing a hydraulic transmission path between reservoir nodes based on the difference between the hydraulic influence values ​​of adjacent reservoir nodes, includes:

[0018] The rising and falling phases of the time-series fluctuation amplitude of the water level measurement parameter are weighted according to the weight ratio to obtain the weighted rising phase time-series fluctuation amplitude and falling phase time-series fluctuation amplitude.

[0019] The weighted time-series fluctuation amplitude of the rising phase and the time-series fluctuation amplitude of the falling phase are directionally superimposed to generate a hydraulic impact value characterizing the reservoir node's influence by the external environment.

[0020] Calculate the absolute difference of the hydraulic influence values ​​of adjacent reservoir nodes, compare the absolute difference with a preset generation threshold for hydraulic conduction paths, and set the preset hydraulic conduction paths whose absolute difference exceeds the generation threshold as candidate conduction paths;

[0021] Based on the consistency between the connection strength of the candidate conduction path and the fluctuation direction of the water level measurement parameters within the current time window, the candidate conduction paths that meet the activation conditions are selected to obtain the activated candidate conduction path set.

[0022] The activated candidate conduction path set is superimposed with the preset conduction path in the cross-reservoir dynamic correlation map to obtain the hydraulic conduction path.

[0023] Optionally, determining the water level imbalance risk level of the cross-basin reservoir group through the mapping relationship between the cross-reservoir dynamic correlation map and the abnormal water level fluctuation range includes:

[0024] The water flow path nodes in the cross-reservoir dynamic correlation map are dynamically matched with the water level anomaly fluctuation interval to determine the correspondence between the connection strength of the water flow path nodes and the duration of the water level anomaly fluctuation interval.

[0025] Based on the correspondence, the spatial distribution range and temporal cumulative effect of the abnormal water level fluctuation interval in the cross-reservoir dynamic correlation map are statistically analyzed.

[0026] The spatial distribution range, the cumulative effect over time, and the preset threshold are compared and analyzed. Based on the comparison and analysis results, the risk level range of water level imbalance of the cross-basin reservoir group is divided. When the spatial distribution range and the cumulative effect over time both exceed the preset threshold, it is determined to be a high-risk level.

[0027] Optionally, the step of performing spatiotemporal matching between the real-time water level signal and a preset water level fluctuation feature database to extract abnormal water level fluctuation intervals associated with water flow path nodes in the cross-reservoir dynamic correlation map includes:

[0028] The real-time water level signal is dynamically divided into continuous fluctuation segments according to a preset time window. The length of the preset time window is adaptively adjusted according to the connection density of the water flow path nodes in the cross-reservoir dynamic correlation map.

[0029] The continuous fluctuation segment is matched with the fluctuation pattern in the preset water level fluctuation feature library in a spatiotemporal order. Based on the matching results, abnormal fluctuation segments in the continuous fluctuation segment whose matching degree with the fluctuation pattern exceeds a preset threshold are selected.

[0030] The abnormal fluctuation segment is mapped to the water flow path node, and the correlation influence intensity of the abnormal fluctuation segment on the adjacent reservoir node is calculated by the connection strength of the mapped water flow path node.

[0031] Based on the spatial distribution range and cumulative duration of the correlation influence intensity in the cross-reservoir dynamic correlation map, the water level anomaly fluctuation range associated with the water flow path node is obtained.

[0032] Optionally, the step of mapping the abnormal fluctuation segment to the water flow path node and calculating the correlation influence strength of the abnormal fluctuation segment on adjacent reservoir nodes based on the connection strength of the mapped water flow path nodes includes:

[0033] The spatiotemporal location distribution of the abnormal fluctuation segment and the water flow path node are compared and overlapped, and the water flow path node with an overlap ratio exceeding a preset ratio is selected as the target mapping node to which the abnormal fluctuation segment belongs.

[0034] Based on the connection strength of the target mapping node, calculate the initial impact of the abnormal fluctuation segment on the target mapping node;

[0035] Based on the connection strength ratio between the target mapping node and the adjacent reservoir nodes, the initial influence amount is allocated to each adjacent reservoir node, and the allocated amounts of the initial influence amount on each adjacent reservoir node are accumulated to obtain the associated influence strength.

[0036] Optionally, generating control instructions based on the water level imbalance risk level includes:

[0037] Based on the risk range corresponding to the water level imbalance risk level, the flood discharge priority order of each reservoir node is determined, and the flood discharge allocation ratio of each reservoir node is calculated according to the difference between the current water storage capacity and the preset safety capacity of each reservoir node in the flood discharge priority order.

[0038] Based on the connection strength and spatial distribution of water flow path nodes in the cross-reservoir dynamic correlation map, an optimized flood discharge path scheme is generated.

[0039] Based on the matching relationship between the flood discharge allocation ratio and the historical flood discharge records for the same period, and combined with the water level imbalance risk level, a water storage capacity adjustment ratio is generated. The water storage capacity adjustment ratio is then linked and bound to the flood discharge path optimization scheme to generate control instructions.

[0040] Secondly, this application provides an artificial intelligence-based water level monitoring system for water conservancy projects, comprising:

[0041] The acquisition module acquires meteorological parameters, soil moisture parameters, and water level measurement parameters of the cross-basin reservoir group. The water level measurement parameters are generated by magnetic levitation displacement sensing and have a linear mapping relationship with water level changes.

[0042] The coding module, based on the dynamic change patterns of the meteorological parameters and soil moisture parameters, obtains the correlation strength coefficient reflecting the hydraulic interaction between reservoirs, and jointly encodes the correlation strength coefficient with the water level measurement parameters to generate a dynamic correlation map across reservoirs.

[0043] The extraction module synchronously collects real-time water level signals from each reservoir, performs spatiotemporal matching of the real-time water level signals with a preset water level fluctuation feature library, and extracts the abnormal water level fluctuation range associated with the water flow path nodes in the cross-reservoir dynamic association map.

[0044] The mapping module determines the water level imbalance risk level of the cross-basin reservoir group by mapping the cross-reservoir dynamic correlation map with the abnormal water level fluctuation range.

[0045] The generation module generates control instructions based on the water level imbalance risk level. The control instructions include a flood discharge path optimization scheme and a water storage capacity adjustment ratio.

[0046] Thirdly, embodiments of this application provide a computing device, including a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are invoked and executed by the processing component to implement an artificial intelligence-based water level monitoring method for water conservancy projects as described in the first aspect above.

[0047] Fourthly, embodiments of this application provide a computer storage medium storing a computer program, which, when executed by a computer, implements an artificial intelligence-based water level monitoring method for water conservancy projects as described in the first aspect.

[0048] In this embodiment, meteorological parameters, soil moisture parameters, and water level measurement parameters of a cross-basin reservoir group are acquired. The water level measurement parameters are generated through magnetic levitation displacement sensing and have a linear mapping relationship with water level changes. This enables non-contact, high-precision acquisition of water level parameters through magnetic levitation displacement sensing. Combined with multi-source data fusion of meteorological and soil moisture data, this overcomes the limitations of traditional single-point monitoring and improves the real-time performance and environmental adaptability of data acquisition. Based on the dynamic change patterns of the meteorological and soil moisture parameters, a correlation strength coefficient reflecting the hydraulic interaction between reservoirs is obtained. This correlation strength coefficient is jointly encoded with the water level measurement parameters to generate a cross-reservoir dynamic correlation map. Based on the dynamic weight allocation of the correlation strength coefficient and the joint encoding of the water level parameters, a spatiotemporal topological model reflecting the hydraulic interaction of the reservoir group can be constructed, realizing the quantitative expression and visualization of complex hydrological correlations. By synchronously acquiring real-time water level signals from each reservoir, these real-time water level signals are compared with preset water level data. The system performs spatiotemporal matching of water level fluctuation feature databases to extract abnormal water level fluctuation intervals associated with water flow path nodes in the cross-reservoir dynamic correlation map. This enables real-time spatiotemporal matching of water level signals with the feature database, accurately locating abnormal fluctuation areas associated with water flow path nodes, and solving the problem of delayed detection of sudden water level changes in traditional methods. Through the mapping relationship between the cross-reservoir dynamic correlation map and the abnormal water level fluctuation intervals, the system determines the water level imbalance risk level of the cross-basin reservoir group. Based on the dynamic mapping relationship between the map and the abnormal fluctuation intervals, a multi-dimensional risk assessment model can be established, enabling graded early warning and source tracing analysis of water level imbalance risks in the cross-basin reservoir group. Control instructions are generated based on the water level imbalance risk level. These instructions include flood discharge path optimization schemes and water storage capacity adjustment ratios. By combining the risk level with the spatial distribution characteristics of the dynamic map, the system generates flood discharge path optimization schemes and water storage capacity adjustment instructions, improving the accuracy and execution efficiency of multi-objective coordinated control.

[0049] Furthermore, based on the dynamic weighting ratio of the instantaneous change rate of meteorological parameters and the cumulative soil moisture, the temporal fluctuation amplitude of water level parameters is superimposed with the weights to generate a hydraulic impact value. A hydraulic transmission path is constructed according to the difference in impact values ​​between adjacent nodes, and the path connection strength is dynamically adjusted in conjunction with the synchronous change trend of water level. Finally, a dynamic correlation map across reservoirs is generated by data binding of node coordinates, transmission paths, and adjustment strengths. This method can effectively reflect the changes in hydraulic impact between different reservoir nodes caused by external environmental factors. By constructing and dynamically adjusting hydraulic transmission paths, it not only improves the understanding of the interaction relationships between reservoir nodes but also provides data support for precise regulation, enhancing the scientificity and rationality of coordinated water level regulation across watershed reservoir groups.

[0050] These or other aspects of this application will become more apparent in the following description of the embodiments. Attached Figure Description

[0051] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0052] Figure 1 A flowchart of an artificial intelligence-based water level monitoring method for water conservancy projects provided in this application is shown;

[0053] Figure 2 This application provides a schematic diagram of the structure of an artificial intelligence-based water level monitoring system for water conservancy projects.

[0054] Figure 3 A schematic diagram of the structure of a computing device provided in this application is shown. Detailed Implementation

[0055] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings.

[0056] In some of the processes described in the specification, claims, and accompanying drawings of this application, multiple operations appearing in a specific order are included. However, it should be clearly understood that these operations may not be executed in the order they appear herein, or may be executed in parallel. The operation numbers, such as 101, 102, etc., are merely used to distinguish different operations and do not themselves represent any execution order. Furthermore, these processes may include more or fewer operations, and these operations may be executed sequentially or in parallel. It should be noted that the descriptions such as "first," "second," etc., in this document are used to distinguish different messages, devices, modules, etc., and do not represent a chronological order, nor do they limit "first" and "second" to different types.

[0057] Researchers have found that existing methods for monitoring water levels in inter-basin reservoir groups struggle to dynamically quantify the hydraulic interactions between reservoirs and lack multi-source data fusion mechanisms. This leads to delays in identifying the propagation paths of sudden water level changes and insufficient precision in multi-objective coordinated regulation. Therefore, this paper proposes an AI-based water level monitoring method for hydraulic engineering. Specifically, firstly, a data foundation is constructed by collecting meteorological parameters, soil moisture parameters, and water level measurement parameters generated by magnetic levitation displacement sensing from inter-basin reservoir groups. Next, correlation strength coefficients are calculated based on these parameters and jointly encoded with the water level measurement parameters to generate a dynamic correlation map across reservoirs, while simultaneously identifying abnormal water level fluctuation ranges. Then, based on the mapping relationship between the dynamic correlation map and the abnormal water level fluctuation ranges, the risk level of water level imbalance is assessed and determined. Finally, regulation commands are automatically generated according to the risk level to optimize flood discharge paths and adjust water storage capacity, achieving scientific and rational water level regulation.

[0058] This method generates a dynamic correlation map by fusing magnetic levitation displacement sensing with meteorological and soil parameters, and achieves precise positioning of abnormal fluctuation ranges by combining real-time water level signal spatiotemporal matching. Based on the risk level, it adaptively generates flood discharge and water storage control commands to achieve scientific and rational control of water levels in the reservoir group, improve the ability to cope with water level changes and extreme weather events, and ensure the safe and efficient use of water resources.

[0059] 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 some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0060] Figure 1 This application provides a flowchart of an artificial intelligence-based water level monitoring method for water conservancy projects, as shown in the embodiments. Figure 1 As shown, the method includes:

[0061] 101. Obtain meteorological parameters, soil moisture parameters, and water level measurement parameters of a cross-basin reservoir group. The water level measurement parameters are generated by magnetic levitation displacement sensing and have a linear mapping relationship with water level changes.

[0062] In this step, meteorological parameters refer to data reflecting changes in the atmospheric environment, including rainfall intensity, wind speed, and evaporation.

[0063] Soil moisture parameters refer to the water content data collected by soil sensors, which characterize the surface infiltration and water storage capacity.

[0064] Water level measurement parameters refer to the non-contact water level signal generated by the magnetic levitation displacement sensing device. The electrical signal is generated by the Hall element triggered by the vertical displacement of the magnetic levitation ball. The intensity of the electrical signal is linearly mapped to the water level change.

[0065] Magnetic levitation displacement sensing is a technology that uses changes in magnetic fields to accurately measure the position or displacement of an object. It is applied in scenarios that require high-precision position detection, and in this solution, it is used to measure changes in water level.

[0066] The linear mapping relationship of water level changes refers to the direct and proportional relationship between the output signal measured by the magnetic levitation displacement sensor and the actual water level.

[0067] In this embodiment, meteorological parameters (precipitation, evaporation, etc.) are first acquired through meteorological stations (such as automatic rain gauges, temperature and humidity sensors) and satellite remote sensing. Soil moisture parameters are then acquired using soil moisture sensors (such as TDR probes) or radar inversion technology, and water level measurement parameters are recorded in real time using a magnetic levitation displacement sensor (based on the Hall effect principle). The analog signals (such as voltage) output by the sensors are then digitized through a conversion module and calibrated as water level height values. Kalman filtering is applied to both meteorological and soil moisture parameters to eliminate noise (e.g., removing abnormal evaporation data points). Finally, linear regression is applied to the water level measurement parameters to verify the linear mapping relationship (R² > 0.99 is considered valid).

[0068] In a reservoir group in a certain province (B), an automatic weather station and soil moisture sensor array covering the entire watershed have been deployed. These devices monitor meteorological data such as precipitation, temperature, and wind speed, as well as soil moisture content information in real time, and upload the data to a central database via wireless communication modules. Simultaneously, magnetic levitation displacement sensors are installed in each reservoir to accurately measure water level changes. For example, on a certain day, the water level in Reservoir C rose rapidly due to continuous rainfall; the magnetic levitation displacement sensor captured this change in real time and transmitted the relevant data to the system. After initial cleaning and formatting, all data is integrated into the Global Positioning System (GPS), providing a clear spatial distribution map for subsequent analysis.

[0069] 102. Based on the dynamic variation patterns of the meteorological parameters and soil moisture parameters, a correlation strength coefficient reflecting the hydraulic interaction between reservoirs is obtained. The correlation strength coefficient is then jointly encoded with the water level measurement parameters to generate a dynamic correlation map across reservoirs.

[0070] In this step, hydraulic interaction refers to the mutual influence and interaction between different water bodies in a water resource system, especially in a cross-basin or multi-reservoir system, due to the exchange of water flow and volume.

[0071] The correlation strength coefficient refers to the degree of influence of meteorological and soil moisture parameters on the hydraulic interaction between reservoirs. It is dynamically calculated from the instantaneous rate of change (meteorology) and the cumulative amount (soil moisture).

[0072] Joint encoding refers to integrating multiple different types of data sources or data features into a unified representation through specific methods or algorithms, so as to facilitate further processing, analysis, or model training.

[0073] The cross-reservoir dynamic association graph is a graph structure that describes the hydraulic connection relationship of a reservoir group, with nodes representing reservoirs and edge weights representing association strength coefficients.

[0074] In this embodiment, firstly, meteorological and soil moisture parameters are extracted from the data obtained in step 101, and time series analysis algorithms are used to identify their changing patterns over time. Then, based on these changing patterns, the correlation strength coefficients between different reservoirs are calculated. This step typically involves statistical methods or machine learning algorithms (e.g., decision trees, random forests) to quantify the degree of water flow interaction between reservoirs. Finally, combining water level measurement parameters, deep learning techniques (e.g., convolutional neural networks) are applied for joint encoding to construct a cross-reservoir dynamic correlation map reflecting the complex relationships between reservoirs. This map not only shows the physical connections between reservoirs but also reflects the complex dynamic interactions between them.

[0075] Using the historical data from step 101, the system performed a time-series analysis on precipitation, soil moisture, and water level changes in the B reservoir group. For example, it was found that specific seasonal rainfall patterns cause Reservoir D to transfer more runoff to downstream Reservoir E. Based on this, the system calculated the correlation strength coefficient between the reservoirs and combined it with water level measurement parameters to generate a cross-reservoir dynamic correlation map using a deep learning model. The cross-reservoir dynamic correlation map shows that when Reservoir D discharges water, it has a significant impact on the water level of Reservoir E, while Reservoir F is relatively independent. The cross-reservoir dynamic correlation map lays the foundation for the next step of extracting abnormal fluctuation ranges.

[0076] 103. Synchronously collect real-time water level signals from each reservoir, perform spatiotemporal matching of the real-time water level signals with a preset water level fluctuation feature library, and extract the abnormal water level fluctuation range associated with the water flow path nodes in the cross-reservoir dynamic correlation map.

[0077] In this step, the water level fluctuation feature library consists of predefined normal / abnormal water level patterns (such as flood peak waveforms and drought-period flat curves).

[0078] Real-time water level signals refer to the information on the current water level height obtained directly from water bodies (such as rivers, lakes, and reservoirs) through various sensors and technologies.

[0079] An abnormal water level fluctuation range refers to a time period that deviates from the feature database pattern (such as the water level rise rate exceeding the threshold for two consecutive hours).

[0080] In this embodiment, real-time water level signals from each reservoir are first synchronized via IoT devices at a frequency of minutes. Then, a dynamic time warping algorithm is used to match the real-time water level signals with a preset water level fluctuation feature database. If the matching discrepancy exceeds a threshold, it is marked as an abnormal interval (e.g., a sudden rise in water level of 1.8m within 2 hours). Finally, a dynamic correlation map across reservoirs is extracted, and the upstream nodes of the abnormal reservoirs are traced back (e.g., the discharge of water from Reservoir A causes an anomaly in Reservoir C), locating the abnormal water level fluctuation intervals along key water flow paths.

[0081] One day, the water level of Reservoir C rose rapidly due to torrential rain upstream. The system simultaneously collected real-time water level signals from all reservoirs and performed spatiotemporal matching with a pre-set water level fluctuation feature database. Using sliding window technology and the isolated forest algorithm, the system identified that the water level of Reservoir C exceeded the normal fluctuation range and marked it as an abnormal fluctuation interval. Further analysis using dynamic correlation graphs revealed that this was a chain reaction caused by a large-scale discharge from Reservoir D upstream within a short period. This result was then used to assess the overall risk level.

[0082] 104. By mapping the cross-reservoir dynamic correlation map to the abnormal water level fluctuation range, determine the water level imbalance risk level of the cross-basin reservoir group;

[0083] In this step, the risk level of water level imbalance refers to the risk level (such as low / medium / high) classified according to the degree of abnormal fluctuation and the topological structure of the map.

[0084] In this embodiment, firstly, the topological importance of each reservoir node in the cross-reservoir dynamic correlation map is calculated based on an algorithm (e.g., the weight of a key node is 0.9). Then, the abnormal amplitude (Z value), correlation strength, and node importance of the abnormal water level fluctuation range are integrated, and the risk is quantified using a weighted formula (risk value = node weight × abnormal Z value × correlation strength). Finally, the risk level is classified according to a preset threshold (e.g., >0.8 is high risk), and the water level imbalance risk level of the reservoir group is output.

[0085] Based on the results of step 103, the system combines the dynamic correlation map across reservoirs and uses a Bayesian network model to assess the overall risk of the B reservoir group. For example, it finds that abnormal fluctuations in Reservoir C may lead to a dam overflow risk at its downstream Reservoir G, while other reservoirs are within a safe range. Ultimately, the system classifies the risk levels into high risk (Reservoir C), medium risk (Reservoir G), and low risk (the remaining reservoirs), and issues an early warning to the management team, recommending that priority measures be taken to alleviate the pressure on Reservoir C.

[0086] 105. Generate control instructions based on the water level imbalance risk level, the control instructions including flood discharge path optimization scheme and water storage capacity adjustment ratio.

[0087] In this step, the control instructions are specific operational guidelines or commands generated in the water resources management system based on the analysis of current hydrological conditions, weather forecasts, and historical data. They aim to optimize the allocation and management of water resources, ensure the safe operation of water conservancy facilities, and address potential risks such as floods or droughts.

[0088] The flood discharge path optimization plan refers to dynamically adjusting the opening of the flood discharge gates and the target reservoir based on the risk level.

[0089] The water storage capacity adjustment ratio refers to the redistribution of water storage based on the load of upstream and downstream reservoirs (such as expanding the capacity of downstream reservoirs by 5%).

[0090] In this embodiment, firstly, the correlation strength coefficients in the dynamic correlation graph are converted into path weights (weight = 1 / correlation strength). An algorithm is then used to search the graph for the optimal flood discharge path that has the least impact on downstream areas (e.g., avoiding hub node C and selecting the A→D→river path). Next, a linear programming model is constructed, using the total reservoir capacity safety threshold as a constraint, to dynamically allocate the adjustment ratio of each reservoir's water storage capacity according to risk level. Finally, the system automatically controls the gate opening, executes the optimized flood discharge path, adjusts the pumping rate to achieve the water storage capacity target, and generates control commands.

[0091] Based on the risk level classification in step 104, the system automatically generates control instructions. For example, for Reservoir C, which is classified as high-risk, it is recommended to increase its discharge to Reservoir H downstream, while reducing the water storage of other reservoirs to make room. Furthermore, the system optimizes the discharge path, prioritizing drainage through channels with less downstream impact. These instructions are sent directly to the operating facilities of the relevant reservoirs via the automated control system, successfully preventing the possibility of Reservoir C overflowing and ensuring the safe operation of the entire reservoir group.

[0092] In summary, steps 101 to 105 form a complete closed-loop management process, from data collection and dynamic correlation analysis to risk assessment and control command generation. In practical scenarios, this method can monitor water level changes across river basin reservoirs in real time, quickly identify potential risks, and generate scientifically sound control plans, thereby significantly improving the efficiency and safety of water resource management and ensuring the smooth implementation of flood control and drought relief efforts.

[0093] To further improve the dynamism and environmental adaptability of the cross-reservoir dynamic correlation map, the weight ratios are first dynamically adjusted based on the instantaneous change rate of meteorological parameters and the cumulative soil moisture, enhancing the model's response to sudden weather events and long-term infiltration. Next, the temporal fluctuation amplitude of water level parameters is superimposed with the weights to generate hydraulic impact values ​​characterizing external environmental influences, and a preliminary hydraulic transmission path is constructed by combining the impact value differences between adjacent nodes. Furthermore, the path connection strength is dynamically corrected based on the synchronicity of water level changes to ensure the timeliness and accuracy of the transmission relationship. Finally, the reservoir geographic coordinates, corrected paths, and intensity data are bound together to form a cross-reservoir dynamic correlation map, providing spatial topological support for watershed collaborative management. In some embodiments, step 102, which involves jointly encoding the correlation strength coefficient with the water level measurement parameters to generate the cross-reservoir dynamic correlation map, includes:

[0094] 201. Based on the instantaneous rate of change of meteorological parameters and the cumulative amount of soil moisture parameters, dynamically allocate the weight ratio of meteorological parameters and soil moisture parameters in the correlation strength coefficient.

[0095] In step 201, the instantaneous rate of change of meteorological parameters refers to the real-time gradient of changes in meteorological elements such as temperature and rainfall. The cumulative amount of soil moisture parameters refers to the integral value of soil moisture parameters over a set time period (e.g., 24 hours). Weight allocation refers to adjusting the contribution ratio of meteorological and soil moisture to the correlation strength through an adaptive algorithm.

[0096] In this embodiment, a sliding window mechanism is first used to calculate the instantaneous rate of change of meteorological parameters (such as the hourly rainfall difference value). Simultaneously, an integral algorithm is used to accumulate soil moisture parameter sensor data over time, calculating the accumulated amount. Next, normalization is applied to map the two types of parameters to the [0,1] interval, and weight ratios are dynamically allocated based on an adaptive weighting model (such as the entropy weighting method). For example, when the meteorological rate of change exceeds a threshold, its weight ratio increases to 70%, while the soil moisture weight decreases to 30%. The final weight ratio is output for subsequent calculation of hydraulic impact values.

[0097] 202. The temporal fluctuation amplitude of the water level measurement parameters is superimposed with the weight ratio to generate a hydraulic influence value characterizing the influence of the external environment on the reservoir node, and a hydraulic transmission path between reservoir nodes is constructed based on the difference between the hydraulic influence values ​​of adjacent reservoir nodes.

[0098] In step 202, the temporal fluctuation amplitude refers to the standard deviation or range of the water level measurement parameters over time, reflecting the degree of drastic change in reservoir storage. The hydraulic impact value is a comprehensive index that integrates environmental weights and water level fluctuations, quantifying the intensity of the external environment's influence on the reservoir node. The hydraulic transmission path refers to the directed edge constructed based on the gradient difference in hydraulic impact values ​​between adjacent reservoirs, characterizing the direction of water potential energy transfer.

[0099] In this embodiment, the temporal fluctuation amplitude of the reservoir water level sensor is first extracted. After filtering and denoising, the temporal fluctuation amplitude of the water level measurement parameter (such as the variance of water level change over 24 hours) is calculated. This fluctuation amplitude is then linearly superimposed with the weight ratio output in step 201. Subsequently, the spatial adjacency relationship of the reservoir nodes is determined based on the triangulation algorithm, and the difference (ΔH) in the hydraulic influence value between adjacent nodes is calculated. If the difference is greater than a preset threshold, a hydraulic transmission path is constructed from the high-value node to the low-value node, and an initial connection strength (such as the absolute value of ΔH) is assigned.

[0100] 203. Based on the synchronous change trend of the water level measurement parameters in the hydraulic transmission path, the connection strength of the hydraulic transmission path is dynamically corrected to obtain the corrected connection strength;

[0101] In step 203, the synchronous change trend refers to the correlation of multiple reservoir water level parameters over time, which is quantified using covariance or dynamic time warping algorithms. Connectivity strength correction refers to strengthening or weakening the weight of the transmission path based on the synchronous analysis results.

[0102] In this embodiment, for reservoir nodes with hydraulic transmission paths, time-series segments of their water level measurement parameters are extracted, and a correlation coefficient algorithm is used to calculate the synchronization index. If the correlation coefficient > 0.8, the connection strength is enhanced using an exponential weighting method (e.g., strength = original strength). (1 + correlation coefficient)). If the correlation coefficient is less than 0.3, an attenuation function is used to reduce the intensity. Simultaneously, time lag analysis (such as cross-correlation function) is introduced to determine the transmission delay and dynamically adjust the path direction. For example, if the water level change in reservoir A lags behind that in reservoir B by 2 hours, the transmission path direction is corrected to A→B. Finally, the corrected connection strength is obtained.

[0103] 204. Bind the geographical coordinates, hydraulic transmission paths and corrected connection strength of the reservoir nodes to generate a dynamic cross-reservoir association map.

[0104] In step 204, geographic location coordinates refer to a set of values ​​used to determine the precise location of reservoir nodes on the Earth's surface. Data binding encapsulates spatial coordinates, path topology, and intensity values ​​into a graph data structure. The cross-reservoir dynamic association map is a multidimensional relationship network stored in a graph database, supporting real-time updates and visual queries.

[0105] In this embodiment, the geographical coordinates of the reservoir nodes are first imported into a geographic information system and spatially aligned with the hydraulic conduction path corrected in step 203. Next, a graph modeling tool is used to construct a weighted directed graph from node attributes (coordinates, current water level) and edge attributes (connection strength, conduction direction). Finally, an interactive cross-reservoir dynamic graph is generated using WebGL technology, supporting the display of the coupling relationship between environmental parameters and hydraulic conduction by time slices.

[0106] Here is a specific example:

[0107] During a prolonged summer drought in a reservoir group B in a certain province, the system monitored abnormal water level fluctuations in 15 reservoirs due to agricultural irrigation. In step 201, meteorological parameters showed that the instantaneous change rate of rainfall over 30 consecutive days was only 2 mm / h (lower than the historical average of 8 mm / h), while the cumulative soil moisture in the irrigated area reached 1200 mm (exceeding the threshold of 800 mm). The system dynamically allocated soil moisture weight to 65% and meteorological weight to 35% using the entropy weight method. Step 202, based on water level sensor data, found that the daily water level fluctuation standard deviation of upstream reservoir P was 0.8 (influence value 78) due to farmland pumping, while downstream reservoir Q had smaller fluctuations due to river replenishment (influence value 52). After determining that the two were adjacent through triangulation, a hydraulic transmission path P→Q was generated (initial strength 26). Step 203 further employed dynamic time warping analysis to reveal the strong synchronicity (correlation coefficient 0.91) and 40-minute lag between the water levels of P and Q during the daily peak irrigation period at 3:00 AM. The connection strength was corrected to 50, and the delayed transmission relationship was marked. Step 204 imports the BeiDou coordinates and the corrected path into the platform, generates a thermal layer showing the deep red high-intensity path from P to Q, and after overlaying the soil moisture distribution, it locks the core area affected by agricultural water use. Based on this, the water resources department diverts 2 million cubic meters of water to Q, successfully alleviating the risk of drying up in the P reservoir area.

[0108] In summary, steps 201 to 204, through dynamic weight allocation, hydraulic transmission path construction and intensity correction, and multi-dimensional data binding, achieve accurate quantification and visualization of cross-reservoir correlations. In this embodiment, the system successfully identified the main risk transmission paths during a typhoon and issued an early warning of the overflow risk of downstream reservoir Y two hours in advance. Compared to static correlation models, this method improves the accuracy of environmental coupling analysis, enhances emergency response efficiency, and provides real-time decision support for integrated watershed management.

[0109] To further improve the temporal sensitivity and dynamic screening accuracy of hydraulic transmission path construction, the rising and falling phases of water level fluctuations are first weighted separately to distinguish the differentiated driving effects of the external environment on water storage and loss. Then, hydraulic influence values ​​are generated through directional superposition to highlight the net water storage trend. The absolute difference between the influence values ​​of adjacent nodes is calculated and compared with a threshold to screen candidate paths with significant transmission potential. Combining the consistency of water level fluctuation direction within the current time window, logically conflicting paths are eliminated and effective transmission relationships are activated. Finally, the activated paths are superimposed with historical preset paths to form a hydraulic transmission network that combines dynamic response and empirical patterns. In some embodiments, step 202 involves superimposing the temporal fluctuation amplitude of the water level measurement parameters with the weight ratio to generate hydraulic influence values ​​characterizing the influence of the external environment on reservoir nodes. Based on the differences between the hydraulic influence values ​​of adjacent reservoir nodes, hydraulic transmission paths between reservoir nodes are constructed, including:

[0110] 301. The rising and falling phases of the time-series fluctuation amplitude of the water level measurement parameter are weighted according to the weight ratio to obtain the weighted rising phase time-series fluctuation amplitude and the falling phase time-series fluctuation amplitude.

[0111] In step 301, the temporal fluctuation amplitude of the rising phase refers to the fluctuation amount of a continuous rising interval in the water level time series curve (e.g., the range of water level changes within a period with a slope > 0). The temporal fluctuation amplitude of the falling phase refers to the fluctuation amount of a continuous falling interval in the water level time series curve (e.g., the range of water level changes within a period with a slope < 0). Weighted processing refers to assigning differentiated weights to the rising / falling phases according to weight ratios. For example, a high meteorological weight strengthens the rising phase (heavy rainfall inflow), while a high soil moisture weight strengthens the falling phase (infiltration loss).

[0112] In this embodiment, the inflection point of the temporal fluctuation amplitude curve of the water level measurement parameters is first detected using the differential method, dividing the complete sequence into alternating rising and falling phases. The fluctuation amplitude (e.g., the difference between the maximum and minimum values ​​within the phase) is calculated for each phase. Then, based on the weighting ratios dynamically allocated in step 201 (e.g., meteorological weight α, soil moisture weight β), the amplitude of the rising phase is multiplied by (α + β × 0.5), and the amplitude of the falling phase is multiplied by (β + α × 0.5), respectively, reflecting the dominant difference in the influence of environmental parameters on water level rise and fall. Finally, the weighted temporal fluctuation amplitudes of the rising and falling phases are output.

[0113] 302. The weighted time-series fluctuation amplitude of the rising phase and the time-series fluctuation amplitude of the falling phase are directionally superimposed to generate a hydraulic impact value characterizing the reservoir node's influence by the external environment.

[0114] In step 302, directional superposition refers to combining the weighted amplitudes of the rising and falling phases according to their physical meaning. For example, positive superposition of rising amplitudes reflects the inflow contribution, while negative superposition of falling amplitudes reflects the loss effect. The hydraulic impact value refers to the scalar value generated by directional superposition, with the positive and negative signs indicating the net water storage trend of the reservoir node (positive for increased water storage, negative for increased loss).

[0115] In this embodiment, the weighted amplitudes of the rising and falling phases output in step 301 are summed. For example, if a reservoir has two rising phases (weighted amplitudes +8m and +5m) and one falling phase (weighted amplitude -6m) within 12 hours, the hydraulic impact value is +7m. This value, using both sign and numerical representation, characterizes the hydraulic impact of the reservoir node on the external environment and is used for subsequent transmission path construction.

[0116] 303. Calculate the absolute difference of the hydraulic influence values ​​of adjacent reservoir nodes, and compare the absolute difference with the preset generation threshold of hydraulic conduction paths. Set the preset hydraulic conduction paths whose absolute difference exceeds the generation threshold as candidate conduction paths.

[0117] In step 303, the absolute difference refers to the absolute difference in hydraulic influence values ​​between adjacent reservoir nodes, used to quantify the difference in conduction potential energy. The generation threshold refers to a preset conduction path triggering threshold (e.g., ΔH≥5m) to filter out low-intensity associations. The candidate conduction path refers to the set of directed edges that satisfy ΔH≥th threshold, initially characterizing the potential direction of hydropower transfer.

[0118] In this embodiment, the adjacency relationship of reservoir nodes is first determined based on a spatial proximity algorithm. For each pair of adjacent reservoir nodes, ΔH = |H1 - H2| is calculated. If ΔH ≥ the generation threshold T (e.g., 4m), a candidate transmission path from the high-H-value node to the low-H-value node is generated, and ΔH is used as the initial connection strength. For example, if reservoir A has H = +10m and adjacent reservoir B has H = +3m, ΔH = 7m ≥ T, and a candidate transmission path A→B is generated with a strength of 7.

[0119] 304. Based on the consistency between the connection strength of the candidate conduction path and the fluctuation direction of the water level measurement parameters within the current time window, the candidate conduction paths that meet the activation conditions are selected to obtain the activated candidate conduction path set.

[0120] In step 304, the consistency of fluctuation direction refers to the matching of water level change trends at both ends of the candidate path within the current time window (e.g., both rising or both falling). The activation condition refers to the logical matching of path strength and fluctuation direction (e.g., upstream rising and downstream rising, or upstream falling and downstream lagging in its decline). The candidate transmission path set refers to the set of potential hydraulic transmission paths initially selected based on certain standards and calculation methods during the analysis of dynamic correlations within the reservoir system and between adjacent reservoir nodes. Connection strength, in the context of hydraulic transmission paths between reservoir nodes, refers to a quantitative indicator measuring the degree of interaction or influence between adjacent reservoir nodes. Water level measurement parameters within the time window refer to the dataset obtained by monitoring and recording reservoir water levels within a specific time period (i.e., the "time window").

[0121] In this embodiment, firstly, the consistency of the direction of water level measurement parameter changes at both ends of the candidate transmission path within the most recent time window (e.g., 3 hours) is extracted (upward movement is marked as +1, downward movement as -1). If the directions are consistent (e.g., A↑ and B↑), the candidate transmission path is retained and the original connection strength is maintained. Then, if the directions are opposite but the downstream lags behind the upstream (the lag time is detected by cross-correlation analysis to be ≤1 hour), the strength is adjusted according to the lag coefficient (e.g., 0.7); otherwise, the candidate transmission path is eliminated. For example, in the candidate path C→D, C recently increased (+1), D decreased (-1), and there is no lag correlation, so it is determined to be an invalid path. Finally, the set of activated candidate transmission paths is obtained.

[0122] 305. The activated candidate conduction path set is superimposed with the preset conduction path in the cross-reservoir dynamic correlation map to obtain the hydraulic conduction path.

[0123] In step 305, the preset conduction path refers to a fixed conduction relationship (such as a natural river channel connection) predefined based on historical hydrological data or human experience. The hydraulic conduction path refers to the effective path for water flow to be transmitted or interact with each other between different reservoir nodes in a reservoir system, determined based on the analysis of water level changes and their influencing factors.

[0124] In this embodiment, a preset conduction path (such as a fixed connection E→F) is first loaded from the cross-reservoir dynamic correlation map. Then, the candidate conduction path set activated in step 304 is superimposed with the preset conduction path. If both a dynamic conduction path and a preset conduction path exist for the same node pair, the dynamic conduction path is retained and its intensity is updated; otherwise, only a single type exists. Finally, the hydraulic conduction path is obtained. For example, if the dynamic path G→H (intensity 9) conflicts with the preset conduction path H→G (intensity 5), only the G→H path is retained in the final map.

[0125] Here is a specific example:

[0126] In an emergency scenario for the B reservoir group in a certain province in response to Typhoon Haikui, reservoir J ​​experienced drastic water level fluctuations due to heavy rainfall caused by the typhoon. In step 301, the system weighted the temporal fluctuation amplitude based on a 90% meteorological weight and a 10% soil moisture weight, calculating a weighted amplitude of 15m for the rising phase (dominated by heavy rainfall) and a weighted amplitude of 2m for the falling phase (with minimal infiltration influence). The adjacent reservoir K, due to lower rainfall intensity, showed an amplitude of 6m for the rising phase and 8m for the falling phase (significantly influenced by soil moisture). In step 302, hydraulic impact values ​​were generated through directional superposition. The net water storage trend of J was +13m (15-2), while the impact value of K due to the loss effect was -2m (6-8). Step 303 calculated the absolute difference between the two, ΔH = 15m (exceeding the threshold by 4m), generating a candidate transmission path J→K and assigning it an initial intensity of 15. Step 304 further verified that within the current time window, the water level of J continued to rise, while K rose synchronously with a 1-hour lag due to the opening of the floodgate. The direction was consistent and the lag was reasonable, thus activating the J→K path. In step 305, the system dynamically covers the preset dry season water diversion path K→J, and prioritizes the high-intensity transmission path J→K. Based on this, the flood control command center initiates emergency flood diversion from J to K, reducing the water level of reservoir J ​​by 1.5 meters within 2 hours, successfully mitigating the risk of dam failure and ensuring downstream safety.

[0127] In summary, steps 301 to 305, through temporal fluctuation directional weighting, dynamic path activation screening, and multi-source path fusion, enable this solution to accurately identify the J→K abnormal transmission path in typhoon emergency scenarios, covering historical countercurrent preset paths. This shortens the flood diversion decision response time and successfully prevents downstream towns from being flooded. Compared to traditional static models, the false alarm rate of dynamic transmission paths is reduced, providing highly reliable topological support for basin-wide coordinated regulation under extreme weather conditions.

[0128] To further improve the dynamics and spatial linkage of the risk assessment of water level imbalance in cross-basin reservoir groups, the flow path nodes in the dynamic map are first matched with water level anomaly intervals to establish a correlation between path intensity and anomaly duration. The spatial distribution density and temporal cumulative effect of abnormal fluctuations are statistically analyzed to quantify the risk diffusion range and persistence. By comparing the spatial distribution range and temporal cumulative effect with preset thresholds, low, medium, and high risk levels are classified. In particular, when both indicators exceed limits simultaneously, a high risk is determined, achieving multi-dimensional risk coupling analysis and providing a basis for tiered management. In some embodiments, step 104, which involves determining the water level imbalance risk level of the cross-basin reservoir group through the mapping relationship between the cross-reservoir dynamic correlation map and the water level anomaly fluctuation intervals, includes:

[0129] 401. Dynamically match the water flow path nodes in the cross-reservoir dynamic association map with the abnormal water level fluctuation interval to determine the correspondence between the connection strength of the water flow path nodes and the duration of the abnormal water level fluctuation interval.

[0130] In step 401, the water flow path node refers to the node in the dynamic correlation map that represents the hydraulic conduction relationship between reservoirs, including connection strength (quantifying conduction capacity) and conduction direction attributes. The abnormal water level fluctuation interval refers to the continuous time period during which the reservoir water level exceeds the historical normal range (e.g., ±2 standard deviations), recording its start and end times and fluctuation amplitude. Dynamic matching refers to aligning the time attributes of the water flow path nodes (e.g., the activation period of the conduction path) with the time window of the abnormal water level interval, establishing a mapping relationship between path strength and abnormal duration.

[0131] In this embodiment, firstly, all active water flow path nodes (e.g., A→B) in the cross-reservoir dynamic correlation map are extracted, and their connection strength and activation time period are recorded. Simultaneously, abnormal water level fluctuation intervals (e.g., fluctuations exceeding the threshold for 3 consecutive hours) are detected from water level monitoring data for each reservoir. Using a time series alignment algorithm (e.g., dynamic time warping), the activation time periods of the path nodes are matched with the abnormal intervals. For each path node, the overlap duration between its activation time period and the abnormal intervals of upstream and downstream nodes is calculated, and a mapping table is established to record the correspondence between path connection strength and overlap duration.

[0132] 402. Based on the aforementioned correspondence, statistically analyze the spatial distribution range and temporal cumulative effect of the abnormal water level fluctuation interval in the cross-reservoir dynamic correlation map.

[0133] In step 402, the spatial distribution range refers to the geographical coverage of abnormal water level fluctuations within the watershed, calculated through the spatial density of abnormal reservoir nodes. The temporal cumulative effect refers to the superposition effect of the persistence and transmission delay of abnormal fluctuations over time, quantified through weighted accumulation of abnormal duration.

[0134] In this embodiment, firstly, taking the abnormal reservoir nodes within the abnormal water level fluctuation range as the center, a kernel density estimation algorithm is used to generate the spatial distribution range, calculate the density of abnormal nodes per unit area (e.g., 8 nodes / 100km²), and combine it with path connectivity strength weighting (density value × average strength). For each abnormal node, the overlap duration of its associated paths is accumulated (e.g., reservoir X is associated with 3 paths, with a total overlap duration of 6 hours), and an exponential decay function is introduced to calculate the time cumulative effect. The spatial distribution range and time cumulative effect are generated, identifying high-risk cluster areas.

[0135] 403. Compare and analyze the spatial distribution range, the time cumulative effect and the preset threshold, and divide the water level imbalance risk level range of the cross-basin reservoir group according to the comparison and analysis results. When the spatial distribution range and the time cumulative effect both exceed the preset threshold, it is determined to be a high-risk level.

[0136] In step 403, the preset thresholds include spatial density thresholds (e.g., 5 per 100 km²) and time accumulation thresholds (e.g., 8 hours), set based on historical disaster data. The risk level ranges are divided into high risk (both exceeding the threshold), medium risk (one exceeding the threshold and lasting >30 minutes), and low risk (one exceeding the threshold and lasting <10 minutes).

[0137] In this embodiment, the spatial distribution range and cumulative temporal effect of the abnormal fluctuation interval are first compared and analyzed with a preset threshold. If the preset threshold indicates a spatial density of 5 or more abnormal nodes per 100 square kilometers, and a cumulative temporal effect of 8 hours or more, it is marked as a high-risk level. Conversely, if only one or both thresholds are not exceeded, the risk level is classified as medium or low based on the specific circumstances.

[0138] Here is a specific example:

[0139] During the response to Typhoon Mangkhut, three out of six reservoirs (X, Y, and Z) in a reservoir group B of a certain province experienced significant abnormal water level fluctuations. Following step 401, two transmission paths were first identified and activated in the cross-reservoir dynamic correlation map: →Y (connection strength 18, activation period 12:00-15:00) and Y→Z (connection strength 12, activation period 13:30-16:00). Further analysis revealed that the abnormal fluctuation period for reservoir X was 12:00-14:30, for reservoir Y it was 13:00-16:00, and for reservoir Z it was 14:00-17:00. Matching results showed that the overlap time between the X→Y path and the abnormal fluctuation period was 2.5 hours, while the overlap time for the Y→Z path was 2 hours. Next, in step 402, the spatial density of the abnormal fluctuation range within the entire reservoir group was calculated to be 8 abnormal nodes per 100 square kilometers (threshold set to 5), and the time cumulative effect reached 10 hours (threshold set to 8 hours). These data indicate that not only is the affected area extensive, but the abnormal fluctuation also lasts for a relatively long time, posing a threat to the stability of the entire system. Finally, in step 403, the above spatial distribution range and time cumulative effect were compared and analyzed with the preset risk level thresholds. Since both indicators exceeded their respective thresholds, the system automatically determined that this event belonged to the high-risk level and triggered the overall flood diversion plan. Accordingly, the water resources management system initiated emergency dispatch measures for reservoirs X, Y, and Z, adjusting their capacity to alleviate flood pressure and ensure the safety of downstream areas. This series of precise and timely response measures effectively mitigated the impact of the disaster and protected the safety of people's lives and property.

[0140] In summary, steps 401 to 403, by combining the mapping relationship between the dynamic correlation map of cross-reservoirs and the abnormal water level fluctuation range, can accurately identify and quantitatively assess areas of water level imbalance risk in cross-basin reservoir groups. This method not only improves the accuracy of water resource management but also enhances the ability to respond to sudden hydrological events, helping to formulate more scientific and reasonable reservoir scheduling strategies and ensuring the safety and stability of regional water resources.

[0141] To further improve the spatiotemporal correlation and dynamic adaptability of water level anomaly fluctuation interval extraction, the time window length is first adaptively adjusted according to the connection density of the water flow path nodes to achieve high-frequency monitoring of high-density areas and global coverage of low-density areas. The segmented fluctuation fragments are dynamically time-normalized and matched with anomaly patterns in a preset feature library to filter out highly similar anomaly fragments. By mapping to water flow path nodes and calculating the correlation influence strength, the anomaly fluctuation interval is defined by combining spatial distribution and cumulative duration, ensuring the spatiotemporal correlation and consistency of the anomaly detection logic. In some embodiments, step 103, which involves spatiotemporally matching the real-time water level signal with a preset water level fluctuation feature library to extract water level anomaly fluctuation intervals associated with the water flow path nodes in the cross-reservoir dynamic correlation map, includes:

[0142] 501. The real-time water level signal is dynamically divided into continuous fluctuation segments according to a preset time window, and the length of the preset time window is adaptively adjusted according to the connection density of the water flow path nodes in the cross-reservoir dynamic correlation map.

[0143] In step 501, the real-time water level signal refers to the time-series data collected in real time by the reservoir water level sensor, with a time resolution down to the minute level. The preset time window refers to the duration of the dynamically segmented data period, initially set to 1 hour, and adaptively adjusted based on the connection density (number of conduction paths per unit area) of the water flow path nodes. A continuous fluctuation segment refers to a continuous sequence of water level changes segmented from the real-time water level signal within a specific time window. Connection density refers to the number of active water flow paths per unit geographical area; higher density indicates more complex conduction relationships, requiring a shorter time window to improve analysis accuracy.

[0144] In this embodiment, firstly, currently active water flow path nodes are extracted from the cross-reservoir dynamic correlation map, and the spatial connectivity density of these nodes is calculated. Spatial connectivity density reflects the number of active water flow paths per unit geographical area; higher density indicates more complex transmission relationships. Based on the spatial connectivity density value, the time window length is dynamically adjusted. If the connectivity density is high (e.g., more than 8 paths per 100 square kilometers), a shorter time window (e.g., 0.5 hours) is set to improve analysis accuracy. If the connectivity density is low, a longer time window (e.g., 1 hour) is used. Next, the real-time water level signal is segmented into continuous fluctuation segments using a sliding window method. For example, for a 24-hour dataset, if a 0.5-hour time window is used for segmentation, 48 segments can be obtained, each containing a 30-minute water level change curve.

[0145] 502. Match the continuous fluctuation segment with the fluctuation pattern in the preset water level fluctuation feature library in a spatiotemporal order. Based on the matching results, filter out the abnormal fluctuation segments in the continuous fluctuation segment whose matching degree with the fluctuation pattern exceeds a preset threshold.

[0146] In step 502, the fluctuation pattern refers to historical abnormal fluctuation patterns (such as sudden rises and gradual falls) stored in the preset feature library, including morphological curves and temporal characteristics. The water level fluctuation feature library is a dataset containing various typical water level change patterns, used to identify and classify real-time monitored reservoir water level changes. Segment-by-segment matching refers to aligning real-time fluctuation segments with the feature library patterns in chronological order and calculating similarity. The matching threshold refers to a preset similarity judgment standard used to filter abnormal segments. An abnormal fluctuation segment refers to a continuous water level change data segment with significant abnormal characteristics identified during reservoir water level monitoring by comparing and analyzing it with typical patterns in the preset water level fluctuation feature library.

[0147] In this embodiment, various typical abnormal fluctuation patterns are read from a preset water level fluctuation feature library as templates. Then, for each continuous fluctuation segment, a dynamic time warping algorithm is used to match it with the abnormal fluctuation pattern template, and the similarity is calculated. Based on the segment-by-segment matching results, it is possible to identify which continuous fluctuation segments are closest to the known abnormal fluctuation patterns. If the minimum distance between a continuous fluctuation segment and the template is lower than a set threshold (e.g., 0.3), the segment is marked as an abnormal fluctuation segment.

[0148] 503. Map the abnormal fluctuation segment to the water flow path node, and calculate the correlation influence intensity of the abnormal fluctuation segment on the adjacent reservoir node through the connection strength of the mapped water flow path node;

[0149] In step 503, the correlation influence intensity refers to the degree of transmission influence of abnormal fluctuations on adjacent reservoirs through water flow path nodes, which is calculated jointly by path connection strength and abnormal amplitude. Abnormal fluctuation segment mapping refers to associating abnormal segments with the water flow path nodes of their respective reservoirs and spreading the influence along the transmission direction.

[0150] In this embodiment, firstly, the abnormal fluctuation segments marked in the previous step are matched with the water flow path nodes in the cross-reservoir dynamic correlation map to identify the specific reservoir locations and upstream and downstream correlation paths of these abnormal fluctuation segments. Then, the correlation influence intensity on each path is calculated according to the formula based on the path connection strength of the mapped water flow path nodes and the abnormal fluctuation amplitude.

[0151] 504. Based on the spatial distribution range and cumulative duration of the correlation influence intensity in the cross-reservoir dynamic correlation map, the abnormal water level fluctuation range associated with the water flow path node is obtained.

[0152] In step 504, the spatial distribution range refers to the geographical area covered by the impact of the abnormal fluctuations, which is quantified by kernel density estimation. The cumulative duration refers to the duration of the abnormal fluctuations' propagation along their impact path, calculated as a weighted total duration. The water level abnormal fluctuation interval refers to a period of time with significant abnormal characteristics identified through monitoring and analysis of the reservoir's water level within a specific time period.

[0153] In this embodiment, firstly, the spatial distribution range of the geographical area affected by the abnormal fluctuations is calculated using a kernel density estimation method to quantify the impact range of the abnormal fluctuations. Next, for each affected water flow path, the duration of its abnormal fluctuations is accumulated, and a weighted cumulative duration is calculated, taking into account the decreasing impact over time. Finally, if both the calculated spatial distribution density and the cumulative duration exceed a preset threshold, the area is determined to be an abnormal water level fluctuation interval.

[0154] Here is a specific example:

[0155] A reservoir group B in a certain province experienced regional torrential rains during the plum rain season, causing a significant rise in the water level of reservoir X. According to step 501, due to the high connection density (12 reservoirs per 100 square kilometers), a 0.5-hour time window was selected, and the real-time data of reservoir X was divided into multiple segments. One segment showed a water level rise of 1.2 meters between 12:00 and 12:30. In step 502, this data was compared with a "sudden rise" pattern, and the result showed that its distance was 0.25, below the set threshold, and therefore it was marked as an anomaly. In step 503, this anomalous fluctuation was mapped to the water flow paths X→Y and X→Z, and the respective impact intensity was calculated. Finally, in step 504, after evaluating spatial density and cumulative duration, an anomalous fluctuation range was identified, and coordinated flood discharge measures were initiated accordingly, effectively reducing the flood risk. This method not only improved the accuracy of early warning but also accelerated the response speed, greatly enhancing flood control capabilities.

[0156] In summary, steps 501 to 504, by performing spatiotemporal matching of real-time water level signals with a pre-defined water level fluctuation feature database, can accurately identify abnormal water level fluctuation ranges in the dynamic correlation map across reservoirs. This provides an efficient method for monitoring and providing early warning of potential water level imbalance risks. This method not only improves the accuracy and timeliness of water resource management but also enhances the response capability to emergencies, contributing to the development of more scientific and rational scheduling strategies and ensuring the safe operation of reservoir groups.

[0157] To further improve the accuracy and spatial transmission of the calculation of the impact intensity of abnormal fluctuations, firstly, water flow path nodes highly correlated with the abnormal segment are screened by calculating the spatiotemporal overlap to ensure the accuracy of the mapping target. The initial impact is calculated based on the connection strength of the target node and the amplitude of the abnormal fluctuation, reflecting the intensity of the direct effect. Furthermore, the initial impact is allocated according to the connection strength ratio between the target node and adjacent nodes, and the associated impact intensity is obtained by accumulating the multi-path contribution values, realizing the spatial diffusion modeling and quantitative evaluation of the abnormal transmission effect. In some embodiments, step 503, which maps the abnormal fluctuation segment to the water flow path node and calculates the associated impact intensity of the abnormal fluctuation segment on adjacent reservoir nodes based on the connection strength of the mapped water flow path node, includes:

[0158] 601. The spatiotemporal location distribution of the abnormal fluctuation segment and the water flow path node are compared and overlapped, and the water flow path node with an overlap ratio exceeding a preset ratio is selected as the target mapping node to which the abnormal fluctuation segment belongs.

[0159] In step 601, the spatiotemporal location distribution refers to the geographical location (latitude and longitude) and activation time period of the water flow path node in the dynamic correlation map. Overlap refers to the percentage of intersection between the spatiotemporal range of the abnormal fluctuation segment (e.g., 12:30-13:30, coordinate range) and the spatiotemporal range of the water flow path node. The target mapping node refers to the water flow path node with an overlap of ≥50% with the abnormal fluctuation segment.

[0160] In this embodiment, firstly, the spatiotemporal range information, including the specific activation time period and geographical coordinate range, is extracted from the abnormal fluctuation segment. Then, this information is compared with the spatiotemporal location distribution of all water flow path nodes in the dynamic association map. The temporal overlap duration and spatial intersection area between each water flow path node and the abnormal fluctuation segment are calculated. Finally, based on the calculated temporal overlap duration and spatial intersection area, the overlap degree between the abnormal fluctuation segment and the water flow path node is evaluated. If the overlap degree reaches or exceeds 50%, the node is marked as a target mapping node.

[0161] 602. Calculate the initial impact of the abnormal fluctuation segment on the target mapping node based on the connection strength of the target mapping node;

[0162] In step 602, the initial influence quantity refers to the direct influence intensity of the abnormal fluctuation segment on the target mapping node, which is calculated jointly by the node connection strength and the abnormal fluctuation amplitude. The connection strength refers to the quantified value of the conduction capacity of the water flow path node.

[0163] In this embodiment, the maximum water level change amplitude (e.g., a water level rise of 1.5 meters within 2 hours) is first extracted from the abnormal fluctuation segment. Then, based on the connection strength of the target mapping node corresponding to the maximum water level change amplitude, the initial influence is calculated. Finally, combined with the connection strength of the target mapping node (e.g., a strength value of 20), the initial influence is calculated by multiplication (20 × 1.5 = 30).

[0164] 603. Based on the connection strength ratio between the target mapping node and the adjacent reservoir nodes, the initial influence amount is allocated to each adjacent reservoir node, and the allocated amounts of the initial influence amount on each adjacent reservoir node are accumulated to obtain the associated influence strength.

[0165] In step 603, the connection strength ratio refers to the proportion of connection strength between the target mapping node and each adjacent node. The allocation amount refers to the value of the abnormal fluctuation impact distributed among different water flow path nodes, calculated according to a specific algorithm or rule. The associated impact strength refers to the cumulative value of the initial impact allocated to adjacent nodes, reflecting the transmission effect of the abnormal fluctuation.

[0166] In this embodiment, the proportion of each path is first calculated based on the connection strength ratio of all adjacent paths of the target mapping node. For example, if the target mapping node has two adjacent paths with strengths of 12 and 8 respectively, the proportions are 60% and 40%. Then, the initial influence (e.g., 30) is proportionally allocated to the adjacent reservoir nodes (30 × 60% = 18, 30 × 40% = 12). Finally, the initial influence received by each adjacent reservoir node from different paths is summed (e.g., if a node has received 15 through other paths, the total intensity is 18 + 15 = 33) to obtain the final associated influence intensity.

[0167] Here is a specific example:

[0168] During a typhoon response period for reservoir group B in a certain province, reservoir P experienced an abnormal water level fluctuation (rising 2 meters) between 12:00 and 13:00. First, the system extracted the spatiotemporal range of this abnormal segment (12:00-13:00, region A) and compared it with the activation period (12:00-14:00) and coverage area (region B) of the water flow path node P→S. After calculating a 1-hour temporal overlap and an 80% spatial intersection, P→S was marked as the target mapping node. Next, based on the connection strength of the P→S node (18) and the abnormal amplitude of 2 meters, the initial impact was calculated to be 36. Then, according to the connection strength ratio (55.6% and 44.4%) of the adjacent nodes S→T (intensity 10) and S→U (intensity 8) of P→S, the initial impact was allocated as follows: S→T received 20 and S→U received 16. Finally, the historical impact of the S→T node was accumulated to 15, and the total correlation strength reached 35, triggering an automatic flood discharge command. The water level of reservoir P was reduced by 1.8 meters within 2 hours, effectively mitigating the risk of dam failure.

[0169] In summary, steps 601 to 603, by mapping anomalous fluctuation segments to water flow path nodes and calculating the initial impact and distribution based on node connectivity, enable this scheme to accurately assess the transmission effect of anomalous fluctuations on the entire reservoir system. This method not only improves the understanding of the impact of anomalous fluctuations but also provides strong support for developing precise emergency response strategies. In typhoon scenarios, the anomalous transmission effect of P→S→T / U was successfully quantified, enabling downstream reservoirs to initiate flood diversion measures ahead of schedule, reducing direct economic losses and improving the accuracy and efficiency of emergency response. Overall, this method significantly enhances water resource management and flood early warning capabilities.

[0170] To further improve the accuracy and coordination of reservoir group control commands, the flood discharge priority is first determined based on the risk level, and the flood discharge ratio is allocated based on the difference between the current reservoir storage capacity and the safety threshold to ensure rapid flood discharge in key areas. Optimized flood discharge paths are generated based on the path connectivity strength and spatial distribution in a dynamic map, balancing efficiency and risk avoidance. Historical flood discharge data from the same period is combined with current needs to dynamically adjust the reservoir safety capacity threshold, and the adjustment ratio is bound to the optimized path, forming a complete control command set for coordinated storage and discharge, achieving global optimized scheduling of cross-reservoir resources. This method achieves coordinated linkage between flood discharge priority, path optimization, and historical data, ensuring the scientific and targeted nature of control measures and improving the emergency response efficiency of cross-basin reservoir groups. In some embodiments, step 105, generating control commands based on the water level imbalance risk level, includes:

[0171] 701. Based on the risk range corresponding to the water level imbalance risk level, determine the flood discharge priority order of each reservoir node, and calculate the flood discharge allocation ratio of each reservoir node according to the difference between the current water storage capacity and the preset safety capacity of each reservoir node in the flood discharge priority order.

[0172] In step 701, the flood discharge priority order refers to the ranking of reservoirs according to their urgency of flood discharge based on risk level, with high-risk reservoirs discharging floodwater first. Risk intervals are typically used to describe the degree of risk of reservoir water level imbalance, and corresponding control measures are formulated accordingly. The difference between the preset safe capacity and the current water storage capacity of the reservoir refers to the difference between the current water storage capacity and the preset safe water storage capacity; the larger the difference, the greater the flood discharge required. The flood discharge allocation ratio refers to the proportion of flood discharge allocated to each reservoir according to priority and capacity difference.

[0173] In this embodiment, risk zones are first divided according to the risk level of water level imbalance (e.g., high risk corresponds to emergency flood discharge, while low risk corresponds to slow discharge). Then, the difference between the current water storage capacity and the preset safety capacity of each reservoir node is calculated; a larger difference indicates a greater flood discharge volume is required. Next, based on the flood discharge priority order, a higher proportion of flood discharge is allocated to high-risk nodes. For example, if Reservoir X has a high risk level and a capacity difference of 1 million cubic meters, while Reservoir Y has a medium risk level and a difference of 500,000 cubic meters, then the flood discharge proportion for X is higher than that for Y. Finally, the flood discharge allocation proportion for each reservoir node is obtained.

[0174] 702. Based on the connection strength and spatial distribution of the water flow path nodes in the cross-reservoir dynamic correlation map, generate an optimized flood discharge path scheme;

[0175] In step 702, the flood discharge path optimization scheme refers to selecting the flood discharge route with the highest conduction efficiency based on the intensity and spatial distribution of water flow paths in the dynamic correlation map. Connection strength refers to the quantified value of the conduction capacity of path nodes; the higher the strength, the higher the flood discharge efficiency.

[0176] In this embodiment, all possible flood discharge path nodes and their connection strengths are first extracted from the cross-reservoir dynamic correlation map. Then, based on the spatial distribution data, high-risk areas or congested paths are excluded. For example, if a path has low connection strength and passes through a geologically vulnerable area, it is excluded. Next, graph theory algorithms (such as shortest path or maximum flow algorithms) are used to optimize path selection, prioritizing paths with high connection strength and reasonable geographical distribution. Finally, an optimized flood discharge path scheme is generated.

[0177] 703. Based on the matching relationship between the flood discharge allocation ratio and the historical flood discharge records for the same period, and combined with the water level imbalance risk level, generate a water storage capacity adjustment ratio, associate and bind the water storage capacity adjustment ratio with the flood discharge path optimization scheme, and generate control instructions.

[0178] In step 703, the water storage capacity adjustment ratio refers to dynamically adjusting the reservoir's water storage safety threshold based on the matching results of historical flood discharge records and the current risk level. Association binding refers to linking the flood discharge path with the adjusted water storage capacity to ensure the synergy of control commands. The flood discharge path optimization scheme refers to selecting the optimal flood discharge path and flow allocation strategy in water resource management and flood early warning systems to effectively address the risks caused by excessively high reservoir water levels, by analyzing the connection strength, spatial distribution, and other relevant factors of water flow path nodes. Control commands are specific operational orders issued to the reservoir system (including a single reservoir or a group of reservoirs) to guide on-site management personnel in carrying out corresponding adjustment and control actions.

[0179] In this embodiment, the current flood discharge allocation ratio is first matched with historical flood discharge records for the same period to analyze the flood discharge effect under similar risk levels. For example, if historical data shows that a reservoir can effectively lower the water level by discharging 50% of its water volume during high-risk periods, a similar ratio is adopted. Then, the ratio is adjusted based on the risk level of water level imbalance, such as increasing the flood discharge ratio during high-risk periods. Finally, the water storage capacity adjustment ratio is bound to the path optimization scheme to generate control instructions (such as "Reservoir X discharges 30% of its water volume via path X→Y→Z, with a flow rate limit of 100 cubic meters per second"), ensuring that the flood discharge volume and path scheme are executed in a coordinated manner.

[0180] Here is a specific example:

[0181] During Typhoon Haiyan, three reservoirs in a certain province's B reservoir group, namely Reservoirs X, Y, and Z, were identified as high-risk. In step 701, based on the risk level ranking (X > Y > Z) and the difference between the current water storage capacity and the safe capacity of each reservoir (X exceeds the limit by 8 million cubic meters, Y by 6 million cubic meters, and Z by 2 million cubic meters), the flood discharge ratio was allocated according to priority: X accounted for 50% of the total discharge, Y for 30%, and Z for 20%. In step 702, based on the cross-reservoir dynamic correlation map, high-connection-strength paths were extracted (X→A strength 25, Y→B strength 18, Z→C strength 10). Combined with spatial distribution analysis, X→A (downstream is a flood discharge area, efficiency 90%) and Y→B (avoiding residential areas, efficiency 85%) were prioritized as flood discharge paths. Step 703 involves matching historical high-risk data from the same period and lowering the safety capacity of X and Y by 15%, ultimately generating control instructions: X will release 4 million cubic meters of floodwater (8 million × 50%) through the X→A route at a 50% ratio, and Y will release 1.8 million cubic meters of floodwater (6 million × 30%) through the Y→B route, ensuring that the flood discharge volume and route plan are accurately linked, and achieving rapid risk response and efficient resource allocation.

[0182] In summary, steps 701 to 703, through multi-level coordinated flood discharge priority calculation, path optimization, and historical data matching, significantly improve the efficiency of water level imbalance control in cross-basin reservoir groups. In typhoon scenarios, the system accurately allocates flood discharge volume and selects the optimal path, avoiding downstream flooding risks while reducing resource waste. Compared to traditional methods, the flood discharge response time is shortened, the risk level is reduced by one level, effectively ensuring the safety of the reservoir group and downstream areas, and providing intelligent decision support for flood control and disaster reduction.

[0183] Figure 2 This application provides a schematic diagram of the structure of an artificial intelligence-based water level monitoring system for water conservancy projects, as shown in the embodiments. Figure 2 As shown, the system includes:

[0184] The acquisition module 21 acquires meteorological parameters, soil moisture parameters, and water level measurement parameters of the cross-basin reservoir group. The water level measurement parameters are generated by magnetic levitation displacement sensing and have a linear mapping relationship with water level changes.

[0185] The encoding module 22 obtains the correlation strength coefficient reflecting the hydraulic interaction between reservoirs based on the dynamic change law of the meteorological parameters and soil moisture parameters, and jointly encodes the correlation strength coefficient with the water level measurement parameters to generate a cross-reservoir dynamic correlation map.

[0186] Extraction module 23 synchronously collects real-time water level signals from each reservoir, performs spatiotemporal matching of the real-time water level signals with a preset water level fluctuation feature library, and extracts the abnormal water level fluctuation range associated with the water flow path nodes in the cross-reservoir dynamic association map.

[0187] The mapping module 24 determines the water level imbalance risk level of the cross-basin reservoir group by mapping the cross-reservoir dynamic correlation map with the abnormal water level fluctuation range.

[0188] The generation module 25 generates control instructions based on the water level imbalance risk level. The control instructions include a flood discharge path optimization scheme and a water storage capacity adjustment ratio.

[0189] Figure 2 The aforementioned artificial intelligence-based water level monitoring system for water conservancy projects can perform... Figure 1 The implementation principle and technical effects of the artificial intelligence-based water level monitoring method for water conservancy projects described in the illustrated embodiment will not be repeated here. The specific methods by which each module and unit of the artificial intelligence-based water level monitoring system for water conservancy projects in the above embodiments are described in detail in the embodiments related to this method, and will not be elaborated upon here.

[0190] In one possible design, Figure 2 The water level monitoring system for hydraulic engineering based on artificial intelligence, as shown in the embodiment, can be implemented as a computing device, such as... Figure 3 As shown, the computing device may include a storage component 31 and a processing component 32;

[0191] The storage component 31 stores one or more computer instructions, wherein the one or more computer instructions are invoked and executed by the processing component 32.

[0192] The processing component 32 is used for the above Figure 1 The embodiment describes an artificial intelligence-based water level monitoring method for water conservancy projects.

[0193] The processing component 32 may include one or more processors to execute computer instructions to complete all or part of the steps in the above-described method. Alternatively, the processing component may be implemented as one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field-programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to perform the above-described method.

[0194] Storage component 31 is configured to store various types of data to support operations at the terminal. The storage component can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0195] Of course, computing devices may also include other components, such as input / output interfaces, display components, communication components, etc.

[0196] Input / output interfaces provide interfaces between processing components and peripheral interface modules, which can be output devices, input devices, etc.

[0197] The communication components are configured to facilitate wired or wireless communication between computing devices and other devices.

[0198] The computing device can be a physical device or an elastic computing host provided by a cloud computing platform. In this case, the computing device can refer to a cloud server, and the aforementioned processing components, storage components, etc., can be basic server resources rented or purchased from the cloud computing platform.

[0199] This application also provides a computer storage medium storing a computer program, which, when executed by a computer, can perform the above-described functions. Figure 1 The embodiment shown is a water level monitoring method for water conservancy projects based on artificial intelligence.

[0200] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0201] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0202] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0203] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A method for monitoring water levels in hydraulic engineering based on artificial intelligence, characterized in that, include: Meteorological parameters, soil moisture parameters, and water level measurement parameters of a cross-basin reservoir group are obtained. The water level measurement parameters are generated by magnetic levitation displacement sensing and have a linear mapping relationship with water level changes. Based on the dynamic variation patterns of the meteorological and soil moisture parameters, a correlation strength coefficient reflecting the hydraulic interaction between reservoirs is obtained. This correlation strength coefficient is then jointly encoded with the water level measurement parameters to generate a cross-reservoir dynamic correlation map. This process includes: dynamically allocating the weight ratios of the meteorological and soil moisture parameters in the correlation strength coefficient according to the instantaneous change rate of the meteorological parameters and the cumulative amount of the soil moisture parameters; superimposing the temporal fluctuation amplitude of the water level measurement parameters with the weight ratios to generate a hydraulic influence value characterizing the influence of the external environment on reservoir nodes; constructing hydraulic transmission paths between reservoir nodes based on the differences in hydraulic influence values ​​between adjacent reservoir nodes; dynamically correcting the connection strength of the hydraulic transmission path based on the synchronous change trend of the water level measurement parameters in the hydraulic transmission path to obtain the corrected connection strength; and binding the geographical coordinates of the reservoir nodes, the hydraulic transmission path, and the corrected connection strength to generate a cross-reservoir dynamic correlation map. Real-time water level signals from each reservoir are collected synchronously, and the real-time water level signals are spatiotemporally matched with a preset water level fluctuation feature library to extract abnormal water level fluctuation ranges associated with water flow path nodes in the cross-reservoir dynamic association map. The risk level of water level imbalance of the cross-basin reservoir group is determined by the mapping relationship between the cross-reservoir dynamic correlation map and the abnormal water level fluctuation range. Based on the water level imbalance risk level, control instructions are generated, which include flood discharge path optimization schemes and water storage capacity adjustment ratios. The step of superimposing the temporal fluctuation amplitude of the water level measurement parameters with the weight ratio to generate a hydraulic influence value characterizing the influence of the external environment on the reservoir nodes, and constructing a hydraulic transmission path between reservoir nodes based on the difference between the hydraulic influence values ​​of adjacent reservoir nodes, includes: The rising and falling phases of the time-series fluctuation amplitude of the water level measurement parameter are weighted according to the weight ratio to obtain the weighted rising phase time-series fluctuation amplitude and falling phase time-series fluctuation amplitude. The weighted time-series fluctuation amplitude of the rising phase and the time-series fluctuation amplitude of the falling phase are directionally superimposed to generate a hydraulic impact value characterizing the reservoir node's influence by the external environment. Calculate the absolute difference of the hydraulic influence values ​​of adjacent reservoir nodes, compare the absolute difference with a preset generation threshold for hydraulic conduction paths, and set the preset hydraulic conduction paths whose absolute difference exceeds the generation threshold as candidate conduction paths; Based on the consistency between the connection strength of the candidate conduction path and the fluctuation direction of the water level measurement parameters within the current time window, the candidate conduction paths that meet the activation conditions are selected to obtain the activated candidate conduction path set. The activated candidate conduction path set is superimposed with the preset conduction path in the cross-reservoir dynamic correlation map to obtain the hydraulic conduction path.

2. The method according to claim 1, characterized in that, The process of determining the water level imbalance risk level of the cross-basin reservoir group by mapping the cross-reservoir dynamic correlation map to the abnormal water level fluctuation range includes: The water flow path nodes in the cross-reservoir dynamic correlation map are dynamically matched with the water level anomaly fluctuation interval to determine the correspondence between the connection strength of the water flow path nodes and the duration of the water level anomaly fluctuation interval. Based on the correspondence, the spatial distribution range and temporal cumulative effect of the abnormal water level fluctuation interval in the cross-reservoir dynamic correlation map are statistically analyzed. The spatial distribution range, the cumulative effect over time, and the preset threshold are compared and analyzed. Based on the comparison and analysis results, the risk level range of water level imbalance of the cross-basin reservoir group is divided. When the spatial distribution range and the cumulative effect over time both exceed the preset threshold, it is determined to be a high-risk level.

3. The method according to claim 1, characterized in that, The step of performing spatiotemporal matching between the real-time water level signal and a preset water level fluctuation feature database to extract abnormal water level fluctuation intervals associated with water flow path nodes in the cross-reservoir dynamic correlation map includes: The real-time water level signal is dynamically divided into continuous fluctuation segments according to a preset time window. The length of the preset time window is adaptively adjusted according to the connection density of the water flow path nodes in the cross-reservoir dynamic correlation map. The continuous fluctuation segment is matched with the fluctuation pattern in the preset water level fluctuation feature library in a spatiotemporal order. Based on the matching results, abnormal fluctuation segments in the continuous fluctuation segment whose matching degree with the fluctuation pattern exceeds a preset threshold are selected. The abnormal fluctuation segment is mapped to the water flow path node, and the correlation influence intensity of the abnormal fluctuation segment on the adjacent reservoir node is calculated by the connection strength of the mapped water flow path node. Based on the spatial distribution range and cumulative duration of the correlation influence intensity in the cross-reservoir dynamic correlation map, the water level anomaly fluctuation range associated with the water flow path node is obtained.

4. The method according to claim 3, characterized in that, The step of mapping the abnormal fluctuation segment to the water flow path node and calculating the correlation influence strength of the abnormal fluctuation segment on adjacent reservoir nodes based on the connection strength of the mapped water flow path nodes includes: The spatiotemporal location distribution of the abnormal fluctuation segment and the water flow path node are compared and overlapped, and the water flow path node with an overlap ratio exceeding a preset ratio is selected as the target mapping node to which the abnormal fluctuation segment belongs. Based on the connection strength of the target mapping node, calculate the initial impact of the abnormal fluctuation segment on the target mapping node; Based on the connection strength ratio between the target mapping node and the adjacent reservoir nodes, the initial influence amount is allocated to each adjacent reservoir node, and the allocated amounts of the initial influence amount on each adjacent reservoir node are accumulated to obtain the associated influence strength.

5. The method according to claim 1, characterized in that, The process of generating control instructions based on the water level imbalance risk level includes: Based on the risk range corresponding to the water level imbalance risk level, the flood discharge priority order of each reservoir node is determined, and the flood discharge allocation ratio of each reservoir node is calculated according to the difference between the current water storage capacity and the preset safety capacity of each reservoir node in the flood discharge priority order. Based on the connection strength and spatial distribution of water flow path nodes in the cross-reservoir dynamic correlation map, an optimized flood discharge path scheme is generated. Based on the matching relationship between the flood discharge allocation ratio and the historical flood discharge records for the same period, and combined with the water level imbalance risk level, a water storage capacity adjustment ratio is generated. The water storage capacity adjustment ratio is then linked and bound to the flood discharge path optimization scheme to generate control instructions.

6. An artificial intelligence-based water level monitoring system for water conservancy projects, applied to the artificial intelligence-based water level monitoring method for water conservancy projects according to any one of claims 1-5, characterized in that, include: The acquisition module acquires meteorological parameters, soil moisture parameters, and water level measurement parameters of the cross-basin reservoir group. The water level measurement parameters are generated by magnetic levitation displacement sensing and have a linear mapping relationship with water level changes. The coding module obtains the correlation strength coefficient reflecting the hydraulic interaction between reservoirs based on the dynamic change patterns of the meteorological parameters and soil moisture parameters. The correlation strength coefficient is then jointly coded with the water level measurement parameters to generate a dynamic correlation map across reservoirs. The extraction module synchronously collects real-time water level signals from each reservoir, performs spatiotemporal matching of the real-time water level signals with a preset water level fluctuation feature library, and extracts the abnormal water level fluctuation range associated with the water flow path nodes in the cross-reservoir dynamic association map. The mapping module determines the water level imbalance risk level of the cross-basin reservoir group by mapping the cross-reservoir dynamic correlation map with the abnormal water level fluctuation range. The generation module generates control instructions based on the water level imbalance risk level. The control instructions include a flood discharge path optimization scheme and a water storage capacity adjustment ratio.

7. A computing device, characterized in that, It includes a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are invoked and executed by the processing component to implement the artificial intelligence-based water level monitoring method for water conservancy projects as described in any one of claims 1 to 5.

8. A computer storage medium, characterized in that, The device contains a computer program that, when executed by a computer, implements an artificial intelligence-based water level monitoring method for water conservancy projects as described in any one of claims 1 to 5.

Images