Water resource fine management method and system integrated with water behavior identification

Abstract

The application discloses a water resource fine management method and system integrated with water behavior identification, relates to the technical field of water resource management, and comprises the following steps: calling water use event records, mining space-time water use characteristics to construct a dynamic water image; reconstructing a fog computing topology according to resource request conditions, establishing a resource field to determine a resource scene; deploying a self-defined scheduler according to the dynamic water image, executing a game to determine a water use strategy; and standardizing the water use strategy reconstruction, executing scheduling and control management according to a water resource management system. The application solves the technical problem that the traditional water resource management mode does not fully utilize big data processing technology and is difficult to cope with complex water use scenes, achieves the technical effects of utilizing big data processing technology, accurately identifying various water use behaviors, realizing resource real-time dynamic scheduling, efficiently integrating and analyzing multi-region water use data, improving the water resource fine management level, and effectively coping with complex water use scenes.

Description

Technical Field

[0001] This invention relates to the field of water resource management technology, and in particular to a refined water resource management method and system that integrates water use behavior recognition. Background Technology

[0002] In the field of water resource management, accurate analysis of water use behavior and efficient resource allocation are crucial. With the increasing demand for refined water resource management, traditional management methods have revealed many shortcomings.

[0003] Current water resource management relies heavily on simple rules or human experience, failing to fully utilize big data processing technologies. Traditional methods struggle to deeply mine and analyze massive amounts of water usage data, and cannot accurately identify complex water use behaviors such as regular, irregular, and flexible water use. Furthermore, the lack of dynamic resource scheduling models based on big data processing prevents real-time adjustments to resource allocation based on spatiotemporal changes in water use behavior. Moreover, in mixed water use environments across multiple regions and scenarios, traditional methods cannot leverage big data to comprehensively integrate and analyze water use data from different areas, resulting in low resource scheduling efficiency and failing to meet the demands of modern, refined water resource management. Summary of the Invention

[0004] This application provides a refined water resource management method and system that integrates water use behavior recognition, which is used to solve the technical problem that traditional water resource management methods do not make full use of big data processing technology and are difficult to cope with complex water use scenarios.

[0005] The first aspect of this application provides a method for refined water resource management integrating water use behavior recognition. The method includes: retrieving water use event records from a water use area, mining spatiotemporal water use characteristics and deconstructing the behavior to construct a dynamic water use profile, wherein regularity, irregularity, and elastic scaling are used as construction constraints; establishing a resource field based on the water use area and performing dynamic resource isolation and elastic scaling constraints based on resource request conditions and reconstructing the cloud-based fog computing topology to determine resource scenarios; deploying a custom scheduler in the cloud based on the dynamic water use profile, performing a first-level partitioned game and a second-level refined game based on the resource scenarios to determine water use strategies; standardizing and reconstructing the water use strategies, and performing resource scheduling and infrastructure control management of the water use area according to the water resource management system.

[0006] The second aspect of this application provides a refined water resource management system integrating water use behavior recognition. The system includes: a water use profile construction module, used to retrieve water use event records of the water use area, mine spatiotemporal water use characteristics and perform behavioral deconstruction to construct a dynamic water use profile, wherein regularity, irregularity, and elastic scaling are used as construction constraints; a resource scenario determination module, used to establish a resource field based on the water use area and perform dynamic resource isolation and elastic scaling constraints based on resource request conditions and reconstructing the cloud-based fog computing topology to determine the resource scenario; a water use strategy determination module, used to deploy a custom scheduler in the cloud based on the dynamic water use profile, and perform a first-level partitioned game and a second-level refined game based on the resource scenario to determine the water use strategy; and a water use strategy reconstruction module, used to standardize and reconstruct the water use strategy, and perform resource scheduling and infrastructure control management of the water use area according to the water resource management system.

[0007] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0008] This application retrieves water usage event records from water-using areas, mines spatiotemporal water usage characteristics, and deconstructs behavioral patterns to construct a dynamic water usage profile constrained by regularity, irregularity, and elastic scaling. Based on resource request conditions, it reconstructs the cloud-based fog computing topology, establishing a resource field encompassing spatial three dimensions, temporal dimensions, and entropy dimensions. This enables dynamic resource isolation and elastic scaling constraints to determine resource scenarios. A custom scheduler is deployed based on the dynamic water usage profile, executing a first-level partitioned game and a second-level refined game to determine water usage strategies. After standardizing and reconstructing the water usage strategies, the water resource management system executes resource scheduling and infrastructure control management, forming a closed loop through synchronous monitoring, anomaly tracing, and feedback adjustment. This solution achieves refined analysis of water usage behavior and spatiotemporal collaborative resource scheduling, improving water resource allocation efficiency and system adaptability. It utilizes big data processing technology to accurately identify various water usage behaviors, achieve real-time dynamic resource scheduling, efficiently integrate and analyze multi-regional water usage data, improve the level of refined water resource management, and effectively address complex water usage scenarios. Attached Figure Description

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

[0010] Figure 1 This is a flowchart illustrating the water resource management method with integrated water use behavior recognition provided in this application embodiment.

[0011] Figure 2This is a schematic diagram of the structure of the integrated water use behavior recognition water resource refinement management system provided in the embodiments of this application.

[0012] Figure labeling: Water usage profile construction module 1, resource scenario determination module 2, water usage strategy determination module 3, water usage strategy reconstruction module 4. Detailed Implementation

[0013] This application provides a refined water resource management method and system that integrates water use behavior recognition, which is used to solve the technical problem that traditional water resource management methods do not make full use of big data processing technology and are difficult to cope with complex water use scenarios.

[0014] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0015] It should be noted that the terms "first," "second," etc., in the specification and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or modules not explicitly listed or inherent to such processes, methods, products, or devices.

[0016] Example 1, as Figure 1 As shown, a water resource management method integrating water use behavior recognition is described, wherein the method includes:

[0017] Step A100: Retrieve water use event records for the water use area, mine spatiotemporal water use characteristics and deconstruct behavior to construct a dynamic water use profile, with regularity, irregularity, and elasticity as constraints for construction.

[0018] In this embodiment, water use event records refer to digital records of water-related events occurring within a water use area, including information such as time dimensions (e.g., specific time periods and periodic characteristics of water use), spatial dimensions (e.g., water use locations and regional distribution), and water use values ​​(e.g., quantitative data such as flow rate and frequency). Spatiotemporal water use characteristics are spatial and temporal features mined by periodically clustering water use event records, traversing each event class, and mapping between classes at a preset frequency. Dynamic water use profiling uses regularity, irregularity, and flexible scales as constraints to deconstruct the water use feature structure in three spatial dimensions and time dimensions, establishing a dynamic model of its time series mapping.

[0019] Specifically, after retrieving water use event records of the water use area, those skilled in the art can perform periodic clustering of the water use event records to determine N event classes, traverse the N event classes at a preset frequency to mine the spatiotemporal water use characteristics under inter-class periodic mapping to determine the water use characteristic structure, and then construct a dynamic water use profile through behavioral deconstruction analysis. The specific steps are explained in detail in A110-A130.

[0020] Step A200: Based on the resource request conditions and reconstructing the fog computing topology in the cloud, establish a resource field based on the water usage area and perform dynamic resource isolation and elastic scaling constraints to determine the resource scenario.

[0021] In this embodiment, resource request conditions refer to the water resource demand conditions of the water-using area, such as water consumption, water usage time period, and water usage type. Fog computing topology refers to a cloud-based distributed computing architecture that reconstructs the node connection relationships after dynamically partitioning the water-using area, i.e., reconstructing fog nodes.

[0022] Optionally, firstly, the water usage area is dynamically partitioned according to the resource request conditions to determine the water usage partition; then, the fog computing topology is reconstructed based on the water usage partition to determine the reconstructed fog nodes. The specific steps are explained in detail in A210-A220.

[0023] Next, the resource field includes three spatial dimensions, a time dimension, and an entropy dimension. Based on the water use zoning, the resource field is dynamically isolated to determine the isolated resource field. Then, the isolated resource field is constrained by the elastic scale to determine the requested resource field, and the association between the reconstructed fog node and the requested resource field is established to determine the resource scenario. The specific steps are explained in detail in A230-A260.

[0024] Step A300: Deploy a custom scheduler in the cloud based on the dynamic water usage profile, and perform a first-level partition game and a second-level refined game based on the resource scenario to determine the water usage strategy.

[0025] In this embodiment, the custom scheduler is built based on the dynamic water usage profile and performs sample-driven supervised training under resource scenario switching. It can be divided into a logically related upper-level scheduler and an autonomous scheduler and deployed in the cloud.

[0026] In one embodiment of this application, firstly, a custom scheduler is constructed by performing sample-driven supervised training based on the dynamic water profile as the underlying basis. After splitting it into a host scheduler and an autonomous scheduler, it is deployed in the cloud, and the association between the autonomous scheduler and the fog computing topology is established. The specific steps are described in detail in A310-A330.

[0027] Next, the total amount of water resources based on the time slice is determined as the resource pool. The upper-level scheduler is then triggered to perform resource game between water use zones to determine the zone resource pool. The autonomous scheduler is then triggered to perform refined water use game within the zone to determine the water use strategy. The specific steps are explained in detail in A340-A360.

[0028] Step A400: Standardize and reconstruct the water use strategy, and implement resource scheduling and infrastructure control management of the water use area according to the water resource management system.

[0029] In this embodiment of the application, the water resource management system is a system used to perform resource scheduling and infrastructure control management in water use areas, and can identify and respond to standard water use strategies after conversion according to their standardized dimensions.

[0030] Specifically, the water use strategy is determined by aggregation and splicing based on the water use area. The strategy is converted into a standard water use strategy according to the standardized dimensions of the water resources management system. The system then identifies and responds to the strategy. The specific steps are detailed in A410-A430.

[0031] Furthermore, step A100 in the method provided in this application embodiment includes:

[0032] A110: Perform periodic clustering on the water usage event records to determine N event classes, where N is the number of periodic cycles.

[0033] A120: Traverse the N event classes, perform spatiotemporal water use feature mining under inter-class periodic mapping, and determine the water use feature structure, wherein a preset frequency is used as the mining constraint.

[0034] A130: Based on the aforementioned water usage characteristic structure, a dynamic water usage profile is constructed through behavioral deconstruction analysis.

[0035] In this embodiment, periodic clustering refers to clustering water use event records according to time periods, dividing water use events into N event classes to extract the periodic patterns of water use events. Inter-class periodic mapping refers to comparing and analyzing the spatiotemporal features of the same event class within different periods at a preset frequency when traversing the N event classes, and mining the differences and correlations in spatiotemporal water use features between classes (between different periods).

[0036] Specifically, firstly, water usage event records are periodically clustered to determine N event categories. In practice, based on the timestamp characteristics of historical water usage data, the k-means clustering algorithm is used to divide water usage events into N categories according to natural cycles such as daily, weekly, and monthly. The algorithm process is as follows:

[0037] First, based on the timestamp features of water usage event records, such as specific dates and times, periodic attributes in the time dimension are extracted. For example, if the period is daily, the number of hours per day is extracted; if the period is weekly, the combination of day of the week and hour is extracted. This is then transformed into a numerical feature vector, such as a two-dimensional vector [day of the week, hour]. Next, the k-means algorithm is used, setting the number of clusters k to the number of periodic cycles N (e.g., N=14 in a bi-weekly model), and randomly initializing N centroids to represent N potential time period categories. The algorithm calculates the Euclidean distance between the feature vector of each water usage event and each centroid, assigning the event to the category of the nearest centroid, and updating the centroid position based on the mean of the feature vectors of all events within the category. This assignment and update process is repeated iteratively until the centroid position no longer changes significantly or the preset number of iterations is reached. Finally, water usage events are divided into N event categories, each corresponding to a specific water usage pattern within a specific time period, such as weekday morning peak hours and weekend midday lulls (i.e., a stable state where water demand is neither peak nor trough), thus achieving automated extraction and grouping of periodic patterns in water usage events.

[0038] Next, the N event classes are traversed at a preset frequency (e.g., every 15 minutes) to mine spatiotemporal water use characteristics under inter-class periodic mapping. Spatially, the water use area is subdivided into several spatial units using GIS grid partitioning technology, such as workshop A and workshop B in a factory area. Then, data such as water flow and equipment start-up / shutdown times for each unit within the corresponding event class time interval are statistically analyzed. Temporally, the temporal fluctuations of the same event class within adjacent periods are analyzed, such as the difference in water use between Monday of this week and Monday of last week, forming a three-dimensional feature matrix containing spatial location, time slice, and water use value. For example, after mining, it was found that in the event class of 9-11 am on weekdays, the average water flow of workshop A is 20 m³ / h, and the data fluctuation for the same time period for three consecutive weeks is less than 5%, confirming that this time period represents a stable high-frequency water use characteristic structure.

[0039] Finally, the water use characteristic structure is deconstructed using spatial three-dimensional and temporal dimensions to determine regular, irregular, and elastic structures. Based on the water use characteristic structure, a water use profile of the assembly is constructed. Using the three structures, three sub-level water use profiles are constructed (the elastic structure is determined based on the spatiotemporal tensor of water use events). A time-series mapping between the assembly and sub-level water use profiles is established as a dynamic water use profile. The specific steps are explained in detail in A131-A133.

[0040] Through a progressive processing flow of periodic clustering for event classification, high-frequency spatiotemporal feature mining, and multi-dimensional behavior deconstruction, the raw water use data is transformed into a structured and dynamic digital model, enabling accurate identification of regular / irregular / flexible water use behaviors and providing a refined underlying basis for subsequent resource field construction and game-theoretic scheduling.

[0041] Furthermore, step A100 in the method provided in this application embodiment includes:

[0042] A131: The water-use characteristic structure is deconstructed using spatial three-dimensional and temporal dimensions to determine regular, irregular, and elastic structures.

[0043] A132: Construct an assembly water usage profile based on the water usage characteristic structure, and construct a sub-level water usage profile based on the regular structure, irregular structure and elastic structure, wherein the sub-level water usage profile includes three items, and the elastic structure is determined based on the spatiotemporal tensor of water usage events.

[0044] A133: Establish a time-series mapping between the water usage profile of the assembly and the water usage profile of the sub-stage, as the dynamic water usage profile.

[0045] Optionally, firstly, using spatial three-dimensional X, Y, Z coordinates or regional grid encoding combined with a time dimension (such as minute-level time intervals, daily / weekly / monthly cycles) as the analysis framework, the water use characteristic structure is deconstructed in multiple dimensions. Utilizing the aforementioned spatial grid partitioning technique, such as dividing the water use area into 100m × 100m grid cells, and combining it with a time series decomposition algorithm (such as STL seasonal decomposition), the water use characteristic structure is decomposed into:

[0046] Regular structure: A water usage pattern that repeats within a fixed spatiotemporal unit, such as the average water flow rate of each unit in a residential community from 7:00 to 9:00 every day being 5 m³ / h, with a fluctuation coefficient of <8% for 30 consecutive days; Irregular structure: Sudden water usage events that deviate from the regular pattern, such as a sudden increase in flow rate to 3 times the average level during a certain period due to equipment cleaning in a factory, with no historical records for the same period; Flexible structure: Based on the spatiotemporal tensor, i.e., the rate of change of water consumption with respect to spatiotemporal variables, such as a flexible water usage pattern determined by ΔQ / ΔT (temperature change) and ΔQ / ΔS (spatial distance), such as the seasonal fluctuation characteristic of water consumption increasing by 1.5% in a certain area for every 1°C increase in summer temperature.

[0047] Secondly, a two-layer water usage profile system is constructed based on the deconstructed feature structure. The assembly water usage profile modeling process is as follows:

[0048] The system's water consumption profile constructs a global feature model through spatiotemporal data aggregation algorithms. First, the spatiotemporal data of the water-consuming area is standardized, using spatial grids (e.g., 100m×100m units) and time slices (e.g., 1 hour / slice) as basic units, summarizing statistics such as total water consumption and average flow within each unit. Time series analysis (e.g., moving average, seasonal decomposition) extracts the overall regional water consumption trend, generating fluctuation curves at monthly / weekly / daily scales. For example, fitting a polynomial function identifies the increasing or decreasing trend of water consumption over time. In the spatial dimension, spatial interpolation techniques (e.g., Kriging interpolation) convert discrete water consumption data into a continuous thermal distribution layer, marking high water-consuming areas (e.g., industrial plants, commercial centers) and their spatiotemporal distribution characteristics. Finally, the system profile is presented in the form of a spatiotemporal matrix and statistical charts, intuitively reflecting the global water consumption pattern, such as the seasonal trend of a park's monthly water consumption being 20% ​​higher in summer than in winter, or the diurnal difference characteristic of a region where nighttime water consumption accounts for less than 15% of daytime water consumption.

[0049] In the process of sub-level water use profiling, sub-level water use profiling is modeled for three types of deconstruction features respectively:

[0050] Regular structural profiling: Based on the results of periodic clustering, such as N=14 event categories, the spatiotemporal data within each periodic category is templated and modeled. For example, for the weekday morning rush hour event category, the mean and standard deviation of water flow in each spatial unit during that time period are calculated to form a standardized water flow sequence template. If the average flow rate of a certain floor from 7:00 to 9:00 is 3 m³ / h and the fluctuation is <5%, then this pattern is defined as a regular structural feature.

[0051] Irregularity Profiling: First, anomaly detection algorithms are employed, such as the Isolation Forest algorithm, to identify anomalies in water usage data in a high-dimensional space. This algorithm constructs multiple random forests and uses the path length of sample points within the forest to determine the degree of anomaly, effectively capturing sudden water usage events. Simultaneously, a dynamic threshold method is used, setting a threshold range that dynamically adjusts over time based on the statistical characteristics of historical water usage data (such as mean and standard deviation). When real-time monitored indicators such as water flow and frequency exceed the threshold range, the anomaly event identification mechanism is triggered. After identifying the anomaly, spatiotemporal trajectory tracking techniques, such as the Kalman filter algorithm, are used to recursively filter the spatiotemporal data of the event, predicting and updating the event's start time, spatial location (such as pipeline node coordinates), and diffusion path parameters. The optimal estimate is calculated iteratively to track the event's dynamics. Finally, combined with the spatial analysis functions of GIS maps, the spatiotemporal trajectory data of the anomaly event is converted into a visualized leak diffusion heat map. Color gradients and contour lines are used to mark the affected area and water flow direction, achieving accurate location and dynamic display of irregular water usage events.

[0052] Resilient Structure Profiling: Based on the random forest learning algorithm, a correlation model between water usage variables and environmental factors is established. First, external variables such as temperature, humidity, and holidays undergo feature engineering, transforming them into numerical features (e.g., temperature values, binary holiday identifiers), which are then combined with water usage data (e.g., flow rate, time period) to form a training dataset. Supervised learning is performed on the training set using the random forest algorithm, leveraging the ensemble learning capability of multiple decision trees to capture non-linear correlations. For example, by calculating the feature importance scores of each environmental factor, temperature is identified as having the highest weight in influencing water usage. After model training, inputting real-time environmental variables outputs predicted water usage values. The deviation coefficient is calculated by comparing these values ​​with actual values, such as a linear relationship where water usage increases by 1.5% for every 1°C increase in temperature. The model's goodness of fit is measured using a coefficient of determination R² = 0.82. To address seasonal and sudden changes, a flexible scale constraint mechanism is introduced. A calibration threshold is set based on the fluctuation range of historical water use data (such as ±20% of the water use in the same period of the past 3 years). When the prediction deviation exceeds the threshold, the model output weight is automatically adjusted to ensure that the prediction results not only conform to real-time environmental changes, but also do not deviate from historical statistical patterns, ultimately achieving accurate characterization and dynamic response to flexible water use patterns.

[0053] Finally, a time-series mapping relationship is established between the overall system profile and the sub-level profiles. Using timestamp alignment technology, detailed features of the sub-level profiles (such as irregular water usage peaks during a certain period) are embedded into the corresponding time nodes of the overall system profile. A sliding window algorithm (e.g., updating data from the previous 7 days daily) is then used to dynamically refresh the profiles. For example, if a region experiences a surge in water usage due to elasticity for three consecutive days, based on temperature rise predictions, the overall system profile will automatically adjust the water usage baseline for that period and trigger an increase in the weight of elasticity features in the sub-level profiles.

[0054] By employing a technical approach that combines spatiotemporal cross-analysis, extraction of three types of structural features, dual-layer profiling modeling, and dynamic correlation of time series data, abstract water usage data is transformed into a behavioral model with physical meaning. This enables hierarchical analysis of water usage behavior, including regularity, irregularity, and elasticity, thereby improving the accuracy of water usage profiling and supporting differentiated resource scheduling strategies.

[0055] Furthermore, step A200 in the method provided in this application embodiment includes:

[0056] A210: Based on the resource request conditions, the water use area is dynamically partitioned to determine the water use partition.

[0057] A220: Based on the water usage partition, reconstruct the fog computing topology and determine the reconstructed fog nodes.

[0058] In this embodiment of the application, the reconstructed fog node refers to the node formed after adjusting and reconstructing the fog computing topology in the cloud based on the result of dynamically partitioning the water use area (determining the water use area).

[0059] Specifically, firstly, the water use area is dynamically partitioned based on resource request conditions. The system collects resource request data of the water use area in real time, such as the water demand characteristics of different time periods (e.g., a certain time span in the evening) and different functional areas (commercial area, residential area, factory area, mixed area), such as peak flow, duration, and spatial distribution. Through GIS spatial analysis algorithms, the water use area is divided into several water use zones with similar demand patterns. The specific process is as follows:

[0060] First, spatiotemporal water usage data for the water-using area is collected, such as water flow, timestamps, and geographic coordinates at each location. The data is preprocessed to remove outliers and redundant information, forming a structured dataset. Then, temporal features (such as time periods and cycles) and spatial features (such as latitude and longitude, and regional functional attributes) are extracted to construct a multidimensional feature matrix. Spatial clustering algorithms (such as DBSCAN density clustering) are used to group water-using locations. Based on a set spatial distance threshold (such as 500 meters) and temporal pattern similarity (such as peak water usage overlap > 80%), the water-using area is divided into several clusters.

[0061] Based on this, the spatial overlay analysis function of GIS is used to overlay the clustering results with regional land use data (such as the distribution of commercial land, residential land, and industrial land), assigning each cluster a clear functional attribute, thus forming water use zones such as central business districts, residential areas, and industrial parks. For example, due to the dense water use activities such as office work and catering during weekdays, the clusters in the commercial area show an average water flow rate >20 m³ / h during the daytime and a concentrated spatial distribution; while the residential area shows a peak water use rate from 19:00 to 22:00 in the evening, with a flow fluctuation coefficient <15%.

[0062] Finally, the rationality of the zoning was verified by using GIS spatiotemporal analysis tools (such as time series animation and thermal distribution evolution) to ensure that the water use patterns of each zone are consistent with actual needs.

[0063] Secondly, the fog computing topology is reconstructed based on the water usage zoning results. The fog computing topology is based on a cloud-distributed architecture, with each water usage zone corresponding to one or more reconstructed fog nodes. The system dynamically adjusts the computing resource allocation of nodes based on the resource complexity of the zone (such as the computational load of water usage scheduling strategies and real-time requirements): zones with high demand and complex structures (such as commercial areas) are allocated more computing nodes or higher-performance resources, while zones with low demand and simple patterns (such as residential areas at night) have their node configuration optimized to save resources. For example, the storage and computing capabilities of nodes are reallocated through a load balancing algorithm, matching the processing pressure of each reconstructed fog node with the zone's demand, forming a dynamic mapping relationship between zone and node. The specific process is as follows:

[0064] In the process of reconstructing the load balancing algorithm for fog nodes, the resource demand weight of each partition is first quantified by monitoring indicators such as water usage data flow and scheduling strategy computational complexity in real time. For example, the weight of the commercial area is set to 0.8 due to its high water usage density, while the weight of the residential area is set to 0.3. Subsequently, a dynamic load balancing algorithm is adopted to allocate resources to each reconstructed fog node according to the current load status of the fog node, including CPU utilization, memory usage, and task queue length, to match the partition demand. For example, when a surge in water scheduling requests is detected in the central commercial area, the algorithm automatically increases the computing resource ratio of the fog node corresponding to that partition from 30% to 50%, while dynamically migrating 20% ​​of idle resources from nodes in low-load partitions (such as residential areas at night) to supplement the load. By establishing a dynamic mapping model between partition demand weight and node resource quantity, the algorithm continuously iterates and optimizes the task allocation between nodes, ensuring that the processing pressure of each reconstructed fog node is dynamically balanced with the partition demand, avoiding resource overload or idle states. Ultimately, this forms an adaptive partition-node elastic resource allocation mechanism that adapts to changes in water demand, improving the overall scheduling efficiency and stability of the cloud fog computing topology.

[0065] By employing a demand-driven approach of dynamic partitioning, topology reconstruction, and elastic resource allocation, we have achieved precise matching between fog computing resources and water demand, thereby improving resource utilization efficiency and supporting real-time, refined scheduling.

[0066] Furthermore, step A200 in the method provided in this application embodiment includes:

[0067] A230: The resource field includes three spatial dimensions, a time dimension, and an entropy dimension, wherein the entropy dimension includes structural entropy and resource entropy.

[0068] A240: Based on the water use zoning, dynamically isolate the resource fields to determine the isolated resource fields.

[0069] A250: Combine the aforementioned elastic scale to constrain the isolated resource field and determine the requested resource field.

[0070] A260: Establish the association between the reconstructed fog node and the requested resource field, and determine the resource scenario.

[0071] In this embodiment, entropy is one of the dimensions included in the resource field, which includes structural entropy and resource entropy. The larger the structural entropy, the more complex the execution and setup of resource scheduling; the greater the resource fluctuation within a partition and the more complex the water usage architecture, the greater the corresponding resource entropy.

[0072] Specifically, firstly, a multi-dimensional resource field is constructed, encompassing three spatial dimensions, a temporal dimension, and an entropy dimension. The three spatial dimensions correspond to the geographical coordinates or grid partitions of the water use area, such as three-dimensional spatial units divided by the X, Y, and Z axes. The temporal dimension covers full-scale time series from second-level time periods to monthly / quarterly cycles. The entropy dimension quantifies system complexity through structural entropy and resource entropy: structural entropy reflects the execution complexity of resource scheduling strategies, such as the connection complexity of pipeline topology; a larger value indicates more complex scheduling logic. Resource entropy characterizes the degree of resource fluctuation and the complexity of the water use architecture within a partition; for example, the resource entropy value increases when the daily water consumption fluctuation coefficient is >20%. For instance, in an industrial park, the structural entropy value reaches 0.8 (range 0-1) due to interleaved production lines, while in a residential area, the resource entropy value is only 0.3 due to stable water use patterns.

[0073] Secondly, dynamic resource isolation is implemented based on water use zoning. According to the dynamic zoning results, such as commercial areas, industrial areas, residential areas, and mixed areas, IoT devices such as smart valves and flow meters are used to divide the water resources of each zone into independent isolated resource fields, avoiding cross-regional interference. For example, during peak water use periods, the isolated resource fields in commercial areas are physically isolated from those in industrial areas, ensuring that the emergency water needs of commercial areas are prioritized.

[0074] Then, an elastic scaling constraint is introduced to optimize the isolated resource field. The elastic scaling is based on sub-order features in the dynamic water use profile, such as the spatiotemporal tensor threshold corresponding to the elastic structure, to flexibly adjust the boundary and capacity of the isolated resource field. For example, when the summer temperature exceeds 35°C, according to the correlation model in the elastic structure profile that water consumption increases by 1.5% for every 1°C increase in temperature, the upper limit of the isolated resource field capacity in the residential area is increased by 10%. At the same time, the pipeline flow is monitored in real time by pressure sensors to ensure that it does not exceed the elastic scaling threshold, such as 120% of the historical maximum flow, thus forming a requested resource field.

[0075] Finally, a dynamic relationship is established between the reconstructed fog nodes and the requested resource fields. Each node in the fog computing topology corresponds to a requested resource field in a specific partition. Resource demands within the field are matched in real time through edge computing, such as matching node computing power with the partition's water usage scheduling complexity. For example, commercial areas, due to their high structural entropy, are allocated two nodes to process scheduling tasks in parallel, while residential areas can meet their needs with a single node. This ultimately forms a three-in-one resource scenario of partition-resource field-fog node, achieving precise mapping of water usage demands.

[0076] By constructing a multi-dimensional resource field, implementing dynamic isolation, elastic constraints, and node association technologies, static resource management is upgraded to a dynamic adaptive mode, thereby improving the accuracy of resource allocation and enhancing the robustness of the system in complex scenarios.

[0077] Furthermore, step A300 in the method provided in this application embodiment includes:

[0078] A310: Based on the dynamic water usage profile, perform sample-driven supervised training under resource scenario switching to construct the custom scheduler.

[0079] A320: The custom scheduler is split into partition scheduling and intra-region autonomous scheduling to determine the logically related upper-level scheduler and autonomous scheduler.

[0080] A330: Deploy the host scheduler and the autonomous scheduler in the cloud, and establish the association between the autonomous scheduler and the fog computing topology.

[0081] In this embodiment, the upper-level scheduler is a split from a custom scheduler built based on dynamic water usage profiles. It is responsible for performing resource game theory between water usage zones based on resource scenarios, determining the zone resource pool, and is a logically related upper-level scheduling component deployed in the cloud. The autonomous scheduler is a lower-level scheduling component that is logically related to the upper-level scheduler after the custom scheduler is split. Based on the zone resource pool and reconstructed fog nodes, it performs refined water usage game theory within the zone to determine the water usage strategy, and at the same time establishes a connection with the fog computing topology to achieve autonomous scheduling within the zone.

[0082] Specifically, when building a custom scheduler based on dynamic water usage profiles, the first step is to extract spatiotemporal feature data from the dynamic water usage profiles. This includes periodic water usage templates reflecting the regular structure of N event classes determined through periodic clustering, features of irregular structural abnormal events (such as the spatiotemporal trajectory of pipeline leaks), and environmental correlation models based on spatiotemporal tensors for elastic structures (such as the correlation coefficient between temperature and water consumption). This forms a training sample set that includes time (such as time periods and cycles), space (such as regional grid coordinates), and entropy dimensions (structural entropy and resource entropy).

[0083] Then, through sample-driven supervised training, the gradient boosting tree learning algorithm is used to train the samples, enabling the scheduling model to learn the optimal scheduling strategy under different resource scenarios. Specifically, samples are input into the gradient boosting tree algorithm for supervised training. Multiple decision trees are built iteratively, with each tree learning the residuals of the preceding model, gradually optimizing the scheduling strategy prediction for different resource scenarios (peak / off-peak / valley). For example, for the weekday morning peak scenario, the model takes the spatiotemporal distribution of residential water use (e.g., average flow rate of 5 m³ / h in residential areas from 7:00 to 9:00) and industrial water use characteristics (e.g., flow fluctuation coefficient of >20% in factories) as input, and outputs strategy rules that prioritize residential water use and stagger industrial water use to the night. By adjusting hyperparameters such as tree depth and learning rate through cross-validation, the model achieves a prediction accuracy of over 90% for scheduling strategies in various scenarios on the test set, ultimately constructing a custom scheduler that can dynamically respond to water use patterns.

[0084] Next, the custom scheduler is split into a supervising scheduler and an autonomous scheduler. The supervising scheduler is responsible for global resource coordination across zones, performing game-theoretic scheduling between zones based on resource scenarios, such as water allocation between commercial and residential areas. The autonomous scheduler performs fine-grained scheduling within a single zone, combining real-time data from the fog computing topology (such as pipeline pressure and equipment status). By utilizing a hierarchical control architecture to decouple global and local scheduling logic, the system response speed is improved. For example, when a sudden surge in water usage is detected in an industrial zone, the autonomous scheduler can trigger a local emergency scheduling strategy within 50ms, while simultaneously synchronizing information with the supervising scheduler to adjust the global resource pool.

[0085] Finally, both types of schedulers are deployed in the cloud, and the association between the autonomous scheduler and the fog computing topology is established. Elastic scaling of the scheduler is achieved through containerized deployment technologies in the cloud (such as Kubernetes), while the autonomous scheduler is connected in real-time to fog nodes (such as edge servers deployed in industrial parks) using edge computing interfaces, forming a collaborative mechanism of cloud decision-making and edge execution. For example, the upper-level scheduler issues partitioned resource pool quotas through a cloud API. After receiving the instructions, the autonomous scheduler combines real-time water usage data collected by fog nodes, such as flow sensor data updated every minute, to dynamically adjust valve openings and equipment start / stop within the zone, achieving precise allocation of water resources.

[0086] By employing a technical approach that combines dynamic profiling training, hierarchical splitting of the scheduler, and cloud-edge collaborative deployment, scheduling strategies are deeply integrated with real-time water usage behavior. By leveraging a data-driven intelligent scheduling model and a distributed computing architecture, the real-time performance of water resource scheduling and the efficiency of resource allocation are improved.

[0087] Furthermore, step A300 in the method provided in this application embodiment includes:

[0088] A340: Determine the resource pool, wherein the resource pool is the total amount of water resources based on time slices.

[0089] A350: Based on the resource pool, the host scheduler is triggered to perform resource game between water use zones under the resource scenario to determine the zone resource pool.

[0090] A360: Using the partitioned resource pool and the reconstructed fog node, trigger the autonomous scheduler to perform refined water use game within the region and determine the water use strategy.

[0091] In one embodiment, firstly, a time-slice-based resource pool is determined. Water resources are divided into fixed-length time slices (e.g., 1 hour / slice) according to the time dimension. Through historical water consumption data statistics and real-time monitoring, the total water resources within each time slice are calculated as the global resource pool. For example, if a city's daily water supply is 100,000 tons, after dividing it into 24 time slices, the initial value of the resource pool for each time slice is approximately 4,167 tons. This value is dynamically adjusted based on real-time rainfall, pipeline pressure, and other data; for example, during heavy rain, the resource pool capacity may be temporarily reduced by 20% during a certain period.

[0092] Secondly, during the execution of the first-level partitioned game, the upper-level scheduler first triggers resource game among water-using zones based on resource scenarios (such as weekday morning peak hours and weekend off-peak hours). Each water-using zone (such as commercial zone, industrial zone, and residential zone) acts as a game participant, submitting resource allocation requests to the system based on its own water demand (such as the commercial zone requesting 30% of the resource pool from 9-11 am) and global constraints (such as the total resource pool limit). The system performs global optimization through a game theory algorithm (such as Nash equilibrium). This algorithm aims to maximize overall resource utilization efficiency and introduces historical water use efficiency (such as water consumption per unit of industrial output and peak water consumption patterns of residents) as weight parameters to calculate the optimal allocation scheme for each zone. For example, when the commercial zone and the industrial zone simultaneously request resources during peak hours, the upper-level scheduler, based on historical data showing that industrial water use efficiency is higher (e.g., water consumption per unit of output is 30% lower than that of the commercial zone), sets the partitioned resource pool ratio to 40% for the commercial zone and 50% for the industrial zone, with the remaining 10% as an emergency reserve to ensure adjustment capabilities in the event of a sudden water use incident. This process achieves a reasonable allocation of global resources through dynamic game theory, balancing the water demand of different zones with the overall stability of the system.

[0093] Finally, a second-level refined game theory is implemented. The autonomous scheduler combines the regional resource pool quotas with real-time data from the reconstructed fog nodes (such as flow and pressure sensor data from each network node within the region) to perform refined water allocation within the region. For example, in a residential area, the autonomous scheduler allocates 60% of the regional resource pool to high-priority users, 30% to regular users, and uses 10% as a dynamic adjustment based on user water usage profiles, such as high-priority users accounting for 15%. Fuzzy control algorithms are used to adjust valve openings in real time to ensure that water pressure at each user end remains stable within ±5% of the error range, while avoiding network overload.

[0094] By combining global resource balancing with local dynamic response through time-slice resource pool modeling, inter-regional game optimization, and real-time refined control within each region, the system ultimately achieves the effects of improving the fairness of water resource allocation, reducing system energy consumption, and enhancing emergency response capabilities through a two-layer game strategy and real-time data driven by fog computing.

[0095] Furthermore, step A400 in the method provided in this application embodiment includes:

[0096] A410: Perform aggregation and splicing based on the water use area on the water use strategy to determine the regional water use strategy.

[0097] A420: Based on the standardized dimensions of the water resources management system, the regional water use strategy is converted to determine the standard water use strategy.

[0098] A430: The water resource management system identifies and responds to the standard water use strategy.

[0099] Optionally, firstly, perform aggregation and stitching based on water use zones. Collect independent water use strategies for each water use zone (such as commercial, residential, and industrial zones), extract key parameters such as water consumption, scheduling time periods, and priorities, and use spatial data fusion technology (such as vector overlay) to integrate the zone strategies into a regional overall strategy. For example, the strategies of prioritizing water supply of 200 m³ / h during peak hours in the commercial zone and supplying 150 m³ / h during off-peak hours in the residential zone can be stitched together in time series to form a 24-hour continuous water supply plan for the region. Conflict detection algorithms (such as intersection analysis) can then be used to eliminate inconsistencies between zone strategies, ensuring the spatiotemporal continuity of the regional strategy.

[0100] Secondly, a standardized water use strategy is established based on standardized dimensions. For heterogeneous dimensions such as cubic meters per second and tons per hour used by different infrastructures (e.g., pipe networks, reservoirs) and water use types (e.g., industrial and domestic water use), a unified dimension conversion model is developed. For example, the cooling flow rate of industrial water equipment (10 tons per minute) is converted to a unified dimension of 600 cubic meters per hour. Simultaneously, strategy parameters (e.g., priority levels) are encoded into numerical variables that the system can recognize, such as setting high priority to 1 and medium priority to 2. Through standardization, a standard strategy file conforming to the interface specifications of the water resources management system, such as JSON format, is generated to ensure data format consistency.

[0101] Finally, the water resource management system identifies and responds to standard strategies. The system reads the standard strategy file via an API interface, parses the scheduling instructions, such as pressurizing the commercial area's pipe network to 0.4 MPa from 8:00 to 10:00, and converts them into control signals for the underlying equipment, such as starting the booster pump and adjusting valve openings. Simultaneously, integrated real-time monitoring modules (such as pressure sensors and flow meters) provide feedback on the execution status, forming a closed loop of strategy formulation-execution-feedback. For example, when the system identifies the nighttime pressure reduction water supply instruction in the standard strategy, it automatically adjusts the variable frequency pump speed, reducing the pipe network pressure from 0.35 MPa to 0.25 MPa, and transmits pressure data back in real time via IoT terminals, ensuring that the strategy execution accuracy error is <2%.

[0102] By employing a technical approach that integrates regional strategies, standardizes metric transformations, and seamlessly connects systems, the efficiency of strategy processing is improved, the error rate is reduced, and ultimately, through a unified metric system and standardized data format technology, the system's compatibility and overall scheduling coordination are enhanced.

[0103] Furthermore, step A500 in the method provided in this application embodiment includes:

[0104] A510: Simultaneously monitor water resources management to determine management response data.

[0105] A520: Identify the management response data, locate the abnormal management data and perform source tracing processing to determine the cause of the abnormality.

[0106] A530: Based on the aforementioned causes of the anomalies, implement feedback adjustments for water resource management.

[0107] In one embodiment, water resource management monitoring is conducted simultaneously. Through an IoT sensor network deployed in the water-using area, such as ultrasonic flow meters, pressure transmitters, and water quality monitors, data on pipeline flow, water pressure, water quality parameters, and equipment status are collected in real time at a frequency of seconds or minutes, forming a multi-dimensional management response dataset. For example, a pressure sensor deployed in a residential community's pipeline network uploads a pressure value (unit: MPa) every 10 seconds. When pressure fluctuations exceed ±5% during peak water consumption periods (such as 18:00-20:00 in the evening), the data automatically triggers an anomaly warning.

[0108] Secondly, the system identifies and traces abnormal data. It uses an isolated forest learning algorithm to compare real-time data with historical benchmarks, identifying anomalies exceeding preset thresholds (the algorithm process is the same as step A132). Through a data tracing engine, the system traces the associated equipment (such as valves and pumps) and water usage events (such as fire-fighting water and sudden industrial water usage) along the pipeline network topology to pinpoint the cause of the anomaly. For example, if an industrial zone detects that water usage suddenly reaches twice the daily peak at 2 AM, the system analyzes the equipment operation logs for that period and determines that a production line malfunction caused a cooling water leak.

[0109] Finally, feedback adjustments are implemented. Based on the cause of the anomaly, the system automatically triggers adjustment strategies: if it's a device malfunction, such as abnormal valve opening, instructions are sent from the cloud to the edge controller to remotely adjust device parameters or activate backup equipment; if it's uneven resource allocation, the elastic resource pool (reserving 10%-15% of emergency water volume) is invoked for secondary scheduling, and the allocation of resource pool quotas in different zones is redistributed through optimization algorithms. For example, when a fire drill causes a sudden drop in water pressure in a commercial area, the system temporarily allocates 5% of the water volume from the idle resource pool of a nearby residential area, restoring normal water supply within 30 minutes.

[0110] Through a closed-loop mechanism of dynamic monitoring, intelligent source tracing, and flexible adjustment, refined and adaptive water resource management has been achieved, effectively reducing leakage rates and enhancing the system's anti-interference capabilities.

[0111] In summary, the water resource management method with integrated water use behavior recognition provided in this application has the following technical effects:

[0112] This application obtains a dynamic water use profile by retrieving water use event records from water use areas, and then processing them through spatiotemporal water use feature mining and behavior deconstruction. It calculates resource scenarios and water use strategies, and makes adjustments based on the results of dynamic isolation and elastic constraints of resource fields, regional game theory, and refined game theory. This achieves refined water resource management, making resource scheduling and infrastructure control management in water use areas more precise and efficient. It achieves the technical effect of using big data processing technology to accurately identify various water use behaviors, realize real-time dynamic resource scheduling, efficiently integrate and analyze water use data from multiple regions, improve the level of refined water resource management, and effectively cope with complex water use scenarios.

[0113] Example 2, as Figure 2 As shown, based on the same inventive concept as in Embodiment 1 above, this application provides a refined water resource management system integrating water use behavior recognition, the system comprising:

[0114] Water usage profile construction module 1 is used to retrieve water usage event records of water usage areas, mine spatiotemporal water usage characteristics and perform behavioral deconstruction to construct a dynamic water usage profile, wherein regularity, irregularity, and elasticity are used as construction constraints.

[0115] Resource scenario determination module 2 is used to determine resource scenarios by establishing a resource field based on water use areas and performing dynamic resource isolation and elastic scaling constraints based on resource request conditions and reconstructing the fog computing topology in the cloud.

[0116] The water use strategy determination module 3 is used to deploy a custom scheduler in the cloud based on the dynamic water use profile, and perform a first-level partition game and a second-level fine-grained game based on the resource scenario to determine the water use strategy.

[0117] Water use strategy reconstruction module 4 is used to standardize and reconstruct the water use strategy, and to perform resource scheduling and infrastructure control management of the water use area according to the water resource management system.

[0118] Furthermore, the water profile construction module 1 is used to perform the following steps:

[0119] The water usage event records are periodically clustered to determine N event classes, where N is the number of periodic cycles. The N event classes are traversed, and spatiotemporal water usage features are mined under inter-class periodic mapping to determine the water usage feature structure, where a preset frequency is used as the mining constraint. Based on the water usage feature structure, a dynamic water usage profile is constructed by performing behavioral deconstruction analysis.

[0120] Furthermore, the water profile construction module 1 is used to perform the following steps:

[0121] The water usage characteristic structure is deconstructed using spatial and temporal dimensions to determine regular, irregular, and elastic structures. An assembly water usage profile is constructed based on the water usage characteristic structure, and a sub-level water usage profile is constructed based on the regular, irregular, and elastic structures. The sub-level water usage profile includes three components, and the elastic structure is determined based on the spatiotemporal tensor of water usage events. A time-series mapping between the assembly water usage profile and the sub-level water usage profile is established as the dynamic water usage profile.

[0122] Furthermore, the resource scenario determination module 2 is used to perform the following steps:

[0123] Based on the resource request conditions, the water usage area is dynamically partitioned to determine the water usage partitions; based on the water usage partitions, the fog computing topology is reconstructed to determine the reconstructed fog nodes.

[0124] Furthermore, the resource scenario determination module 2 is used to perform the following steps:

[0125] The resource field comprises three spatial dimensions, a temporal dimension, and an entropy dimension, wherein the entropy dimension includes structural entropy and resource entropy; based on the water usage zoning, the resource field is dynamically isolated to determine isolated resource fields; combined with the elastic scale, the isolated resource fields are constrained to determine requested resource fields; the association between the reconstructed fog nodes and the requested resource fields is established to determine the resource scenario.

[0126] Furthermore, the water use strategy determination module 3 is used to perform the following steps:

[0127] Based on the dynamic water usage profile, sample-driven supervised training under resource scenario switching is performed to construct the custom scheduler; the custom scheduler is split into partition scheduling and intra-region autonomous scheduling to determine the logically related upper-level scheduler and autonomous scheduler; the upper-level scheduler and the autonomous scheduler are deployed in the cloud, and the association between the autonomous scheduler and the fog computing topology is established.

[0128] Furthermore, the water use strategy determination module 3 is used to perform the following steps:

[0129] A resource pool is determined, wherein the resource pool is the total amount of water resources based on time slices; based on the resource pool, the upper-level scheduler is triggered to perform resource game between water use zones under the resource scenario to determine the zone resource pool; the autonomous scheduler is triggered with the zone resource pool and the reconstructed fog node to perform refined water use game within the zone to determine the water use strategy.

[0130] Furthermore, the water use strategy reconfiguration module 4 is used to perform the following steps:

[0131] The water use strategy is aggregated and spliced ​​based on the water use area to determine the regional water use strategy; the regional water use strategy is transformed according to the standardized dimensions of the water resource management system to determine the standard water use strategy; the water resource management system identifies and responds to the standard water use strategy.

[0132] Furthermore, the water use strategy reconfiguration module 4 is used to perform the following steps:

[0133] Simultaneously monitor water resources management to determine management response data; identify the management response data, locate abnormal management data and perform source tracing to determine the cause of the abnormality; and adjust water resources management based on the cause of the abnormality.

[0134] The integrated water use behavior recognition water resource management system provided in the embodiments of the present invention can execute the integrated water use behavior recognition water resource management method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the method.

[0135] Although this application makes various references to certain modules in the system according to the embodiments of this application, any number of different modules can be used and run on user terminals and / or servers. The various units and modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of each functional unit are only for easy distinction between each other and are not used to limit the scope of protection of this invention.

[0136] The specific embodiments described above do not constitute a limitation on the scope of protection of this application. Those skilled in the art should understand that various modifications, combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this application should be included within the scope of protection of this application. In some cases, the actions or steps described in this application can be performed in a different order than that shown in the embodiments and still achieve the desired results. Furthermore, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

Claims

1. A water resource management method integrating water use behavior recognition, characterized in that, The method includes: Retrieve water use event records from water use areas, mine spatiotemporal water use characteristics and deconstruct behaviors to construct a dynamic water use profile, with regularity, irregularity, and flexible scale as constraints for construction. Based on resource request conditions and reconstructing the cloud-based fog computing topology, a resource field based on water usage areas is established, and dynamic resource isolation and elastic scaling constraints are applied to determine resource scenarios. Based on the dynamic water usage profile, a custom scheduler is deployed in the cloud, and a first-level partition game and a second-level refined game are performed based on the resource scenario to determine the water usage strategy; The water use strategy is standardized and restructured, and resource scheduling and infrastructure control management of the water use area are implemented according to the water resource management system. By mining spatiotemporal water use characteristics and deconstructing their behavior, a dynamic water use profile is constructed, including: The water usage event records are periodically clustered to determine N event classes, where N is the number of periodic cycles. Traverse the N event classes, perform spatiotemporal water use feature mining under inter-class periodic mapping, and determine the water use feature structure, wherein a preset frequency is used as the mining constraint; Based on the aforementioned water usage characteristic structure, a dynamic water usage profile is constructed through behavioral deconstruction analysis. By conducting behavioral deconstruction analysis, a dynamic water usage profile is constructed, including: The water-use characteristic structure is deconstructed using spatial three-dimensional and temporal dimensions to determine regular, irregular, and elastic structures. A water usage profile of the assembly is constructed based on the water usage characteristic structure, and a sub-level water usage profile is constructed based on the regular structure, irregular structure and elastic structure. The sub-level water usage profile includes three items, and the elastic structure is determined based on the spatiotemporal tensor of water usage events. Establish a time-series mapping between the water usage profile of the assembly and the water usage profile of the sub-stage, as the dynamic water usage profile; Reconstructing the fog computing topology in the cloud, including: Resource request conditions refer to the water demand conditions of a water-using area; Based on the resource request conditions, the water use area is dynamically partitioned to determine the water use zones; Based on the water usage zoning, the fog computing topology is reconstructed to determine the reconstructed fog nodes; The resource field includes three spatial dimensions, a time dimension, and an entropy dimension, wherein the entropy dimension includes structural entropy and resource entropy; Structural entropy reflects the execution complexity of resource scheduling strategies, while resource entropy characterizes the degree of resource fluctuation and water use architecture complexity within a partition. Based on the water use zoning, the resource fields are dynamically isolated to determine the isolated resource fields; By combining the aforementioned elastic scale, constraints are applied to the isolated resource field to determine the requested resource field; Establish the association between the reconstructed fog node and the requested resource field to determine the resource scenario.

2. The water resource refined management method with integrated water use behavior recognition as described in claim 1, characterized in that, Deploying a custom scheduler in the cloud based on the dynamic water usage profile includes: Based on the dynamic water usage profile, sample-driven supervised training under resource scenario switching is performed to construct the custom scheduler; The custom scheduler is split into partition scheduling and intra-region autonomous scheduling to determine the logically related upper-level scheduler and autonomous scheduler; The host scheduler and the autonomous scheduler are deployed in the cloud, and the association between the autonomous scheduler and the fog computing topology is established.

3. The water resource refined management method with integrated water use behavior recognition as described in claim 2, characterized in that, Based on the aforementioned resource scenario, a first-level partitioning game and a second-level refined game are performed to determine the water use strategy, including: Determine the resource pool, wherein the resource pool is the total amount of water resources based on time slices; Based on the resource pool, the host scheduler is triggered to perform resource game between water use zones under the resource scenario and determine the zone resource pool; Using the partitioned resource pool and the reconstructed fog node, the autonomous scheduler is triggered to perform refined water use game within the region and determine the water use strategy.

4. The water resource refined management method with integrated water use behavior recognition as described in claim 1, characterized in that, The standardized restructuring of the water use strategy includes: The water use strategy is aggregated and stitched based on the water use area to determine the regional water use strategy; Based on the standardized dimensions of the water resources management system, the water use strategy for the region is converted to determine the standard water use strategy; The water resource management system identifies and responds to the standard water use strategy.

5. The water resource refined management method with integrated water use behavior recognition as described in claim 1, characterized in that, After implementing resource allocation and infrastructure control management in the water use area, the following is included: Simultaneously monitor water resources management to determine management response data; Identify the management response data, locate abnormal management data and perform source tracing processing to determine the cause of the abnormality; Based on the aforementioned causes of the anomalies, feedback adjustments will be made to water resource management.

6. A refined water resource management system integrating water use behavior recognition, characterized in that: The system for implementing the integrated water use behavior identification method for refined water resource management according to any one of claims 1-5, the system comprising: The water usage profile construction module is used to retrieve water usage event records of water usage areas, mine spatiotemporal water usage characteristics and perform behavioral deconstruction to construct dynamic water usage profiles, with regularity, irregularity, and flexible scale as construction constraints. The resource scenario determination module is used to determine the resource scenario by establishing a resource field based on the water use area and performing dynamic resource isolation and elastic scaling constraints based on the resource request conditions and reconstructing the fog computing topology in the cloud. The water use strategy determination module is used to deploy a custom scheduler in the cloud based on the dynamic water use profile, and perform a first-level partition game and a second-level fine-grained game based on the resource scenario to determine the water use strategy. The water use strategy reconstruction module is used to standardize and reconstruct the water use strategy, and to perform resource scheduling and infrastructure control management of the water use area according to the water resource management system.

Images