Urban water prediction scheduling method based on user behavior modeling and storage device

Abstract

The invention discloses an urban water consumption prediction scheduling method based on user behavior modeling and a storage device, and relates to the technical field of water consumption prediction scheduling, and the method comprises the steps: building a vibration cascade relation through dividing multiple stages of water consumption units according to the user cluster topology of a target city; behavior phase change points are determined, a water consumption behavior portrait sequence is constructed, and a scheduling decision maker is developed; a water consumption task is received through an urban water system, a scheduling decision maker is triggered, and a prediction scheduling strategy is determined; and according to the urban water system, executing issuing execution management of the prediction scheduling strategy. The technical problems of large water consumption prediction error and low scheduling management efficiency caused by extensive division of urban water consumption units and lack of accuracy and real-time performance of a scheduling strategy in the prior art are solved, and the technical effects of realizing division of the water consumption units, formulating a prediction scheduling strategy and improving urban water consumption prediction accuracy and scheduling management efficiency are achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of water use prediction and scheduling, and in particular to an urban water use prediction and scheduling method and a storage device based on user behavior modeling. Background Art

[0002] With the acceleration of urbanization, urban water use continues to expand, and water demand is becoming more diverse and dynamic. Existing urban water forecasting and scheduling technologies often rely on traditional statistical models or simple time series analysis. These technologies suffer from crude water unit divisions, an inability to accurately characterize differences in user behavior and the dynamic impact of environmental factors, and a difficulty in effectively capturing phase transitions in water use behavior. Furthermore, scheduling decisions lack a deep fusion of real-time sensor data and behavioral patterns, resulting in low water use forecast accuracy and lagging scheduling strategies. These technologies fail to meet the needs of refined urban water management, leading to the risk of water resource waste and supply-demand imbalance.

[0003] The existing technology has technical problems such as extensive division of urban water use units and lack of accuracy and real-time scheduling strategies, which lead to large water use prediction errors and low scheduling management efficiency. Summary of the Invention

[0004] This application provides an urban water use prediction and scheduling method and storage device based on user behavior modeling, which is used to solve the technical problems in the existing technology of extensive urban water use unit division, lack of accuracy and real-time scheduling strategy, resulting in large water use prediction errors and low scheduling management efficiency.

[0005] In view of the above problems, the present application provides an urban water use prediction and scheduling method and storage device based on user behavior modeling.

[0006] In a first aspect, the present application provides a method for predicting and scheduling urban water use based on user behavior modeling, the method comprising:

[0007] Aiming at the user cluster topology of the target city, a vibration cascade relationship is established by dividing the water use units into multiple levels, wherein the vibration cascade relationship is the voiceprint sensing topology corresponding to the water use unit network at different cluster granularities; the behavior phase change point is determined by the time-sharing behavior phase change and the external environment behavior phase change, and a water use behavior portrait sequence is constructed. Combined with the voiceprint fingerprint library, a scheduling decision maker is developed in the urban water system, wherein the voiceprint fingerprint library is constructed by the pipeline vibration characteristics-water flow behavior; water use tasks are received through the urban water system, task interpretation and decision-making based on the vibration cascade relationship are performed, directional sensing at the task cluster granularity is performed, and the scheduling decision maker is transmitted back and triggered, and the signal normal and abnormal binary classification and the behavior phase are judged in parallel to form a behavior state space and make long-term and short-term decisions to determine the prediction scheduling strategy; according to the urban water system, the issuance and execution management of the prediction scheduling strategy is executed.

[0008] A second aspect of the present application provides a storage device, comprising: a memory for storing executable instructions; and a processor for implementing an urban water use prediction and scheduling method based on user behavior modeling when executing the executable instructions stored in the memory.

[0009] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0010] Based on the user cluster topology of the target city, the system divides the user into multiple levels of water use units and establishes a vibration cascade relationship. By combining time-sharing behavior phase changes with external environmental behavior phase changes, the system identifies behavioral phase change points and constructs a water use behavior profile sequence. Combined with a voiceprint fingerprint library, a scheduling decision-maker is developed within the urban water system. The system receives water use tasks, interprets them, makes decisions based on the vibration cascade relationship, and performs directional sensing at the task cluster granularity. This feedback triggers the scheduling decision-maker, which performs binary signal classification based on signal regularity and concurrent determination of behavior phases, constructs a behavior state space, makes long-term and short-term decisions, and determines a predictive scheduling strategy. The system then manages the issuance and execution of this predictive scheduling strategy based on the urban water system. This approach achieves the technical effect of dividing water use units and formulating predictive scheduling strategies, improving urban water use prediction accuracy and scheduling management efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0012] Figure 1 A flow chart of a method for predicting and scheduling urban water use based on user behavior modeling provided in an embodiment of the present application;

[0013] Figure 2 A schematic diagram of the structure of a storage device provided in an embodiment of the present application.

[0014] Description of the reference numerals: input device 201 , processor 202 , memory 203 , output device 204 . DETAILED DESCRIPTION

[0015] This application provides an urban water use prediction and scheduling method and storage device based on user behavior modeling, which is used to solve the technical problems in the existing technology of extensive urban water use unit division, lack of accuracy and real-time scheduling strategy, resulting in large water use prediction errors and low scheduling management efficiency.

[0016] The following will be combined with the accompanying drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making any creative work are within the scope of protection of this application.

[0017] Example 1, as Figure 1 As shown, the present application provides a method for predicting and scheduling urban water use based on user behavior modeling, the method comprising:

[0018] Step S100: For the user cluster topology of the target city, a vibration cascade relationship is established by dividing the water use units into multiple levels, wherein the vibration cascade relationship is the voiceprint sensing topology corresponding to the water use unit network at different cluster granularities.

[0019] Specifically, based on the user cluster topology of the target city, multi-level water use units are divided from small to large according to the cluster granularity, such as the first-level water use unit with the household as the smallest cluster granularity, the second-level water use unit with the building as the granularity, and finally the N-level water use unit with the region as the granularity. For each level of water use unit, the voiceprint sensor array distributedly deployed in the urban water pipe network is matched to establish a cascade relationship between the water use unit and the sensor array. For example, the first-level water use unit corresponds to the first-level sensor array deployed in the user's home, and the second-level water use unit corresponds to the second-level sensor array deployed in the building's main water pipe, forming a multi-level vibration cascade relationship from household to region. This vibration cascade relationship is essentially a mapping of the water use unit network and the voiceprint sensor topology at different cluster granularities, which can realize the perception and transmission of vibration signals from micro-user water use behavior to macro-regional water use trends.

[0020] Step S200: Determine the behavior phase change point and construct a water use behavior portrait sequence based on the time-sharing behavior phase change and the external environment behavior phase change. Combined with the voiceprint fingerprint library, a scheduling decision maker is developed in the urban water system. The voiceprint fingerprint library is constructed by pipeline vibration characteristics-water flow behavior.

[0021] Specifically, the system first explores phase transition points in water use behavior from both temporal and environmental dimensions. By analyzing historical water use records in target cities, the system identifies periodic phase transition points in user water use behavior (e.g., morning rush hour washing, evening showering, etc.) based on daily, weekly, and monthly cycles. Furthermore, the system considers phase transition points caused by external environmental factors such as seasonal temperature fluctuations and sudden rainfall events (e.g., water consumption surges during summer heat waves, and pressure changes in the water supply network during rainstorms). These two types of phase transition points are then fitted to form a complete set of behavioral phase transition points. Based on these phase transition points, a sequence of water use behavior profiles is constructed, distinguishing, for example, between "morning water use profiles" and "weekday water use profiles." This is combined with a voiceprint fingerprint library constructed by correlating pipeline vibration characteristics with water flow behavior (e.g., vibration frequency characteristics corresponding to different water flow rates). Generative adversarial training is then used to determine signal discriminators (distinguishing normal from abnormal water use signals) and behavioral phase discriminators (identifying phase transitions in water use behavior). Finally, a scheduling decision maker is integrated into the urban water system for subsequent signal analysis and scheduling decisions.

[0022] Step S300: Receive water use tasks through the urban water system, perform task interpretation and make decisions based on vibration cascade relationships, perform directional sensing at the task cluster granularity, transmit back and trigger the scheduling decision maker, perform signal normal and abnormal binary classification and parallel judgment of behavioral phase, construct behavioral state space and make long-term and short-term decisions, and determine the predictive scheduling strategy.

[0023] Specifically, when the urban water system receives a water use task (such as global water supply scheduling or local pipeline maintenance), it first interprets the task type and locates the optimal cluster granularity for the target water use unit (e.g., local tasks correspond to building-level water use units, global tasks correspond to regional-level water use units). Based on the vibration cascade relationship, it matches the sensor array corresponding to the target cascade, and then directionally collects and transmits the pipeline water ripple vibration signal. The signal discriminator in the scheduling decision maker is triggered to classify the vibration signal as normal / abnormal. Simultaneously, the phase discriminator is triggered based on the task scenario, and the current water use behavior profile (e.g., whether it is in the morning rush hour phase) is determined based on the behavioral phase transition point. The signal classification results are integrated with the behavioral profile to construct a behavioral state space. Using algorithms such as long short-term memory networks (LSTMs), water use trends within the state space are analyzed to develop a predictive scheduling strategy covering both short-term scheduling (e.g., real-time water pressure adjustment) and long-term planning (e.g., regional water quota allocation).

[0024] Step S400: Executing the issuance and execution management of the forecast scheduling strategy according to the urban water system.

[0025] Specifically, based on the control architecture of the urban water system, the determined prediction and scheduling strategies are distributed to execution units at all levels, such as sending water pressure regulation instructions to regional water supply stations and sending valve opening and closing control signals to local pipe networks. During this process, the multi-cluster partitioning mechanism enables different tasks to dynamically adapt to the corresponding unit cluster architecture and sensor topology (for example, emergency repair tasks automatically switch to the building-level sensor topology in the fault area), and uses phase change factors (such as sudden changes in water consumption and sudden drops in ambient temperature) as a benchmark to switch directional behavior profiles in real time (for example, switching from conventional water use profiles to anti-freeze emergency profiles). By dynamically reconstructing the sensor topology and updating the phase state profiles, the behavior state space is continuously updated to ensure that water use prediction and scheduling decisions are always based on the most accurate current water use status, thus realizing intelligent scheduling and dynamic management of urban water use.

[0026] In one possible implementation, step S100 further includes:

[0027] Step S110: deploying a voiceprint sensor array, wherein the voiceprint sensor array is distributedly deployed in the city water network.

[0028] Step S120: Acquire the user cluster topology of the target city, and establish a vibration cascade relationship between the user cluster topology and the voiceprint sensor array.

[0029] Specifically, a distributed deployment of voiceprint sensor arrays is carried out at key nodes and user terminals in the urban water network. This involves installing vibration sensors at user terminals (such as taps and water meters), building mains, and regional water supply trunks, forming a multi-level sensor network covering the entire network from user terminals to the main trunk lines of the urban water supply network. This voiceprint sensor array is connected to the urban water system's data center via wired or wireless communication, collecting vibration signals generated by water flow in the pipelines in real time, providing raw data support for subsequent water use behavior analysis and scheduling decisions.

[0030] The target city's user cluster topology is obtained through the city's Geographic Information System (GIS) and water management database. This topology reflects the distribution of users at the region, building, and unit levels. Based on the different cluster granularities of the user cluster topology (such as household, building, community, and region), sensors at each level of the voiceprint sensor array are matched with corresponding water use units to establish a vibration cascade relationship. For example, the primary sensor array deployed in the user's home is cascaded with the household-level water use unit, and the secondary sensor array on the building's main water pipe is cascaded with the building-level water use unit. This allows the water use vibration signal of each water use unit to be collected in real time by the corresponding sensor array, forming a vibration signal transmission link from micro-user water use behavior to macro-regional water use conditions, laying the foundation for subsequent task interpretation and directional sensing based on vibration cascade relationships.

[0031] In one possible implementation, step S100 further includes:

[0032] Step S130: Divide the user cluster topology into multiple levels of water use units based on the cluster granularity.

[0033] Step S140: determining a primary water use unit for the multi-level water use units, wherein the primary water use unit is the smallest cluster granularity, with households as water use units.

[0034] Step S150: for the primary water use unit, determining a primary sensor array in the voiceprint sensor array, wherein the primary sensor array is deployed in the primary water use unit.

[0035] Step S160: establishing a cascade connection between the primary water-using unit and the primary sensor array as a first vibration cascade relationship.

[0036] Specifically, the multi-level water use units are divided according to the cluster granularity of the user cluster topology. Specifically, the water use units are divided hierarchically according to the spatial distribution and hierarchical characteristics of the user cluster topology of the target city, and the cluster granularity is adjusted from small to large. First, the household is used as the most basic bottom unit, and then it is aggregated upward into water use units of different granularities such as buildings, communities, and regions, forming a multi-level architecture from micro users to macro regions. In the division process, the physical topological structure of the urban water supply network and the aggregation characteristics of user water use behavior are combined to ensure that water use units at all levels can not only independently reflect the water use characteristics of specific clusters, but also form a water use trend transmission chain from the bottom to the top level as a whole, laying the foundation for the subsequent establishment of vibration cascade relationships and the realization of water use monitoring and scheduling decisions at different granularities.

[0037] Determining the first-level water use unit for multi-level water use units specifically refers to defining the household with the smallest cluster granularity as the first-level water use unit in the multi-level water use unit system based on the user cluster topology. This process takes the household or individual user as the basic unit, combines the user profile and geographic distribution information in the urban water management system, defines the boundaries of each first-level water use unit (such as the specific residential number, the physical location corresponding to the water use account), and associates its water use attributes (such as the water meter number, the type of water use equipment, etc.). The first-level water use unit serves as the basis of the entire water use unit hierarchy. It can be aggregated into larger-granularity water use units such as buildings and communities, and can be refined downward to the behavioral monitoring of different water use terminals in the household, ensuring that the entire chain of data collection and analysis from micro-user water use behavior to macro-regional scheduling is logically coherent and traceable.

[0038] Determining the first-level sensor array in the voiceprint sensor array for the first-level water-using unit specifically refers to locating and selecting the sensor device combination that directly serves the first-level water-using unit in the voiceprint sensor array distributedly deployed in the urban water pipe network. Among them, the first-level water-using unit uses the household as the smallest cluster granularity, so the first-level sensor array is usually deployed at the water terminal node inside the user's home, such as near the household water pipe, water meter, faucet, water heater and other equipment. By installing vibration sensors, pressure sensors and other equipment, the pipeline vibration signals generated by the water use of a single household are collected in real time. These dispersedly deployed sensors form an array based on wireless or wired communication protocols, forming a full coverage perception of the water use behavior of the first-level water-using unit, ensuring that the frequency, amplitude and other vibration characteristics generated by the water flow during indoor water use can be accurately captured, providing a data collection basis for the subsequent establishment of a vibration cascade relationship between the first-level water-using unit and the sensor array.

[0039] A cascade of primary water use units and primary sensor arrays is established as the first vibration cascade relationship. This maps the physical location and signal transmission between these household-level primary water use units and the primary sensor arrays deployed within the user's home. Using the city water system's Geographic Information System (GIS) and sensor network topology data, a household's water use unit (e.g., specific residential unit number) is mapped to a combination of vibration sensors (i.e., the primary sensor array) deployed at their water meter, faucet, and other locations. Whenever a household uses water, the primary sensor array collects pipeline vibration signals in real time and transmits them back to the city water system via a communication link, forming a direct cascade link between household water use and vibration signal perception. This first vibration cascade relationship forms the underlying foundation of a multi-level vibration cascade system, ensuring that it can be subsequently aggregated to form cascade relationships between water use units and sensor arrays at larger granularities, such as buildings and regions. This enables signal transmission and coordinated monitoring from micro-level user water use behavior to macro-level regional water use status.

[0040] In one possible implementation, step S130 further includes:

[0041] Step S131: traverse the multi-level water use units, perform matching and cascading based on the voiceprint sensor array, and determine the Nth vibration cascade relationship, wherein the Nth vibration cascade relationship uses the maximum cluster granularity as the water use unit.

[0042] Step S132: adding the first vibration cascade relationship to the Nth vibration cascade relationship into the vibration cascade relationship.

[0043] Specifically, after completing the cascade of the first-level water use unit (with households as the smallest cluster granularity) and the first-level sensor array, all divided multi-level water use units are traversed in ascending order of cluster granularity (such as households, buildings, communities, and regions). For each level of water use unit, the corresponding level of sensor combination is matched in the distributed voiceprint sensor array. For example, the second-level water use unit (building level) corresponds to the second-level sensor array deployed on the building's main water pipe, the third-level water use unit (community level) corresponds to the third-level sensor array of the community water supply main pipe, and so on, until the N-level water use unit with the largest cluster granularity (such as a specific area in a city) is matched to the N-level sensor array of the regional water supply pipeline. Through this layer-by-layer matching approach, a vibration cascade relationship is established between water-using units at all levels and the corresponding sensor arrays. The Nth vibration cascade relationship is the mapping of the water-using unit with the largest cluster granularity and the regional sensor array, forming a full-level vibration signal perception link from microscopic single households to macroscopic regions, ensuring that the urban water system can realize the cascade collection and transmission of vibration signals based on water-using units of different granularities.

[0044] The first vibration cascade relationship through the Nth vibration cascade relationship are added to the vibration cascade relationship database. The established vibration cascade relationships, from the first vibration cascade relationship corresponding to the smallest cluster granularity (the mapping of the first-level water use unit to the first-level sensor array) to the Nth vibration cascade relationship corresponding to the largest cluster granularity (e.g., the region) (the mapping of the Nth-level water use unit to the Nth-level sensor array), are hierarchically integrated into the vibration cascade relationship database of the urban water system. This operation forms a complete multi-level vibration cascade system, including mapping relationships at various levels, such as household-to-first-level sensor array, building-to-second-level sensor array, and community-to-third-level sensor array. This system records and manages the cascade relationships between water use units and voiceprint sensor arrays at different cluster granularities. Through this integrated vibration cascade relationship, the urban water system can implement targeted sensing based on water use task requirements (e.g., local pipeline monitoring or global water supply scheduling). This provides a multi-level infrastructure support for subsequent task interpretation, signal acquisition, and scheduling decisions based on the vibration cascade relationship.

[0045] In one possible implementation, step S200 further includes:

[0046] Step S210: Generate adversarial training using normal water usage and abnormal water usage to determine a signal discriminator.

[0047] Step S220: Generate adversarial training using the behavior phase transition points and the behavior portrait sequence to determine a behavior phase discriminator.

[0048] Step S230: constructing the scheduling decision maker according to the signal discriminator and the phase discriminator, wherein the voiceprint fingerprint library is built into the scheduling decision maker.

[0049] Specifically, generative adversarial training is conducted on normal water use versus abnormal water use (e.g., water ripples and flow vibrations caused by user water use, or vibrations caused by inherent pipe anomalies) to determine the signal discriminator. Specifically, pipeline vibration signals generated by daily user water use (such as washing and showering) in normal water use scenarios, as well as vibration signals caused by non-user water use behaviors such as pipe leaks and equipment failures in abnormal water use scenarios, are used as training samples. Training is performed using a generative adversarial network (GAN) architecture. The generator simulates various types of water use vibration signals, while the discriminator performs dual judgments on the input signals: distinguishing whether the signal is real-world or simulated by the generator, and identifying whether the signal is normal user water use vibration or abnormal pipe vibration. Through iterative adversarial game play between the generator and the discriminator, the discriminator gradually learns and grasps the differences in frequency, amplitude, waveform, and other characteristics between normal and abnormal vibration signals. Ultimately, a signal discriminator with accurate discrimination is developed, capable of performing binary classification of normal and abnormal water use on the target vibration signals transmitted back in real time.

[0050] Generative adversarial training is performed on behavioral phase transition points and the behavioral profile sequences to determine a behavioral phase discriminator. Based on the identified behavioral phase transition points (including a first-class behavioral phase transition point mined through periodic time-sharing, such as morning peak water use periods and evening peak water use periods, and a second-class behavioral phase transition point mined based on seasonal environmental evolution and sudden environmental changes, such as surges in water use during high temperatures in summer and sudden changes in pipe network pressure during rainstorms) and constructed water use behavior profile sequences (such as morning water use profiles, weekday water use profiles, and emergency water use profiles), a generative adversarial network (GAN) training framework is used to determine the behavioral phase discriminator. The generator simulates and generates water use behavior feature sequences under different behavioral phases. These feature sequences contain various parameters related to the behavioral phase transition points, such as water use changes, water use time distribution, and environmental factors. The discriminator analyzes the input behavioral feature sequence and determines the specific behavioral phase it corresponds to (e.g., whether it is in the phase corresponding to a certain phase transition point). During the training process, the generator is continuously optimized to generate a behavioral feature sequence that is closer to the real situation, while the discriminator continuously improves its ability to identify different phase characteristics. Through the confrontation between the two, the discriminator is eventually able to accurately identify the phase of the current water use behavior based on the specific identification elements (i.e., phase change elements) of the behavioral phase change point, realize dynamic switching and precise judgment of the water use behavior portrait, and thus determine the behavioral phase discriminator.

[0051] A scheduling decision maker is constructed based on a signal discriminator and a phase discriminator, with a voiceprint fingerprint library embedded within it. The signal discriminator (for distinguishing normal from abnormal water usage signals) and the behavioral phase discriminator (for identifying the current water usage behavior phase), developed through generative adversarial training, serve as core algorithmic modules and are integrated into the urban water system's scheduling decision maker. Furthermore, a voiceprint fingerprint library, constructed by correlating pipeline vibration characteristics with water flow behavior (storing feature templates of vibration signals for different water usage scenarios, such as user washing and pipe leaks), is embedded within the scheduling decision maker as a reference database for signal discrimination. The scheduling decision maker then performs feature matching on real-time vibration signals using the voiceprint fingerprint library. This combines the normal / abnormal classification results of the signal discriminator with the phase determination of the behavioral phase discriminator (e.g., whether the water usage is during the morning rush hour), forming a collaborative decision-making mechanism based on signal status and behavioral phase.

[0052] In one possible implementation, step S220 further includes:

[0053] Step S221: call the water consumption records of the target city, set the user's water consumption behavior cycle and perform periodic time-sharing mining to determine a type of behavior phase change point.

[0054] Step S222: mining two types of behavior phase change points based on seasonal environmental evolution and sudden environmental changes.

[0055] Step S223: fitting the first type of behavior phase transition point and the second type of behavior phase transition point to determine the behavior phase transition point.

[0056] Specifically, the water use records of the target city are called up, the user water use behavior cycle is set, and periodic time-sharing mining is performed to determine a type of behavioral phase change point. The historical water use records of the target city are obtained through the urban water system database, covering data such as user water consumption and water equipment usage in different time periods, so as to set user water use behavior cycles such as days, weeks, and months. The water use data within each cycle is segmented and mined according to the time dimension. For example, the fluctuation of water consumption is counted in hours, and the peak and valley periods of water use with repetitive and regularity are identified, such as the peak of water use for washing and rinsing from 6 to 8 in the morning and the peak of water use for showering from 19 to 21 in the evening. These key time points of water use behavior changes that appear repeatedly in a fixed period are a type of behavioral phase change point, which reflects the periodic law of the time distribution of users' daily water use behavior and provides key nodes based on the time dimension for the subsequent construction of water use behavior portrait sequences.

[0057] Using seasonal environmental evolution and sudden environmental changes, we identify two types of behavioral phase transition points. Starting from the impact of the external environment on water use behavior, we identify water use behavior mutation nodes along two dimensions: First, we analyze the impact of seasonal environmental evolution on water use, such as the surge in water use due to showers and landscaping during high temperatures in summer (June-August), and the need to adjust water use patterns due to anti-freezing measures during low temperatures in winter (December-February). We identify seasonal phase transition points based on these factors. Second, we focus on sudden environmental changes, such as pressure fluctuations in the water supply network caused by heavy rain or the centralized water storage required by users after sudden water outages. By comparing water use data before and after sudden environmental events, we identify water use behavior mutation points caused by sudden environmental changes. These phase transition points reflect the interference of non-periodic environmental factors on water use behavior. Together with the first type of behavioral phase transition points (periodic time-sharing phase transition points), they constitute a complete set of behavioral phase transition points, providing a key environmental dimension for subsequent water use behavior profiling and phase discrimination.

[0058] Fitting first- and second-type behavioral phase transition points to identify behavioral phase transition points is accomplished by using regression fitting algorithms in machine learning (such as support vector regression). Time series data for first-type behavioral phase transition points (e.g., water consumption and timestamps corresponding to daily, weekly, or monthly peak water consumption periods) and environmentally relevant data for second-type behavioral phase transition points (e.g., sudden changes in water consumption corresponding to environmental factors such as seasonal temperature and sudden rainfall) are input into a multidimensional feature space to construct a coupled model of time, water consumption, and environmental factors. Feature engineering is used to standardize the characteristic parameters of the two phase transition points (e.g., periodic weights and environmental impact coefficients). Grid search is then used to optimize model hyperparameters, enabling the model to identify phase transition thresholds that simultaneously meet periodic patterns and environmental interference (e.g., a 20% sudden increase in water consumption between 7 PM and 9 PM in summer is considered a phase transition point). The model then outputs a fused set of behavioral phase transition points, enabling precise identification of turning points in user water use behavior.

[0059] In one possible implementation, step S220 further includes:

[0060] Step S224: determining a phase change element for the behavior phase change point, wherein the phase change element is a specific identification element of the behavior switching.

[0061] Step S225: Based on the phase change elements, conduct switching training of behavior phase recognition and discrimination and water use behavior profiling to construct the phase discriminator.

[0062] Specifically, phase transition factors are identified for each behavioral phase transition point. These phase transition factors are used to identify specific behavioral transitions. Key characteristic parameters that characterize the transition in water use behavior are extracted from these identified phase transition points. These phase transition factors include temporal characteristics, such as the specific time period and periodicity of the phase transition point; environmental characteristics, such as thresholds for seasonal changes in environmental factors like temperature and rainfall; and behavioral characteristics, such as fluctuations in water use and changes in the mix of water-using devices. For example, for a phase transition point associated with peak evening water use due to high summer temperatures, phase transition factors might include the specific time period (7 PM to 9 PM), the ambient temperature threshold (≥30°C), and the increase in water use compared to the daily average (≥20%). Together, these factors constitute specific indicators for identifying the transition from a normal state to a high-load summer state, providing a key basis for subsequent identification of behavioral phases and the transition of water use behavior profiles.

[0063] Based on these phase transition factors, a generative adversarial algorithm is employed to train behavioral phase identification and switching between water use behavior profiles. A phase discriminator is constructed, and the model is trained using the phase transition factors as input using the Generative Adversarial Network (GAN) framework. The generator is responsible for generating simulated behavioral phase feature sequences that strive to approximate the actual water use behavior phase. The discriminator determines whether the input feature sequence originates from real data or is generated by the generator, and identifies the specific behavioral phase to which it corresponds. During training, the generator is continuously optimized to generate more realistic phase features, while the discriminator strengthens its ability to distinguish between the actual and generated phases. This adversarial game between the two enables the discriminator to accurately distinguish different behavioral phases. For example, when fed phase transition factors such as high summer temperatures, evening hours, and a surge in water use, the discriminator can accurately identify the high summer evening water use phase and trigger the corresponding water use behavior profile switching. This results in a highly robust phase discriminator that dynamically identifies and switches between water use behavior phases.

[0064] In one possible implementation, step S300 further includes:

[0065] Step S310: Determine a water use task, wherein the water use task is any combination of a global task-a local task, or an actively initiated task-a passively generated task.

[0066] Step S320: locating a target water use unit by interpreting the water use task, wherein the target water use unit is the optimal cluster granularity that matches the task.

[0067] Step S330: According to the target water-using unit, a target cascade relationship is matched and determined in the vibration cascade relationship.

[0068] Specifically, identifying water use tasks requires clarifying their combined attributes across spatial and triggering mechanisms. The spatial dimension encompasses both global and local tasks. Global tasks are city-wide water management tasks, such as citywide water resource allocation and overall pressure control in the urban water supply network. Local tasks focus on specific areas of the city, such as pipeline maintenance in a specific community or water use monitoring in a specific industrial park. The triggering mechanism dimension encompasses proactive and reactive tasks. Proactive tasks are initiated based on pre-set plans or objectives, such as scheduling plans developed before the summer peak water use or annual water supply facility maintenance plans. Reactive tasks are triggered by emergencies or external events, such as emergency repairs following a water supply network leak or troubleshooting after user reports of water use anomalies. Water use tasks can be any combination of these two dimensions, such as global proactive tasks or local reactive tasks. By clarifying the combined attributes of tasks, a clear logical foundation is provided for subsequent task interpretation and the identification of target water use units.

[0069] By interpreting the water use task, the target water use unit is located. The target water use unit is the optimal cluster granularity for the task. After clarifying the combined attributes of the water use task (e.g., global - proactive initiation, local - passive generation), the specific requirements of the task are analyzed, such as the spatial scope, time requirements, and accuracy indicators. Then, within the multi-level water use unit (e.g., household, building, community, region, etc.), the cluster granularity that best meets the task requirements is matched. For example, for the local - passive generation task of leak detection in a community water supply network, the optimal cluster granularity is the building or household level, as it requires precise location to a specific building or household. The target water use unit is the specific building or household within the community. For the global - proactive initiation task of citywide summer peak water use scheduling, which requires water resource allocation at the city level, the optimal cluster granularity is the city or region level, with the target water use unit being the entire city or a specific water supply area. This process ensures that subsequent sensing and decision-making can be performed at the most appropriate scale by accurately matching task requirements with the granularity of water use units, thereby improving task execution efficiency and accuracy.

[0070] Based on the target water-using unit, the target cascade relationship is matched and determined within the vibration cascade relationship. After locating the target water-using unit that matches the water-using task (e.g., at a cluster granularity such as household, building, community, or region), the cascade mapping relationship corresponding to the target water-using unit is retrieved from a pre-built vibration cascade relationship database. The vibration cascade relationship records the correspondence between water-using units at each level and the distributed voiceprint sensor arrays. For example, when the target water-using unit is a community, the cascade relationship between the community-level water-using unit and the sensor array deployed on the community water supply main must be matched. If the target water-using unit is a single household, the cascade relationship between the household-level water-using unit and the first-level sensor array installed on the household's internal pipes must be matched. This matching mechanism accurately determines the required sensing path and signal acquisition range for the task, providing infrastructure support for the subsequent invocation of the target sensor array for directional vibration signal acquisition and triggering the scheduling decision maker to perform signal discrimination and phase analysis, ensuring that the entire water-using forecasting and scheduling process operates efficiently at the corresponding cluster granularity.

[0071] In one possible implementation, step S300 further includes:

[0072] Step S340: Based on the target cascade relationship, the target sensor array is called to sense the water ripples in the pipeline and determine the target vibration signal.

[0073] Step S350: transmitting the target vibration signal back to trigger the signal discriminator to discriminate between normal water usage signals and abnormal water usage signals.

[0074] Step S360: Based on the task scenario of the water use task, trigger the cascade judgment based on the phase discriminator to determine the target behavior profile.

[0075] Step S370: constructing the behavior state space according to the normal water usage signal and the target behavior portrait.

[0076] Step S380: Based on the behavior state space, water usage forecasting, scheduling, analysis and management are performed.

[0077] Specifically, after clarifying the mapping relationship between the target water use unit and the soundprint sensor array (i.e., the target cascade relationship), the target sensor array deployed at the corresponding location, such as the household-level sensor array, building-level sensor array, or regional-level sensor array, is called based on this relationship. These soundprint sensor arrays, distributed throughout the urban water pipe network, will sense the vibration of the water flow in the pipe in real time. By capturing the vibration frequency, amplitude, waveform, and other characteristics of the pipe wall, the water ripple vibration is converted into a corresponding electrical signal. These electrical signals are then pre-processed through filtering and amplification to remove noise interference, thereby determining the target vibration signal that accurately represents the current water use status, providing reliable raw data support for subsequent signal discrimination and water use behavior analysis.

[0078] The target vibration signal, collected and pre-processed by the target sensor array, is transmitted back to the dispatch decision maker via the city water system's communication network, automatically activating a built-in signal discriminator. This signal discriminator, built on a generative adversarial network (GAN) mechanism and trained using extensive historical data, accurately identifies the differences in vibration signatures between normal water use scenarios (such as washing and showering) and abnormal water use scenarios (such as pipe leaks, valve failures, and equipment anomalies). Upon inputting the target vibration signal, the signal discriminator compares it with pre-stored standard vibration templates in a voiceprint fingerprint library (covering characteristic parameters such as frequency, amplitude, and waveform for different water use behaviors). Using a pattern recognition algorithm, it performs a binary classification of normal / abnormal. For example, if the signal signature matches the high-frequency vibration template corresponding to a pipe leak above a preset threshold, it is identified as an abnormal water use signal; if it matches the low-frequency, stable vibration signature of daily washing, it is identified as a normal water use signal. This provides critical signal classification information for subsequent water use scheduling and anomaly warnings.

[0079] Based on the specific water usage scenario (such as global peak water usage scheduling or local pipeline leak detection), the phase discriminator's cascaded decision mechanism is activated. The phase discriminator combines identified phase transition factors (such as time, ambient temperature, and water consumption fluctuation amplitude, which are specific identification factors for behavior switching) to perform a multi-level decision on the phase of the current water usage behavior. For example, in a global peak water usage scheduling scenario during high summer temperatures, if the time is detected to be between 7 PM and 9 PM, the ambient temperature exceeds 30°C, and water consumption increases by 20% compared to the daily average, the phase discriminator uses a cascaded decision mechanism to sequentially match the corresponding phase transition factors, ultimately determining the target behavior profile as a summer evening high-load water usage profile. This cascaded decision process, leveraging the pre-stored behavioral phase transition rules and water usage behavior profile sequences in the phase discriminator, achieves a precise mapping from the task scenario to the specific behavior profile, providing a behavioral model-level basis for the subsequent construction of the behavior state space and the formulation of predictive scheduling strategies.

[0080] Based on the normal water use signal and the target behavior profile, the behavioral state space is constructed. The normal water use signal features (such as time-domain and frequency-domain characteristic parameters such as vibration frequency, amplitude, and waveform) output by the signal discriminator are multi-dimensionally fused with the target behavior profile (such as the morning washing profile and the emergency water use profile) determined by the phase discriminator to construct a behavioral state space in the spatiotemporal dimensions. This space is based on coordinates such as time, water consumption, vibration characteristics, environmental factors, and behavioral phases. Each dimension corresponds to a different characteristic parameter. For example, the time dimension is refined to hours and days, the vibration characteristic dimension includes dominant frequency and energy distribution, and the environmental factor dimension covers temperature, humidity, and other parameters. By associating and mapping the feature vectors of the normal water use signal with the labels of the target behavior profile, specific coordinate points or regions are formed in the state space, each of which represents a specific water use behavior pattern. For example, the low-frequency vibration signal of the faucet detected between 6 and 8 in the morning (normal water usage signal) is combined with the "morning washing portrait" to construct a corresponding state point in the behavioral state space, intuitively representing the typical water usage behavior characteristics of users during this period, and providing a structured data model for subsequent water use prediction and scheduling analysis based on the state space.

[0081] Based on the behavioral state space, a long short-term memory (LSTM) neural network is employed for water use forecasting, scheduling, analysis, and management. The constructed behavioral state space data (including multidimensional features such as time series, vibration characteristics, and behavioral phase labels) is used as input for the LSTM model. Leveraging its ability to process long-term dependencies, the LSTM model learns the temporal periodicity and environmental sensitivity of water use behavior. For example, the model is fed with behavioral state space data from historically high summer evening water use phases (e.g., vibration signal characteristics, temperature parameters, and water consumption fluctuations between 7 PM and 9 PM). Through training, the LSTM network is trained to capture the changing patterns of water consumption during this phase. During forecasting, the model combines the current behavioral state space point with future timestamps and environmental forecast data (e.g., weather forecast temperature) to generate a series of water consumption forecasts for the future. This is then used to formulate scheduling strategies. When a peak water use forecast is predicted for a particular area, the LSTM model's output triggers the city water system to preemptively adjust pump station pressures, optimize pipe network flow distribution, or activate backup water sources. This achieves a closed-loop management system from data modeling to scheduling execution, improving water use efficiency and the accuracy of system responses.

[0082] Example 2: Figure 2 3 is a schematic structural diagram of a storage device provided by an embodiment of the present invention, showing a block diagram of an exemplary storage device suitable for implementing an embodiment of the present invention. Figure 2The storage device shown is merely an example and should not limit the functionality or scope of use of the embodiments of the present invention. The storage device is implemented as a general-purpose computing device, whose components may include, but are not limited to, an input device 201, a processor 202, a memory 203, and an output device 204. There may be one or more processors 202; processor 202 executes the software programs, instructions, and modules stored in memory 203 to perform various functional applications and data processing of the computer device, thereby implementing the aforementioned urban water use forecasting and scheduling method based on user behavior modeling.

[0083] It should be noted that the order in which the embodiments of the present application are presented is for illustrative purposes only and does not necessarily represent the superiority or inferiority of the embodiments. Furthermore, the foregoing descriptions of specific embodiments of this specification are provided. Furthermore, the processes depicted in the accompanying drawings do not necessarily require the specific order or sequential sequence shown to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0084] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application should be included in the scope of protection of the present application.

[0085] This specification and drawings are merely illustrative of the present application and are intended to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Obviously, those skilled in the art may make various modifications and variations to this application without departing from the scope of this application. Thus, this application is intended to include such modifications and variations as fall within the scope of this application and its equivalents.

Claims

1. Urban water use prediction and scheduling method based on user behavior modeling, characterized by: The method comprises: Based on the user cluster topology of the target city, a vibration cascade relationship is established by dividing the water use unit into multiple levels. The vibration cascade relationship is the voiceprint sensing topology corresponding to the water use unit network at different cluster granularities. By combining time-based behavior phase changes with external environmental behavior phase changes, we can determine the behavior phase change points and construct a water use behavior portrait sequence. Combined with the voiceprint fingerprint library, we can develop a scheduling decision maker for the urban water system. The voiceprint fingerprint library is constructed from pipeline vibration characteristics and water flow behavior. The city water system receives water use tasks, interprets the tasks and makes decisions based on vibration cascade relationships, performs directional sensing at the task cluster granularity, transmits back and triggers the scheduling decision maker, performs signal normality and abnormality binary classification and parallel judgment of behavior phase, constructs a behavior state space, makes long-term and short-term decisions, and determines the prediction scheduling strategy; According to the urban water system, the forecast scheduling strategy is issued and executed.

2. The urban water use prediction and scheduling method based on user behavior modeling according to claim 1, characterized in that: Deploy a voiceprint sensor array, wherein the voiceprint sensor array is distributedly deployed in the city water network; Obtain the user cluster topology of the target city, and establish a vibration cascade relationship between the user cluster topology and the voiceprint sensor array.

3. The urban water use prediction and scheduling method based on user behavior modeling according to claim 2, characterized in that: By dividing the water use units into multiple levels, a vibration cascade relationship is established, including: Divide the user cluster topology into multiple levels of water use units based on the cluster granularity. For the multi-level water use units, a first-level water use unit is determined, wherein the first-level water use unit is the smallest cluster granularity, and the household is the water use unit; For the primary water use unit, determining a primary sensor array in the voiceprint sensor array, wherein the primary sensor array is deployed in the primary water use unit; A cascade connection between the primary water-using unit and the primary sensor array is established as a first vibration cascade relationship.

4. The urban water use prediction and scheduling method based on user behavior modeling according to claim 3 is characterized in that: Traversing the multi-level water use units, performing matching and cascading based on the voiceprint sensor array, and determining an Nth vibration cascade relationship, wherein the Nth vibration cascade relationship uses the largest cluster granularity as the water use unit; The first vibration cascade relationship to the Nth vibration cascade relationship are added to the vibration cascade relationship.

5. The urban water use prediction and scheduling method based on user behavior modeling according to claim 1, characterized in that: Develop a scheduling decision maker, including: Generate adversarial training using normal water use versus abnormal water use to determine the signal discriminator; Generate adversarial training using the behavioral phase transition points and the behavioral portrait sequence to determine a behavioral phase discriminator; The scheduling decision maker is constructed according to the signal discriminator and the phase discriminator, wherein the voiceprint fingerprint library is built into the scheduling decision maker.

6. The urban water use prediction and scheduling method based on user behavior modeling according to claim 5 is characterized in that: Determination of behavioral phase transition points, including: Call the water use records of the target city, set the user's water use behavior cycle and conduct periodic time-sharing mining to determine a type of behavior phase change point; Using seasonal environmental evolution and sudden environmental changes to explore two types of behavioral phase transition points; The first type of behavior phase transition point and the second type of behavior phase transition point are fitted to determine the behavior phase transition point.

7. The urban water use prediction and scheduling method based on user behavior modeling according to claim 6 is characterized in that: Determine the behavior phase discriminator, including: Determining a phase change element for the behavior phase change point, wherein the phase change element is a specific identification element for the behavior switching; According to the phase change elements, the switching training of behavior phase recognition and water use behavior profiling is carried out to construct the phase discriminator.

8. The urban water use prediction and scheduling method based on user behavior modeling according to claim 5, characterized in that: Perform task interpretation and decision making based on vibration cascades, including: Determine a water use task, wherein the water use task is any combination of a global task and a local task, or an actively initiated task and a passively generated task; By interpreting the water use task, a target water use unit is located, wherein the target water use unit is the optimal cluster granularity that matches the task; According to the target water usage unit, a target cascade relationship is matched and determined in the vibration cascade relationship.

9. The urban water use prediction and scheduling method based on user behavior modeling according to claim 8, characterized in that: Perform directional sensing at the task cluster granularity, transmit back and trigger the scheduling decision maker, perform signal normality and abnormality binary classification and behavioral phase determination in parallel, including: According to the target cascade relationship, the target sensor array is called to sense the water ripples in the pipeline and determine the target vibration signal; The target vibration signal is transmitted back to trigger the signal discriminator to discriminate between normal water use signals and abnormal water use signals; Based on the task scenario of the water use task, triggering the cascade judgment based on the phase discriminator to determine the target behavior profile; constructing the behavior state space according to the normal water use signal and the target behavior portrait; Based on the behavior state space, water use forecasting, scheduling, analysis and management are carried out.

10. A storage device, characterized in that: The storage device includes: a memory for storing executable instructions; The processor is configured to implement the urban water use prediction and scheduling method based on user behavior modeling according to any one of claims 1 to 9 when executing the executable instructions stored in the memory.

Images