A building water pipe health monitoring system based on real-time data

Abstract

This invention relates to the field of intelligent monitoring technology, specifically to a building water pipe health monitoring system based on real-time data. The system includes a flow direction deviation identification module, a leakage path confirmation module, a pressure difference fluctuation marking module, an active water usage screening module, and a node sampling and control module. In this invention, the flow direction diffusion trajectory is identified through the spatial distribution relationship between nodes; pressure imbalance sections are extracted by combining pressure difference sequence changes; periodic high-frequency water usage characteristics are superimposed to lock in concentrated periods of behavior; and the sampling frequency and monitoring density are dynamically adjusted by integrating the intersection areas of multiple features. This enables trend tracking of the target state, refined identification of abnormal signals, and real-time control of monitoring strategies. A data-driven closed-loop linkage mechanism is constructed to achieve rapid identification of weak anomalies in the building's internal pipe network and accurate location of leakage paths, improving the system's perception depth, response efficiency, and risk response capabilities in multi-factor coupled scenarios.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent monitoring, and in particular to a building water pipe health monitoring system based on real-time data. BACKGROUND

[0002] The technical field of intelligent monitoring belongs to the key technical direction of the cross-fusion of information perception and data analysis, mainly including automatic, real-time and continuous data collection, analysis and identification of physical entities or environment to master the target state and behavior change. The core matters of this field cover sensor network layout, signal acquisition, data transmission, real-time processing, state evaluation and abnormal identification, etc., and are widely used in urban infrastructure, environmental protection, industrial equipment, medical health and other scenes, supporting the digital management and safety protection of equipment operation. This technical field relies on stable perception system and accurate data analysis capability, and is an important supporting technology system for realizing intelligent perception and early warning control. Among them, the traditional building water pipe health monitoring system refers to the system technology for carrying out structural health evaluation on the running state of the building internal water supply and drainage pipe network. The technical matters to be solved are efficient detection and identification of the running abnormal state such as damage, aging and leakage of the water pipe system. The traditional building water pipe health monitoring usually adopts manual inspection, local installation of pressure gauges or uses flow anomaly detection method for state monitoring. Manual inspection judges the key nodes of water pipes according to building drawings and experience and checks regularly. The pressure gauge method detects anomalies according to the change of node water pressure. The flow detection method judges the leakage position according to the deviation between water consumption statistics and actual use.

[0003] The existing technology takes node fixed-point data as the basis for state evaluation, ignores the dynamic process of flow evolution in the pipe network, cannot reveal the trend expansion direction in continuous anomalies in real time, and the judgment of water pressure or flow anomaly highly depends on instantaneous deviation statistics. In the scene of slow change or local anomaly, it is easy to be covered by normal fluctuations. The lack of spatial coverage leads to incomplete identification of risk areas. The single time period division limits the extraction accuracy of periodic behavior. The monitoring strategy cannot be actively adjusted according to the complex state intersection. When multiple abnormal factors overlap, there is a lack of unified judgment mechanism to support multi-dimensional information linkage reasoning, which limits the early warning timeliness and lags behind the abnormal state development process. SUMMARY

[0004] The purpose of the present application is to solve the shortcomings in the prior art and to provide a building water pipe health monitoring system based on real-time data.

[0005] In order to achieve the above purpose, the technical scheme adopted by the present application is as follows: a building water pipe health monitoring system based on real-time data comprises:

[0006] The flow direction deviation identification module obtains flow direction data of a continuous time period in the water supply network, constructs a direction change sequence, screens a pipe section area with continuous deviation, extracts a time segment with continuous decrease in combination with water pressure trend in the area, and if the time segment overlaps with a direction deviation period, marks the pipe section as a controlled monitoring area, and generates a deviation trend concentrated section;

[0007] The leakage path confirmation module extracts node spatial distribution of the deviation trend concentrated section, tracks water flow direction change over time, judges adjacent time period spreading trend, if the boundary expands outward, then delimits a propagation channel and marks it as a suspected leakage path, and generates a suspicious path flow trajectory;

[0008] The differential pressure fluctuation marking module extracts water pressure change data in the suspicious path flow trajectory, searches whether a pressure difference forms a continuous paragraph of increase and decrease, excavates a continuous segment with unevenness, then marks it as a fluctuation strong area, records the corresponding time, and generates a pressure abnormal segment;

[0009] The active water screening module extracts water flow change in the pressure abnormal segment, marks a time period with frequent fluctuation, if the performance shows consistency in multiple days, records it as a periodic water concentrated period, and generates a node repeated high-frequency time window.

[0010] As a further scheme of the application, the deviation trend concentrated section includes a continuous direction deviation pipe section, a water pressure synchronous decrease section, and a direction and water pressure overlapping period, the suspicious path flow trajectory includes node spatial distribution characteristics, a propagation trajectory path, and a boundary extension trend, the pressure abnormal segment includes node pressure difference distribution, an increase and decrease fluctuation paragraph, and a continuous uneven state section, and the node repeated high-frequency time window includes a high change time period, a periodic consistency fluctuation interval, and an active water concentrated node.

[0011] As a further scheme of the application, the flow direction deviation identification module includes:

[0012] The direction sequence construction submodule obtains flow direction data of a pipe section in a continuous time period in a building water supply network, constructs a direction change sequence according to direction change of adjacent time points, records cumulative number of direction changes of the pipe section and change interval time, counts direction change trend and fluctuation frequency in each time period, and generates a direction change trend;

[0013] The continuous deviation screening submodule judges whether there is a continuous state of same direction deviation in a continuous time based on the direction change trend, screens an area meeting a set change continuity standard, forms a preliminary deviation section set, calls flow direction change intensity, change frequency and fluctuation stability value in each set, calculates a deviation intensity index, judges in combination with a threshold value, and obtains a direction continuous deviation section.

[0014] The water pressure overlap judgment submodule calls pipe section data in the direction continuous deviation section, collects water pressure trend data of a corresponding time period, extracts a slope of the water pressure sequence and calculates a continuous fluctuation amplitude, screens a time segment with a continuous downward trend, judges whether the time segment has an interval intersection with the direction continuous deviation time period, and establishes a deviation trend concentrated section.

[0015] As a further scheme of the present application, the leakage path confirmation module comprises:

[0016] The node distribution extraction submodule obtains three-dimensional coordinates of all nodes in the deviation trend concentrated section, extracts positions of the nodes under a pipe section topological relationship, calls coordinate difference values, connection relationships and spatial distribution range values between the nodes, calculates distances between node pairs and statistics local density change values, and establishes node spatial density.

[0017] The path propagation judgment submodule collects water flow direction data between the nodes in a continuous time period according to the node spatial density, arranges direction change values of the nodes in time sequence, compares direction continuity and time synchronicity of adjacent nodes, identifies whether an extension trend path is formed, calculates a path extension trend value, marks a suspected path when the path extension trend value is lower than an extension trend threshold, and generates a trajectory extension trend.

[0018] The leakage channel generation submodule calls a node path set in the trajectory extension trend that meets a continuous decreasing feature, judges boundary node positions, screens whether a path tail end continuously expands to the periphery, merges the paths and marks path numbers and node sequences, and establishes a suspected path flow trajectory.

[0019] As a further scheme of the present application, the differential pressure fluctuation marking module comprises:

[0020] The water pressure change extraction submodule obtains water pressure change data of the nodes in a continuous time period in the suspected path flow trajectory, extracts water pressure record values according to node numbers and time sequences, calculates water pressure difference values of adjacent time points and constructs a water pressure difference change sequence of the nodes, and generates a node water pressure difference value sequence.

[0021] The gradient sequence identification submodule identifies whether there are continuous direction-consistent increasing and decreasing sections according to the node water pressure difference value sequence, judges whether a change trend meets a gradient rule standard, extracts a continuous time and a direction consistency value, and obtains a differential pressure imbalance interval.

[0022] The abnormal range marking submodule calls the differential pressure imbalance interval, screens a segment with a continuous time period feature and meeting an imbalance strength condition, extracts corresponding start and end times and node numbers, marks as a fluctuation strong area, and establishes a pressure abnormal segment.

[0023] As a further scheme of the present application, the active water screening module comprises:

[0024] The water flow change extraction sub-module obtains node numbers and corresponding time period data in the pressure anomaly segment, collects hourly water flow data for consecutive days, calculates the flow change amplitude within an hour, counts the daily water flow fluctuation frequency, and constructs a node water flow fluctuation sequence to obtain a node water flow fluctuation amount;

[0025] The time period feature recognition sub-module extracts a time period with frequent and high-amplitude repeated occurrence in each day according to the node water flow fluctuation amount, screens an interval that meets the consistent time position within consecutive dates, marks a time period with periodic change rule, and obtains a repeated fluctuation time period interval.

[0026] The node time window locking sub-module calls the repeated fluctuation time period interval, merges nodes corresponding to time segments with consistent periodic characteristics, outputs the node number and high-frequency time period correspondence, and establishes a node repeated high-frequency time window.

[0027] As a further scheme of the present application, the system further comprises a node sampling regulation module.

[0028] The node sampling regulation module identifies a monitoring area covered by the node repeated high-frequency time window, matches whether the key nodes in the suspicious path are included in combination with the pressure anomaly segment, adjusts the sampling interval of all nodes and improves the frequency if the result areas overlap, and generates a monitoring intensity adjustment distribution map.

[0029] The monitoring intensity adjustment distribution map comprises an adjusted sampling interval node, an improved monitoring frequency area, and a result overlapping monitoring block.

[0030] As a further scheme of the present application, the node sampling regulation module comprises:

[0031] The area range recognition sub-module obtains a node set corresponding to the node repeated high-frequency time window, and locates the monitoring area boundary covered based on the node number, extracts all node numbers in the area in combination with the pipe network spatial topology information, and generates a high-frequency area node set.

[0032] The key node intersection judgment sub-module matches whether there is an overlap with the node numbers in the pressure anomaly segment according to the high-frequency area node set, judges whether the key node sequence in the suspicious path flow trajectory is included in the overlapping area, screens the node range appearing in the same area in the screening result, and obtains an overlapping node identification result.

[0033] The sampling strategy adjustment sub-module calls the overlapping node identification result, sets a sampling strategy adjustment identifier for all nodes in the identified overlapping area, shortens the current sampling time interval, forms a sampling frequency variation sequence of each node, and establishes a monitoring intensity adjustment distribution map.

[0034] Compared with the prior art, the application has the advantages and positive effects that:

[0035] In the application, the diffusion trajectory is identified through the spatial distribution relationship between nodes, the pressure imbalance section is extracted combined with the pressure difference sequence change, the behavior concentrated period is locked by superimposing the periodic high-frequency water use characteristics, the sampling frequency and monitoring density are dynamically adjusted in the intersection region of multiple types of features, the trend tracking of the target state, the refined identification of abnormal signals and the real-time regulation of monitoring strategies are realized, the closed-loop linkage mechanism under the data driving is constructed, the weak abnormality of the building internal pipe network is quickly identified and the leakage path is accurately positioned, and the perception depth, response efficiency and risk response capability of the system in the multi-factor coupling scene are improved. BRIEF DESCRIPTION OF DRAWINGS

[0036] Figure 1 The system flowchart of the application is shown in the figure;

[0037] Figure 2 The flowchart of the flow direction deviation identification module of the application is shown in the figure;

[0038] Figure 3 The flowchart of the leakage path confirmation module of the application is shown in the figure;

[0039] Figure 4 The flowchart of the pressure difference fluctuation marking module of the application is shown in the figure;

[0040] Figure 5 The flowchart of the active water screening module of the application is shown in the figure;

[0041] Figure 6 The flowchart of the node sampling regulation module of the application is shown in the figure. DETAILED DESCRIPTION

[0042] In order to make the purpose, technical scheme and advantages of the application more clear and explicit, the application will be further described in detail below in combination with the drawings and examples. It should be understood that the specific examples described herein are only used to explain the application and not to limit the application.

[0043] In the description of the application, it should be understood that the orientations or positional relationships indicated by the terms "length", "width", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer" and the like are based on the orientations or positional relationships shown in the drawings, and are only for the convenience of describing the application and simplifying the description, and therefore cannot be understood as indicating or implying that the devices or elements referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as limiting the application. In addition, in the description of the application, the meaning of "multiple" is two or more, unless otherwise specifically limited.

[0044] Please refer to Figure 1A building water pipe health monitoring system based on real-time data comprises:

[0045] The flow direction offset identification module obtains flow rate direction data of a continuous time period in a building water supply network, constructs a direction transition sequence, screens pipe segment areas where continuous offset occurs, and then extracts time segments where water pressure continuously decreases according to the synchronous recorded water pressure trend in the areas, judges whether the two types of phenomena overlap, and if so, marks the pipe segment as a controlled monitoring area and generates an offset trend concentrated segment;

[0046] The leakage path confirmation module extracts the spatial distribution characteristics of all nodes in the offset trend concentrated segment, tracks the continuous change process of the flow direction between the nodes in time sequence, compares whether there is a gradually spreading propagation track in adjacent time periods, and if so, delimits a propagation channel and marks it as a suspected leakage path, and generates a suspicious path flow track;

[0047] The differential pressure fluctuation marking module extracts the continuous change result of the node water pressure in the suspicious path flow track, retrieves the pressure difference between the node water pressure change points, judges whether there is a sequence paragraph of step-by-step increasing and decreasing, and if so, marks it as a fluctuation strong area and marks the corresponding time range, and generates a pressure abnormal segment;

[0048] The active water screening module extracts the hourly flow change amplitude of each day in the multi-day continuous data for the nodes in the pressure abnormal segment, marks the time period where the fluctuation is frequent and repeatedly high, and if the time period remains consistent in multiple days, records it as a periodic water use concentrated period and locks the corresponding node, and generates a node repeated high-frequency time window;

[0049] The node sampling control module identifies the monitoring area covered by the node repeated high-frequency time window, combines the pressure abnormal segment, and matches whether the key nodes in the suspicious path flow track are included in the range, and if so, adjusts the current sampling interval of all nodes in the area, increases the monitoring frequency, and generates a monitoring intensity adjustment distribution map.

[0050] The offset trend concentrated segment includes the continuous direction offset pipe segment, the water pressure synchronous decline segment, and the direction and water pressure overlap period. The suspicious path flow track includes the node spatial distribution characteristics, the propagation track path, and the boundary extension trend. The pressure abnormal segment includes the node pressure difference distribution, the increasing and decreasing fluctuation paragraph, and the continuous uneven state segment. The node repeated high-frequency time window includes the high-variation time period, the periodic consistency fluctuation interval, and the active water concentrated node. The monitoring intensity adjustment distribution map includes the adjusted sampling interval node, the improved monitoring frequency area, and the result overlapping monitoring block.

[0051] Please refer to Figure 2The flow direction offset identification module comprises:

[0052] The direction sequence construction submodule obtains flow velocity direction data of the pipe section in the building water supply network in a continuous time period, constructs a direction change sequence according to the direction change of adjacent time points, records the cumulative number of direction changes and the change interval time, counts the direction change trend and the fluctuation frequency in each period of time, and generates a direction change trend;

[0053] Obtaining the flow velocity direction data of the pipe section in the building water supply network in a continuous time period, first, the data collected by the pipe section at different monitoring points in the building at each hour time scale is numbered and recorded, and a time sequence data stream is established in turn. The data stream takes pipe section ID, time point and flow velocity direction as basic dimensions. By recording the direction data sequence of a pipe section at 0 point, 1 point, 2 point and the like, for example, if the direction sequence of a pipe section is recorded as [1, 1, -1, -1, 1, 1] in 6 consecutive hours, it can be judged that the direction change occurs at the 2nd hour and the 4th hour according to the product of the direction of adjacent time points being negative. It is determined that the direction has been reversed, and a direction change sequence is constructed as [0, 0, 1, 0, 1, 0], wherein 1 represents a direction change, and 0 represents no change. The cumulative number of direction changes is 2 times, and the direction change interval time is 2 hours and 2 hours. Further, the time period is divided, for example, every 3 hours is a statistical period. The direction change trend in 0-3 hours is determined by the difference between the direction values to determine whether the overall direction change presents a centralized offset trend. If the change value is in a single direction, it is marked as a one-way offset, otherwise it is marked as a fluctuation state. At the same time, the fluctuation frequency, i.e. the number of direction reversals in a 3-hour period, is counted. Here, it is 1 time. The trend is marked as moderate fluctuation. The direction change trend is finally output as an index in each cycle, which serves as the basis for subsequent continuous offset analysis.

[0054] The continuous offset screening submodule judges whether there is a continuous state of same-direction offset in a continuous time period based on the direction change trend, screens regions that meet the set change continuity standard, forms a preliminary offset section set, calls the flow velocity direction change intensity, change frequency and fluctuation stability value in each section set, and uses the formula:

[0055] ;

[0056] The operation obtains the offset intensity index , judges in combination with the threshold value, and obtains the direction continuous offset section;

[0057] wherein, represents the offset intensity index of the i-th pipe section, represents the direction change value of the i-th pipe section at time t, represents the direction change amplitude of the i-th pipe section at time t, This represents the frequency of directional change of the i-th pipe segment at time t. This represents the moving average of the directional trend value of the i-th pipe segment at time t. Let be the average trend value of the i-th pipe segment at the initial time. Indicates the total number of time periods. This serves as the baseline for the offset intensity threshold.

[0058] Based on the trend of directional change, it is determined whether there is a continuous state of unidirectional shift within a continuous time period. Regions that meet the set criteria for the persistence of change are selected. The intensity, frequency, and stability values ​​of the velocity direction change within each segment are retrieved using the following formula:

[0059] ;

[0060] During implementation, for a sample pipe segment B, the continuous time period is set to... Take the direction change values ​​at three different times. for directional change range for Frequency of directional changes for Trend value moving average for Initial trend value It is 0.8;

[0061] First, calculate the molecule:

[0062] ;

[0063] Then calculate the square root of the first part of the denominator, the sum of squared frequencies:

[0064] ;

[0065] Then calculate the sum of the trend values ​​and the initial value:

[0066] ;

[0067] ;

[0068] The combined denominators are:

[0069] ;

[0070] The final offset strength index is:

[0071] ;

[0072] If a offset strength threshold is set Therefore, , so the area does not meet the continuous deviation determination criterion.

[0073] The advantage of the formula is that by taking the fluctuation of change frequency and the change stability of trend moving average into account, the trend strength indicator is effectively quantified by the product of the direction change value and the absolute magnitude.

[0074] The water pressure overlap judgment submodule calls the pipe segment data in the direction continuous deviation section, collects the water pressure trend data in the corresponding time period, extracts the slope of the water pressure sequence and calculates the continuous fluctuation amplitude, filters the time segments with continuous downward trend, judges whether the time segments have interval intersection with the direction continuous deviation time section, and establishes the deviation trend concentration section;

[0075] The water pressure trend sequence corresponding to the direction deviation time section is collected by calling the pipe segment data in the direction continuous deviation section. First, the hourly water pressure value is linearly fitted, and the first derivative approximate slope is calculated, that is, the local downward or upward slope is calculated by dividing the water pressure difference between the two time points by the time difference. For example, the water pressures at four time points in the deviation section are [0.48, 0.46, 0.42, 0.38] MPa, and the corresponding slopes are (-0.02, -0.04, -0.04) MPa / h. The continuous downward trend is recorded as a continuous negative value. The fluctuation amplitude of the time period is calculated, that is, the difference between the maximum value and the minimum value in the time period is found. In this example, it is 0.48-0.38=0.10 MPa. If the fluctuation amplitude is greater than the set value 0.05 MPa and the continuous downward duration is greater than 2 hours, the water pressure change in this section is marked as a significant downward trend section. Then, the time index of the water pressure downward trend section is matched with the direction continuous deviation time section in interval, and the intersection is used to judge whether there is a common time section. For example, the direction deviation section occurs in [1h, 4h], and the water pressure downward section occurs in [2h, 5h]. The intersection is [2h, 4h], and the non-empty intersection indicates that the two trend sections have spatial overlap relationship, so the time section is determined as the deviation trend concentration section.

[0076] Please refer to Figure 3 The leakage path confirmation module comprises:

[0077] The node distribution extraction submodule obtains the three-dimensional coordinates of all nodes in the deviation trend concentration section, extracts the position of the node under the pipe segment topology relationship, calls the coordinate difference between the nodes, the connection relationship and the spatial distribution range value, calculates the distance between the node pairs and the local density change value, and establishes the node spatial density.

[0078] Upon receiving the label of the section of the bias trend cluster, the three-dimensional coordinate information of all nodes in the section is extracted, each node having a unique node number and three-dimensional spatial coordinates, including lateral coordinates, longitudinal coordinates, and height coordinates. For example, the coordinates of node N01 are (12.3, 45.1, 3.0), and the coordinates of node N02 are (12.8, 45.3, 3.1). The system reads the coordinate data of all nodes in batches and constructs a node position information table. Then, the topological structure of the pipe section in the building water supply network is called to determine the connection relationship between nodes and establish node pairs. The three-dimensional coordinate difference between the two adjacent nodes in the node pair is further extracted, and the relative position and distribution direction of the nodes in space are determined based on the difference content. For example, if the X coordinate difference is greater than 0.5 meters and the Y coordinate difference is less than 0.2 meters, it is determined that the nodes have a dominant distribution trend along the X axis. The actual connection distance between each pair of nodes is estimated, the total connection length between the current node and its neighbor nodes is calculated, and the number of neighbor nodes connected to each node is recorded. The node number in the space is calculated based on the space range, for example, a 2m x 2m square space unit as the basic unit, to calculate the local space density value. Then, the node number in multiple adjacent space units is differentially processed to determine the gradient change of the node density. For example, the node numbers in adjacent units are 5 and 2, and the density change value is 3, which reflects whether there is a significant concentration or sparse trend in the spatial distribution of nodes. If the density change value exceeds a set standard value, such as 2.5, it is determined that the location is a space density mutation point, which will be used as an important basis for subsequent path direction judgment. After completing the node density statistics, the system finally outputs a node space density mapping result, which associates each node with its density value, local neighbor information, and connection distance value, providing complete spatial distribution reference data for the next step of trajectory extension trend judgment.

[0079] The path propagation judgment submodule collects the water flow direction data between nodes in a continuous time period based on the node space density, arranges the direction change values of the nodes in time sequence, compares the direction continuity and time synchronization of adjacent nodes, and identifies whether an extension trend path is formed. The formula is:

[0080] ;

[0081] The operation obtains the path extension trend value When is lower than the extension trend threshold , it is marked as a suspected path, and a trajectory extension trend is generated.

[0082] wherein, represents the path extension trend value of the k time period, represents the flow direction of node r in the k time period, flow direction of the previous time period, distance between the rth node and the previous node, number of participating nodes in the current time period, fixed threshold value for the extension trend judgment.

[0083] According to the node spatial density, the water flow direction data between nodes in the continuous time period is collected. First, the flow direction values in different time periods are numbered. In each time period k, each node r has a direction change value , which is defined as +1 indicating the continuation of the previous moment direction, -1 indicating the direction reversal, and 0 indicating no data or uncertain change in direction. At the same time, a consistency discrimination variable is set, which is 1 if the direction is the same as the previous moment, otherwise it is 0. Each two nodes r and r-1 have a distance . According to the data, the path extension trend value is calculated using the following formula:

[0084] ;

[0085] Now take a set of path examples for calculation: the path is composed of three nodes N01, N02, and N03, and the spatial distance between the nodes is: In time period k=1, the direction change value of the three nodes is , , and the consistency with the previous moment direction is 1, i.e. , so the numerator is ; the denominator is ;

[0086] Therefore, the path extension trend value is:

[0087] ;

[0088] If the extension trend threshold value is set to 1.6, this value is less than the threshold value, so the path is marked as a suspected path, and the trajectory extension trend is generated. If the direction values of another path N05 to N07 are [1, -1, -1], and the consistency is [1, 0, 1], then

[0089] ;

[0090] The total path distance is m, so the path trend value is 0 / 2.1=0. This type of path trend value is significantly lower than the standard threshold value, and it is determined as a non-extension path. In this formula, reflects the direction change characteristics, reflects the time continuity, The formula is beneficial in that by normalizing the direction consistency and the path space length, the fusion discrimination ability of the path trend in both time and space dimensions is strengthened.

[0091] The leakage channel generation submodule calls the node path set that meets the continuously decreasing feature in the trajectory extension trend, judges the boundary node position, filters whether the path tail end continuously expands to the periphery, merges the paths and marks the path number and node sequence, and establishes a suspicious path flow trajectory;

[0092] After receiving the trajectory extension trend path set, first, the node subsequence in the path that meets the continuously decreasing feature is retrieved. The judgment of this feature is based on the change of the node direction value in the continuous time period. If the subsequent node direction value in the path continuously decreases or remains negative compared to the previous node direction value, for example, the node direction record in a certain path is [1, 0, -1, -1], it is judged that this section has a direction attenuation trend. The system records this path section as a decreasing trend section. Then, the position of the tail node of the path section is judged. The three-dimensional coordinate data of the node is called to evaluate whether the node is close to the spatial boundary of the building water supply network. The judgment basis is whether the horizontal or vertical coordinate is in the adjacent area of the overall boundary coordinate range. For example, if a building is in the X-axis range of 0 to 20 meters, more than 19 meters can be regarded as the boundary. If the X coordinate of the tail node is 19.4 meters, the system marks the node as a boundary node. After confirming that it is at the boundary, the system also needs to judge whether it has the trend of continuing to extend outside the boundary. The judgment standard is whether there is an uncovered space area in its adjacent direction or whether the spatial distance between it and other isolated nodes is less than 1.5 meters. If any of the above conditions is met, it is determined that the path tail end has the possibility of continuing to expand. The system marks it as an expandable path. Then, the system cross-compares all the path sets marked as expandable to find out whether there is a close situation between the spatial positions of the path tail ends. For example, if the spatial distance between the two path tail ends is less than 1 meter, it is considered as a mergable path section. The system will merge the path numbers in the order of priority according to the number sequence. After merging, the node sequence of the merged path will be sequentially integrated, and a new path number will be assigned to the merged path. For example, after the merging of path P03 and path P05, the path number is still P03, and the node sequence is merged as [N01, N02, N03, N08, N09]. After completing the merging and node sequence registration of all paths, the system finally marks each path as a separate suspicious path number and outputs the complete node number sequence contained in the path. This sequence is used as the main input data source for subsequent leakage analysis.

[0093] Please refer to Figure 4 , the differential pressure fluctuation marking module comprises:

[0094] The water pressure change extraction submodule obtains water pressure change data of nodes in the suspicious path flow trajectory in a continuous time period, extracts water pressure record values according to node numbers and time sequences, calculates water pressure difference values of adjacent time points and constructs a water pressure difference change sequence of the nodes, and generates a node water pressure difference value sequence;

[0095] After obtaining the water pressure change data of nodes in the suspicious path flow trajectory in a continuous time period, first, the water pressure monitoring data associated with each node is located through the node number, and then the historical record value sequence corresponding to the node is extracted in time sequence. For example, for the node with the number N05, the continuously recorded water pressure values in the time period T0 to T5 are 0.51 MPa, 0.48 MPa, 0.44 MPa, 0.39 MPa, 0.37 MPa and 0.34 MPa in turn. Then, the system constructs the water pressure difference value between adjacent time points according to the time sequence, and the calculation process is to subtract the water pressure value at the previous time point from the water pressure value at the current time point to obtain a set of water pressure difference change data. The difference value of T1 is -0.03 MPa, T2 is -0.04 MPa, T3 is -0.05 MPa, T4 is -0.02 MPa, and T5 is -0.03 MPa. This calculation result is written into the water pressure difference value sequence database with the node number as the index. The difference value sequence is constructed to establish a set of water pressure difference record arrays for each node. This array is used for subsequent identification of the consistency of the trend direction and judgment of the imbalance degree. The difference value unit is megapascal (MPa), and the storage precision is three decimal places. Each difference value data has its time point label and node number, thereby forming a complete node water pressure difference sequence record set.

[0096] The gradient sequence identification submodule identifies whether there is a continuously directional consistent increasing and decreasing section according to the node water pressure difference value sequence, judges whether the change trend meets the gradient rule standard, and extracts the duration and directional consistency value to obtain the pressure difference imbalance interval.

[0097] According to the node water pressure difference value sequence, a continuous trend identification task is performed. Firstly, the direction of each node difference value sequence is classified in time sequence. The system determines the sign of each water pressure difference value. If it is less than zero, it is marked as a decreasing trend. If it is greater than zero, it is marked as an increasing trend. In order to exclude the interference of fluctuations, the system continuously searches for three or more segments with the same direction as the candidate variable sequence. Taking the N05 node difference value sequence as an example, the difference values in T1 to T3 are -0.03 MPa, -0.04 MPa and -0.05 MPa. The continuous three values are all less than zero, so the system marks this section as a decreasing section. Further analysis of the difference value of T4 is -0.02 MPa, which is still negative. T5 is -0.03 MPa. It is judged to be consistent in direction, forming a decreasing sequence with a total length of 5 time periods. The system calculates the direction consistency value. The consistency ratio is 5 / 5=1.00 by calculating the number of consistent items divided by the total length. This value will be compared with the set threshold value. The consistency threshold value is set to 0.75. This value is set by referring to the common fluctuation ratio in the historical water pressure sequence, to ensure that the captured sequence has certain stability and clear trend. If the consistency value is not less than the standard and the duration is not less than 3 time points, it is determined that the sequence is a valid variable trend sequence, and the starting time is marked as T1, the ending time is marked as T5, and the corresponding node is N05. The system records this section as a pressure difference imbalance interval. This information will be used for subsequent screening and judgment of the abnormal range labeling submodule.

[0098] The abnormal range labeling submodule calls the pressure difference imbalance interval, screens the segments with continuous time period characteristics and meets the imbalance strength condition, extracts the corresponding start and end time and node number, and marks it as a strong fluctuation area. The pressure abnormal segment is established.

[0099] After receiving the differential pressure imbalance interval, first, secondary screening is conducted for each imbalance interval to determine whether it meets the abnormality determination standard. The judgment conditions are divided into two aspects: one is that the duration must be no less than the set time, and the second is that the direction consistency value needs to reach or exceed the set reference value. The duration threshold is set to 3 consecutive periods, which is derived from the analysis of the periodic variation trend of the actual building water supply system. If the pressure difference fluctuation can remain consistent in the same direction for 3 hours, it may reflect potential abnormal changes. Similarly, the direction consistency threshold is 0.75, which is derived from the experience comparison of the average fluctuation structure stability in the past leakage test data. In the specific screening steps, the system calculates the total duration of each pressure difference interval and calculates the proportion of negative direction difference value. If the water pressure change direction of N05 node is decreasing from T1 to T5 for five consecutive periods, the system determines that the interval duration is 5 and the consistency is 1.00, which meets the set conditions. The system then extracts the start time T1 and the end time T5 of the segment, reads the node number N05 corresponding to the interval, and combines the above information to construct complete pressure fluctuation section identification information. This information is composed of node number, start and end time, water pressure difference sequence, etc., and is uniformly given the classification label of fluctuation strong area. The system then writes the data of the abnormal section into the pressure abnormal segment database, which can be called by the scheduling module or the early warning system later.

[0100] Please refer to Figure 5 , the active water screening module includes:

[0101] The water flow change extraction submodule obtains the node number and corresponding time period data in the pressure abnormal segment, collects continuous multi-day hourly flow data, calculates the flow change amplitude within the hour, and constructs the node flow fluctuation sequence to obtain the node flow fluctuation amount;

[0102] After obtaining the node number and corresponding time period data in the pressure anomaly segment, the system first locates the node number involved in each abnormal segment, and extracts the original water flow monitoring data of the node for not less than three days before and after the abnormal segment, with a data granularity of hourly level, i.e. 24 groups of data per day, recording the total water flow per hour from 0 to 24 hours, with a unit of cubic meters per hour (m³ / h). After the system classifies and summarizes the water flow per hour according to the node and date, the absolute value of the difference between the water flow values of adjacent two hours in each day is taken as the flow variation amplitude, for example, the water flow data of node N08 from 8 to 10 hours in D1 day is 0.9, 1.5, and 0.8, the variation amplitude in the 8 to 9 hour period is 0.6 m³ / h, and the variation amplitude in the 9 to 10 hour period is 0.7 m³ / h. The system sets the fluctuation amplitude threshold value as 0.5 m³ / h, which is the average fluctuation average value obtained by long-term observation and statistics of water flow data under normal operation of the building plus two standard deviations, reflecting the upper limit of non-abnormal fluctuation in most cases. Therefore, when the hourly variation amplitude exceeds 0.5, the system records the period as a fluctuation event. After performing the same operation on all hour periods in each day, the system calculates the water flow fluctuation frequency per day and arranges the multi-day data to form the water flow fluctuation frequency sequence corresponding to each node. At the same time, the average daily fluctuation value is calculated as the water flow fluctuation of the node, for example, the fluctuation frequencies of N08 node for three consecutive days are 4, 5, and 4 times, and the average fluctuation frequency is 4.33 times / day. This value is compared with the set fluctuation reference value (set as 4.0 times / day) to determine whether the water flow change of the node is significant. If the average value is greater than the reference value, the node is recorded as a water flow anomaly feature node and its fluctuation frequency sequence is output for subsequent periodic screening.

[0103] The time period feature recognition submodule extracts the time period with frequent and high amplitude repeated occurrence according to the water flow fluctuation of the node, screens the interval with consistent time position in consecutive days, marks the time period with periodic change rule, and obtains the repeated fluctuation interval.

[0104] According to the node water flow fluctuation data, first, the fluctuation events of each day are located in units of hours, and a fluctuation mapping diagram in the time dimension is constructed. Mark the hours from 0 to 23 in each day as "1" if they are marked as fluctuation, and mark the non-fluctuation period as "0". After the construction is completed, the system will compare the fluctuation marks of the same hours in multiple days vertically by node unit. For example, node N08 has fluctuation marks of "1" from 8 to 10 in D1, D2 and D3, so it is preliminarily judged that there is a fluctuation concentration trend in this time period. The system extracts all 3-hour window intervals by sliding for each node, and counts the number of times each 3-hour interval is identified as a high fluctuation period in three consecutive days. If a 3-hour interval is identified as a fluctuation period for 3 times in 3 days, the system preliminarily marks it as a periodic candidate period. To ensure the accuracy of identification, the system also needs to evaluate whether the fluctuation amplitude is significant, that is, whether the average fluctuation amplitude in all candidate time periods exceeds the system's set significant fluctuation amplitude threshold. The threshold is set to 0.45 m³ / h, which is based on the 80% quantile of the historical fluctuation mean of each node in the sampling building. The system calculates the average fluctuation amplitude of each hour in the candidate time period. If the average fluctuation amplitude of the time period in three days is 0.6, 0.7 and 0.5 respectively, the mean is 0.6 m³ / h, which meets the threshold requirement. Finally, the system marks the time period from 8 to 10 as a repeated fluctuation period interval, records the node number, time period start and end, continuous fluctuation days and average amplitude, and provides time window locking.

[0105] The node time window locking submodule calls the repeated fluctuation period interval, merges the nodes corresponding to the time slices with consistent periodic characteristics, outputs the node number and high-frequency period correspondence, and establishes the node repeated high-frequency time window;

[0106] After receiving the repeated fluctuation period interval information, all identified repeated fluctuation time periods are aggregated by taking the node number as the primary key. For the same node that appears in multiple days with the same start and end time of the repeated segment, the system merges it into a unified time window. At the same time, for the segments that are not exactly the same but have more than 90% time intersection, the system aligns and maps them into a standard window. For example, node N08 has high-frequency fluctuation in 8-10, 8-9 and 9-10 in three days respectively, and the intersection segment is 9-10, with a time overlap rate of 100%. The system finally takes 9-10 as the repeated high-frequency time window of the node. Then the system establishes a one-to-one correspondence between the node and the time window, and outputs the result as a time window index table. Each record includes node number, time window start and end time, number of consecutive days identified, time consistency ratio, average fluctuation amplitude, etc. The table structure is stored for subsequent spatial correlation and periodic event aggregation processing. This operation finally establishes a stable and periodic high-frequency behavior in the time scale, which helps to unify the identification of building internal water cycle and abnormal behavior overlap characteristics.

[0107] Referring to Figure 6 , the node sampling regulation module comprises:

[0108] The regional range identification submodule obtains a node set corresponding to the node repeated high-frequency time window, and locates the monitored region boundary based on the node number, extracts all node numbers in the region in combination with the pipe network spatial topology information, and generates a high-frequency regional node set;

[0109] After obtaining the node set corresponding to the node repeated high-frequency time window, firstly, all high-frequency time window records are classified and aggregated according to the node number, the system identifies the frequency of all nodes appearing in the repeated fluctuation period, and extracts nodes with an appearance rate greater than 80% in multi-day high-frequency fluctuation as a core node set, then according to the node number, the corresponding monitoring point position information is queried, the three-dimensional topology information in the building hydraulic monitoring system is called, the spatial position of each node in the pipe network is obtained, and the boundary coordinate value of these nodes in the plane layout diagram is calculated, the boundary calculation method is to extract the minimum and maximum values of all node X, Y coordinates to form a closed boundary region, then all other node numbers intersecting or located inside the boundary of the region are found in combination with the topology structure diagram, these nodes are defined as associated nodes under the influence of the high-frequency time window, in this way, the system can expand the influence range of the core node in space to form a complete regional coverage diagram, finally, the number set of all associated nodes in the region is output as the high-frequency regional node set, this set not only contains the periodic fluctuation frequent nodes, but also covers the surrounding nodes affected by the topology connection relationship.

[0110] The key node intersection judgment submodule matches whether there is an overlap with the node number in the pressure anomaly segment according to the high-frequency regional node set, judges whether the overlapping region includes the key node sequence in the suspicious path flow trajectory, and selects the node range appearing in the same region, to obtain the overlapping node identification result;

[0111] After receiving the node set of the high-frequency area, the node numbers are compared one by one. First, the system reads the node number list involved in all the abnormal pressure segments identified in the previous stage as the comparison reference set, and then calls the key node number sequence extracted in the suspicious path flow trajectory. The system cross-filters these two sets with the current area node set in turn. First, it is determined whether the node numbers of the abnormal pressure segment appear in the current area set. If they do, it is further determined whether they overlap with the key nodes in the suspicious path. For example, if the current area contains nodes N07, N08, and N10, and the abnormal segment nodes are N08 and N09, and the suspicious path nodes are N08, N10, and N11, then the overlapping nodes N08 and N10 are identified. This judgment is completed using the set intersection method to ensure that the matching occurs only within the same area. By setting the judgment threshold to "at least two key node numbers overlapping" as the effective intersection condition, false positives caused by accidental repetition are avoided. This threshold is set based on the common minimum overlap of key nodes in the same path in actual scene analysis. After filtering, the system records the overlapping node numbers and their area labels, and outputs the overlapping node identification result as the pre-judgment basis for sampling strategy adjustment.

[0112] The sampling strategy adjustment submodule calls the overlapping node identification result to set the sampling strategy adjustment flag for all nodes in the overlapping area, shorten the current sampling interval, form the sampling frequency variation sequence of each node, and establish the monitoring intensity adjustment distribution map.

[0113] After calling the overlapping node identification result, the system first obtains all the identified overlapping node numbers and extracts their current sampling interval values. The basic sampling interval of a building monitoring system is usually 10 minutes. If the node belongs to the overlapping area, the system will adjust its sampling interval to 5 minutes. This sampling adjustment strategy is based on the on-site sampling resource load capacity and data mutation response requirements. The 5-minute interval is the recommended value found by the system after long-term testing, which can balance real-time performance and data load. The system uses node numbers as the main index to record the adjusted sampling interval, the adjusted sampling interval, and the effective timestamp for each node with adjusted sampling strategy, forming the sampling frequency variation sequence of each node. For example, node N08 originally has a sampling interval of 10 minutes. After being identified as an overlapping node, it is adjusted to 5 minutes. Then the system draws a monitoring intensity adjustment distribution map based on the spatial position of the node. This map uses color to mark the sampling frequency levels of different nodes. Nodes with shortened sampling intervals are marked in red, and the rest are marked in gray. This map is part of the system's local high-density monitoring response basis and is displayed simultaneously in the dispatch center.

[0114] The above merely describes the preferred embodiments of the present application, and is not intended to limit the present application in other forms. Any skilled person in the art can modify or change the disclosed technical content into equivalent embodiments with equivalent changes, and apply them to other fields. However, any simple modification, equivalent change and modification made to the above embodiments according to the technical essence of the present application, without departing from the technical solution content of the present application, still falls within the protection scope of the present application.

Claims

1. A real-time data based building plumbing health monitoring system, characterized in that, The system comprises: The flow direction offset identification module obtains flow direction data of a continuous time period in a water supply network, constructs a direction transition sequence, screens a continuously offset pipe segment area, extracts a continuously decreasing time segment in combination with water pressure trends in the area, and if the time segment overlaps with a direction offset period, marks the pipe segment as a controlled monitoring area, generates an offset trend concentrated segment, and The leakage path confirmation module extracts node spatial distribution of the offset trend concentrated segment, tracks water flow direction changes over time, judges adjacent time period spreading trends, and if the boundary expands outward, it draws a propagation channel and marks it as a suspected leakage path, and generates a suspicious path flow trajectory. The differential pressure fluctuation marking module extracts water pressure change data in the suspicious path flow trajectory, searches whether the pressure difference forms a continuous paragraph of increasing and decreasing, excavates a continuous segment of imbalance, marks it as a fluctuation strong area, records the corresponding time, and generates a pressure abnormal segment. The active water screening module extracts water flow changes in the pressure abnormal segment, marks time periods with frequent fluctuations, and if consistency is shown over multiple days, records it as a periodic water concentration period, and generates a node repeated high-frequency time window. The offset trend concentrated segment includes a continuously direction offset pipe segment, a water pressure synchronous decrease segment, and a direction and water pressure overlapping period, the suspicious path flow trajectory includes node spatial distribution characteristics, propagation trajectory path, and boundary extension trend, the pressure abnormal segment includes node pressure difference distribution, increasing and decreasing fluctuation paragraph, and continuous imbalance state segment, and the node repeated high-frequency time window includes high-variation time period, periodic consistency fluctuation interval, and active water concentration node. The leakage path confirmation module comprises: The node distribution extraction submodule obtains three-dimensional coordinates of all nodes in the offset trend concentrated segment, extracts the position of the node under the topology relationship of the pipe segment, calls the coordinate difference value, connection relationship and spatial distribution range value between nodes, calculates the distance between node pairs and the local density change value, and establishes the node spatial density. The path propagation judgment submodule collects water flow direction data between nodes in a continuous time period according to the node spatial density, arranges the direction change value of the node in time sequence, compares the direction continuity and time synchronization of adjacent nodes, identifies whether an extension trend path is formed, calculates the path extension trend value, and marks it as a suspected path when the value is lower than the extension trend threshold, and generates a trajectory extension trend. The leakage path confirmation module comprises:

2. The real-time data based building plumbing health monitoring system of claim 1, wherein, The direction sequence construction submodule obtains flow direction data of a pipe segment in a continuous time period in a building water supply network, constructs a direction transition sequence according to direction changes at adjacent times, records the cumulative number of direction changes of the pipe segment and the change interval time, counts the direction change trend and fluctuation frequency in each period of time, and generates a direction transition trend. ​ The continuous deviation screening submodule determines whether there is a continuous state of same-direction deviation in continuous time based on the direction transition trend, screens areas that meet the set change continuity standard to form a preliminary deviation segment set, calls the flow direction change intensity, change frequency and fluctuation stability value in each segment set, calculates the deviation intensity index, judges in combination with the threshold value, and obtains the direction continuous deviation section; the water pressure overlap judgment submodule calls the pipe segment data in the direction continuous deviation section, collects the water pressure trend data in the corresponding time period, extracts the slope of the water pressure sequence and calculates the continuous fluctuation amplitude, screens the time segments of continuous downward trend, judges whether the time segments have an interval intersection with the direction continuous deviation time period, and establishes the concentrated deviation trend section.

3. The real-time data based building plumbing health monitoring system of claim 1, wherein, The differential pressure fluctuation marking module comprises: The water pressure change extraction submodule obtains the water pressure change data of the nodes in the suspicious path flow trajectory in a continuous time period, extracts the water pressure record value according to the node number and time sequence, calculates the water pressure difference value of adjacent time points and constructs the water pressure difference change sequence of the nodes, and generates the node water pressure difference value sequence; The gradient sequence identification submodule identifies whether there is a continuously increasing and decreasing section with the same direction according to the node water pressure difference value sequence, judges whether the change trend meets the gradient rule standard, extracts the duration and direction consistency value, and obtains the differential pressure imbalance interval; The abnormal range marking submodule calls the differential pressure imbalance interval, screens the segments that have continuous time period characteristics and meet the imbalance intensity condition, extracts the corresponding start and end time and node number, marks as a strong fluctuation area, and establishes the pressure abnormal segment.

4. The real-time data based building plumbing health monitoring system of claim 1, wherein, The active water screening module comprises: The water flow change extraction submodule obtains the node number and corresponding time period data in the pressure abnormal segment, collects the hourly flow data of continuous days, calculates the flow change amplitude within an hour, counts the daily flow fluctuation frequency and constructs the node flow fluctuation sequence, and obtains the node flow fluctuation amount; The time period feature identification submodule extracts the time period with frequent and high-amplitude repeated fluctuations in each day according to the node flow fluctuation amount, screens the intervals that meet the consistent time position in continuous dates, marks the time period with periodic change rule, and obtains the repeated fluctuation time period interval; The node time window locking submodule calls the repeated fluctuation time period interval, merges the nodes corresponding to the time segments with consistent periodic characteristics, outputs the node number and high-frequency time period correspondence relationship, and establishes the node repeated high-frequency time window.

5. The real-time data based building plumbing health monitoring system of claim 1, wherein, The system further comprises a node sampling control module: The node sampling control module identifies the monitoring area covered by the node repeated high-frequency time window, matches whether the key nodes in the suspicious path are included in combination with the pressure abnormal segment, adjusts the sampling interval of all nodes and improves the frequency if the result areas overlap, and generates a monitoring intensity adjustment distribution map; The monitoring intensity adjustment distribution map comprises an adjusted sampling interval node, an improved monitoring frequency area and a result overlapping monitoring block.

6. The real-time data based building plumbing health monitoring system of claim 5, wherein, The node sampling control module comprises: The regional range identification submodule obtains a node set corresponding to the node repeated high-frequency time window, and locates the monitored regional boundary based on the node number, extracts all node numbers in the region in combination with the pipe network spatial topological information, and generates a high-frequency regional node set; The key node intersection judgment submodule matches whether there is an overlapping node number in the pressure abnormal segment based on the high-frequency regional node set, judges whether the overlapping region includes a key node sequence in the suspicious path flow trajectory, screens the node range appearing in the same region, and obtains an overlapping node identification result; The sampling strategy adjustment submodule calls the overlapping node identification result, sets a sampling strategy adjustment identifier for all nodes in the identified overlapping region, shortens the current sampling time interval, forms a sampling frequency variation sequence of each node, and establishes a monitoring intensity adjustment distribution diagram.

Images