An intelligent water meter operation monitoring method and system based on an internet of things

Abstract

The application discloses a kind of based on Internet of Things's intelligent water meter operation monitoring method and system, it is related to intelligent water meter technical field, the method is by constructing disturbance consistency score index Ipd and behavior atlas variation score Simv, and introduce multi-parameter coupling backtracking risk level score function Rtrace, disturbance mutation intensity, periodic stability, flow consistency and historical behavior deviation degree can be considered comprehensively, effectively improve the identification precision of non-natural disturbance water flow and potential abnormal water use behavior.Especially in the case where disturbance anomaly and behavior atlas anomaly are coupled, the backtracking risk level score function Rtrace is compared twice with the risk interval threshold, which can achieve more differentiated risk judgment and risk level division, which is beneficial to accurately lock illegal water taking, backflow and other abnormal water use behaviors under the condition of low misjudgment rate, and enhances the intelligent perception ability and behavior modeling reliability of urban water supply system.

Description

Technical Field

[0001] This invention relates to the field of smart water meter technology, specifically to a method and system for monitoring the operation of smart water meters based on the Internet of Things. Background Technology

[0002] With the widespread deployment of IoT technology in urban infrastructure, smart water systems, as a core component of next-generation urban water resource management, are continuously improving their sensing capabilities and operational monitoring accuracy. Especially in urban water supply network management scenarios, smart water meters, as edge sensing terminal devices, have gradually replaced traditional mechanical meters, enabling refined collection and remote control of water usage behavior, flow fluctuations, and operational status. This paper presents an operational monitoring method based on the data sensing capabilities of existing smart water meters, combined with an IoT platform for hydrodynamic disturbance identification and behavior analysis. More specifically, it involves an intelligent monitoring method that can identify and respond to abnormal backflow paths, illegal water extraction, and unauthorized branch connections in the pipe network without requiring additional dedicated pressure sensors or flow direction identification hardware.

[0003] Currently, the identification and handling of issues such as illegal water extraction, backflow pollution, and unauthorized pipeline connections in water supply systems mainly rely on the deployment of costly hardware such as differential pressure sensors, one-way check valves, and physical flow direction calibration devices. These devices are not only expensive to install and maintain, and difficult to achieve comprehensive coverage, but also cannot meet the monitoring needs of high-density areas or older residential communities with complex pipe network structures and frequent multi-point interference. Meanwhile, while traditional smart water meters have flow rate and total volume monitoring functions, they lack the ability to sensitively identify minute disturbances and unstable reverse disturbances, making it impossible to accurately determine whether there is backflow-type abnormal behavior or overlapping multi-source water use paths, resulting in significant lag and blind spots in the identification of illegal activities.

[0004] The aforementioned problems primarily stem from the lack of sufficient multi-dimensional modeling capabilities for hydraulic disturbance patterns in current pipeline monitoring mechanisms. This is especially true in practical applications, such as during off-peak water usage periods at night, at the end of branch networks, or in nodes with multiple users connected in parallel. Sudden changes in water pressure, reversed flow direction, or frequency disturbances are easily overlooked, and traditional systems cannot effectively collect and analyze these weak signals. Once illegal reconnection of pipelines, backflow, or unauthorized pump connections occur, it can easily lead to metering errors in neighboring households, water backflow contamination risks, and system hydraulic imbalances. This not only damages the metering accuracy and economic benefits of water supply companies but may also pose a serious threat to regional public health and pipeline stability. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a smart water meter operation monitoring method and system based on the Internet of Things, which solves the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution, comprising the following steps:

[0007] S1. Set up a collection point at the smart water meter to collect disturbance data in real time and transmit the disturbance data to the Internet of Things (IoT) platform. In the IoT platform, perform feature extraction and preprocessing on the disturbance data to obtain a standardized disturbance dataset.

[0008] S2. Calculate the perturbation consistency score index Ipd based on the standardized perturbation dataset, and preset the perturbation threshold Ith. Perform a preliminary comparison and evaluation between the perturbation consistency score index Ipd and the perturbation threshold Ith.

[0009] S3. Based on the preliminary comparison and evaluation results, trigger the behavior backtracking process, extract the historical normal cycle behavior map, analyze the current cycle behavior feature group Bnow, and call the normal cycle behavior feature group Bref in the normal cycle behavior map for comparison, and output the mutation map similarity score Simv.

[0010] S4. Combine the perturbation consistency score index Ipd with the mutation map similarity score Simv to calculate the backtracking risk score function Rtrace.

[0011] S5. Preset a risk range threshold, and then compare and evaluate the risk range threshold with the backtracking risk level score Rtrace to determine the risk level of the current user's water usage behavior. The risk level will trigger corresponding response control.

[0012] Preferably, S1 includes S11 and S12;

[0013] S11. Two data collection points are set in the water flow channel structure of the smart water meter, including a first disturbance data collection point and a second disturbance data collection point;

[0014] The first disturbance data acquisition point is located in the central area of ​​the water inlet guide cavity of the water meter body, and is used to collect the initial disturbance information of the water flow;

[0015] The second disturbance data acquisition point is located downstream of the outlet pressure stabilizing chamber and near the outlet valve, and is used to collect data on the water flow behavior after disturbance.

[0016] Each collection point is equipped with a high-sensitivity electromagnetic flow velocity sensor, a piezoelectric water hammer sensor, an array of ultrasonic reflection sensors, and a disturbance frequency capture module to collect disturbance data in real time. The disturbance data includes the disturbance flow velocity change sequence Vdiff, the direction reflection offset vector PhΔ, and the disturbance frequency distribution function Ff. The disturbance data is then combined into a disturbance raw data frame and cached in the edge data control module of the smart water meter.

[0017] S12. The original disturbance data frame is packaged and transmitted through the wireless communication module configured inside the smart water meter. The wireless communication module uses the NB-IoT protocol to send the upload data message containing timestamp, data type identifier and disturbance parameter value group to the remote IoT platform.

[0018] Preferably, S1 further includes S13;

[0019] S13. After receiving the uploaded data packet on the Internet of Things platform, the disturbance data is unpacked and extracted and the features are extracted to obtain the disturbance feature set. The disturbance feature set includes the disturbance mutation gradient Dg, the disturbance fluctuation period Tp and the flow direction variation index Rrev.

[0020] The disturbance abrupt change gradient Dg analyzes the disturbance velocity change sequence Vdiff through a sliding time window. Within the current sliding time window, it identifies the maximum absolute increase in the velocity change value per unit time. Then, it performs statistical analysis on the entire disturbance velocity change sequence Vdiff and calculates the standard deviation of the change amplitude. Finally, it adds the maximum absolute increase in the velocity change value per unit time to the standard deviation to obtain the disturbance abrupt change gradient Dg.

[0021] The disturbance fluctuation period Tp is obtained by performing spectral analysis on the disturbance frequency distribution function Ff to extract the peak amplitude position corresponding to its dominant frequency; then, the corresponding angular frequency is derived based on the dominant frequency, where the angular frequency is calculated by multiplying 2π by the dominant frequency; and finally, the disturbance period Tp is obtained by calculating the corresponding disturbance period using the reciprocal of the angular frequency.

[0022] The flow direction variation index Rrev is obtained by using the time series of the direction reflection offset vector PhΔ as input parameters. Within a set observation period T, the first derivative of the direction reflection offset vector PhΔ on a continuous time slice is calculated using the numerical difference method to obtain the phase offset rate change sequence. The absolute value of the value at each time point in the phase offset rate change sequence is taken, and the absolute value sequence within the entire sliding time window is numerically integrated to obtain the cumulative amount of the reverse trend fluctuation within the sliding time window. The cumulative amount of the reverse trend fluctuation is divided by the duration of the period T to obtain the average reverse offset degree per unit time. The result obtained is the flow direction variation index Rrev.

[0023] Furthermore, based on the minimum and maximum values ​​of each parameter in the historical sample data, the three parameters of the disturbance feature set are subjected to min-max normalization to eliminate the influence of physical dimensions and form a standardized disturbance dataset.

[0024] Preferably, S2 includes S21;

[0025] S21. Construct a disturbance consistency score calculation model in the Internet of Things platform, and input the standardized disturbance dataset acquired in real time into the constructed disturbance consistency score calculation model to calculate and output the disturbance consistency score index Ipd, which measures the non-natural disturbance characteristics of the local water flow state.

[0026] The perturbation consistency score index Ipd is calculated and output using the following perturbation consistency score calculation model;

[0027] Within the set observation time window, the first derivative of all water flow rate data in the disturbance velocity change sequence Vdiff is calculated to obtain the velocity change rate sequence. The velocity change rate sequence is then multiplied point-to-point with the sine function corresponding to the disturbance main frequency signal to form the disturbance derivative modulation sequence.

[0028] Perform a square integration operation on the perturbation derivative modulation sequence to obtain the perturbation modulation energy integral value;

[0029] Based on the integral value of the disturbance modulation energy, a disturbance intensity suppression factor, which is formed by adding the absolute value of the disturbance abrupt gradient Dg to a constant 1, is introduced as a denominator. The ratio of the disturbance intensity suppression factor is multiplied by the reciprocal of the natural logarithm of the disturbance fluctuation period Tp to obtain the disturbance consistency score index Ipd.

[0030] Preferably, S2 further includes S22;

[0031] S22. The calculated disturbance consistency score index Ipd is compared and evaluated with the set disturbance threshold Ith to determine the consistency of the current water flow disturbance. The disturbance threshold Ith is an empirical dynamic statistical threshold. Based on a large number of historical disturbance consistency score indices Ipd, a score lower bound with a confidence level of 95% is selected from the disturbance consistency score index Ipd samples of historical normal disturbance behavior as the initial threshold benchmark.

[0032] When the disturbance consistency score index Ipd > the disturbance threshold Ith, the current water flow disturbance behavior is consistent in cycle and stable in direction. Only the behavior log is recorded and routine monitoring continues.

[0033] When the perturbation consistency score index Ipd ≤ perturbation threshold Ith, it is determined that there is a risk of perturbation consistency loss in the current water flow behavior, and the behavior backtracking process is triggered.

[0034] Preferably, S3 includes S31;

[0035] S31. After the initial comparison and evaluation of the triggered behavior backtracking process, the historical normal cycle behavior map is extracted from the historical behavior map database in the Internet of Things platform. The historical normal cycle behavior map is constructed by monitoring the water use behavior samples of multiple users at different times, under different water pressures and different living habits. Then, the sample data in the historical normal cycle behavior map is normalized to form a cycle behavior feature template with a unified structure.

[0036] Within the current behavior cycle, continuous operating data of the water meter is collected to construct the current cycle behavior feature group Bnow. The current cycle behavior feature group Bnow includes instantaneous water usage duration, maximum peak flow velocity, flow velocity change rate slope, intraday water usage frequency, water usage interval distribution, and water usage rhythm cycle.

[0037] From the historical normal cycle behavior map, retrieve the cycle behavior feature template that matches the current user category, house type, water meter model and typical water use pattern, and extract the normal cycle behavior feature group Bref.

[0038] Preferably, S3 further includes S32;

[0039] S32. Based on the obtained values ​​of each corresponding feature dimension in the current cycle behavior feature group Bnow and the normal cycle behavior feature group Bref, perform item-by-item difference processing to obtain the absolute value sequence of behavior deviation. Summate the absolute value sequence of behavior deviation to obtain the total value of behavior deviation. Divide the total value of behavior deviation by the sum of the cumulative reference values ​​of all feature dimensions in the normal cycle behavior feature group Bref to obtain the variation map similarity score Simv, which measures the degree of deviation of the current water use behavior from the standard behavior map in the overall behavior feature space.

[0040] Preferably, S4 includes S41;

[0041] S41. Based on the perturbation consistency score index Ipd, the mutation map similarity score Simv, and the cumulative number of abnormal behaviors Tact within the current detection period, calculate the backtracking risk level score Rtrace to quantitatively analyze the current user's abnormal water usage behavior.

[0042] Preferably, S5 includes S51 and S52;

[0043] S51. Extract the historical backtracking risk level score Rtrace, and perform frequency distribution analysis on all historical backtracking risk level scores Rtrace. Set the critical value of the historical backtracking risk level score Rtrace of normal users and the historical backtracking risk level score Rtrace of abnormal users as the risk interval threshold.

[0044] The risk interval thresholds include a first risk interval threshold F1 and a second risk interval threshold F2;

[0045] Among them, the first risk interval threshold F1 is the critical value of the historical backtracking risk level score Rtrace for normal users, and the second risk threshold F2 is the critical value of the historical backtracking risk level score Rtrace for abnormal users.

[0046] S52. Based on the real-time acquired retrospective risk level score Rtrace and the risk interval threshold, a secondary comparison and evaluation is performed to determine the risk level of the current user's water usage behavior, and corresponding differentiated response control is implemented based on the risk level. The specific evaluation content is as follows.

[0047] If the risk level score Rtrace is less than the first risk interval threshold F1, it is determined to be a level 1 risk behavior, the behavior log is recorded and normal operation is maintained.

[0048] If the first risk interval threshold F1 is between the retrospective risk level score Rtrace and the second risk interval threshold F2, it is determined to be a level 2 risk behavior, and the behavior trend tracking mechanism is activated, extending the data collection cycle by 50% and recording the behavior vector changes in real time.

[0049] If the risk level score Rtrace is greater than or equal to the second risk interval threshold F2, it is determined to be a level 3 risk behavior. The highest level response scheme in the behavior trigger strategy library is invoked to generate a set of processing commands including water meter control instructions, abnormal alarm instructions and platform write-back instructions.

[0050] The set of processing commands is executed after being parsed in the smart water meter controller;

[0051] Among them: the water meter control command locks abnormal users, prohibits the reporting and settlement of user water use records, and cuts off the water supply to the current user;

[0052] Anomaly alarm commands send warning information to the administrator's port via a remote IoT platform;

[0053] The platform write-back command is used to update the status of the user behavior graph and synchronize abnormal behavior information to the audit database for administrator audit review.

[0054] A smart water meter operation monitoring system based on the Internet of Things includes a disturbance data acquisition module, a disturbance consistency analysis module, a similarity analysis module, a comprehensive analysis module, and a risk level response control module.

[0055] The disturbance data acquisition module collects disturbance data in real time by setting up collection points at the smart water meter and transmits the disturbance data to the Internet of Things (IoT) platform. The IoT platform performs feature extraction and preprocessing on the disturbance data to obtain a standardized disturbance dataset.

[0056] The disturbance consistency analysis module calculates the disturbance consistency score index Ipd based on a standardized disturbance dataset and presets a disturbance threshold Ith, and then performs a preliminary comparison and evaluation between the disturbance consistency score index Ipd and the disturbance threshold Ith.

[0057] The similarity analysis module triggers a behavior backtracking process based on the preliminary comparison and evaluation results, extracts historical normal cycle behavior maps, analyzes the current cycle behavior feature group Bnow, and calls the normal cycle behavior feature group Bref in the normal cycle behavior map for comparison, outputting a variant map similarity score Simv.

[0058] The comprehensive analysis module calculates the perturbation consistency score index Ipd and the mutation map similarity score Simv together and outputs the backtracking risk score function Rtrace.

[0059] The risk level response control module determines the risk level of the current user's water usage behavior by comparing the risk range threshold with the retrospective risk level score Rtrace, based on a preset risk range threshold. The module then executes corresponding response controls based on the risk level.

[0060] This invention provides a method and system for monitoring the operation of smart water meters based on the Internet of Things (IoT). It has the following beneficial effects:

[0061] (1) This method sets two collection points in the water flow channel structure of a smart water meter, and integrates a high-sensitivity electromagnetic flow velocity sensor, a piezoelectric water hammer sensor, an array-type ultrasonic reflection sensor, and a disturbance frequency capture module at each collection point. This invention can comprehensively sense disturbance data such as the disturbance velocity change sequence Vdiff, the direction reflection offset vector PhΔ, and the disturbance frequency distribution function Ff. The original data frame is cached by the edge data control module and wirelessly transmitted to the Internet of Things platform via the NB-IoT protocol. Finally, the disturbance mutation gradient Dg, the disturbance fluctuation period Tp, and the flow direction variation index Rrev are extracted and normalized on the platform side to form a standardized disturbance dataset. This scheme effectively avoids the limitations of traditional schemes that rely on independent pressure sensors for anomaly identification, and has the advantages of simple deployment, rich sensing dimensions, and high identification accuracy.

[0062] (2) This method constructs a disturbance consistency scoring index Ipd and a behavior map variation scoring index Simv, and introduces a multi-parameter coupled backtracking risk level scoring function Rtrace. It can comprehensively consider the disturbance mutation intensity, periodic stability, flow direction consistency, and historical behavior deviation, effectively improving the identification accuracy of non-natural disturbance water flow and potential abnormal water use behavior. Especially in the case of disturbance anomaly coupled with behavior map anomaly, the scoring model can achieve more discriminative risk judgment, which is conducive to accurately locking abnormal water use behaviors such as illegal water extraction and backflow under low false judgment conditions, and enhancing the intelligent perception capability and behavior modeling reliability of urban water supply system.

[0063] (3) Based on the backtracking risk level score Rtrace and historical threshold modeling, this method proposes a differentiated response mechanism for risk levels one to three. Combined with the behavior triggering strategy library, it automatically generates response instruction sets, realizing intelligent linkage control of the entire process from behavior recording and data sampling extension to water supply cut-off and automatic distribution of audit tasks. Especially in high-risk situations, the system can automatically lock abnormal users and simultaneously push alarm signals and write back the platform status, significantly improving the real-time performance and security of platform operation and maintenance, and ensuring that water supply behavior still has a robust and controllable closed-loop processing capability in complex urban scenarios. Attached Figure Description

[0064] Figure 1 This is a schematic diagram of the operation monitoring method for smart water meters based on the Internet of Things according to the present invention;

[0065] Figure 2 This is a schematic diagram illustrating the steps of an IoT-based smart water meter operation monitoring system according to the present invention.

[0066] Figure 3 This is a schematic diagram of the module composition and data flow path of an IoT smart water meter monitoring system. Detailed Implementation

[0067] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. Example

[0068] Please see Figure 1 and Figure 3 This invention provides a method for monitoring the operation of smart water meters based on the Internet of Things. To achieve the above objectives, this invention employs the following technical solution, comprising the following steps:

[0069] S1. Set up a collection point at the smart water meter to collect disturbance data in real time and transmit the disturbance data to the Internet of Things (IoT) platform. In the IoT platform, perform feature extraction and preprocessing on the disturbance data to obtain a standardized disturbance dataset.

[0070] S2. Calculate the perturbation consistency score index Ipd based on the standardized perturbation dataset, and preset the perturbation threshold Ith. Perform a preliminary comparison and evaluation between the perturbation consistency score index Ipd and the perturbation threshold Ith.

[0071] S3. Based on the preliminary comparison and evaluation results, trigger the behavior backtracking process, extract the historical normal cycle behavior map, analyze the current cycle behavior feature group Bnow, and call the normal cycle behavior feature group Bref in the normal cycle behavior map for comparison, and output the mutation map similarity score Simv.

[0072] S4. Combine the perturbation consistency score index Ipd with the mutation map similarity score Simv to calculate the backtracking risk score function Rtrace.

[0073] S5. Preset risk range threshold, and then compare and evaluate the risk range threshold with the backtracking risk level score Rtrace to determine the risk level of the current user's water use behavior. The risk level will trigger the corresponding response control.

[0074] In this embodiment, the method sets up inlet and outlet water collection points inside the smart water meter structure, integrates multimodal sensors to collect disturbance parameters such as disturbance flow velocity change sequences, directional reflection offset vectors, and disturbance frequency functions, and uploads them to the Internet of Things platform via an NB-IoT wireless communication module to achieve low-latency, high-fidelity raw disturbance data transmission. On the platform, core disturbance features such as disturbance mutation gradient, disturbance fluctuation period, and flow direction variation index are extracted to construct a standardized disturbance dataset. Then, a disturbance consistency score index Ipd is calculated and initially compared with the dynamic disturbance threshold Ith to identify whether there is a consistency risk in the current water flow behavior. When a potential risk is identified, a historical behavior map comparison mechanism is further triggered. Based on the deviation analysis between the current periodic behavior feature group Bnow and the historical periodic behavior feature group Bref, a variation map similarity score Simv is calculated. Combined with the abnormal behavior repetition frequency Tact, a comprehensive backtracking risk level score Rtrace is generated. Finally, by setting risk interval thresholds obtained from historical statistics, the risk level score Rtrace is compared and graded to determine the risk level of user behavior and match corresponding differentiated response strategies, such as log recording, behavior tracking, or abnormal water outage control. Through the organic integration of the above processes, this invention not only achieves dual identification of abnormal disturbances and behavioral variations without relying on new pressure hardware, but also constructs a risk response mechanism with continuous adaptive capabilities. This method has advantages such as high identification accuracy, low deployment cost, wide applicability, and strong system stability. It is particularly suitable for dealing with hidden high-risk scenarios in urban water supply networks, such as illegal water extraction, pipeline backflow, and unauthorized connections, significantly improving the behavior perception capability and security response level of smart water meters, and achieving a closed-loop management goal from data perception to risk control. Example

[0075] Please see Figure 1 Specifically: S1 includes S11 and S12;

[0076] S11. Two data collection points are set in the water flow channel structure of the smart water meter, including a first disturbance data collection point and a second disturbance data collection point;

[0077] The first disturbance data acquisition point is located in the central area of ​​the water inlet guide cavity of the water meter body, and is used to collect the initial disturbance information of the water flow;

[0078] The second disturbance data collection point is located downstream of the outlet pressure stabilizing chamber and near the outlet valve, and is used to collect data on the water flow behavior after disturbance.

[0079] Each collection point is equipped with a high-sensitivity electromagnetic flow velocity sensor, a piezoelectric water hammer sensor, an array of ultrasonic reflection sensors, and a disturbance frequency capture module to collect disturbance data in real time. The disturbance data includes the disturbance flow velocity change sequence Vdiff, the direction reflection offset vector PhΔ, and the disturbance frequency distribution function Ff. The disturbance data is then combined into a disturbance raw data frame and cached in the edge data control module of the smart water meter.

[0080] S12. The original data frame of the disturbance is packaged and transmitted through the wireless communication module configured inside the smart water meter. The wireless communication module uses the NB-IoT protocol to send the uploaded data message containing the timestamp, data type identifier and disturbance parameter value group to the remote IoT platform, ensuring the continuity of data transmission, low latency and integrity of data structure.

[0081] S13. After receiving the uploaded data packet on the IoT platform, unpack the packet to extract the disturbance data and perform feature extraction to obtain the disturbance feature set. The disturbance feature set includes the disturbance mutation gradient Dg, the disturbance fluctuation period Tp, and the flow direction variation index Rrev.

[0082] The disturbance abrupt change gradient Dg is obtained by analyzing the disturbance velocity change sequence Vdiff through a sliding time window. Within the current sliding time window, the maximum absolute increase in velocity change value per unit time is identified. The standard deviation of the change amplitude is calculated by statistically analyzing the entire disturbance velocity change sequence Vdiff. The maximum absolute increase in velocity change value per unit time is added to the standard deviation to obtain the disturbance abrupt change gradient Dg, which is used to characterize the severity of water flow disturbance.

[0083] The disturbance fluctuation period Tp is obtained by performing spectral analysis on the disturbance frequency distribution function Ff to extract the peak amplitude position corresponding to its dominant frequency; then, the corresponding angular frequency is derived from the dominant frequency, where the angular frequency is calculated by multiplying 2π by the dominant frequency; and finally, the disturbance period is calculated using the reciprocal of the angular frequency to obtain the disturbance fluctuation period Tp, which is used to quantify the stable rhythm of the repeated disturbances.

[0084] The flow direction variability index Rrev is obtained by using the time series of the direction reflection offset vector PhΔ as input parameters. Within a set observation period T, the first derivative of the direction reflection offset vector PhΔ on a continuous time slice is calculated using the numerical difference method to obtain the phase offset rate change sequence. The absolute value of the phase offset rate change sequence at each time point is taken, and the absolute value sequence within the entire sliding time window is numerically integrated to obtain the cumulative amount of the reverse trend fluctuation within the sliding time window. The cumulative amount of the reverse trend fluctuation is divided by the duration of the period T to obtain the average reverse offset degree per unit time. The result obtained is the flow direction variability index Rrev, which characterizes whether the water flow has the behavior of unstable and repeated changes in direction.

[0085] Furthermore, based on the minimum and maximum values ​​of each parameter in the historical sample data, the three parameters of the perturbation feature set are subjected to min-max normalization to eliminate the influence of physical dimensions and form a standardized perturbation dataset, which is used as the input feature set in the subsequent model recognition process.

[0086] In this embodiment, the method establishes a first disturbance data acquisition point and a second disturbance data acquisition point in the inlet and outlet water channel structure of the smart water meter, respectively, to acquire initial disturbance information and disturbance behavior results. The two acquisition points integrate a high-sensitivity electromagnetic flow velocity sensor, a piezoelectric water hammer sensor, an array-type ultrasonic reflection sensor, and a disturbance frequency capture module, enabling multi-dimensional real-time acquisition of disturbance data. The original disturbance data frame is then cached in the edge control module, ensuring the integrity and timeliness of the original data structure. Subsequently, the data is packaged and uploaded to a remote IoT platform using an NB-IoT wireless communication module, including timestamps and parameter value groups, achieving highly reliable, low-latency data link communication. After receiving the disturbance data, the platform further performs disturbance feature extraction processing to obtain a disturbance feature set. Among them, the disturbance mutation gradient Dg is modeled by summing the absolute maximum value and standard deviation of the flow velocity change, reflecting the intensity of the disturbance; the disturbance fluctuation period Tp is based on the spectral peak analysis and angular frequency derivation of the disturbance frequency function Ff, quantifying the stability of the disturbance rhythm; and the flow direction variation index Rrev uses the direction reflection offset vector PhΔ as input, and calculates the unstable change trend in the flow direction by combining the first derivative and integral methods, which has clear physical meaning and behavioral interpretability. After the above three disturbance parameters are extracted, a standardized disturbance dataset with cross-user universality and high robustness is formed based on the minimum-maximum normalization strategy of historical samples, providing comparable and controllable input feature support for subsequent identification modeling and risk calculation. Example

[0087] Please see Figure 1 Specifically: S2 includes S21;

[0088] S21. Construct a disturbance consistency score calculation model in the Internet of Things platform, and input the standardized disturbance dataset acquired in real time into the constructed disturbance consistency score calculation model to calculate and output the disturbance consistency score index Ipd, which measures the non-natural disturbance characteristics of the local water flow state.

[0089] The perturbation consistency score index Ipd is calculated and output using the following perturbation consistency score calculation model;

[0090] Within the set observation time window, the first derivative of all water flow rate data in the disturbance velocity change sequence Vdiff is calculated to obtain the velocity change rate sequence. The velocity change rate sequence is then multiplied point-to-point with the sine function corresponding to the disturbance main frequency signal to form the disturbance derivative modulation sequence.

[0091] The integral value of the perturbation modulation energy is obtained by performing a square integration operation on the perturbation derivative modulation sequence;

[0092] Based on the integral value of the disturbance modulation energy, a disturbance intensity suppression factor, which is formed by adding the absolute value of the disturbance abrupt gradient Dg to a constant 1, is introduced as a denominator. The ratio of the disturbance intensity suppression factor is multiplied by the reciprocal of the natural logarithm of the disturbance fluctuation period Tp to obtain the disturbance consistency score index Ipd, which is used to comprehensively evaluate the stability, periodic repeatability and directional consistency of the water flow in the current time interval.

[0093] ;

[0094] In the formula, T represents the length of the observation time window, V(t) represents the water flow velocity at time t, which is extracted from the perturbation velocity change sequence Vdiff, ln represents the natural logarithm function, sin represents the sine function, w represents the dominant frequency, and dt represents the time integral function.

[0095] The derivation logic and physical meaning of the formula: Represents the integral of the perturbation modulation energy, where, It represents the rate of change of water flow velocity within the observation time window. It is the first derivative of the perturbation velocity change sequence Vdiff and reflects the trend of velocity change. The modulation factor of the analog signal represents the periodic disturbance mode under the influence of the dominant frequency w in the current system. Multiplying these two quantities, squaring them, and integrating them over the time interval is equivalent to calculating the energy intensity under the modulation of the interference frequency during that time period. In a physical sense, the higher this energy integral is, the stronger the disturbance response of the water flow at that frequency, which may represent some kind of unstable event. This represents the disturbance intensity suppression factor. When the disturbance amplitude itself is already very severe, the amplitude of the integral energy result is suppressed to prevent accidental drastic changes from causing misjudgment of the scoring result. Add 1 as a benchmark term to avoid the denominator being zero. This represents the periodic instability factor. It uses the natural logarithm function for compression mapping to avoid amplifying the disturbance fluctuation period Tp value, which would lead to an excessively low score. The physical meaning is: the more irregular the period, that is, the larger the disturbance fluctuation period Tp, the greater the unnatural possibility of the disturbance, and the lower the final score.

[0096] Formula Derivation Background and Technical Source: This formula integrates the following three theories, including background signal energy spectrum analysis theory, normalized stability modeling method, and hydrodynamic physics inspired model;

[0097] Among them, the background signal energy spectrum analysis theory treats the perturbation derivative sequence as a quasi-continuous signal and constructs a modulation energy window through a frequency modulation function. It is a transform domain energy modeling method and is widely used in sound signal recognition and mechanical vibration fault detection.

[0098] Normalized stability modeling methods achieve dimensionless elimination and engineering robustness by nonlinearly compressing the disturbance amplitude and period parameters, such as by using logarithmic functions and amplitude suppression functions, to control the influence range of different disturbance sources on the model.

[0099] Hydrodynamic physics-inspired models combine the response characteristics of water flow under unsteady conditions, such as small water hammer, pulsating back pressure, and changes in the backflow field. They use derivatives and frequency modulation functions to simulate the system response behavior, making them a simulation method with physical interpretability.

[0100] S22. The calculated disturbance consistency score index Ipd is compared and evaluated with the set disturbance threshold Ith to determine the consistency of the current water flow disturbance. The disturbance threshold Ith is an empirical dynamic statistical threshold. Based on a large number of historical disturbance consistency score indices Ipd, a score lower bound with a confidence level of 95% is selected from the historical normal disturbance behavior disturbance consistency score index Ipd samples as the initial threshold benchmark. It is periodically corrected according to environmental parameters such as season, water pressure, and user group structure to form a dynamically updated score judgment threshold, so as to ensure that the system has stable recognition performance in multiple environments and scenarios.

[0101] When the disturbance consistency score index Ipd > the disturbance threshold Ith, the current water flow disturbance behavior cycle is consistent and the direction is stable. The water flow is identified as being in a natural disturbance state, and only the behavior log is recorded. Routine monitoring continues.

[0102] When the perturbation consistency score index Ipd ≤ perturbation threshold Ith, it is determined that there is a risk of perturbation consistency loss in the current water flow behavior, and the behavior backtracking process is triggered.

[0103] In this embodiment, the method constructs a calculation model for the disturbance consistency score index Ipd and introduces a dynamic disturbance threshold Ith for preliminary comparative evaluation, thereby achieving quantitative identification of unnatural disturbance characteristics in local water flow states. Specifically, the method first establishes a disturbance consistency score calculation model on the IoT platform side, using a standardized disturbance dataset acquired in real time as input. Based on the coupling relationship between the disturbance velocity change sequence Vdiff and the dominant frequency signal, a disturbance consistency score index Ipd is constructed through composite modeling steps such as derivative modulation, energy integration, disturbance intensity suppression, and periodic instability compression. This index characterizes the changing trend and periodic matching degree of the water flow state within the observation time interval. This score index comprehensively considers the magnitude of the disturbance energy, the concentration of the frequency response, and the intensity of the disturbance itself, possessing clear physical interpretability and sensitive response characteristics. After calculating the disturbance consistency score index Ipd, it is further compared and evaluated with a preset disturbance threshold Ith. The disturbance threshold Ith is an empirical dynamic statistical threshold, which is periodically adjusted based on the confidence lower bound of the historical disturbance consistency score index Ipd, and environmental variables such as season, water pressure, and user group structure to ensure the dynamic adaptability and stability of the threshold. The implementation of the above steps achieves a closed-loop design for the entire process from raw disturbance data to initial assessment of disturbance consistency risk, significantly improving the sensitivity and accuracy of identifying micro-features of water flow disturbances. Example

[0104] Please see Figure 1 Specifically: S3 includes S31;

[0105] S31. After the initial comparison and evaluation of the triggered behavior backtracking process, the historical normal cycle behavior map is extracted from the historical behavior map database in the Internet of Things platform. The historical normal cycle behavior map is constructed by monitoring the water use behavior samples of multiple users at different times, under different water pressures and different living habits. Then, the sample data in the historical normal cycle behavior map is normalized to form a cycle behavior feature template with a unified structure.

[0106] Within the current behavior cycle, continuous operating data of the water meter is collected to construct the current cycle behavior feature group Bnow. The current cycle behavior feature group Bnow includes instantaneous water usage duration, maximum peak flow velocity, flow velocity change rate slope, intraday water usage frequency, water usage interval distribution, and water usage rhythm cycle.

[0107] The system retrieves periodic behavior feature templates that match the current user category, apartment type, water meter model, and typical water usage patterns from the historical normal periodic behavior map. It then extracts the normal periodic behavior feature group Bref as the comparison periodic behavior feature template for the current periodic behavior features, which is used for subsequent similarity scoring model calculations.

[0108] S32. Based on the obtained values ​​of each corresponding feature dimension in the current cycle behavior feature group Bnow and the normal cycle behavior feature group Bref, perform item-by-item difference processing to obtain the absolute value sequence of behavior deviation. Then sum the absolute value sequence of behavior deviation to obtain the total value of behavior deviation. Divide the total value of behavior deviation by the sum of the cumulative reference values ​​of all feature dimensions in the normal cycle behavior feature group Bref to obtain the variation map similarity score Simv, which measures the degree of deviation between the current water use behavior and the standard behavior map in the overall behavior feature space.

[0109] The variant map similarity score Simv is calculated and output using the following algorithm formula; ;

[0110] In the formula, N represents the total number of behavioral features in the historical normal cycle behavior map, and Bnow i Bref represents the i-th behavioral feature vector in the normal cyclical behavioral feature group Bnow. i This represents the reference value of the i-th behavior feature vector in the normal periodic behavior feature group Bref, and E represents the denominator correction constant to prevent calculation errors caused by a denominator of 0. This represents the total behavioral deviation value, used to measure the degree of deviation between the current behavior and the template; This represents the sum of cumulative reference values, used to eliminate the direct impact of the absolute value of feature dimensions on the scoring results, achieve normalization, make the scoring range more comparable, and is suitable for comparing behaviors between different users.

[0111] In this embodiment, the method introduces a behavior backtracking process to perform graph-level feature comparison and similarity calculation on user water usage behavior in the current period, thereby achieving quantitative identification of the degree of deviation from abnormal behavior and verification of behavioral background. Specifically, after the perturbation consistency score index Ipd triggers a preliminary risk assessment, historical normal cycle behavior graphs, which have undergone long-term construction and normalization, are first extracted from the historical behavior graph database in the IoT platform to form standardized cycle behavior feature templates. Next, based on the continuous operating data collected by the water meter within the current time period, multi-dimensional behavioral features, including instantaneous water usage duration, maximum peak flow velocity, flow velocity change rate slope, intraday water usage frequency, water usage interval distribution, and water usage rhythm cycle, are extracted to construct the current cycle behavior feature group Bnow, which is then compared item by item with the historical normal cycle behavior feature group Bref. Subsequently, by calculating the absolute value sequence of behavioral deviations and their normalized distance in the feature space, a variation graph similarity score Simv is output, reflecting the overall degree of deviation between the current user behavior and the standard behavior graph. This scoring mechanism not only considers the absolute numerical differences between behavioral characteristics, but also introduces a denominator correction constant E to avoid division by zero errors caused by extreme values ​​or missing values, thus ensuring the stability and comparability of the scoring results. Example

[0112] Please see Figure 1 Specifically: S4 includes S41;

[0113] S41. Based on the perturbation consistency score index Ipd and the mutation map similarity score Simv, as well as the cumulative number of abnormal behaviors Tact in the current detection period, calculate the backtracking risk level score Rtrace to quantitatively analyze the current user's abnormal water use behavior.

[0114] The retrospective risk level score Rtrace is calculated and output using the following algorithm formula; ;

[0115] In the formula, ln represents the natural logarithm function. This represents the behavioral deviation control index, used to enhance the sensitivity of the variation map similarity score Simv, making deviant behaviors more weighted in the score. It is set by the user and has a dimensionless value. This represents the score indicating the coupling between disturbance and behavior. The lower the disturbance score, the more abnormal the hydraulic system. The greater the behavioral deviation, When the value is greater than 1, the influence of the deviation is amplified, demonstrating high sensitivity; this item as a whole reflects the severity of the coupling between "hydraulic instability and behavioral anomaly product", and the superposition effect of the two is significant; The cumulative number of abnormal behaviors (Tact) within the detection period is the cumulative number of times this type of abnormal behavior is triggered within the current time window. The natural logarithm function is used to compress the numerical growth to avoid excessive fluctuations that could affect the score. Physical meaning: the more frequently an anomaly occurs, the higher the risk, which is a trend warning signal.

[0116] By multiplying the disturbance and behavior score by the behavior trend index, a risk level score is finally generated. The higher the backtracking risk level score Rtrace, the more severe the behavioral abnormality, which can be used for classification and triggering control strategies.

[0117] In this embodiment, the method introduces a comprehensive risk quantification model into the IoT platform, constructing a retrospective risk level scoring mechanism Rtrace based on the fusion of disturbance and behavior indicators. This enables quantitative analysis and level assessment of the risk intensity of abnormal user water use behavior. Specifically, the method receives three key input parameters: the disturbance consistency score index Ipd, the variation map similarity score Simv, and the cumulative number of abnormal behaviors Tact within the current detection period. These are then input into the designed retrospective risk level scoring function to generate a unique corresponding retrospective risk level score Rtrace. This scoring function incorporates a behavior deviation control index to adjust the sensitivity of the variation map similarity score Simv to the final scoring result, enhancing its responsiveness to deviation behaviors. Simultaneously, a disturbance and behavior coupled scoring term is constructed, causing the disturbance anomaly degree to multiply and link with the behavior deviation, reflecting the dual risk characteristics of hydraulic instability and behavioral anomalies. Finally, the natural logarithm function is used to nonlinearly compress the cumulative number of abnormal behaviors Tact, avoiding frequent triggers that could exaggerate the overall score, thus balancing trend signals and assessment stability. Example

[0118] Please see Figure 1 Specifically: S5 includes S51 and S52;

[0119] S51. Extract the historical backtracking risk level score Rtrace, and perform frequency distribution analysis on all historical backtracking risk level scores Rtrace. Set the critical value of the historical backtracking risk level score Rtrace of normal users and the historical backtracking risk level score Rtrace of abnormal users as the risk interval threshold.

[0120] The risk interval thresholds include the first risk interval threshold F1 and the second risk interval threshold F2;

[0121] Among them, the first risk interval threshold F1 is the critical value of the historical backtracking risk level score Rtrace for normal users, and the second risk threshold F2 is the critical value of the historical backtracking risk level score Rtrace for abnormal users.

[0122] S52. Based on the real-time acquired retrospective risk level score Rtrace and the risk interval threshold, a secondary comparison and evaluation is performed to determine the risk level of the current user's water usage behavior, and corresponding differentiated response control is implemented based on the risk level. The specific evaluation content is as follows.

[0123] If the risk level score Rtrace is less than the first risk interval threshold F1, it is determined to be a level 1 risk behavior, the behavior log is recorded and normal operation is maintained.

[0124] If the first risk interval threshold F1 is between the retrospective risk level score Rtrace and the second risk interval threshold F2, it is determined to be a level 2 risk behavior, and the behavior trend tracking mechanism is activated, extending the data collection cycle by 50% and recording the behavior vector changes in real time.

[0125] If the risk level score Rtrace is greater than or equal to the second risk interval threshold F2, it is determined to be a level 3 risk behavior. The highest level response scheme in the behavior trigger strategy library is invoked to generate a set of processing commands including water meter control instructions, abnormal alarm instructions and platform write-back instructions.

[0126] The processing command set is executed after being parsed in the smart water meter controller;

[0127] Among them: the water meter control command locks abnormal users, prohibits the reporting and settlement of user water use records, and cuts off the water supply to the current user;

[0128] Anomaly alarm commands send warning information to the administrator's port via a remote IoT platform;

[0129] The platform write-back command is used to update the status of the user behavior graph and synchronize abnormal behavior information to the audit database for administrator audit review.

[0130] In this embodiment, the method performs frequency distribution analysis on historical backtracking risk level scores Rtrace to statistically determine the score thresholds for normal and abnormal users, thereby constructing a multi-segment risk zoning model including a first risk interval threshold F1 and a second risk interval threshold F2. This model dynamically sets risk thresholds through an experience-driven approach, effectively improving the model's environmental adaptability and compatibility with user behavior differences. A secondary comparison and evaluation is performed based on the real-time acquired backtracking risk level scores Rtrace and the aforementioned first and second risk interval thresholds F1 and F2 to intelligently classify the current user's risk level. Based on the evaluation results, a three-level response strategy is implemented. Through the above implementation, not only is accurate identification of the risk level of user water usage behavior achieved, but a complete closed-loop control mechanism encompassing identification, classification, response, and recording is also established. The main objective of this method is to achieve intelligent dynamic management and real-time decision-making intervention for water meter operation risks. The beneficial effects include: improved timeliness and accuracy of abnormal behavior responses, enhanced adaptability of the IoT platform to user behavior in complex scenarios, significantly reduced missed detection and false alarm rates, and ensured the safety and regulatory capacity of the urban water supply system. Example

[0131] Please see Figure 2 and Figure 3A smart water meter operation monitoring system based on the Internet of Things includes a disturbance data acquisition module, a disturbance consistency analysis module, a similarity analysis module, a comprehensive analysis module, and a risk level response control module.

[0132] The disturbance data acquisition module collects disturbance data in real time by setting up collection points at the smart water meter and transmits the disturbance data to the Internet of Things (IoT) platform. The IoT platform performs feature extraction and preprocessing on the disturbance data to obtain a standardized disturbance dataset.

[0133] The perturbation consistency analysis module calculates the perturbation consistency score index Ipd based on a standardized perturbation dataset and presets a perturbation threshold Ith. It then performs a preliminary comparison and evaluation between the perturbation consistency score index Ipd and the perturbation threshold Ith.

[0134] The similarity analysis module triggers the behavior backtracking process based on the preliminary comparison and evaluation results, extracts the historical normal cycle behavior map, analyzes the current cycle behavior feature group Bnow, and calls the normal cycle behavior feature group Bref in the normal cycle behavior map for comparison, and outputs the mutation map similarity score Simv.

[0135] The comprehensive analysis module calculates the perturbation consistency score index Ipd and the variation map similarity score Simv together, and outputs the backtracking risk scoring function Rtrace.

[0136] The risk level response control module determines the risk level of the current user's water usage behavior by pre-setting a risk range threshold and then comparing the risk range threshold with the retrospective risk level score Rtrace. The module then executes corresponding response controls based on the risk level.

[0137] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention.

Claims

1. A method for monitoring the operation of a smart water meter based on the Internet of Things, characterized in that: Includes the following steps: S1. Set up a collection point at the smart water meter to collect disturbance data in real time and transmit the disturbance data to the Internet of Things (IoT) platform. In the IoT platform, perform feature extraction and preprocessing on the disturbance data to obtain a standardized disturbance dataset. S1 also includes S13; S13. After receiving the uploaded data packet on the Internet of Things platform, the disturbance data is unpacked and extracted and the features are extracted to obtain the disturbance feature set. The disturbance feature set includes the disturbance mutation gradient Dg, the disturbance fluctuation period Tp and the flow direction variation index Rrev. The disturbance abrupt change gradient Dg analyzes the disturbance velocity change sequence Vdiff through a sliding time window. Within the current sliding time window, it identifies the maximum absolute increase in the velocity change value per unit time. Then, it performs statistical analysis on the entire disturbance velocity change sequence Vdiff and calculates the standard deviation of the change amplitude. Finally, it adds the maximum absolute increase in the velocity change value per unit time to the standard deviation to obtain the disturbance abrupt change gradient Dg. The disturbance fluctuation period Tp is obtained by performing spectral analysis on the disturbance frequency distribution function Ff to extract the peak amplitude position corresponding to its dominant frequency; then, the corresponding angular frequency is derived based on the dominant frequency, where the angular frequency is calculated by multiplying 2π by the dominant frequency; and finally, the disturbance period Tp is obtained by calculating the corresponding disturbance period using the reciprocal of the angular frequency. The flow direction variation index Rrev is obtained by using the time series of the direction reflection offset vector PhΔ as input parameters. Within a set observation period T, the first derivative of the direction reflection offset vector PhΔ on a continuous time slice is calculated using the numerical difference method to obtain the phase offset rate change sequence. The absolute value of the value at each time point in the phase offset rate change sequence is taken, and the absolute value sequence within the entire sliding time window is numerically integrated to obtain the cumulative amount of the reverse trend fluctuation within the sliding time window. The cumulative amount of the reverse trend fluctuation is divided by the duration of the period T to obtain the average reverse offset degree per unit time. The result obtained is the flow direction variation index Rrev. Furthermore, based on the minimum and maximum values ​​of each parameter in the historical sample data, the three parameters of the disturbance feature set are subjected to min-max normalization to eliminate the influence of physical dimensions and form a standardized disturbance dataset. S2. Calculate the perturbation consistency score index Ipd based on the standardized perturbation dataset, and preset the perturbation threshold Ith. Perform a preliminary comparison and evaluation between the perturbation consistency score index Ipd and the perturbation threshold Ith. S2 includes S21; S21. Construct a disturbance consistency score calculation model in the Internet of Things platform, and input the standardized disturbance dataset acquired in real time into the constructed disturbance consistency score calculation model to calculate and output the disturbance consistency score index Ipd, which measures the non-natural disturbance characteristics of the local water flow state. The perturbation consistency score index Ipd is calculated and output using the following perturbation consistency score calculation model; Within the set observation time window, the first derivative of all water flow rate data in the disturbance velocity change sequence Vdiff is calculated to obtain the velocity change rate sequence. The velocity change rate sequence is then multiplied point-to-point with the sine function corresponding to the disturbance main frequency signal to form the disturbance derivative modulation sequence. Perform a square integration operation on the perturbation derivative modulation sequence to obtain the perturbation modulation energy integral value; Based on the integral value of the disturbance modulation energy, a disturbance intensity suppression factor, which is formed by adding the absolute value of the disturbance abrupt gradient Dg to a constant 1, is introduced as a denominator. The ratio of the disturbance intensity suppression factor is multiplied by the reciprocal of the natural logarithm of the disturbance fluctuation period Tp to obtain the disturbance consistency score index Ipd. S3. Based on the preliminary comparison and evaluation results, trigger the behavior backtracking process, extract the historical normal cycle behavior map, analyze the current cycle behavior feature group Bnow, and call the normal cycle behavior feature group Bref in the normal cycle behavior map for comparison, and output the mutation map similarity score Simv. S4. Combine the perturbation consistency score index Ipd with the mutation map similarity score Simv to calculate the backtracking risk score function Rtrace. S5. Preset risk range threshold, and then compare and evaluate the risk range threshold with the backtracking risk level score Rtrace to determine the risk level of the current user's water use behavior. The risk level will trigger the corresponding response control.

2. The method for monitoring the operation of a smart water meter based on the Internet of Things according to claim 1, characterized in that: S1 includes S11 and S12; S11. Two data collection points are set in the water flow channel structure of the smart water meter, including a first disturbance data collection point and a second disturbance data collection point; The first disturbance data acquisition point is located in the central area of ​​the water inlet guide cavity of the water meter body, and is used to collect the initial disturbance information of the water flow; The second disturbance data acquisition point is located downstream of the outlet pressure stabilizing chamber and near the outlet valve, and is used to collect data on the water flow behavior after disturbance. Each collection point is equipped with a high-sensitivity electromagnetic flow velocity sensor, a piezoelectric water hammer sensor, an array of ultrasonic reflection sensors, and a disturbance frequency capture module to collect disturbance data in real time. The disturbance data includes the disturbance flow velocity change sequence Vdiff, the direction reflection offset vector PhΔ, and the disturbance frequency distribution function Ff. The disturbance data is then combined into a disturbance raw data frame and cached in the edge data control module of the smart water meter. S12. The original disturbance data frame is packaged and transmitted through the wireless communication module configured inside the smart water meter. The wireless communication module uses the NB-IoT protocol to send the upload data message containing timestamp, data type identifier and disturbance parameter value group to the remote IoT platform.

3. The method for monitoring the operation of a smart water meter based on the Internet of Things according to claim 1, characterized in that: S2 further includes S22; S22. The calculated disturbance consistency score index Ipd is compared and evaluated with the set disturbance threshold Ith to determine the consistency of the current water flow disturbance. The disturbance threshold Ith is an empirical dynamic statistical threshold. Based on a large number of historical disturbance consistency score indices Ipd, a score lower bound with a confidence level of 95% is selected from the disturbance consistency score index Ipd samples of historical normal disturbance behavior as the initial threshold benchmark. When the disturbance consistency score index Ipd > the disturbance threshold Ith, the current water flow disturbance behavior is consistent in cycle and stable in direction. Only the behavior log is recorded and routine monitoring continues. When the perturbation consistency score index Ipd ≤ perturbation threshold Ith, it is determined that there is a risk of perturbation consistency loss in the current water flow behavior, and the behavior backtracking process is triggered.

4. The method for monitoring the operation of a smart water meter based on the Internet of Things according to claim 1, characterized in that: S3 includes S31; S31. After the initial comparison and evaluation of the triggered behavior backtracking process, the historical normal cycle behavior map is extracted from the historical behavior map database in the Internet of Things platform. The historical normal cycle behavior map is constructed by monitoring the water use behavior samples of multiple users at different times, under different water pressures and different living habits. Then, the sample data in the historical normal cycle behavior map is normalized to form a cycle behavior feature template with a unified structure. Within the current behavior cycle, continuous operating data of the water meter is collected to construct the current cycle behavior feature group Bnow. The current cycle behavior feature group Bnow includes instantaneous water usage duration, maximum peak flow velocity, flow velocity change rate slope, intraday water usage frequency, water usage interval distribution, and water usage rhythm cycle. From the historical normal cycle behavior map, retrieve the cycle behavior feature template that matches the current user category, house type, water meter model and typical water use pattern, and extract the normal cycle behavior feature group Bref.

5. The method for monitoring the operation of a smart water meter based on the Internet of Things according to claim 4, characterized in that: S3 further includes S32; S32. Based on the obtained values ​​of each corresponding feature dimension in the current cycle behavior feature group Bnow and the normal cycle behavior feature group Bref, perform item-by-item difference processing to obtain the absolute value sequence of behavior deviation. Summate the absolute value sequence of behavior deviation to obtain the total value of behavior deviation. Divide the total value of behavior deviation by the sum of the cumulative reference values ​​of all feature dimensions in the normal cycle behavior feature group Bref to obtain the variation map similarity score Simv, which measures the degree of deviation of the current water use behavior from the standard behavior map in the overall behavior feature space.

6. The method for monitoring the operation of a smart water meter based on the Internet of Things according to claim 1, characterized in that: S4 includes S41; S41. Based on the perturbation consistency score index Ipd, the mutation map similarity score Simv, and the cumulative number of abnormal behaviors Tact within the current detection period, calculate the backtracking risk level score Rtrace to quantitatively analyze the current user's abnormal water usage behavior.

7. The method for monitoring the operation of a smart water meter based on the Internet of Things according to claim 6, characterized in that: S5 includes S51 and S52; S51. Extract the historical backtracking risk level score Rtrace, and perform frequency distribution analysis on all historical backtracking risk level scores Rtrace. Set the critical value of the historical backtracking risk level score Rtrace of normal users and the historical backtracking risk level score Rtrace of abnormal users as the risk interval threshold. The risk interval thresholds include a first risk interval threshold F1 and a second risk interval threshold F2; Among them, the first risk interval threshold F1 is the critical value of the historical backtracking risk level score Rtrace for normal users, and the second risk threshold F2 is the critical value of the historical backtracking risk level score Rtrace for abnormal users. S52. Based on the real-time acquired retrospective risk level score Rtrace and the risk interval threshold, a secondary comparison and evaluation is performed to determine the risk level of the current user's water usage behavior, and corresponding differentiated response control is implemented based on the risk level. The specific evaluation content is as follows. If the risk level score Rtrace is less than the first risk interval threshold F1, it is determined to be a level 1 risk behavior, the behavior log is recorded and normal operation is maintained. If the first risk interval threshold F1 is between the retrospective risk level score Rtrace and the second risk interval threshold F2, it is determined to be a level 2 risk behavior, and the behavior trend tracking mechanism is activated, extending the data collection cycle by 50% and recording the behavior vector changes in real time. If the risk level score Rtrace is greater than or equal to the second risk interval threshold F2, it is determined to be a level 3 risk behavior. The highest level response scheme in the behavior trigger strategy library is invoked to generate a set of processing commands including water meter control instructions, abnormal alarm instructions and platform write-back instructions. The set of processing commands is executed after being parsed in the smart water meter controller; Among them: the water meter control command locks abnormal users, prohibits the reporting and settlement of user water use records, and cuts off the water supply to the current user; Anomaly alarm commands send warning information to the administrator's port via a remote IoT platform; The platform write-back command is used to update the status of the user behavior graph and synchronize abnormal behavior information to the audit database for administrator audit review.

8. An IoT-based smart water meter operation monitoring system, applied to the IoT-based smart water meter operation monitoring method described in any one of claims 1-7, characterized in that: It includes a disturbance data acquisition module, a disturbance consistency analysis module, a similarity analysis module, a comprehensive analysis module, and a risk level response control module; The disturbance data acquisition module collects disturbance data in real time by setting up collection points at the smart water meter and transmits the disturbance data to the Internet of Things (IoT) platform. The IoT platform performs feature extraction and preprocessing on the disturbance data to obtain a standardized disturbance dataset. After receiving the uploaded data packet, the IoT platform unpacks and extracts the disturbance data and performs feature extraction to obtain a disturbance feature set, which includes the disturbance mutation gradient Dg, the disturbance fluctuation period Tp, and the flow direction variation index Rrev. The disturbance abrupt change gradient Dg analyzes the disturbance velocity change sequence Vdiff through a sliding time window. Within the current sliding time window, it identifies the maximum absolute increase in the velocity change value per unit time. Then, it performs statistical analysis on the entire disturbance velocity change sequence Vdiff and calculates the standard deviation of the change amplitude. Finally, it adds the maximum absolute increase in the velocity change value per unit time to the standard deviation to obtain the disturbance abrupt change gradient Dg. The disturbance fluctuation period Tp is obtained by performing spectral analysis on the disturbance frequency distribution function Ff to extract the peak amplitude position corresponding to its dominant frequency; then, the corresponding angular frequency is derived based on the dominant frequency, where the angular frequency is calculated by multiplying 2π by the dominant frequency; and finally, the disturbance period Tp is obtained by calculating the corresponding disturbance period using the reciprocal of the angular frequency. The flow direction variation index Rrev is obtained by using the time series of the direction reflection offset vector PhΔ as input parameters. Within a set observation period T, the first derivative of the direction reflection offset vector PhΔ on a continuous time slice is calculated using the numerical difference method to obtain the phase offset rate change sequence. The absolute value of the value at each time point in the phase offset rate change sequence is taken, and the absolute value sequence within the entire sliding time window is numerically integrated to obtain the cumulative amount of the reverse trend fluctuation within the sliding time window. The cumulative amount of the reverse trend fluctuation is divided by the duration of the period T to obtain the average reverse offset degree per unit time. The result obtained is the flow direction variation index Rrev. Furthermore, based on the minimum and maximum values ​​of each parameter in the historical sample data, the three parameters of the disturbance feature set are subjected to min-max normalization to eliminate the influence of physical dimensions and form a standardized disturbance dataset. The disturbance consistency analysis module calculates the disturbance consistency score index Ipd based on a standardized disturbance dataset and presets a disturbance threshold Ith, and then performs a preliminary comparison and evaluation between the disturbance consistency score index Ipd and the disturbance threshold Ith. A disturbance consistency score calculation model is constructed in the Internet of Things platform, and the standardized disturbance dataset acquired in real time is input into the constructed disturbance consistency score calculation model to calculate and output the disturbance consistency score index Ipd, which measures the non-natural disturbance characteristics of local water flow state. The perturbation consistency score index Ipd is calculated and output using the following perturbation consistency score calculation model; Within the set observation time window, the first derivative of all water flow rate data in the disturbance velocity change sequence Vdiff is calculated to obtain the velocity change rate sequence. The velocity change rate sequence is then multiplied point-to-point with the sine function corresponding to the disturbance main frequency signal to form the disturbance derivative modulation sequence. Perform a square integration operation on the perturbation derivative modulation sequence to obtain the perturbation modulation energy integral value; Based on the integral value of the disturbance modulation energy, a disturbance intensity suppression factor, which is formed by adding the absolute value of the disturbance abrupt gradient Dg to a constant 1, is introduced as a denominator. The ratio of the disturbance intensity suppression factor is multiplied by the reciprocal of the natural logarithm of the disturbance fluctuation period Tp to obtain the disturbance consistency score index Ipd. The similarity analysis module triggers a behavior backtracking process based on the preliminary comparison and evaluation results, extracts historical normal cycle behavior maps, analyzes the current cycle behavior feature group Bnow, and calls the normal cycle behavior feature group Bref in the normal cycle behavior map for comparison, outputting a variant map similarity score Simv. The comprehensive analysis module calculates the perturbation consistency score index Ipd and the mutation map similarity score Simv together and outputs the backtracking risk score function Rtrace. The risk level response control module determines the risk level of the current user's water usage behavior by pre-setting a risk range threshold and then comparing the risk range threshold with the retrospective risk level score Rtrace. The module then executes corresponding response controls based on the risk level.

Images