A digital twin intelligent water-saving system based on a distributed fiber optic sensor network

Abstract

This invention discloses a digital twin intelligent water-saving system based on a distributed optical fiber sensor network, belonging to the field of water-saving management technology. Specifically, it includes: a distributed optical fiber sensor network laid along the water pipeline in a tree-like topology; a water use event identification module that collects optical fiber vibration signals, converts them into digital waveforms, and identifies trigger points through amplitude thresholds; a water supply topology construction module that constructs a water supply dependency topology graph containing nodes and directed edges based on the spatial location and time delay of the trigger points; a hydraulic model generation module that overlays the topology graph with a digital map of the water pipeline to generate a dynamic hydraulic balance model containing node water use intensity weight coefficients; and an intelligent water-saving control module that sends periodic intermittent closing commands to the intelligent water valves of low-weight nodes based on changes in node weights, with the command duration proportional to the weight difference between adjacent nodes. This invention achieves refined intelligent water-saving management for centralized water use scenarios.

Description

Technical Field

[0001] This invention relates to the field of water conservation management technology, specifically to a digital twin intelligent water conservation system based on a distributed optical fiber sensor network. Background Technology

[0002] The water supply networks in urban centralized water supply systems are becoming increasingly complex, with water lines in schools, residential areas, industrial parks, and business centers exhibiting multi-level branching characteristics. Distributed fiber optic sensing technology, with its long-distance continuous sensing capabilities, has been applied to the field of pipeline physical condition monitoring, locating leaks by capturing abnormal vibration signals. However, existing technologies mainly focus on post-incident leak handling and lack effective technical means for systemic water conservation in pipeline networks.

[0003] Current intelligent water-saving systems primarily rely on two basic architectures: one based on discretely installed intelligent metering terminals, which collect data from water-using endpoints to construct water usage pattern analysis models; the other employs fixed pressure sensor arrays to locate leak areas based on pipeline pressure fluctuations. Some systems introduce a digital twin framework, mapping the physical pipeline structure to a computational model and adjusting valve openings or pump station power based on simulation results. Related research proposes using multi-type sensor data to construct a digital mirror of the pipeline network, and achieving water supply equipment control through hydraulic model calculations.

[0004] Existing methods still suffer from the following technical limitations: First, point-based sensor deployment requires substantial hardware, significantly increasing system construction costs and resulting in insufficient monitoring continuity. Second, the digital twin model's topology relies on pre-set pipeline network drawings, and cannot autonomously update hierarchical relationships when the actual water supply path changes due to valve switching. More importantly, current technology only triggers a response when a clear leak signal or abnormal water consumption is detected, failing to implement pre-emptive water-saving intervention based on the dynamic topology of the water supply network. Especially in multi-branch pipeline network environments, existing solutions have not yet overcome the technical bottleneck of automatically constructing real-time water supply dependency structures through continuous sensor data, and also lack an effective mechanism to directly translate dynamic topology relationships into water-saving control strategies. These technical limitations hinder a fundamental improvement in water-saving efficiency. Summary of the Invention

[0005] The purpose of this invention is to provide a digital twin intelligent water-saving system based on a distributed optical fiber sensor network, solving the following technical problems:

[0006] Existing point-based sensing deployment methods are costly, and digital twin topologies rely on pre-defined drawings that cannot be automatically updated, lacking proactive intervention based on dynamic topologies.

[0007] The objective of this invention can be achieved through the following technical solutions:

[0008] A digital twin intelligent water-saving system based on a distributed optical fiber sensor network includes:

[0009] The distributed optical fiber sensor network consists of a tree-like topology formed by a single trunk optical fiber and multiple branch optical fibers, and is continuously laid along the physical path of the water pipeline.

[0010] The water use event identification module is used to collect fiber optic vibration signals in real time and convert them into digital waveform sequences. By setting a waveform amplitude threshold, the trigger point of the water use event can be identified.

[0011] The water supply topology construction module is used to automatically construct a water supply dependency topology graph between water-using units based on the spatial location of the water use event trigger point and the time delay relationship between adjacent trigger points. Nodes in the topology graph represent water-using units, and directed edges in the topology graph represent water supply directions.

[0012] The hydraulic model generation module is used to spatially overlay the water supply dependency topology map with the preset digital map of water transmission pipelines to generate a dynamic hydraulic balance model, which includes the weight coefficient of water intensity of each node.

[0013] The intelligent water-saving control module is used to send a series of periodic intermittent closing commands to the intelligent water valves associated with nodes with low weight coefficients, based on the real-time changing trend of the node weight coefficients in the dynamic hydraulic balance model, without the occurrence of physical leakage. The duration of the command sequence is proportional to the difference in weight coefficients between adjacent nodes.

[0014] As a further aspect of the present invention: the process of constructing the water supply dependency topology graph in the water supply topology construction module is as follows:

[0015] When a water use event trigger point is detected at the end of a branch fiber, the first vibration peak time of the water use event trigger point is recorded; if the sensing point on the main fiber closest to the end of the branch fiber has a trigger point with a vibration waveform similarity exceeding the threshold within a set time window, it is determined that the branch node depends on the main node for water supply.

[0016] For branch nodes at the same level, if the triggering time interval between two branch nodes is less than the propagation time of the vibration wave in the pipeline, and the waveform envelope shape of the downstream branch node matches the attenuation shape of the upstream branch node, then a directed edge from the upstream branch node to the downstream branch node is established.

[0017] After all dependencies are determined, node groups with the same water supply source are merged to form a hierarchical topology. The top node of the hierarchical topology is the main water supply inlet of the park, and the bottom node of the hierarchical topology is the terminal water unit.

[0018] As a further aspect of the present invention: the method for determining the similarity of the vibration waveforms is as follows:

[0019] The original vibration waveforms of a fixed time length before and after the trigger point of the branch node are taken as the matching sequence, and the vibration waveforms of the candidate nodes of the trunk fiber corresponding to the time window are taken as the reference sequence. The length of the time window is calculated by dividing the physical distance between the branch fiber and the trunk fiber by the water hammer wave velocity under the pipe diameter.

[0020] The sequence to be matched and the reference sequence are non-linearly aligned on the time axis. The minimum cumulative distance on the alignment path is calculated. When the minimum cumulative distance is less than the reference value of sound wave propagation loss determined by the pipe material, the waveform similarity is deemed to meet the standard.

[0021] As a further aspect of the present invention: in the hydraulic model generation module, the process of constructing the digital map of the water transmission pipeline is as follows:

[0022] The characteristic vibration modes of pipe fittings are identified by optical fiber vibration signals. These include vortex-induced vibration waveforms at elbows, water hammer wavefront characteristics during valve opening and closing, and specific frequency harmonics during pump startup.

[0023] The location of pipe fittings is determined based on the spatial distribution of characteristic vibration modes; the actual length of the pipe section is inverted using the time difference of vibration wave propagation on the continuous pipe section; the identification results are compared with the design drawings to generate a three-dimensional pipeline model with topological attributes, and the update cycle of the three-dimensional pipeline model is synchronized with the water supply dependent topology map.

[0024] As a further aspect of the present invention: the process of generating a dynamic hydraulic balance model in the hydraulic model generation module is as follows:

[0025] Based on a water supply dependency topology map, the system overlays pipe diameter change points, pump station locations, and reservoir coordinates from a digital map of water pipelines. Each node in the water supply dependency topology map is assigned an initial weight coefficient, the value of which is determined by the type of water-using unit associated with the node. The system continuously monitors the number of water-using events triggered per unit time for each node, calculating water intensity by combining the waveform integral area corresponding to each event trigger. Water intensity data is updated using a sliding time window to update node weight coefficients; nodes with a weight coefficient change rate exceeding a threshold are automatically marked as abnormal nodes. When the weight coefficient ratio between adjacent nodes continuously deviates from the square of the pipe diameter ratio, a topology verification process is triggered.

[0026] As a further aspect of the present invention: the process of generating a periodic intermittent shutdown command sequence in the intelligent water-saving control module is as follows:

[0027] Obtain the weight coefficient change curves of the target node in the dynamic hydraulic balance model for several consecutive time periods, and extract the steady phase and rising phase from the weight coefficient change curves.

[0028] During the stable phase, a shutdown command is generated, and the duration of the shutdown command is equal to the historical average water usage interval of the target node multiplied by the weight coefficient decay factor; a pre-shutdown command is inserted before the rising phase, and the duration of the pre-shutdown command is negatively correlated with the rising slope of the weight coefficient.

[0029] For sibling nodes that rely on the same water supply source, a staggered shutdown timing table is generated after arranging them in ascending order according to their weight coefficients. The time interval between the start of shutdown instructions of adjacent nodes in the staggered shutdown timing table is greater than the time required for water pressure fluctuations to stabilize.

[0030] As a further aspect of the present invention: the weighting coefficient attenuation factor is calculated as follows:

[0031] Establish a correlation function between the target node's weight coefficient and the standard deviation of the weight coefficients of other nodes at the same level. When the standard deviation is less than a threshold, a fixed attenuation factor is used.

[0032] When the standard deviation is greater than the threshold, the attenuation factor is dynamically adjusted according to the deviation of the target node from the average weight coefficient. That is, when the weight coefficient of the target node is lower than the average weight coefficient, the attenuation factor is increased, and when the weight coefficient of the target node is higher than the average weight coefficient, the attenuation factor is decreased. The adjustment range of the attenuation factor is inversely proportional to the hierarchical depth from the target node to the root node in the water supply dependency topology graph.

[0033] As a further aspect of the present invention, it also includes a meteorological data fusion module, specifically:

[0034] Real-time access to rainfall intensity, evaporation, and sunshine duration data from regional meteorological stations, converting the rainfall intensity, evaporation, and sunshine duration data into regional water demand compensation coefficients;

[0035] When the regional water demand compensation coefficient is consistently lower than the threshold, a virtual water source node is added to the dynamic hydraulic balance model. The weight coefficient of the virtual water source node is negatively correlated with the regional water demand compensation coefficient. A bidirectional connection edge is established between the virtual water source node and the physical water storage tank node, triggering a smart water valve opening priority redistribution strategy.

[0036] As a further aspect of the present invention: the function of the virtual water source node is as follows:

[0037] When the regional water demand compensation coefficient decreases, the virtual water source node automatically increases its weight coefficient value. The weight coefficient increase value is passed from top to bottom according to the hierarchical structure of the water supply dependence topology. During the transmission process, the weight coefficient increase value decreases equally according to the number of out-degrees of the node as it passes through a node in the water supply dependence topology.

[0038] When the deviation between the actual water level sensor data of the physical water storage node and the theoretical water level calculated by the virtual water source node exceeds the tolerance limit, the water supply dependency topology reconstruction process is triggered.

[0039] The beneficial effects of this invention are:

[0040] This invention replaces traditional point-based sensors with a distributed fiber optic sensor network of a tree-like topology continuously laid along the water pipeline, significantly reducing hardware deployment costs while achieving real-time monitoring of the entire water pipeline, thus solving the problem of discontinuous monitoring by point sensors. Based on the spatial location and time delay relationship of water usage event trigger points, a dynamic water supply dependency topology map is automatically constructed. This map can autonomously update its hierarchical structure according to changes in the actual water supply path, breaking the dependence of digital twin topologies on preset maps and forming a real-time water supply structure that matches the actual pipeline network. This topology map is overlaid with a digital map of the water pipeline to generate a dynamic hydraulic balance model. The node weight coefficients are dynamically adjusted based on water usage intensity to accurately map the physical pipeline network status. Based on the trend of weight coefficient changes, periodic intermittent shutdown commands proportional to the weight difference between adjacent nodes are sent to low-weight nodes. This peak-shaving control achieves pre-emptive water-saving intervention, changing the existing technology's passive response mode that only occurs after leakage or exceeding limits. Meteorological data is introduced to generate regional water demand compensation coefficients, and the water supply strategy is dynamically adjusted through virtual water source nodes, improving the system's adaptability to environmental changes. This invention comprehensively improves water-saving efficiency through the synergistic effect of the above technologies, and is applicable to refined water management in various scenarios such as campuses, residential areas, and industrial parks. Attached Figure Description

[0041] The invention will now be further described with reference to the accompanying drawings.

[0042] Figure 1 This is a schematic diagram of the modules of the present invention. Detailed Implementation

[0043] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0044] Please see Figure 1 As shown, this invention is a digital twin intelligent water-saving system based on a distributed optical fiber sensor network, comprising:

[0045] The distributed optical fiber sensor network adopts a tree topology design, consisting of a single trunk optical fiber and multiple branch optical fibers, which are continuously laid along the physical path of the water transmission pipeline. The trunk optical fiber covers the main water transmission line, and the branch optical fibers extend to each terminal water-using unit, forming a seamless monitoring network covering the entire area. This network can comprehensively capture vibration signals within the pipeline, providing basic data for subsequent analysis.

[0046] The water usage event identification module is responsible for real-time acquisition of vibration signals transmitted through the distributed fiber optic network. These signals originate from various dynamic changes in water flow within the pipeline, such as water flow impact when terminal water units are turned on, water hammer effects caused by valve opening and closing, and continuous disturbances during the operation of water-using equipment. The module converts the acquired analog vibration signals into quantifiable and analyzable digital waveform sequences. During this process, basic signal noise reduction processing is performed simultaneously to filter out irrelevant interference such as environmental noise. Then, the module identifies valid water usage events based on preset waveform amplitude thresholds. Different water usage behaviors in different scenarios will produce characteristic vibration amplitudes. For example, the waveform amplitude when high-flow water-using equipment is started is usually higher than that of daily low-flow water usage. The module sets differentiated threshold standards accordingly. When the vibration waveform amplitude at a certain location exceeds the corresponding threshold, it is determined as a water usage event trigger point. The module not only records the precise time of the event but also, combined with the spatial distribution information of fiber optic sensor points, accurately marks the specific location of the trigger point within the physical pipeline, providing accurate event data for subsequent topology construction.

[0047] The water supply topology construction module operates based on the spatial location of water usage event trigger points and the time delay relationship between adjacent trigger points. The system first determines the specific location of each trigger point within the physical pipeline. Then, by analyzing the time difference between adjacent trigger points—for example, the vibration signal delay between a trigger point at the end of a branch fiber and the corresponding sensor point on the main fiber—it determines the water supply direction and constructs a water supply dependency topology graph. Nodes in the graph represent water-using units, and directed edges indicate the water supply direction, ultimately forming a hierarchical structure from the main water supply inlet to the terminal water-using units, which can be dynamically adjusted according to changes in the actual water supply path.

[0048] The hydraulic model generation module spatially overlays a water supply dependency topology map with a pre-set digital map of water transmission pipelines. The digital map contains physical information such as pipe diameter change points, pump station locations, and reservoir coordinates. The fusion of these two maps generates a dynamic hydraulic balance model. Each node in the model is assigned a weight coefficient, which is updated in real-time based on indicators reflecting water usage intensity, such as the number of water usage events per unit time and the integral area of ​​the waveform, ensuring that the virtual model accurately maps the real-time state of the physical pipeline network.

[0049] The intelligent water-saving control module operates based on the real-time changing trends of node weight coefficients in the dynamic hydraulic balance model. When a node with a low weight coefficient is identified—that is, a unit with low water intensity or room for optimization—the module sends a series of periodic intermittent shut-off commands to its associated intelligent water valve. The duration of the commands is related to the difference in weight coefficients between adjacent nodes; the larger the difference, the longer the shut-off time. Simultaneously, the module uses peak-shaving control to avoid water pressure fluctuations caused by simultaneous valve closures at nodes in the same area, proactively intervening to reduce ineffective water consumption in the absence of physical leakage.

[0050] In the water supply topology construction module, the process of constructing the water supply dependency topology graph is as follows:

[0051] When a water usage event trigger point is detected at the end of a branch fiber optic cable, the module immediately records the first peak vibration time at that trigger point. This time marker is a crucial benchmark for determining the water flow direction, as the water flow propagating in the pipeline generates vibration waves, and the time difference from the source to the end directly reflects the path correlation. Subsequently, the system focuses on the sensor point on the trunk fiber closest to the end of the branch fiber optic cable, monitoring its vibration signal changes within a set time window. The time window setting must be combined with the pipeline's physical characteristics to ensure it covers a reasonable duration for the vibration wave to travel from the trunk to the branch, avoiding both missing effective signals due to an excessively short window and introducing irrelevant interference due to an excessively long window. If a trigger point with a vibration waveform similarity exceeding a threshold appears at the trunk sensor point within the window, it indicates a clear correlation between the two vibration signals. This determines that the branch node depends on the trunk node for water supply, thus establishing a basic connection from the trunk node to the branch node in the topology graph.

[0052] For branch nodes at the same level, the module employs a more refined association determination logic. When two branch nodes at the same level experience water usage event triggers successively, the system first calculates the trigger time interval between the two and compares it with the propagation time of the vibration wave in the pipeline. If the trigger time interval is less than the propagation time, it indicates that the vibration signal of the former node has sufficient time to propagate to the latter node, suggesting a possible water supply association. Next, the system performs morphological analysis on the vibration waveforms of the two nodes, focusing on comparing the waveform envelope shape of the downstream branch node with the waveform attenuation shape of the upstream branch node. Since the waveform exhibits a specific attenuation pattern due to factors such as friction and pipe diameter changes when water propagates in the same level pipeline, if the two shapes match, a direct water supply transmission relationship can be confirmed, and a directed edge from the upstream branch node to the downstream branch node can be established in the topology graph.

[0053] Once the dependencies between all nodes are determined, the module will merge nodes with the same water supply source. For example, if multiple branch nodes all depend on the same trunk node for water supply, they will be grouped into the same node group, thus simplifying the topology and highlighting the hierarchical relationship. In the final hierarchical topology, the top-level node corresponds to the main water supply inlet of the campus, serving as the source of the entire water supply network; the bottom-level nodes correspond to various terminal water-using units, such as dormitory faucets and workshop equipment interfaces; and the intermediate-level nodes are distributed sequentially according to the branching relationships of the actual water supply paths, ensuring that the topology diagram can completely and accurately reflect the hierarchical architecture of the physical pipe network.

[0054] Determining the similarity of vibration waveforms is a crucial step in ensuring the accurate establishment of water supply dependencies. Through refined waveform comparison, the influence of interference signals on the correlation determination is eliminated. Specifically:

[0055] First, the original vibration waveforms of a fixed time length before and after the trigger point of the branch node are extracted as the matching sequence. This sequence contains the complete vibration process caused by the water use event and is the basis for feature extraction. Simultaneously, the vibration waveforms of the candidate nodes in the main fiber corresponding to the time window are extracted as the reference sequence. The time window length of the reference sequence needs to be determined comprehensively based on the physical distance between the branch fiber and the main fiber and the water hammer wave velocity under the pipe diameter to ensure that the two sequences are comparable in the time dimension and can accurately reflect the propagation correlation of the vibration waves.

[0056] Subsequently, the system performs nonlinear time-axis alignment processing on the extracted sequences to be matched and the reference sequence. Since vibration waves may shift on the time axis due to differences in attenuation rates when propagating in pipelines of different materials and diameters, nonlinear alignment eliminates this shift, ensuring precise temporal correspondence between the characteristic waveforms of the two sequences. Based on this, the minimum cumulative distance along the alignment path is calculated. This value reflects the overall similarity between the two waveforms; the smaller the value, the closer the waveform characteristics. When the minimum cumulative distance is less than the sound wave propagation loss benchmark determined by the pipe material, the waveform similarity is considered to be satisfactory. Different pipe materials result in different sound wave propagation loss characteristics. This benchmark value fully considers the influence of the material on the vibration signal, ensuring that the similarity judgment standard matches the actual pipeline characteristics, thus providing a reliable basis for accurately determining water supply dependence.

[0057] In the hydraulic model generation module, the process of constructing the digital map of the water transmission pipeline is as follows:

[0058] This system utilizes fiber optic vibration signals to identify the characteristic vibration modes of pipe fittings. Different pipe fittings exhibit unique vibration characteristics under the influence of water flow: at elbows, abrupt changes in water flow direction create periodic vortex shedding, resulting in vortex-induced vibration waveforms with specific frequency fluctuations and amplitudes that increase and decrease systematically with water flow velocity; when valves open and close, the rapid change in water flow conditions triggers water hammer, characterized by sudden increases and decreases in vibration amplitude, forming steep waveform fronts, with significant differences in wavefront polarity between opening and closing; when a water pump starts, the combined effect of motor operation and water flow generates harmonics of specific frequencies, whose frequencies are directly related to pump power and speed, forming identifiable characteristic spectra. The system accurately identifies these characteristic vibration modes by performing spectral analysis and waveform morphology comparison on the vibration signals acquired via fiber optics, thereby distinguishing different types of pipe fittings.

[0059] After identifying characteristic vibration patterns, the system locates pipe fittings based on the spatial distribution of these vibration signals. The continuous deployment of the fiber optic sensor network ensures that each vibration signal corresponds to specific physical coordinates. By matching the location information of characteristic vibration patterns, the specific installation points of fittings such as elbows, valves, and pumps within the actual pipeline can be determined. For continuous pipeline sections, the system uses the time difference of vibration wave propagation within the pipeline to invert the actual length: the time interval between the vibration wave propagating from one end of the section to the other, combined with the propagation characteristics of vibration waves in this type of pipeline, can be used to calculate the actual length of the pipeline, effectively correcting potential dimensional deviations in the design drawings.

[0060] Finally, the system compares all identified pipe fitting locations, pipe lengths, and other information with the original design drawings, filling in missing details or correcting deviations to generate a 3D pipeline model containing topological attributes. These topological attributes refer to the consistency between the connection relationships between pipe fittings and segments in the model and the actual pipe network, such as the connection angle between elbows and adjacent straight pipes, and the connection method between valves and upstream / downstream pipes. The update cycle of the 3D pipeline model is synchronized with the water supply-dependent topology map, ensuring that when the physical pipe network changes due to maintenance or renovation, the virtual model can promptly reflect these adjustments, providing accurate basic data for the subsequent construction of the hydraulic model.

[0061] The process of generating a dynamic hydraulic balance model by the hydraulic model generation module is to integrate the physical characteristics of the pipeline network with the real-time water usage status to construct a virtual model that can reflect the dynamic operation of the pipeline network.

[0062] This process is based on a water supply dependency topology map, which clearly presents the hierarchical relationships and water supply directions of each water-using unit. On this basis, the system overlays key physical information from the digital map of the water pipeline: pipe diameter change points mark abrupt changes in the pipe cross-section, directly affecting water flow velocity and pressure distribution; pump station locations are associated with the water supply power source, and their operating status determines the pressure level of the pipeline network; the coordinates of the reservoirs reflect the spatial distribution of water reserves, affecting the water supply stability of local areas. This overlay of information endows the model with the basic properties of the physical pipeline network, providing parameter support for the simulation of hydraulic states.

[0063] The next step in model building is to assign initial weight coefficients to each node in the water supply dependency topology graph. The initial values ​​are set based on the type of water-using unit associated with each node: for example, dormitory nodes have lower initial weight coefficients because of their regular daily water usage and relatively stable single-use volume; while factory nodes in industrial parks have relatively higher initial weight coefficients because they may involve high-volume equipment water usage. This setting allows the model to initially match the water usage characteristics of different scenarios.

[0064] Subsequently, the system calculates water usage intensity by statistically analyzing the number of water usage events triggered per unit time for each node in real time, and combining this with the waveform integral area corresponding to each trigger. The waveform integral area comprehensively reflects the flow rate and duration of a water usage event; a larger integral area indicates higher water usage intensity. Water usage intensity data is updated with node weight coefficients using a sliding time window. The length of the time window is set according to the frequency of changes in the water usage scenario, ensuring that the weight coefficients promptly reflect recent changes in water usage status.

[0065] When the rate of change of the weight coefficient of a node exceeds a threshold, the system automatically marks it as an abnormal node, indicating that the node may have abnormal water usage behavior or changes in the pipeline network status. More importantly, the system continuously monitors the weight coefficient ratio between adjacent nodes: since pipe diameter directly affects water flow carrying capacity, under normal circumstances, the weight coefficient ratio of adjacent nodes should be consistent with the square of the pipe diameter ratio. This relationship reflects the intrinsic correlation between pipe diameter and flow rate. If this ratio continues to deviate, it indicates that the model's topology may not match the actual water supply path. The system will then trigger a topology verification process to re-examine the dependencies between nodes, ensuring that the dynamic hydraulic balance model can always accurately map the operating status of the physical pipeline network.

[0066] In the intelligent water-saving control module, the process of generating a periodic intermittent shutdown command sequence is as follows:

[0067] First, the system obtains the weight coefficient variation curves of the target nodes over several consecutive time periods in the dynamic hydraulic balance model. These curves record the fluctuations in water usage intensity at different times. For example, the campus dormitory node may exhibit periodic differences between weekdays and weekends, while the industrial park workshop node may show regular fluctuations with changes in production shifts. The system extracts the stable and rising phases through curve morphology analysis. The stable phase typically corresponds to periods of stable water usage intensity at the node, such as low and gentle fluctuations in water usage in the dormitory area late at night; the rising phase indicates that water demand is about to increase, such as a gradual increase in water usage intensity before the morning rush hour.

[0068] During the stable phase, the module generates a shutdown command. The duration of the shutdown command is determined by the target node's historical average water usage interval and the weighting coefficient decay factor. The historical average water usage interval reflects the typical duration between two water usage events at that node; for example, an office node might experience concentrated water usage on average every two hours, and this time interval provides a reference for the base duration of the shutdown command. The weighting coefficient decay factor is dynamically adjusted according to the relative level of the node's water usage intensity, ensuring that the shutdown duration matches the node's actual water-saving potential.

[0069] Before the peak water demand arrives, the module inserts a pre-shutdown command. The purpose of the pre-shutdown command is to briefly close the valve before water demand increases, avoiding ineffective water flow before the peak period. Its duration is negatively correlated with the slope of the weighting coefficient: if the curve rises steeply, it indicates that water demand is increasing rapidly, and the pre-shutdown duration needs to be shortened to avoid affecting normal water use; if the rise is gentle, the pre-shutdown duration can be appropriately extended to increase water-saving effects.

[0070] For sibling nodes that rely on the same water source, the module employs a staggered shutdown strategy. The system sorts sibling nodes in ascending order according to their weight coefficients, with nodes having lower weight coefficients entering the shutdown sequence first. In the generated staggered shutdown timing table, the time interval between the shutdown commands of adjacent nodes is set to be greater than the time required for water pressure fluctuations to stabilize. This is because multiple nodes closing valves simultaneously may cause a sudden increase in local network water pressure, leading to pipeline vibration or equipment damage. Staggered control can mitigate such pressure fluctuations through time differences, ensuring stable network operation.

[0071] The calculation of the weighting coefficient attenuation factor is the key to achieving differentiated shutdown control. By dynamically adjusting the factor size, water-saving intervention can be more closely aligned with the actual water use characteristics of the nodes.

[0072] The module first establishes a correlation function between the weight coefficient of the target node and the standard deviation of the weight coefficients of other nodes at the same level. The standard deviation reflects the overall dispersion of water use intensity among nodes at the same level: a small standard deviation indicates that the differences in water use intensity among nodes are small, and the overall situation is relatively balanced; a large standard deviation indicates that there is obvious differentiation in water use intensity.

[0073] When the standard deviation is less than the threshold, the system uses a fixed attenuation factor. In this case, the water usage patterns of nodes at the same level are similar, and a uniform attenuation factor can simplify calculations while ensuring the consistency of control effects. For example, when the water usage intensity of households on the same floor of a residential community is relatively small, a fixed factor can avoid overly complex adjustments.

[0074] When the standard deviation is greater than the threshold, the attenuation factor is dynamically adjusted according to the deviation of the target node from the average weight coefficient. If the weight coefficient of the target node is lower than the average weight coefficient, it indicates that its water intensity is low and there is a large water-saving space. In this case, the attenuation factor is increased to prolong the duration of the shutdown command, thereby enhancing the water-saving effect. If the weight coefficient of the target node is higher than the average weight coefficient, it indicates that its water demand is relatively urgent. The attenuation factor needs to be reduced and the shutdown time shortened to ensure water supply.

[0075] Meanwhile, the adjustment range of the attenuation factor is inversely proportional to the hierarchical depth from the target node to the root node in the water supply dependency topology graph. As the water supply source, nodes with shallow hierarchical depth (such as main pipe nodes close to the main water source) have a wider impact on the entire network, and their adjustment range needs to converge to avoid global water pressure fluctuations. Conversely, nodes with deep hierarchical depth (such as terminal water units) have a limited impact range, and their adjustment range can be appropriately amplified to improve local water-saving efficiency. This hierarchical adjustment logic ensures both water-saving effects and maintains the overall stability of the network operation.

[0076] This invention also includes a meteorological data fusion module. As a key component for the system to adapt to dynamic environmental changes, the meteorological data fusion module deeply couples meteorological factors with the pipeline hydraulic model to achieve environmentally adaptive adjustments to water-saving strategies. Specifically, it includes:

[0077] First, a real-time data connection is established with regional meteorological stations to continuously acquire three core meteorological parameters: rainfall intensity, evaporation, and sunshine duration. These parameters directly affect the actual water demand of the water system: rainfall intensity reflects natural water replenishment capacity; the greater the intensity, the higher the natural moisture content of the surface and vegetation, and the lower the demand for artificial irrigation. Evaporation and sunshine duration together determine the rate of water loss; the greater the evaporation and the more abundant the sunshine, the faster the water loss from the soil and water bodies, and the higher the water demand. The system performs time-series analysis on these parameters, constructs a correlation model based on historical data from the same period, and converts them into a regional water demand compensation coefficient. This coefficient is a quantitative indicator that integrates meteorological conditions and is used to correct the baseline water demand. When meteorological conditions are humid, such as high rainfall and low evaporation, the coefficient value decreases, indicating that the actual water demand is lower than the baseline level; when meteorological conditions are arid, such as no rain and strong sunshine, the coefficient value increases, indicating that the water demand is higher than the baseline level.

[0078] When the regional water demand compensation coefficient remains below the threshold, it indicates that the actual water demand under current meteorological conditions is significantly lower than normal. In this case, the system adds a virtual water source node to the dynamic hydraulic balance model. The virtual water source node is not a physical water source, but a virtual regulating unit used to balance the supply and demand relationship in the model. Its weight coefficient is negatively correlated with the regional water demand compensation coefficient; the lower the compensation coefficient, the higher the weight coefficient of the virtual water source. This setting allows the virtual water source to play a "substitute" role in scenarios with low water demand, reducing dependence on physical water sources. Simultaneously, a bidirectional connection is established between the virtual water source node and the physical reservoir node. This connection is not an actual pipeline connection, but a logical association at the model level, used to transmit supply and demand regulation signals. After the connection takes effect, the system automatically triggers a smart water valve opening priority redistribution strategy: for example, during periods of abundant rainfall, the priority of water valves in irrigation areas is reduced, while the priority of water valves in domestic water use areas remains stable, ensuring that limited physical water resources prioritize meeting core needs.

[0079] The weighting coefficient of virtual water source nodes is propagated from top to bottom according to the hierarchical structure of the water supply dependency topology, forming a regulation gradient from source to end. During the propagation process, the weighting value is equally diminished according to the number of out-degrees of that node. The out-degree of a node refers to the number of downstream nodes it connects to. A higher out-degree means that the node needs to distribute regulation signals to more downstream units. Equal attenuation prevents the regulation force of a single node from excessively affecting the multi-level pipe network. For example, if the main water supply inlet node has a high out-degree, its weighting value will be evenly distributed according to the number of downstream nodes when it is propagated to the next level, ensuring that each branch receives a reasonable regulation amplitude. This propagation mechanism allows the influence of the virtual water source to gradually spread along the pipe network hierarchy, ensuring both the overall regulation and avoiding excessive regulation in local areas.

[0080] The system continuously compares the actual water level sensor data of physical reservoir nodes with the theoretical water level calculated by virtual water source nodes. The theoretical water level is a predicted value derived by combining the regulation effect of the virtual water source, the water demand of the pipeline network, and the replenishment capacity of the physical water source, while the actual water level reflects the actual water reserve status. When the deviation between the two exceeds the tolerance limit, it indicates that there is a significant difference between the model's prediction of the pipeline network's supply and demand relationship and the actual situation, which may be caused by reasons such as hidden leaks in the pipeline network, sensor errors, or discrepancies between the topology and the actual situation. At this time, the system automatically triggers the water supply dependency topology reconstruction process, re-checks the water supply dependency relationship of each node, corrects pipe diameter parameters and water-using unit association information, so that the topology map more accurately reflects the current state of the physical pipeline network, providing a reliable model foundation for the generation of subsequent water-saving strategies.

[0081] By dynamically integrating meteorological data with hydraulic models, the system breaks through the dependence of traditional water-saving strategies on fixed parameters. It can autonomously adjust its regulation logic according to environmental changes, ensuring water use stability while further improving water resource utilization efficiency. This integration mechanism is particularly suitable for scenarios significantly affected by weather, such as campus greening and park irrigation, making water-saving control more aligned with actual demand fluctuations.

[0082] The foregoing has provided a detailed description of one embodiment of the present invention, but this description is merely a preferred embodiment and should not be construed as limiting the scope of the invention. All equivalent variations and modifications made within the scope of the claims of this invention should still fall within the patent coverage of this invention.

Claims

1. A digital twin intelligent water-saving system based on a distributed optical fiber sensor network, characterized in that, include: The distributed optical fiber sensor network consists of a tree-like topology formed by a single trunk optical fiber and multiple branch optical fibers, and is continuously laid along the physical path of the water pipeline. The water use event identification module is used to collect fiber optic vibration signals in real time and convert them into digital waveform sequences. By setting a waveform amplitude threshold, the trigger point of the water use event can be identified. The water supply topology construction module is used to automatically construct a water supply dependency topology graph between water-using units based on the spatial location of the water use event trigger point and the time delay relationship between adjacent trigger points. Nodes in the topology graph represent water-using units, and directed edges in the topology graph represent water supply directions. In the water supply topology construction module, the process of constructing the water supply dependency topology graph is as follows: When a water use event trigger point is detected at the end of a branch fiber, the first vibration peak time of the water use event trigger point is recorded; if the sensing point on the main fiber closest to the end of the branch fiber has a trigger point with a vibration waveform similarity exceeding the threshold within a set time window, it is determined that the branch node depends on the main node for water supply. For branch nodes at the same level, if the triggering time interval between two branch nodes is less than the propagation time of the vibration wave in the pipeline, and the waveform envelope shape of the downstream branch node matches the attenuation shape of the upstream branch node, then a directed edge from the upstream branch node to the downstream branch node is established. After all dependencies are determined, node groups with the same water supply source are merged to form a hierarchical topology. The top node of the hierarchical topology is the main water supply inlet of the park, and the bottom node of the hierarchical topology is the terminal water unit. The hydraulic model generation module is used to spatially overlay the water supply dependency topology map with the preset digital map of water transmission pipelines to generate a dynamic hydraulic balance model, which includes the weight coefficient of water intensity of each node. The intelligent water-saving control module is used to send a series of periodic intermittent closing commands to the intelligent water valves associated with nodes with low weight coefficients, based on the real-time changing trend of the node weight coefficients in the dynamic hydraulic balance model, without the occurrence of physical leakage. The duration of the command sequence is proportional to the difference in weight coefficients between adjacent nodes.

2. The digital twin intelligent water-saving system based on a distributed optical fiber sensor network according to claim 1, characterized in that, The method for determining the similarity of the vibration waveforms is as follows: The original vibration waveforms of a fixed time length before and after the trigger point of the branch node are taken as the matching sequence, and the vibration waveforms of the candidate nodes of the trunk fiber corresponding to the time window are taken as the reference sequence. The time window length is calculated by dividing the physical distance between the branch fiber and the trunk fiber by the water hammer wave velocity under the pipe diameter. The sequence to be matched and the reference sequence are non-linearly aligned on the time axis. The minimum cumulative distance on the alignment path is calculated. When the minimum cumulative distance is less than the reference value of sound wave propagation loss determined by the pipe material, the waveform similarity is deemed to meet the standard.

3. The digital twin intelligent water-saving system based on a distributed optical fiber sensor network according to claim 1, characterized in that, In the hydraulic model generation module, the process of constructing the digital map of the water transmission pipeline is as follows: The characteristic vibration modes of pipe fittings are identified by optical fiber vibration signals. These include vortex-induced vibration waveforms at elbows, water hammer wavefront characteristics during valve opening and closing, and specific frequency harmonics during pump startup. The location of pipe fittings is determined based on the spatial distribution of characteristic vibration modes; the actual length of the pipe section is inverted using the time difference of vibration wave propagation on the continuous pipe section; the identification results are compared with the design drawings to generate a three-dimensional pipeline model with topological attributes, and the update cycle of the three-dimensional pipeline model is synchronized with the water supply dependent topology map.

4. The digital twin intelligent water-saving system based on a distributed optical fiber sensor network according to claim 3, characterized in that, The process of generating a dynamic hydraulic balance model in the hydraulic model generation module is as follows: Based on the water supply dependency topology map, the pipe diameter change points, pump station locations, and reservoir coordinate information from the digital map of water transmission pipelines are overlaid. Each node in the water supply dependency topology is assigned an initial weight coefficient, the value of which is determined by the type of water-using unit associated with the node. The number of water usage events triggered per unit time for each node is counted in real time, and the water usage intensity is calculated by combining the waveform integral area corresponding to each water usage event. The water usage intensity data is updated with node weight coefficients in a sliding time window manner. Nodes with a weight coefficient change rate exceeding the threshold are automatically marked as abnormal nodes. When the weight coefficient ratio between adjacent level nodes continues to deviate from the square of the pipe diameter ratio, the topology verification process is triggered.

5. A digital twin intelligent water-saving system based on a distributed optical fiber sensor network according to claim 1, characterized in that, In the intelligent water-saving control module, the process of generating a periodic intermittent shutdown command sequence is as follows: Obtain the weight coefficient change curves of the target node in the dynamic hydraulic balance model for several consecutive time periods, and extract the steady phase and rising phase from the weight coefficient change curves. During the stable phase, a shutdown command is generated, and the duration of the shutdown command is equal to the historical average water usage interval of the target node multiplied by the weighting coefficient decay factor. A pre-shutdown instruction is inserted before the rising phase, and the duration of the pre-shutdown instruction is negatively correlated with the rising slope of the weight coefficient. For sibling nodes that rely on the same water supply source, a staggered shutdown timing table is generated after arranging them in ascending order according to their weight coefficients. The time interval between the start of shutdown instructions of adjacent nodes in the staggered shutdown timing table is greater than the time required for water pressure fluctuations to stabilize.

6. A digital twin intelligent water-saving system based on a distributed optical fiber sensor network according to claim 5, characterized in that, The weighting coefficient attenuation factor is calculated as follows: Establish a correlation function between the target node's weight coefficient and the standard deviation of the weight coefficients of other nodes at the same level. When the standard deviation is less than a threshold, a fixed attenuation factor is used. When the standard deviation is greater than the threshold, the attenuation factor is dynamically adjusted according to the deviation of the target node from the average weight coefficient. That is, when the target node weight coefficient is lower than the average weight coefficient, the attenuation factor is increased, and when the target node weight coefficient is higher than the average weight coefficient, the attenuation factor is decreased. The adjustment range of the attenuation factor is inversely proportional to the hierarchical depth from the target node to the root node in the water supply dependency topology graph.

7. A digital twin intelligent water-saving system based on a distributed optical fiber sensor network according to claim 1, characterized in that, It also includes a meteorological data fusion module, specifically: Real-time access to rainfall intensity, evaporation, and sunshine duration data from regional meteorological stations, converting the rainfall intensity, evaporation, and sunshine duration data into regional water demand compensation coefficients; When the regional water demand compensation coefficient is consistently lower than the threshold, a virtual water source node is added to the dynamic hydraulic balance model. The weight coefficient of the virtual water source node is negatively correlated with the regional water demand compensation coefficient. A bidirectional connection edge is established between the virtual water source node and the physical water storage tank node, triggering a smart water valve opening priority redistribution strategy.

8. A digital twin intelligent water-saving system based on a distributed optical fiber sensor network according to claim 7, characterized in that, The function of the virtual water source node is as follows: When the regional water demand compensation coefficient decreases, the virtual water source node automatically increases its weight coefficient value. The weight coefficient increase value is passed from top to bottom according to the hierarchical structure of the water supply dependence topology. During the transmission process, the weight coefficient increase value decreases equally according to the number of out-degrees of the node as it passes through a node in the water supply dependence topology. When the deviation between the actual water level sensor data of the physical water storage node and the theoretical water level calculated by the virtual water source node exceeds the tolerance limit, the water supply dependency topology reconstruction process is triggered.

Images