Water meter well remote ponding early warning method and system based on Internet of Things

Abstract

The invention belongs to the technical field of intelligent early warning, and discloses a water meter well remote ponding early warning method and system based on the Internet of Things. The method comprises the steps that collected water meter well measurement data are integrated through an Internet of Things gateway, comprehensive analysis is conducted on the measurement data, comprehensive characteristics are obtained, the measurement data and the comprehensive characteristics serve as input of an accumulated water analysis model, and an accumulated water characteristic mode is obtained; determining a leakage point coordinate through a fusion positioning algorithm in combination with sound wave positioning derivative features in the comprehensive features; performing dynamic analysis on the measurement data to obtain derivative dynamic characteristics, and taking the measurement data and the derivative dynamic characteristics as input of a dynamic characteristic model to obtain a dynamically adjusted graded ponding early warning threshold value; corresponding graded early warning is carried out according to the relation between the actual water accumulation amount and the graded water accumulation early warning threshold value; according to the invention, through multi-dimensional sensor data fusion and intelligent feature engineering, a closed-loop system integrating real-time monitoring, abnormity diagnosis, accurate positioning and grading early warning is constructed.

Description

Technical Field

[0001] This application relates to the field of intelligent early warning technology. More specifically, this application relates to a remote water accumulation early warning method and system for water meter wells based on the Internet of Things. Background Art

[0002] With the acceleration of the urbanization process, the scale of urban water supply pipe networks is constantly expanding. As a key node of the pipe network, water meter wells face the hidden danger of water accumulation caused by pipeline aging, external water seepage, drainage blockage, etc. Traditional manual inspections are inefficient and lack real-time performance. Moreover, the early warning method based on the static threshold of a single sensor is vulnerable to environmental fluctuations, with problems such as high false alarm and missed alarm rates, fuzzy cause identification, and rough leak point positioning.

[0003] Chinese Patent Application with Publication No. CN112712680A discloses a remote voice early warning box, an early warning method, and an intelligent water meter for an intelligent water meter: The remote voice early warning box at least includes an early warning broadcast module, an information processing module, and a communication module. The early warning method includes: The communication module receives the early warning information sent by the intelligent water meter and sends the received early warning information to the information processing module. The early warning information at least includes the remaining prepaid amount and / or the remaining water volume information; The information processing module receives the early warning information, generates early warning broadcast information according to the early warning information, and sends the early warning broadcast information to the early warning broadcast module; The early warning broadcast module receives the early warning broadcast information and broadcasts according to the content of the early warning broadcast information to give an early warning reminder to the user. This invention effectively improves the water use quality and user experience and avoids the predicament or inconvenience caused by water cut-off.

[0004] Although the above method can meet most scenarios, through research and practical application of the above method and the existing technology, it is found that the above method and the existing technology have at least the following partial defects: The data perception dimension of the above method is single and it cannot identify the cause of water accumulation.

[0005] In view of this, this application proposes a remote water accumulation early warning method and system for water meter wells based on the Internet of Things to solve the above problems. Summary of the Invention

[0006] To overcome the above defects of the prior art and to achieve the above object, this application provides the following technical solution: A remote water accumulation early warning method for water meter wells based on the Internet of Things, including the following steps: Integrate and collect the measurement data of the water meter well through the Internet of Things gateway, conduct a comprehensive analysis of the measurement data to obtain comprehensive features, use the measurement data and comprehensive features as the input of the water accumulation analysis model, and obtain the water accumulation feature pattern; The comprehensive features include sensor cross features, temporal dynamic features, spatial correlation features, and environmental features; the sensor cross features include threshold trigger features, correlation features, and physical derivative features; the temporal dynamic features include short-term window features, frequency domain features, and state transition features; the spatial correlation features include spatial gradient features and acoustic positioning derivative features; Combined with the acoustic positioning derivative feature, the leak point coordinates are determined through a fusion positioning algorithm; Perform dynamic analysis on the measurement data to obtain derivative dynamic features. Use the measurement data and derivative dynamic features as the input of the dynamic feature model to obtain a dynamically adjusted hierarchical waterlogging warning threshold; and perform corresponding hierarchical warnings according to the relationship between the actual water accumulation volume and the hierarchical waterlogging warning threshold.

[0007] Furthermore, the method for obtaining the threshold trigger feature includes: Detect whether the change value of the pipeline pressure exceeds the preset pressure threshold, detect whether the difference between the total flow value and the user-end flow exceeds the preset flow threshold; detect whether the water level rising rate exceeds the preset rate threshold; if all three exceed simultaneously, mark the pipeline rupture risk as 1, otherwise as 0; use the pipeline rupture risk as the threshold trigger feature; The method for obtaining the correlation feature includes: Calculate the correlation between different sensors using the Pearson correlation coefficient to obtain the correlation feature.

[0008] Furthermore, the method for obtaining the physical derivative feature includes: Obtain the total flow of the user end and the pipeline inlet flow, and calculate the leakage rate; count the pressure change amount and the flow change amount, and calculate the combined feature; splice the leakage rate and the combined feature to obtain the physical derivative feature.

[0009] Furthermore, the method for obtaining the state transition feature includes: Preset a division threshold, divide the sensor state into normal, warning, and failure, and encode them in sequence. Convert the temporal data of the sensor into a state sequence, count the number of transitions from state a to state b of the sensor, calculate the total residence time of state a, calculate the ratio of the number of transitions to the total residence time to obtain the transition probability matrix; count the continuous residence duration of each state; splice the transition probability matrix and the continuous residence duration to obtain the statistical feature; Count the average time interval from state a to state b, count the number of occurrences of the preset transition mode, and splice the average time interval and the number of occurrences to obtain the temporal dependence feature; Obtain the change rate and change amplitude of the sensor during state transition, and whether the states of other sensors change when the change amplitude of the sensor exceeds the preset change threshold; splice the change rate and change amplitude of the sensor, and whether the states of other sensors change when the change amplitude of the sensor exceeds the preset change threshold to obtain dynamic trend features; Splice the statistical features, time series dependence features and dynamic trend features to obtain state transition features.

[0010] Furthermore, the method for obtaining time series dynamic features includes: Statistically calculate the mean value, standard deviation and change rate of the water level, the minimum value, fluctuation range of the pipeline pressure and the number of times the pressure change amount per unit time is lower than the pressure threshold, and splice the mean value, standard deviation and change rate of the water level, and the minimum value, fluctuation range of the pipeline pressure and the number of times the pressure change amount per unit time is lower than the pressure threshold to obtain short-term window features; Statistically obtain the deviation between the actual temperature value and the mean temperature value in the same historical period, the deviation between the actual humidity value and the mean humidity value in the same historical period, and the proportion of low-frequency vibration energy below the preset low-frequency threshold, and splice the deviation between the actual temperature value and the mean temperature value in the same historical period, the deviation between the actual humidity value and the mean humidity value in the same historical period, and the proportion of low-frequency vibration energy below the preset low-frequency threshold as long-term window features; Extract the main frequency energy of the vibration signal and the high-frequency noise power of the pipeline pressure signal through Fourier transform, and splice the main frequency energy and the high-frequency noise power as frequency domain features; Splice the short-term window features, long-term window features, frequency domain features and state transition features to obtain time series dynamic features.

[0011] Furthermore, the method for obtaining spatial correlation features includes: Calculate the water level difference and distance between adjacent water meter wells, calculate the ratio of the water level difference and distance to obtain the water level gradient; Calculate the pressure difference and distance between the upstream water meter well and the downstream water meter well, calculate the ratio of the pressure difference to the distance to obtain the pressure attenuation rate; Splice the water level gradient and the pressure attenuation rate to obtain spatial gradient features; Detect the time difference between any two water meter wells receiving leakage sound waves, calculate the ratio of the difference in received signal strength to the distance between the two water meter wells to obtain the signal attenuation rate; splice the time difference and the signal attenuation rate to obtain acoustic positioning derivative features; Splice the spatial gradient features and the acoustic positioning derivative features to obtain spatial correlation features.

[0012] Furthermore, the method for obtaining environmental features includes: Statistically count the historical water accumulation times at the same location, splice the coordinates corresponding to the location and the historical water accumulation times as the historical water accumulation feature; statistically count the failure times of the same type of equipment, splice the equipment model and the failure times as the equipment failure feature, and splice the historical water accumulation feature and the equipment failure feature to obtain the environmental feature.

[0013] Further, the method for obtaining the leak point coordinates includes: Obtain the water meter well coordinates, and establish a hyperbolic positioning model based on the time difference of the acoustic positioning derivative feature; Obtain the signal strength received by the water meter well, establish an empirical attenuation model of the signal strength with respect to the distance, and calculate the distance from the leak point to the water meter well through the empirical attenuation model; establish three circle equations with the coordinates of three water meter wells and the corresponding distances; Fuse the hyperbolic positioning model and the three circle equations by the least squares method; Adopt the maximum likelihood estimation or Kalman filter algorithm to iteratively optimize and obtain the leak point coordinates.

[0014] Further, the method for obtaining the derivative dynamic feature includes: Obtain the water level change amount, temperature change amount, and humidity change amount within a preset time period, and calculate the ratios of the water level change amount, temperature change amount, and humidity change amount to the preset time period respectively to obtain the water level change rate, temperature change rate, and humidity change rate; Combine the pressure sudden drop amplitude and the flow rate difference to calculate the leakage probability index; Calculate the difference between the current pipeline pressure and the preset pipeline pressure to obtain the pressure change amplitude; calculate the ratio of the difference to the preset change time to obtain the pressure drop rate; statistically obtain the standard deviation of the pressure fluctuation; Obtain the difference between the user end flow rate and the pipeline inlet flow rate, and calculate the ratio of the difference to the pipeline inlet flow rate to obtain the flow rate difference ratio; Obtain the absolute difference between the current user end flow rate and the user end flow rate at the previous moment, and calculate the ratio of the absolute difference to the user end flow rate at the previous moment to obtain the flow rate mutation coefficient; Obtain the change amount of the water accumulation depth within the preset change time, and calculate the ratio of the change amount of the water accumulation depth to the preset change time to obtain the water accumulation depth change rate; Obtain the amplitude of the high-frequency vibration signal whose vibration signal exceeds the preset high-frequency threshold; Obtain whether the temperature change rate exceeds the preset temperature change rate threshold, if so, mark it as 1, otherwise as 0, to obtain the temperature anomaly index; Obtain whether the humidity change rate exceeds the preset humidity change rate threshold, if so, mark it as 1, otherwise as 0, to obtain the humidity anomaly index; Concatenate the water level change rate, temperature change rate, humidity change rate, standard deviation of pressure fluctuation, proportion of flow difference, flow mutation coefficient, change rate of accumulated water depth, amplitude of high-frequency vibration signal, temperature anomaly index, and humidity anomaly index to obtain derivative dynamic features.

[0015] An Internet of Things-based remote water accumulation warning system for water meter wells implements the Internet of Things-based remote water accumulation warning method for water meter wells, including: Feature analysis module: Integrate the measured data of the water meter well collected through the Internet of Things gateway, comprehensively analyze the measured data to obtain comprehensive features, use the measured data and comprehensive features as the input of the water accumulation analysis model, and obtain the water accumulation feature pattern; The comprehensive features include sensor cross features, time series dynamic features, spatial correlation features, and environmental features; the sensor cross features include threshold trigger features, correlation features, and physical derivative features; the time series dynamic features include short-term window features, frequency domain features, and state transition features; the spatial correlation features include spatial gradient features and acoustic positioning derivative features; Leak point positioning module: Combine the acoustic positioning derivative features and determine the leak point coordinates through the fusion positioning algorithm; Water accumulation warning module: Dynamically analyze the measured data to obtain derivative dynamic features, use the measured data and derivative dynamic features as the input of the dynamic feature model, and obtain a dynamically adjusted hierarchical water accumulation warning threshold; and perform corresponding hierarchical warnings according to the relationship between the actual water accumulation volume and the hierarchical water accumulation warning threshold.

[0016] Technical effects and advantages of the Internet of Things-based remote water accumulation warning method and system of the present application: Through multi-dimensional sensor data fusion and intelligent feature engineering, the present application constructs a closed-loop system integrating real-time monitoring, abnormal diagnosis, precise positioning, and hierarchical warning; through cross-sensor threshold joint judgment, correlation analysis, and time series state transition modeling, it accurately identifies the water accumulation pattern, improves the accuracy of cause diagnosis, and reduces the false alarm rate and missed alarm rate; through the hyperbola-triangle positioning algorithm that fuses the acoustic time difference and signal strength, combined with the least squares and Kalman filter optimization, it improves the leak point positioning accuracy and shortens the maintenance response time; based on the dynamic hierarchical threshold model of derivative dynamic features, it automatically adapts to environmental changes such as seasons and loads, realizes multi-level precise warning, synchronously correlates the historical high-frequency water accumulation area with equipment failure characteristics, promotes the centralized allocation of operation and maintenance resources to high-risk areas, reduces the inspection cost, and significantly improves the intelligent management level of urban water supply pipe networks and the emergency disposal efficiency of water accumulation events. Description of the Drawings

[0017] Figure 1 It is a schematic diagram of the Internet of Things-based remote water accumulation warning method of the present application; Figure 2Schematic diagram of the waterlogging positioning and warning process of this application; Figure 3 Structural diagram of the remote waterlogging warning system for water meter wells based on the Internet of Things of this application. Specific implementation manners

[0018] Next, the technical solutions in the embodiments of this application will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of this application. Obviously, the described embodiments are only a part of the embodiments of this application, rather than all the embodiments. Based on the embodiments in this application, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of this application.

[0019] Embodiment 1 Please refer to Figure 1 、 Figure 2 As shown, this embodiment provides a remote waterlogging warning method for water meter wells based on the Internet of Things, including the following steps: Combined with sensors to collect measurement data of the water meter well, comprehensively analyze the measurement data to obtain comprehensive features, use the measurement data and comprehensive features as the input of the waterlogging analysis model, and obtain the waterlogging feature pattern; such as Pattern 1 (pipe rupture): sudden pressure drop + abnormal flow + continuous water level rise; Pattern 2 (drainage blockage): continuous water level rise + stable humidity + no pressure abnormality; Pattern 3 (external seepage): synchronous change of temperature and humidity + no abnormal pipe parameters.

[0020] By deploying multi-type sensors such as water level, temperature and humidity, pressure, flow, and vibration inside the water meter well, multi-dimensional data such as the waterlogging depth, environmental parameters, and pipeline operation status in the well are collected in real time. After data cleaning, feature engineering and other processes, comprehensive features including cross-sensor association, time-series dynamic trend, and spatial coordination are generated. Inputting the original data and comprehensive features into the waterlogging analysis model can accurately identify feature patterns such as pipe rupture, drainage blockage, and external seepage. On this basis, a remote waterlogging warning system is constructed, which can rely on the dynamic threshold and hierarchical warning mechanism to match the data with the preset pattern in real time, quickly locate the leak point coordinates and visually present them through the GIS map, and synchronously trigger SMS / APP push and maintenance work order generation. This technical system realizes the upgrade from "manual inspection" to "intelligent monitoring", significantly improves the accuracy of waterlogging cause diagnosis and response efficiency, effectively reduces risks such as water resource waste and road collapse caused by pipeline leakage, and provides data-driven decision support for the refined management and preventive maintenance of urban water supply and drainage pipe networks, helping to build an intelligent and intensive drainage safety guarantee system.

[0021] Comprehensive features include sensor cross-features, time series dynamic features and spatial correlation features; sensor cross-features include threshold trigger features, correlation features and physical derivative features; time series dynamic features include short-term window features, frequency domain features and state transition features; spatial correlation features include spatial gradient features and acoustic wave positioning derivative features; Methods for obtaining sensor cross-signal characteristics include: Detect whether the pipeline pressure change value exceeds the preset pressure threshold, detect whether the difference between the total flow value and the user-end flow exceeds the preset flow threshold; detect whether the water level rise rate exceeds the preset rate threshold; if all three are exceeded at the same time, mark the pipeline rupture risk as 1, otherwise 0; use the pipeline rupture risk as the threshold trigger feature; the pipeline rupture risk is judged by multiple thresholds of pressure, flow, and water level, which can accurately capture the coupling relationship of key indicators of high-risk scenarios such as "pipeline rupture", avoid misjudgment of a single sensor, directly output a clear risk status, and provide a strong discriminant Boolean feature for the early warning model; The Pearson correlation coefficient is used to calculate the correlation between different sensors and obtain the correlation characteristics. The Pearson correlation coefficient is used to quantify the linear correlation between sensors, which can mine the potential dependency between data, assist in distinguishing the causes of waterlogging, and enrich the characteristic dimensions of model input. Methods for obtaining physically derived features include: Obtain the total flow at the user end and the flow at the pipeline inlet, and calculate the leakage rate; count the pressure change and flow change, and calculate the joint features; concatenate the leakage rate and the joint features to obtain the physical derivative features.

[0022] The threshold trigger feature jointly determines the risk of pipeline rupture through multiple thresholds of pressure, flow, and water level, directly outputs a clear risk status in the form of Boolean values, accurately captures the coupling relationship of key indicators in high-risk scenarios, effectively avoids misjudgment of a single sensor, provides strong discriminant features for the early warning model, and improves the recognition ability of core risk scenarios such as "pipeline rupture"; the correlation feature uses the Pearson correlation coefficient to quantify the linear correlation between sensors, mines the potential dependencies of the data, and assists in distinguishing the causes of waterlogging (such as pipeline rupture, rainstorm waterlogging and other different scenarios), enriches the feature dimensions of the model input, and enables the early warning model to analyze the causes of waterlogging from a multivariate correlation perspective; the physical derivative feature calculates the joint features of leakage rate, pressure and flow changes and splices them to form an indicator with both physical meaning and comprehensive characterization capabilities, converts the basic sensor data into deep features that reflect the operating status of the pipeline network, and enhances the model's understanding of physical processes related to waterlogging (such as leakage degree and fluid dynamics changes). The three work together to build a multi-dimensional input system from three levels: risk status identification, data correlation analysis, and physical process characterization, which improves the accuracy, robustness, and scenario distinction capabilities of the early warning model, and provides a solid feature foundation for accurate early warning of remote waterlogging in water meter wells.

[0023] Methods for obtaining temporal dynamic characteristics include: The mean, standard deviation and change rate of the water level are counted, and the minimum value, fluctuation range and number of times the pressure change per unit time of the pipeline pressure is lower than the pressure threshold are counted. The mean, standard deviation and change rate of the water level, as well as the minimum value, fluctuation range and number of times the pressure change per unit time of the pipeline pressure is lower than the pressure threshold are spliced to obtain the short-term window characteristics; The short-term window feature integrates the real-time statistics of water level and pipeline pressure, accurately captures the short-term fluctuation trend of water depth and the high-frequency sudden state of pressure anomaly, and provides "immediate abnormal signal" for real-time warning; Statistically obtain the deviation between the actual temperature value and the historical average temperature of the same period, the deviation between the actual humidity value and the historical average humidity of the same period, and the proportion of low-frequency vibration energy below the preset low-frequency threshold. The deviation between the actual temperature value and the historical average temperature of the same period, the deviation between the actual humidity value and the historical average humidity of the same period, and the proportion of low-frequency vibration energy below the preset low-frequency threshold are spliced as long-term window features; The long-term window feature filters short-term noise and mines long-term trend anomalies through the deviation of temperature and humidity from the historical mean value and the proportion of low-frequency vibration energy, thus improving the system's ability to identify gradual waterlogging scenarios. The main frequency energy of the vibration signal and the high-frequency noise power of the pipeline pressure signal are extracted through Fourier transform, and the main frequency energy and high-frequency noise power are spliced as frequency domain features; Frequency domain features extract the vibration main frequency energy and pressure high-frequency noise power through Fourier transform, distinguish different abnormal modes from the frequency dimension, and enhance the feature decoupling capability of complex working conditions; The short-term window features, long-term window features, frequency domain features and state transition features are spliced to obtain time series dynamic features. The time series dynamic features integrate short-term dynamics, long-term trends, frequency domain characteristics and state transition processes (such as the time series dependency of normal → warning) into time series dynamic features, so that the waterlogging analysis model can respond to real-time sudden anomalies (such as a sudden drop in pressure + a sudden increase in water level in a short-term window) and learn long-term evolution laws (such as long-term humidity deviation + low-frequency vibration anomalies indicating pipeline aging and water seepage), so as to more accurately identify the time evolution path of characteristic patterns such as "pipeline rupture-drainage blockage-external water seepage", effectively improve the early warning system's detection sensitivity for multi-time scale anomalies and the dynamic diagnosis capability of the cause of waterlogging, and provide time series feature support with both real-time and predictive properties for remote early warning.

[0024] Methods for obtaining state transition characteristics include: Set a preset division threshold to divide the sensor status into normal, early warning, and failure, and encode them in sequence. Convert the time-series data of the sensor into a status sequence, count the number of transitions from status a to status b, calculate the total residence time of status a, calculate the ratio of the number of transitions to the total residence time, and obtain a transition probability matrix; count the continuous residence duration of each status; splice the transition probability matrix and the continuous residence duration to obtain statistical features; the statistical features quantify the transition rules between states such as "normal - early warning - failure", and combined with the status residence duration, can help the model identify the typical status sequence patterns of the causes of water accumulation. Count the average time interval from status a to status b, count the number of occurrences of a preset transition pattern, and splice the average time interval and the number of occurrences to obtain time-series dependence features; the time-series dependence features capture the time logic of state transitions, and through the high-frequency occurrence of a preset transition pattern (such as a chain transition of "abnormal vibration → sudden pressure drop → sudden water level increase"), accurately locate the time series chain of sudden water leakage events, improving the real-time performance of early warning. Obtain the change rate and change amplitude of the sensor during state transition, and whether the status of other sensors changes when the change amplitude of the sensor exceeds a preset change threshold; splice the change rate and change amplitude of the sensor, and whether the status of other sensors changes when the change amplitude of the sensor exceeds a preset change threshold to obtain dynamic trend features; the dynamic trend features verify the physical consistency of state transitions, ensuring that abnormal states are caused by real water accumulation events rather than noise interference, enhancing the reliability of early warning. Splice the statistical features, time-series dependence features, and dynamic trend features to obtain state transition features. The state transition features integrate the statistical laws, time dependence relationships, and dynamic collaborative trends of state transitions, enabling the water accumulation analysis model to learn the complete evolution path of "normal operation → early warning → failure outbreak", not only capturing current state anomalies in real time, but also predicting the escalation of water accumulation risks through state transition patterns, thereby achieving the precise triggering of hierarchical early warnings (such as mild / moderate / severe), the rapid location of the causes of water leakage points, and the early initiation of emergency responses, significantly enhancing the adaptability of the early warning system to complex working conditions and the full-process closed-loop management ability of water accumulation events.

[0025] The methods for obtaining spatial correlation features include: Calculate the water level difference and distance between adjacent water meter wells, calculate the ratio of the water level difference to the distance, and obtain the water level gradient; Calculate the pressure difference and distance between the upstream water meter well and the downstream water meter well, calculate the ratio of the pressure difference to the distance, and obtain the pressure attenuation rate; Concatenate the water level gradient and the pressure decay rate to obtain the spatial gradient feature; the spatial gradient feature can quantify the spatial distribution anomalies of the water accumulation depth and pipeline pressure, helping the model to identify the spatial aggregation of abnormal water accumulation and the along - route changes of pipeline conditions from the regional scale, and avoiding misjudgment of single - well data; Detect the time difference between any two water meter wells receiving the leakage sound wave, calculate the ratio of the difference in the received signal strength to the distance between the two water meter wells to obtain the signal decay rate; concatenate the time difference and the signal decay rate to obtain the acoustic positioning - derived feature; the acoustic positioning - derived feature is directly related to the propagation characteristics of the leakage sound wave. By using the time difference of the sound wave arrival between multiple wells and the signal strength decay law, combined with the triangulation algorithm, accurately calculate the leak point coordinates, and visually mark them on the GIS map to achieve the integration of "abnormality detection - cause analysis - location positioning"; Concatenate the spatial gradient feature and the acoustic positioning - derived feature to obtain the spatial correlation feature. The spatial correlation feature fuses the spatial distribution gradient and the acoustic positioning information, enabling the water accumulation analysis model to not only judge the spatial correlation of regional water accumulation from a macroscopic level but also accurately lock the leak point position from a microscopic level, significantly improving the adaptability of the early warning system to complex pipe network environments, providing a double - precision basis of "spatial location + abnormality type" for rapid dispatch of maintenance and reduction of leakage impact, and promoting the efficient closed - loop upgrade of water accumulation early warning from "status early warning" to "location disposal".

[0026] The methods for obtaining environmental features include: Count the historical water accumulation times at the same location, concatenate the coordinates corresponding to the location and the historical water accumulation times as the historical water accumulation feature; count the failure times of the same type of equipment, concatenate the equipment model and the failure times as the equipment failure feature, and concatenate the historical water accumulation feature and the equipment failure feature to obtain the environmental feature.

[0027] The historical water accumulation feature quantifies the historical risk level of a specific area, enabling the early warning system to dynamically adjust the monitoring strategy for "old problem points" and identify potential repeated water accumulation hazards in advance; the equipment failure feature mines the common defects of equipment, helping the model to assign a higher anomaly weight to specific types of equipment and achieve "equipment - level precise risk assessment"; Feature concatenation and integration combines the historical water accumulation law of the spatial location with the failure probability of the equipment type, providing a double environmental context of "regional risk + equipment reliability" for the water accumulation analysis model, enabling it to not only identify current real - time data anomalies but also give early warnings in combination with historical high - incidence scenarios and inherent equipment risks, effectively reducing false alarms, and helping to concentrate maintenance resources on high - risk areas and high - defect equipment, improving the economy and preventive maintenance efficiency of the early warning system.

[0028] The training method of the water accumulation analysis model includes: Pre-collect Q groups of waterlogging training data, where the waterlogging training data includes measurement data, comprehensive features, and waterlogging feature patterns.

[0029] Use the measurement data and comprehensive features as the input of the waterlogging analysis model, and the waterlogging feature pattern as the output of the waterlogging analysis model. With the goal of minimizing the error between the output waterlogging feature pattern and the actual waterlogging feature pattern, optimize the network parameters of the waterlogging analysis model through a nature-inspired optimization algorithm, obtain the network parameters corresponding to the minimum error between the waterlogging feature pattern output by the waterlogging analysis model and the actual waterlogging feature pattern, and construct the waterlogging analysis model with the corresponding network parameters as the trained waterlogging analysis model.

[0030] Combine the acoustic positioning derivative features and determine the leak point coordinates through a fusion positioning algorithm; The methods for obtaining the leak point coordinates include: Obtain the water meter well coordinates and establish a hyperbolic positioning model based on the time difference of the acoustic positioning derivative features; Obtain the signal strength received by the water meter well, establish an empirical attenuation model of the signal strength with respect to the distance, and calculate the distance from the leak point to the water meter well through the empirical attenuation model; establish three circle equations with the coordinates of three water meter wells and the corresponding distances; Fuse the hyperbolic positioning model and the three circle equations through the least squares method, such as minimizing the time difference error and the distance error from the leak point to the water meter well; Adopt the maximum likelihood estimation or Kalman filter algorithm to iteratively optimize and obtain the leak point coordinates.

[0031] The hyperbolic positioning model is based on the time difference of the sound wave arriving at multiple wells, constructs geometric constraints, and can achieve leak point spatial positioning with meter-level accuracy, solving the problem that a single sensor "knows the anomaly but does not know the location"; the empirical attenuation model combines empirical formulas to inversely deduce the leak point distance, forming a complement with the hyperbolic model, providing auxiliary constraints when the time synchronization accuracy is insufficient or the acoustic signal is weak, and improving the positioning robustness in complex environments; the fusion algorithm balances the time difference and distance errors through the least squares method, combines the maximum likelihood estimation or Kalman filter iteration optimization, eliminates the multi-solution problem and reduces the influence of noise, ensuring that the leak point coordinates are consistent with the pipe network topology; the accurate leak point coordinates can be directly visualized and marked on the GIS map, automatically generate maintenance work orders with location information, shorten the manual inspection time, help the operation and maintenance team quickly locate the leak point and start the repair, reduce the spread of waterlogging, pipe damage, and water resource waste caused by water leakage, realize the full-process closed-loop of "detection - positioning - disposal", and significantly improve the engineering practicability and emergency response efficiency of the early warning system.

[0032] Perform dynamic analysis on the measurement data to obtain derivative dynamic features. Use the measurement data and derivative dynamic features as the input of the dynamic feature model to obtain a dynamically adjusted hierarchical waterlogging warning threshold; and perform corresponding hierarchical warnings according to the relationship between the actual water accumulation volume and the hierarchical waterlogging warning threshold. Dynamically correcting the hierarchical waterlogging warning threshold based on real-time data and historical rules can solve the problem that traditional static thresholds cannot cope with environmental fluctuations, and significantly reduce the false alarm rate and missed alarm rate; through the multi-level thresholds output by the dynamic feature model, the system can accurately trigger different responses according to the matching relationship between the actual water accumulation volume and the threshold. For example, the first-level warning prompts "Drainage anomaly, it is recommended to conduct inspections", the second-level warning marks "Pipeline parameter anomaly, locate potential leakage points", and the third-level warning directly locks the leakage point coordinates and initiates an emergency response, realizing the upgrade from "extensive warning" to "accurate hierarchical disposal"; the dynamic feature model not only provides real-time thresholds, but also can predict changes in risk levels through dynamically derived features, helping the operation and maintenance team deploy resources in advance, shorten the fault response time, reduce facility damage, traffic impact and water resource waste caused by waterlogging, and provide the core technical support of "data-driven - dynamic decision-making - hierarchical disposal" for the refined management of smart water

[0033] The methods for obtaining derivative dynamic features include: Obtain the water level change amount, temperature change amount and humidity change amount within a preset time period, and calculate the ratios of the water level change amount, temperature change amount and humidity change amount to the preset time period respectively to obtain the water level change rate, temperature change rate and humidity change rate; Calculate the difference between the current pipeline pressure and the preset pipeline pressure to obtain the pressure change amplitude; calculate the ratio of the difference to the preset change time to obtain the pressure drop rate; statistically obtain the standard deviation of pressure fluctuations; Obtain the difference between the user-end flow and the pipeline inlet flow, and calculate the ratio of the difference to the pipeline inlet flow to obtain the flow difference ratio; Obtain the absolute difference between the current user-end flow and the user-end flow at the previous moment, and calculate the ratio of the absolute difference to the user-end flow at the previous moment to obtain the flow mutation coefficient; Obtain the water accumulation depth change amount within a preset change time, and calculate the ratio of the water accumulation depth change amount to the preset change time to obtain the water accumulation depth change rate; Obtain the amplitude of the high-frequency vibration signal when the vibration signal exceeds the preset high-frequency threshold; Obtain whether the temperature change rate exceeds the preset temperature change rate threshold. If it does, mark it as 1, otherwise as 0, to obtain the temperature anomaly index; Obtain whether the humidity change rate exceeds the preset humidity change rate threshold. If it does, mark it as 1, otherwise as 0, to obtain the humidity anomaly index; Concatenate the water level change rate, temperature change rate, humidity change rate, standard deviation of pressure fluctuation, ratio of flow difference, flow mutation coefficient, change rate of accumulated water depth, amplitude of high-frequency vibration signal, temperature anomaly index, and humidity anomaly index to obtain derivative dynamic features.

[0034] Features such as the water level change rate, temperature change rate, humidity change rate, and change rate of accumulated water depth quantitatively measure the evolution speed of the environment and accumulated water in real time, helping the system quickly identify abnormal development trends; features such as the ratio of flow difference and flow mutation coefficient comprehensively judge the leakage possibility from the aspect of the pipeline operation state to avoid misjudgment by a single index; features such as the standard deviation of pressure fluctuation and amplitude of high-frequency vibration signal lock abnormal vibration and pressure instability of the pipeline. Combining the temperature anomaly index and humidity anomaly index, accurately distinguish the causes such as pipeline rupture and leakage, external seepage, and drainage blockage; concatenate continuous dynamic parameters with discrete anomaly indicators to provide rich input dimensions for the dynamic feature model, enabling the system to dynamically adjust the classification threshold according to real-time features, and realizing accurate response from anomaly detection to risk classification. For example, a mild warning indicates a trend anomaly, a moderate warning locates potential problems, and a severe warning directly locks the leak point and initiates emergency disposal, significantly improving the real-time adaptation ability of the warning system to complex working conditions and the full-process closed-loop management efficiency of accumulated water events.

[0035] The training method of the dynamic feature model includes: Pre-collect W groups of dynamic training data, where the dynamic training data includes measurement data, derivative dynamic features, and classification thresholds for accumulated water warnings.

[0036] Use the measurement data and derivative dynamic features as the input of the dynamic feature model, and the classification threshold for accumulated water warnings as the output of the dynamic feature model. With the goal of minimizing the error between the output classification threshold for accumulated water warnings and the actual classification depth of accumulated water, optimize the network parameters of the dynamic feature model through a nature-inspired optimization algorithm to obtain the network parameters corresponding to the minimum error between the classification threshold for accumulated water warnings output by the dynamic feature model and the actual classification threshold for accumulated water warnings (the actual classification threshold for accumulated water warnings is obtained by reviewing historical accumulated water events, extracting critical accumulated water thresholds under different working conditions through numerical simulation or machine learning classification, and comparing real-time accumulated water monitoring data with the historical threshold library, and adjusting the threshold through rolling window update or online learning to adapt to the current working condition (such as increasing the warning sensitivity in the rainy season and reducing the false alarm rate in the dry season)), and construct the dynamic feature model with the corresponding network parameters as the trained dynamic feature model.

[0037] Example 2 This example provides a method for automatically tracing the diffusion trend of accumulated water applied to Example 1, including the following steps: Divide the pipeline into an N×N grid, and obtain the real-time measurement data and dynamic derivative features within the grid; use the measurement data and dynamic derivative features as the input of the water accumulation analysis model to obtain the water accumulation probability. Take the grid with a water accumulation probability greater than the preset probability threshold as the tracing starting point, analyze the hydraulic connectivity between grids through a graph neural network to obtain a hydraulic connectivity weight matrix, combine the hydraulic connectivity, and predict the water accumulation diffusion trend through a bidirectional long short-term memory network. According to the hydraulic connectivity weight matrix and combined with the time decay factor, construct a reverse diffusion tree from the current moment to the historical moment; simulate and obtain the diffusion data under different leak point positions according to the reverse diffusion tree, and perform a spatial correlation analysis on the actual diffusion data and the predicted water accumulation diffusion trend. If the spatial correlation exceeds the correlation threshold, obtain the corresponding leak point position. If there are multiple leak points, use Kalman filtering to fuse multi-sensor data for secondary screening to obtain the leak point position.

[0038] Embodiment 3 Please refer to Figure 3 as shown in the figure. This embodiment provides an Internet of Things-based remote water accumulation warning system for water meter wells, including: Feature analysis module: Integrate the collected measurement data of the water meter well through the Internet of Things gateway, comprehensively analyze the measurement data to obtain comprehensive features, and use the measurement data and comprehensive features as the input of the water accumulation analysis model to obtain the water accumulation feature pattern; The comprehensive features include sensor cross features, time series dynamic features, spatial association features, and environmental features; the sensor cross features include threshold trigger features, correlation features, and physical derivative features; the time series dynamic features include short-term window features, frequency domain features, and state transition features; the spatial association features include spatial gradient features and acoustic positioning derivative features; Leak point positioning module: Combine the acoustic positioning derivative features and determine the leak point coordinates through a fusion positioning algorithm; Water accumulation warning module: Dynamically analyze the measurement data to obtain derivative dynamic features, use the measurement data and derivative dynamic features as the input of the dynamic feature model to obtain a dynamically adjusted hierarchical water accumulation warning threshold; and perform corresponding hierarchical warnings according to the relationship between the actual water accumulation volume and the hierarchical water accumulation warning threshold.

[0039] The above is only the specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present application can easily think of changes or substitutions, which should all be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

[0040] Finally, the above are only the preferred embodiments of the present application and are not intended to limit the present application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A remote waterlogging warning method for water meter wells based on the Internet of Things, characterized in that, It includes the following steps: Integrate the measured data of the water meter well collected through the Internet of Things gateway, conduct comprehensive analysis on the measured data to obtain comprehensive features, use the measured data and comprehensive features as the input of the waterlogging analysis model to obtain the waterlogging feature pattern; The comprehensive features include sensor cross features, time series dynamic features, spatial association features and environmental features; the sensor cross features include threshold trigger features, correlation features and physical derivative features; the time series dynamic features include short-term window features, frequency domain features and state transition features; the spatial association features include spatial gradient features and acoustic positioning derivative features; Combine the acoustic positioning derivative features and determine the leak point coordinates through the fusion positioning algorithm; Conduct dynamic analysis on the measured data to obtain derivative dynamic features, use the measured data and derivative dynamic features as the input of the dynamic feature model to obtain the dynamically adjusted hierarchical waterlogging warning threshold; And conduct corresponding hierarchical warnings according to the relationship between the actual waterlogging volume and the hierarchical waterlogging warning threshold.

2. The remote waterlogging warning method for water meter wells based on the Internet of Things according to claim 1, wherein The method for obtaining the threshold trigger feature includes: Detect whether the change value of the pipeline pressure exceeds the preset pressure threshold, detect whether the difference between the total flow value and the user-end flow exceeds the preset flow threshold; detect whether the water level rising rate exceeds the preset rate threshold; if all three exceed at the same time, mark the pipeline rupture risk as 1, otherwise as 0; use the pipeline rupture risk as the threshold trigger feature; The method for obtaining the correlation feature includes: Calculate the correlation between different sensors using the Pearson correlation coefficient to obtain the correlation feature.

3. The remote waterlogging warning method for water meter wells based on the Internet of Things according to claim 1, characterized in that, The method for obtaining the physical derivative feature includes: Obtain the total flow at the user end and the pipeline inlet flow, and calculate the leakage rate; count the pressure change amount and the flow change amount, and calculate the combined feature; splice the leakage rate and the combined feature to obtain the physical derivative feature.

4. The remote waterlogging warning method for water meter wells based on the Internet of Things according to claim 1, characterized in that, The method for obtaining the state transition feature includes: Preset the division threshold, divide the sensor state into normal, warning and failure, and encode them in order, convert the time series data of the sensor into a state sequence, count the number of transitions of the sensor from state a to state b, calculate the total residence time of state a, calculate the ratio of the number of transitions to the total residence time to obtain the transition probability matrix; count the continuous residence duration of each state; splice the transition probability matrix and the continuous residence duration to obtain the statistical feature; Count the average time interval from state a to state b, count the number of occurrences of the preset transition pattern, and splice the average time interval and the number of occurrences to obtain the time series dependence feature; Obtain the change rate and change amplitude of the sensor when the state transitions, and whether the states of other sensors change when the change amplitude of the sensor exceeds the preset change threshold; splice the change rate and change amplitude of the sensor, and whether the states of other sensors change when the change amplitude of the sensor exceeds the preset change threshold to obtain the dynamic trend feature; Splice the statistical feature, the time series dependence feature and the dynamic trend feature to obtain the state transition feature.

5. The remote waterlogging warning method for water meter wells based on the Internet of Things according to claim 1, characterized in that, The method for obtaining the time series dynamic feature includes: Statistically calculate the mean value, standard deviation, and change rate of the water level, and statistically calculate the minimum value of the pipeline pressure, the fluctuation range, and the number of times the pressure change amount per unit time is lower than the pressure threshold. Concatenate the mean value, standard deviation, and change rate of the water level, as well as the minimum value of the pipeline pressure, the fluctuation range, and the number of times the pressure change amount per unit time is lower than the pressure threshold to obtain short-term window features; Statistically obtain the deviation between the actual temperature value and the mean temperature value in the same historical period, the deviation between the actual humidity value and the mean humidity value in the same historical period, and the proportion of low-frequency vibration energy lower than the preset low-frequency threshold. Concatenate the deviation between the actual temperature value and the mean temperature value in the same historical period, the deviation between the actual humidity value and the mean humidity value in the same historical period, and the proportion of low-frequency vibration energy lower than the preset low-frequency threshold as long-term window features; Extract the main frequency energy of the vibration signal and the high-frequency noise power of the pipeline pressure signal through Fourier transform, and concatenate the main frequency energy and the high-frequency noise power as frequency domain features; Concatenate the short-term window features, long-term window features, frequency domain features, and state transition features to obtain time series dynamic features.

6. The remote waterlogging warning method for water meter wells based on the Internet of Things according to claim 1, characterized in that, The methods for obtaining spatial correlation features include: Calculate the water level difference and the distance between adjacent water meter wells, calculate the ratio of the water level difference to the distance to obtain the water level gradient; Calculate the pressure difference and the distance between the upstream water meter well and the downstream water meter well, calculate the ratio of the pressure difference to the distance to obtain the pressure attenuation rate; Concatenate the water level gradient and the pressure attenuation rate to obtain spatial gradient features; Detect the time difference between any two water meter wells receiving leakage sound waves, calculate the ratio of the difference in the received signal intensity to the distance between the two water meter wells to obtain the signal attenuation rate; concatenate the time difference and the signal attenuation rate to obtain acoustic positioning derivative features; Concatenate the spatial gradient features and the acoustic positioning derivative features to obtain spatial correlation features.

7. The remote waterlogging warning method for water meter wells based on the Internet of Things according to claim 1, characterized in that, The methods for obtaining environmental features include: Statistically calculate the historical waterlogging times at the same location, concatenate the coordinates corresponding to the location and the historical waterlogging times as historical waterlogging features; statistically calculate the failure times of similar equipment, concatenate the equipment model and the failure times as equipment failure features, and concatenate the historical waterlogging features and the equipment failure features to obtain environmental features.

8. The remote waterlogging warning method for water meter wells based on the Internet of Things according to claim 6, wherein, The methods for obtaining the coordinates of the leakage point include: Obtain the coordinates of the water meter well, and establish a hyperbolic positioning model according to the time difference of the acoustic positioning derivative features; Obtain the signal intensity received by the water meter well, establish an empirical attenuation model of the signal intensity with respect to the distance, and calculate the distance from the leakage point to the water meter well through the empirical attenuation model; establish three circle equations with the coordinates of three water meter wells and the corresponding distances; Fuse the hyperbolic positioning model and the three circle equations by the least squares method; Adopt the maximum likelihood estimation or Kalman filtering algorithm to iteratively optimize to obtain the coordinates of the leakage point.

9. The remote waterlogging warning method for water meter wells based on the Internet of Things according to claim 1, characterized in that The methods for obtaining derivative dynamic features include: Obtain the water level change amount, temperature change amount, and humidity change amount within a preset time period, and calculate the ratios of the water level change amount, temperature change amount, and humidity change amount to the preset time period respectively to obtain the water level change rate, temperature change rate, and humidity change rate; Calculate the difference between the current pipeline pressure and the preset pipeline pressure to obtain the pressure change range; calculate the ratio of the difference to the preset change time to obtain the pressure drop rate; statistically obtain the standard deviation of pressure fluctuations. Obtain the difference between the user - end flow rate and the pipeline inlet flow rate, and calculate the ratio of the difference to the pipeline inlet flow rate to obtain the proportion of flow difference. Obtain the absolute difference between the current user - end flow rate and the user - end flow rate at the previous moment, and calculate the ratio of the absolute difference to the user - end flow rate at the previous moment to obtain the flow mutation coefficient. Obtain the change amount of the water accumulation depth within the preset change time, and calculate the ratio of the change amount of the water accumulation depth to the preset change time to obtain the water accumulation depth change rate. Obtain the amplitude of the high - frequency vibration signal whose vibration signal exceeds the preset high - frequency threshold. Obtain whether the temperature change rate exceeds the preset temperature change rate threshold. If so, mark it as 1, otherwise as 0, to obtain the temperature anomaly index. Obtain whether the humidity change rate exceeds the preset humidity change rate threshold. If so, mark it as 1, otherwise as 0, to obtain the humidity anomaly index. Concatenate the water level change rate, temperature change rate, humidity change rate, standard deviation of pressure fluctuations, proportion of flow difference, flow mutation coefficient, water accumulation depth change rate, amplitude of high - frequency vibration signal, temperature anomaly index, and humidity anomaly index to obtain the derived dynamic features.

10. The remote water accumulation warning system for water meter wells based on the Internet of Things implements the remote water accumulation warning method for water meter wells based on the Internet of Things according to any one of claims 1-9, characterized in that Include: Feature analysis module: Integrate the measured data of the water meter well collected through the IoT gateway, comprehensively analyze the measured data to obtain comprehensive features, and use the measured data and comprehensive features as the input of the water accumulation analysis model to obtain the water accumulation feature pattern. The comprehensive features include sensor cross - features, time - series dynamic features, spatial correlation features, and environmental features; the sensor cross - features include threshold - trigger features, correlation features, and physical derivative features; the time - series dynamic features include short - term window features, frequency - domain features, and state - transfer features; the spatial correlation features include spatial gradient features and acoustic positioning derivative features. Leak point location module: Combine the acoustic positioning derivative features and determine the leak point coordinates through the fusion positioning algorithm. Water accumulation warning module: Dynamically analyze the measured data to obtain the derived dynamic features, and use the measured data and the derived dynamic features as the input of the dynamic feature model to obtain the dynamically adjusted graded water accumulation warning threshold. And perform corresponding graded warnings according to the relationship between the actual water accumulation volume and the graded water accumulation warning threshold.

Images