Intelligent water affair dynamic monitoring system and monitoring method based on digital twinning

Abstract

The invention relates to the technical field of intelligent water affairs, in particular to an intelligent water affairs dynamic monitoring system and method based on digital twinning, and aims to detect data drift in real time, calculate a standard deviation change rate to generate abnormal judgment, analyze model stability and drift rate synchronism to output deviation indexes, screen data abnormal items to form quality evaluation, and realize dynamic monitoring of the intelligent water affairs. And comparing node synchronization difference to trigger compensation correction, and generating a monitoring adjustment result through integrity verification. According to the invention, through comparison of the multi-cycle standard deviation change rate and the environmental parameters, sensor drift classification and confidence evaluation are realized, the reliability of the model is enhanced by constructing a composite attenuation coefficient, abnormal sampling is positioned by combining the data change rate and an outlier map, and data synchronization is guaranteed by adopting a difference matrix and state feedback. And version verification and repair records are introduced to realize closed-loop tracing.

Description

Technical Field

[0001] The present invention relates to the field of smart water technology, and in particular to a smart water dynamic monitoring system and monitoring method based on digital twins. Background Art

[0002] The field of smart water technology encompasses urban water supply network operation monitoring, real-time water quality analysis, water resource scheduling and management, and abnormal event warnings. This technology, based on IoT sensing devices, collects network operating parameters through terminal devices such as pressure sensors, flow meters, and water quality testers. This technology integrates with geographic information systems to build hydraulic models, and utilizes data communication networks to enable remote transmission and centralized processing of monitoring information, forming a complete technical architecture from the data acquisition layer to the platform application layer.

[0003] Among them, the smart water dynamic monitoring system based on digital twins refers to the use of three-dimensional modeling technology to build a virtual simulation model that is completely corresponding to the physical pipeline network, using edge computing nodes to timestamp the pump station pressure data, pipeline flow rate data, and water quality pH value data at the millisecond level, and using data assimilation technology to couple the real-time monitoring data of the SCADA system with the hydraulic calculation model. The time series database is used to realize the two-way iterative update of the monitoring data stream and the simulation model parameters, and the WebGL engine is used to realize the multi-dimensional visualization of the pipeline network operation status.

[0004] Traditional centralized data processing architectures lead to delayed data synchronization at edge nodes, resulting in decision-making lags in scenarios where pipe network pressure changes suddenly. Existing digital twin systems lack a dynamic correction mechanism for sensor drift, and model parameters are affected by cumulative errors during long-term operation. For example, unidentified pH drift can cause water quality prediction deviations. Static threshold verification mechanisms struggle to adapt to nonlinear changes in pipe network parameters, and drastic fluctuations in flow data during sudden leaks can easily trigger misjudgments. Fixed compensation thresholds fail to account for temporal and spatial differences in environmental parameters, making local data desynchronization more likely in complex pipe network areas. Model stability assessment relies on a single fluctuation indicator, unrelated to the data drift rate, making it difficult to distinguish between model attenuation and anomalies caused by external interference. Summary of the Invention

[0005] The purpose of the present invention is to solve the shortcomings of the existing technology and propose a smart water dynamic monitoring system and method based on digital twins.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a smart water affairs dynamic monitoring system based on digital twins, the system comprising:

[0007] The sensor drift monitoring module acquires real-time sampling data and model output data, detects data drift, calculates the standard deviation change rate of multiple sampling periods, determines data fluctuations based on water environment characteristics, and generates drift anomaly determination results;

[0008] A model stability analysis module, based on the drift anomaly determination result, analyzes the stability of the model output variables, determines whether there is a synchronous change in combination with the drift rate of the sampled data, and generates a model stability deviation result;

[0009] The data quality monitoring module analyzes the fluctuation trend of water service sampling data based on the stability deviation results of the model, determines whether it exceeds the quality range, uses the volatility and standard deviation methods to filter out abnormal data, and generates data quality fluctuation assessment results;

[0010] The distributed data synchronization module obtains the data quality fluctuation assessment results, compares the data synchronization of the edge computing nodes, calculates the synchronization difference, and triggers the data compensation mechanism if the difference exceeds the threshold, performs correction, and generates a data synchronization consistency result;

[0011] The edge computing verification module verifies the integrity of the edge computing node data based on the data synchronization consistency results, checks and compares whether the data meets the standards, and generates water monitoring data adjustment results.

[0012] As a further solution of the present invention, the drift anomaly judgment result is specifically an abnormal threshold interval, a drift direction identifier, and a confidence rating; the model stability deviation result includes fluctuation tolerance, output deviation, and attenuation coefficient; the data quality fluctuation assessment result includes an abnormal sampling point number, a compliance rate index, and an outlier distribution diagram; the data synchronization consistency result is specifically a difference quantization matrix, a compensation magnitude record, and a node synchronization status code; the water monitoring data adjustment result includes a verification pass identifier, a data correction value, and a node data version number.

[0013] As a further solution of the present invention, the sensor drift monitoring module includes a data synchronization acquisition submodule, a dynamic index calculation submodule, and a drift threshold determination submodule;

[0014] The data synchronization acquisition submodule obtains the real-time sampling data stream and the model output data stream, detects the numerical differences of the conductivity, flow, and pressure parameters in the two sets of data streams, extracts the data points whose difference values ​​exceed the allowable deviation of the industry standard within the continuous sampling period, and generates a drift characteristic signal;

[0015] The dynamic index calculation submodule calls the drift characteristic signal, calculates the standard deviation of the temperature and turbidity parameters in adjacent sampling windows, obtains the relative change percentage of the standard deviation between windows, and generates the window standard deviation change rate;

[0016] The drift threshold determination submodule compares the window standard deviation change rate with the dynamic threshold interval based on the turbidity reference range and pH reference range in the water environment characteristic parameter library. When the continuous window change rate exceeds the threshold, a drift abnormality determination result is generated.

[0017] As a further solution of the present invention, the model stability analysis module includes a variable fluctuation quantification submodule, a drift synchronicity assessment submodule, and a stability deviation generation submodule;

[0018] The variable fluctuation quantification submodule extracts the continuous sampling values ​​of the temperature, pressure, and flow variables in the model output based on the drift anomaly judgment result, calculates the absolute value of the standard deviation of each variable within the set time window, and generates the variable fluctuation amplitude;

[0019] The drift synchronization evaluation submodule calls the variable fluctuation amplitude and the sampling data drift rate, using the formula:

[0020]

[0021] The deviation between the variable fluctuation and the drift rate is obtained by calculation, and the deviation is compared with the synchronization threshold to generate the synchronization deviation coefficient;

[0022] Among them, S sync Represents the degree of synchronization deviation, ΔV i Represents the fluctuation range of the i-th variable, ΔD j represents the drift rate of the jth sampling parameter, n is the total number of model variables, and m is the number of sampling parameters;

[0023] The stability deviation generation submodule calculates the sum of squares of the deviations of each variable and the reference value based on the synchronization deviation coefficient when the coefficient exceeds the stability judgment benchmark to generate a model stability deviation result.

[0024] As a further solution of the present invention, the data quality monitoring module includes a trend baseline construction submodule, an abnormal fluctuation detection submodule, and a quality assessment generation submodule;

[0025] The trend baseline construction submodule obtains the continuous monitoring values ​​of temperature, turbidity and pH value in the water service sampling data based on the stability deviation results of the model, calculates the moving average of each parameter under standard working conditions, and generates a dynamic trend baseline;

[0026] The abnormal fluctuation detection submodule calls the dynamic trend baseline and uses the formula:

[0027]

[0028] The calculation obtains the comprehensive fluctuation amount of each parameter deviating from the baseline in the current sampling window, compares the fluctuation amount with the quality fluctuation threshold, and generates the abnormal fluctuation index;

[0029] Among them, Q anm represents the abnormal fluctuation, x k is the real-time value of the kth parameter, μ k is the dynamic trend baseline mean of the corresponding parameter, σ k is the standard deviation of real-time data, is the baseline standard deviation, p is the number of monitored parameters;

[0030] The quality assessment generation submodule, based on the abnormal fluctuation index, screens the sampling points whose index exceeds the upper limit of the quality interval, calculates the proportion and duration of the exceeding points, and generates the data quality fluctuation assessment results.

[0031] As a further solution of the present invention, the distributed data synchronization module includes a difference quantization submodule, a compensation triggering submodule, and a synchronization verification submodule;

[0032] The difference quantification submodule obtains the data quality fluctuation assessment results, extracts the synchronized timestamp data of the temperature, pressure, and flow parameters between edge nodes, calculates the absolute value of the standard deviation of the corresponding parameter values ​​of adjacent nodes, and generates a synchronization difference coefficient;

[0033] The compensation trigger submodule calls the synchronization difference coefficient and compares each parameter coefficient with the industry standard synchronization threshold. When the number of parameters exceeding the threshold reaches a set ratio, the data compensation queue is activated and a compensation trigger instruction is generated.

[0034] The synchronization verification submodule performs retransmission and verification of the missing data segments based on the compensation trigger instruction, calculates the maximum relative error of the parameter values ​​between the nodes after compensation, and generates a data synchronization consistency result.

[0035] As a further solution of the present invention, the edge computing verification module includes an integrity check submodule, an anomaly location submodule, and a data repair submodule;

[0036] An integrity check submodule extracts hash check values ​​of the flow, pressure, and turbidity parameters in the edge node based on the data synchronization consistency result, calculates the matching percentage between the check value and the standard value, and generates an integrity check rate;

[0037] The abnormality positioning submodule calls the integrity check rate, locates the parameter type and node number whose check rate is lower than the industry standard threshold, calculates the distribution density of abnormal parameters, and generates an abnormal distribution map;

[0038] The data repair submodule performs incremental synchronization and redundancy check of abnormal node data according to the abnormal distribution map, calculates the hash matching degree of the repaired data, and generates the water monitoring data adjustment result.

[0039] The digital twin-based smart water affairs dynamic monitoring method is implemented based on the digital twin-based smart water affairs dynamic monitoring system, and includes the following steps:

[0040] S1: Real-time sampling data is acquired through pressure sensors and water quality detectors. A sliding window algorithm is used to calculate the standard deviation change rate of the data stream over multiple consecutive sampling periods. Based on the turbidity threshold and conductivity reference value in the water environment characteristic parameter library, the data fluctuation amplitude is compared with the preset range to generate a drift anomaly determination result.

[0041] S2: extracting the model output variables based on the drift anomaly determination result, and using the dynamic time warping algorithm to calculate the synchronization index between the variable fluctuation trajectory and the drift rate of the sampled data. When the synchronization index is lower than the set threshold, a model stability deviation result is generated;

[0042] S3: Calling the attenuation coefficient in the stability deviation result of the model, performing double screening based on the standard deviation of the change rate on the water service sampling data, when the data change rate exceeds the preset threshold and the standard deviation is higher than the historical mean multiple threshold, marking the abnormal item and generating the data quality fluctuation assessment result;

[0043] S4: Calculate the cosine similarity of the data sets between edge nodes based on the distribution of abnormal items in the data quality fluctuation assessment results. When the similarity is lower than a set threshold, trigger a data compensation operation based on the Newton interpolation method to generate a data synchronization consistency result including the compensation value.

[0044] S5: Based on the node status code in the data synchronization consistency result, perform CRC32 check and timestamp continuity check on the edge computing node. When the check failure rate exceeds the set threshold or the timestamp interval exceeds the set threshold, generate a water monitoring data adjustment result containing a repair operation record.

[0045] Compared with the prior art, the advantages and positive effects of the present invention are:

[0046] In the present invention, through multi-cycle standard deviation change rate analysis and dynamic comparison of environmental parameters, real-time classification and confidence assessment of sensor drift are achieved, thereby improving the spatiotemporal accuracy of anomaly identification. Model stability judgment integrates the correlation analysis of the output variable fluctuation amplitude and the data drift rate to construct a composite attenuation coefficient index, thereby enhancing the dimension of model reliability verification. A dynamic quality scoring matrix is ​​constructed based on the data change rate and standard deviation, and combined with the outlier distribution map, rapid positioning and quality grading of abnormal sampling points are achieved. Data synchronization between nodes adopts a difference quantization matrix and a status code feedback mechanism, combined with dynamic adjustment of compensation thresholds of water environment parameters, to improve the efficiency of distributed data consistency maintenance. Integrity verification introduces version iteration verification and repair value records to form a closed-loop traceability mechanism for data adjustment, ensuring the reproducibility of the monitoring data correction process. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 is a system flow chart of the present invention;

[0048] Figure 2 This is a flow chart for obtaining the sensor drift monitoring module of the present invention;

[0049] Figure 3 This is a flow chart for obtaining the model stability analysis module of the present invention;

[0050] Figure 4 This is a flowchart of the acquisition of the data quality monitoring module of the present invention;

[0051] Figure 5 This is an acquisition flow chart of the distributed data synchronization module of the present invention;

[0052] Figure 6 This is a flow chart for obtaining the edge computing verification module of the present invention. DETAILED DESCRIPTION

[0053] The technical solution of the present invention is described below in conjunction with the accompanying drawings.

[0054] In the embodiments of the present invention, words such as "exemplarily" and "for example" are used to indicate examples, illustrations, or explanations. Any embodiment or design described as an "exemplary" in the present invention should not be interpreted as being preferred or advantageous over other embodiments or designs. Rather, the use of the word "exemplary" is intended to present concepts in a concrete manner. Furthermore, in the embodiments of the present invention, "and / or" can mean both or either of the two.

[0055] In the embodiments of the present invention, the terms "image" and "picture" may be used interchangeably. It should be noted that, when the distinction between them is not emphasized, their intended meanings are the same. The terms "of," "corresponding," and "corresponding" may be used interchangeably. It should be noted that, when the distinction between them is not emphasized, their intended meanings are the same.

[0056] In the embodiments of the present invention, sometimes a subscript such as W1 may be written as a non-subscript such as W1. When the difference is not emphasized, the meanings to be expressed are the same.

[0057] In order to make the technical problems, technical solutions and advantages to be solved by the present invention clearer, a detailed description will be given below with reference to the accompanying drawings and specific embodiments.

[0058] See also Figure 1 The present invention provides a technical solution: a smart water affairs dynamic monitoring system based on digital twins, the system includes:

[0059] The sensor drift monitoring module acquires real-time sampling data and model output data, detects data drift through the drift identification unit, calculates the standard deviation change rate within multiple consecutive sampling periods, compares the water environment characteristic parameters with the comparison library, and performs drift judgment when the data fluctuation exceeds the set range, triggering the drift anomaly flag and generating the drift anomaly judgment result;

[0060] The model stability analysis module analyzes the stability of relevant variables in the model output based on the drift anomaly judgment results, calculates the fluctuation amplitude and standard deviation of the model output variables, and makes a stability judgment based on the drift rate of the sampled data. If the fluctuation amplitude of the model output variables does not change synchronously with the data drift rate, a model stability deviation result is generated;

[0061] The data quality monitoring module obtains water service sampling data based on the model stability deviation results, analyzes the fluctuation trend of the sampling data, determines whether the data changes exceed the preset quality range, and uses an assessment method based on data change rate and standard deviation to screen out data quality anomalies and generate data quality fluctuation assessment results;

[0062] The distributed data synchronization module obtains the data quality fluctuation assessment results, compares the data synchronization status of edge computing nodes based on the data deviation between nodes, calculates the synchronization difference, and triggers the data compensation mechanism if the difference exceeds the threshold. It performs data correction to ensure data consistency between nodes and generates data synchronization consistency results.

[0063] The edge computing verification module verifies the data of edge computing nodes based on the data synchronization consistency results, performs integrity checks and comparisons on node data, determines whether the data meets the preset integrity standards, performs repair operations when anomalies are found, and generates water monitoring data adjustment results.

[0064] The drift anomaly judgment results include the anomaly threshold range, drift direction identification, and confidence rating. The model stability deviation results include fluctuation tolerance, output deviation, and attenuation coefficient. The data quality fluctuation assessment results include the abnormal sampling point number, compliance rate index, and outlier distribution diagram. The data synchronization consistency results include the difference quantization matrix, compensation magnitude record, and node synchronization status code. The water monitoring data adjustment results include the verification pass mark, data correction value, and node data version number.

[0065] See also Figure 2 ,The sensor drift monitoring module includes a data synchronization acquisition submodule, a dynamic index calculation submodule, and a drift threshold determination submodule;

[0066] The data synchronization acquisition submodule obtains the real-time sampling data stream and the model output data stream, detects the numerical differences of the conductivity, flow, and pressure parameters in the two sets of data streams, extracts the data points whose difference values ​​exceed the allowable deviation of the industry standard within the continuous sampling period, and generates a drift characteristic signal;

[0067] Obtaining the real-time sampling value of the conductivity sensor from the SCADA system C sensor and the hydraulic model output value C model , establish a data cache queue, set T = 5s as the sampling period, when |C sensor (t)-C model (t)|>5μS / cm is recorded as a difference point. For example, in a water plant monitoring, when the sensor records a conductivity value of 520μS / cm, the model output is 500μS / cm, triggering a difference mark. When collecting flow parameters, the electromagnetic flowmeter measured value Q real and the model predicted value Q pred For comparison, when the instantaneous flow difference ΔQ=|Q real -Q pred |>0.2m 3 / s for 3 sampling cycles, and the pressure parameter is collected using the measured value P of the pressure transmitter. sensor and the digital twin pressure value P digital Perform sliding window averaging processing, and set the window width to N = 10 sampling points. When the average pressure difference in the window Generate a drift signature signal.

[0068] The dynamic index calculation submodule calls the drift characteristic signal, calculates the standard deviation of temperature and turbidity parameters in adjacent sampling windows, obtains the relative change percentage of the standard deviation between windows, and generates the window standard deviation change rate;

[0069] Extract temperature parameter sequence {T i |i=1,2,...,n}, set the sampling window length W=30 minutes, each window contains n=60 sampling points (sampling interval is 30 seconds), calculate the average temperature in the window For example, the temperature sequence measured in a certain period is [22.3, 22.1, 22.6, 23.0, 22.4, …, 22.9] ℃ (a total of 60 points), and the calculation is 22.7℃, and then calculate the squared deviation of each data point from the mean The sum is divided by n-1=59 to get the variance Standard deviation When calculating adjacent windows, the next window is measured Rate of change When processing turbidity parameters, take the turbidity value sequence of 6 consecutive windows Calculate the standard deviation for each window For example, the turbidity values ​​in window 1 [2.1, 2.3, 2.0, 2.4, 2.2] are NTU. Window 2 turbidity value [2.5, 2.8, 2.6, 2.7, 2.9] NTU Rate of change According to the preset weight w T :w NTU =7:3 weighted synthesis, R total =59.46%×0.7+(-6.25%)×0.3=40.07%.

[0070] The drift threshold determination submodule compares the window standard deviation change rate with the dynamic threshold interval based on the turbidity reference range and pH reference range in the water environment characteristic parameter library. When the continuous window change rate exceeds the threshold, a drift anomaly determination result is generated;

[0071] Read the current water turbidity reference range NTU from the environmental parameter library range =[0,5] and PH reference range PH range =[6.5,8.5], when the real-time pH value is current =7.2, calculate the turbidity threshold adjustment coefficient k ntu =1+α(PH current -PH base ), where the base pH value is base=7.0, adjustment coefficient α = 0.2 (determined based on historical data regression analysis), substituting into k ntu =1+0.2×(7.2-7.0)=1.04, turbidity change rate threshold R ntu_th =30%×1.04=31.2%, when the turbidity change rate R of a certain window is detected NTU =35%, compared with the threshold value 35%>31.2%, the pH change rate threshold is fixed at R ph_th =15%, when the pH change rates of three consecutive windows are When each window satisfies The dynamic threshold interval of the comprehensive change rate is set to R lower ,R upper ]=[20%,50%],when R is detected total =55%, the execution interval comparison 55%>50%, if the consecutive occurrence count ≥ 2 (preset condition), for example, two consecutive windows R total =51% and (55%), triggering the drift abnormality mark. For the turbidity parameter, when the change rates of 5 consecutive windows are 32%, 33%, 34%, 35%, and 36% respectively, the change rate of each window exceeds the 31.2% threshold, and the cumulative number of exceeding the limit reaches 5 times (exceeding the preset threshold of 3 times), generating a drift abnormality judgment result.

[0072] See also Figure 3 ,The model stability analysis module includes the variable fluctuation quantification submodule, the drift synchronization evaluation submodule, and the ,stability deviation generation submodule;

[0073] The variable fluctuation quantification submodule extracts the continuous sampling values ​​of temperature, pressure, and flow variables in the model output based on the drift anomaly judgment results, calculates the absolute value of the standard deviation of each variable within the set time window, and generates the variable fluctuation amplitude;

[0074] Retrieve the temperature output value of the last 30 minutes from the model database {T k |k=1,2,...,60}, the sampling interval is set to 30 seconds, and the window mean is calculated In the specific implementation, the temperature sequence [22.3, 22.1, 22.6, 23.0, 22.4, …, 22.9] ℃ (a total of 60 points) is measured at a water supply network node, and the sum is obtained as ∑T k =1362℃, average Calculate the squared deviation point by point (22.3-22.7) 2 =0.16, (22.1-22.7) 2 =0.36, the total sum of square deviations of 60 points is 8.214, and the standard deviation When processing the pressure parameters, take the pressure output sequence [0.81, 0.83, 0.80, ..., 0.82] MPa (a total of 60 points) and calculate The sum of squared deviations is 0.15, and the standard deviation When processing traffic parameters, the sequence [2.88, 2.91, 2.85, …, 2.93] m 3 / s calculated to get σ Q =0.18m 3 / s, store the three standard deviation values ​​[0.37, 0.05, 0.18] into the fluctuation range queue.

[0075] Table 1: Statistical characteristics of model variables

[0076]

[0077]

[0078] As shown in Table 1, the statistical characteristics of each variable within the evaluation window are fully displayed, among which the standard deviation data will be used as the benchmark value of the fluctuation range.

[0079] The drift synchronization evaluation submodule uses the variable fluctuation amplitude and the sampling data drift rate, and adopts the formula:

[0080]

[0081] The deviation between the variable fluctuation and the drift rate is obtained by calculation, and the deviation is compared with the synchronization threshold to generate the synchronization deviation coefficient;

[0082] Among them, S sync Represents the degree of synchronization deviation, ΔV i Represents the fluctuation range of the i-th variable, ΔD j represents the drift rate of the jth sampling parameter, n is the total number of model variables, and m is the number of sampling parameters;

[0083] Obtain the temperature sensor drift rate ΔD1 = 0.12 °C / min, the pressure sensor drift rate ΔD2 = 0.09 MPa / min, and the model variable fluctuation amplitude ΔV = [0.37, 0.05, 0.18]. When calculating the numerator, first obtain the drift rate mean. Calculate the absolute deviation item by item: temperature item |0.37-0.105|=0.265, pressure item |0.05-0.105|=0.055, flow item |0.18-0.105|=0.075, sum up to 0.265+0.055+0.075=0.395. When calculating the denominator, first calculate the sum of squares of the model fluctuation 0.37 2 +0.05 2 +0.182 =0.1369+0.0025+0.0324=0.1718, the sum of squares of drift rates is 0.12 2 +0.09 2 =0.0144+0.0081=0.0225, the square root of the product is Final synchronization deviation S sync =0.395 / 0.0623≈6.34, synchronization threshold S th =5.0 is based on the statistical analysis of 200 sets of historical normal data. The maximum deviation under normal working conditions is 4.8. Adding a 4% safety margin, we get 4.8×1.04=5.0. Calculate the synchronization deviation coefficient C sync =6.34 / 5.0=1.268;

[0084] This result indicates that the overall deviation between the model variable fluctuations and the sensor drift rate exceeds the safety threshold of 26.8%, triggering a synchronization anomaly flag. Specifically, the deviation between the temperature fluctuation amplitude of 0.37°C and the drift rate of 0.12°C / min contributes the most (0.265%), accounting for 67% of the total deviation, indicating that temperature parameters are the primary instability factor.

[0085] The stability deviation generation submodule calculates the sum of squares of the deviations of each variable and the benchmark value based on the synchronization deviation coefficient. When the coefficient exceeds the stability judgment benchmark, it generates the model stability deviation result.

[0086] When C sync =1.268>1.0, call the reference value of each variable: Temperature fluctuation reference V T0 =0.3℃, pressure reference V P0 =0.04MPa, flow rate reference V Q0 =0.15m 3 / s, calculate the single item deviation: temperature item d T =0.37-0.3=0.07, pressure term d P =0.05-0.04=0.01, flow term d Q =0.18-0.15=0.03, square sum calculation E=0.07 2 +0.01 2 +0.03 2 =0.0049+0.0001+0.0009=0.0059, deviation threshold E th=0.005 is determined based on the regression analysis of 50 groups of fault cases (the maximum E value under normal conditions is 0.0048, and the minimum E value under fault conditions is 0.0052). Since 0.0059>0.005, the temperature and flow variables are marked as abnormal items. A stability deviation result is generated, including the timestamp "2025-05-26 14:30:00", the abnormal parameter list ["temperature", "flow"], and the deviation value [0.07, 0.03]. This result will be written to the quality monitoring database for subsequent processing.

[0087] See also Figure 4 ,The data quality monitoring module includes a trend baseline construction submodule, an ,abnormal fluctuation detection submodule, and a quality assessment generation submodule;

[0088] The trend baseline construction submodule obtains the continuous monitoring values ​​of temperature, turbidity, and pH value in the water service sampling data based on the model stability deviation results, calculates the moving average of each parameter under standard working conditions, and generates a dynamic trend baseline;

[0089] Retrieve the temperature monitoring value of the water supply network node N-2037 in the last 72 hours from the historical database {T t |t=1,...,864} (sampling interval is 5 minutes), using the sliding window algorithm, the window width is set to 12 sampling points (1 hour), and the hourly moving average is calculated. For example, when n = 14:00, take 12 data points from 13:05 to 14:00 [22.3, 22.1, 22.6, 23.0, 22.4, 22.7, 22.5, 22.9, 22.8, 22.6, 22.7, 22.9]℃, and sum them to get ∑ = 270.9℃, the mean When processing turbidity parameters, take the pH value sequence {PH m |m=1,...,288} (15-minute intervals), the window width is set to 4 points (1 hour), calculate For example, the data at m = 14:00 is [7.2, 7.3, 7.1, 7.0], and the mean is (7.2 + 7.3 + 7.1 + 7.0) / 4 = 7.15. The baseline data is updated every 5 minutes.

[0090] Table 2: Dynamic trend baseline calculation example

[0091] Timestamp Real-time temperature value (℃) Window data range Moving average (℃) 14:00 23.5 13:05-14:00 (12 o'clock) 22.58 14:05 23.6 13:10-14:05 (12 o'clock) 22.64 14:10 23.8 13:15-14:10 (12 o'clock) 22.71

[0092] As shown in Table 2 , the baseline value gradually increases over time, reflecting the natural increase in water temperature in the afternoon and providing a dynamic reference for anomaly detection.

[0093] The abnormal fluctuation detection submodule calls the dynamic trend baseline and uses the formula:

[0094]

[0095] The calculation obtains the comprehensive fluctuation amount of each parameter deviating from the baseline in the current sampling window, compares the fluctuation amount with the quality fluctuation threshold, and generates the abnormal fluctuation index;

[0096] Among them, Q anm represents the abnormal fluctuation, x k is the real-time value of the kth parameter, μ k is the dynamic trend baseline mean of the corresponding parameter, σ k is the standard deviation of real-time data, is the baseline standard deviation, p is the number of monitored parameters;

[0097] Get the data of the current sampling window (14:00-14:30): temperature value x1 = 23.5℃, corresponding baseline mean μ1 = 22.58℃, and calculate the temperature standard deviation σ1 = 0.5℃ (6 sampling points [23.3, 23.5, 23.7, 23.6, 23.4, 23.5]) in real time. (Historical data statistics for the same period), turbidity value x2 = 3.2NTU, μ2 = 2.4NTU, σ2 = 0.4NTU, PH value x3=7.8, μ3=7.15, σ3=0.3, Calculate the square root of the first term: (23.5-22.58) 2 =0.92 2 =0.8464, (3.2-2.4) 2 =0.8 2 =0.64, (7.8-7.15) 2 =0.65 2 =0.4225, the sum is 0.8464+0.64+0.4225=1.9089, the square root is The second calculation: |0.5-0.3|=0.2, |0.4-0.2|=0.2, |0.3-0.1|=0.2, the average is (0.2+0.2+0.2) / 3=0.2, the final Q anm =1.38+0.2=1.58, compared with the quality fluctuation threshold Q th =1.5 (based on the 95th percentile of 1.48 plus 1.3% margin of 300 sets of normal data). The result shows that the comprehensive fluctuation of the current window exceeds the threshold by 5.3%, of which the temperature deviation contributes the most (0.8464 / 1.9089≈44.3%), and the standard deviation change contributes 12.6%. According to the preset rules, when Q anmWhen the value >1.5 lasts for 2 consecutive windows, a level 3 alarm is triggered. Currently, it has lasted for 1 window, and the system will activate the preliminary warning mechanism.

[0098] The quality assessment generation submodule, based on the abnormal fluctuation index, screens sampling points whose index exceeds the upper limit of the quality interval, calculates the proportion and duration of the exceeding points, and generates data quality fluctuation assessment results;

[0099] Continuously monitor the Q of 6 sampling windows (14:00-14:30) anm The values ​​are [1.58, 1.62, 1.67, 1.73, 1.69, 1.75]. Each window is 5 minutes. The number of exceeded points is counted as 6 (threshold 1.5). The exceeded percentage is calculated as 6 / 6 = 100%. The duration is 6 × 5 = 30 minutes. According to the quality assessment rule: when the percentage is ≥ 80% and the duration is ≥ 25 minutes, a serious abnormality report is generated. The maximum deviation value of each parameter is extracted: temperature ΔT max =23.7-22.71=0.99℃, turbidity ΔNTU max =3.5-2.5=1.0NTU, PHΔPH max =7.8-7.2=0.6, generating a data quality fluctuation assessment result including the abnormality level "severe", the influencing parameters ["temperature", "turbidity", "pH"], and the recommended measures ["calibrate the sensor", "check the dosing system"]. The result is transmitted to the central control room via the OPC UA protocol.

[0100] Table 3: Quality Assessment Rating Matrix

[0101]

[0102]

[0103] As shown in Table 3, the current abnormal state matches the red warning condition. The system will perform the preset node isolation operation and send the detailed data packet to the operation and maintenance terminal.

[0104] See also Figure 5 ,The distributed data synchronization module includes a difference quantization submodule, a compensation trigger submodule, and a synchronization verification submodule;

[0105] The difference quantification submodule obtains the data quality fluctuation assessment results, extracts the synchronized timestamp data of temperature, pressure, and flow parameters between edge nodes, calculates the absolute value of the standard deviation of the corresponding parameter values ​​of adjacent nodes, and generates the synchronization difference coefficient;

[0106] Obtain the 2025-05-26T14:00:00Z timestamp data and temperature parameter T from the OPC UA server of edge nodes E-12 (coordinates X:235, Y:178) and E-15 (coordinates X:240, Y:182).E12 =22.5℃(Serial No. SN:TX2025-0526-001) and T E15 =22.8℃(SN:TX2025-0526-002), pressure parameter P E12 =0.82MPa (calibration certificate number CL2025-0456) and P E15 =0.79MPa(CL2025-0457), flow parameter Q E12 =2.9m 3 / s (Instrument ID: FM-8892) and Q E15 =3.1m 3 / s(FM-8893), calculate the absolute value of the standard deviation: temperature term |σ T |=|22.5-22.8|=0.3℃,pressure term|σ P |=|0.82-0.79|=0.03MPa, flow term |σ Q |=|2.9-3.1|=0.2m 3 / s, generating a difference coefficient group [0.3, 0.03, 0.2]. This result shows that the temperature parameter has reached the critical value of the industry standard threshold and requires special attention. For example, in the synchronization of turbidity data between pipe network nodes N-2037 and N-2041, NTU 2037 =2.1NTU (test time 14:00:05) and NTU 2041 =2.5NTU (detection time 14:00:03) calculated |σ NTU |=0.4NTU, exceeding the 0.3NTU threshold by 0.1NTU.

[0107] Table 4: Details of edge node data differences

[0108] parameter Node E-12 data Node E-15 data Collection time difference Absolute difference temperature 22.5℃±0.1 22.8℃±0.1 2 seconds 0.3℃ pressure 0.82MPa±0.01 0.79MPa±0.01 3 seconds 0.03MPa flow <![CDATA[2.9m 3 / s±0.05]]> <![CDATA[3.1m 3 / s±0.05]]> 1 second <![CDATA[0.2m 3 / s]]>

[0109] As shown in Table 4, the temperature parameter difference value accurately reaches the 0.3°C threshold boundary specified by the industry standard GB / T 19837-2025, triggering a yellow warning state.

[0110] The compensation trigger submodule calls the synchronization difference coefficient and compares each parameter coefficient with the industry standard synchronization threshold. When the number of parameters exceeding the threshold reaches a set ratio, the data compensation queue is activated and a compensation trigger instruction is generated.

[0111] Call the synchronization threshold specified in the water industry standard Q / HSW 203-2025: temperature Th T =0.3℃ (based on the verification value of the water supply network thermodynamic model), pressure Th P =0.05MPa (refer to GB 50013-2025 Water Supply Design Code), flow rate ThQ =0.25m 3 / s (calculated based on the flow velocity upper limit for pipes with a DN300 diameter). The error coefficients are compared to the group [0.3, 0.03, 0.2]. Each parameter is judged item by item: 0.3 ≥ 0.3 for temperature, 0.03 < 0.05 for pressure, and 0.2 < 0.25 for flow. One parameter is counted as exceeding the standard, and the excess ratio is calculated as 1 / 3 (≈ 33.3%). This is compared to the set threshold of 30% (set according to the ISO 55000 asset management standard). When 33.3% ≥ 30%, the data compensation queue containing the temperature parameter is activated, generating the instruction code "COMP-TEMP-202505261400." For example, during the monitoring period of 2025-05-26T14:05:00Z, the pH parameter difference is 0.4 (threshold 0.3) and the turbidity difference is 0.25 (threshold 0.3). Two parameters exceed the standard, and the excess ratio is 2 / 3 (≈ 66.7%), triggering an orange alert and activating the dual-parameter compensation queue.

[0112] The synchronization verification submodule, based on the compensation trigger instruction, performs retransmission and verification of the missing data segments, calculates the maximum relative error of the parameter values ​​between nodes after compensation, and generates the data synchronization consistency result;

[0113] When retransmitting temperature data, the original data packet from 14:00:00 to 14:05:00 (checksum SHA-256: a1b2c3d4) is extracted from the local cache of node E-12. This data packet contains six sampling points: [22.5, 22.6, 22.7, 22.5, 22.6, 22.5]°C (timestamp sequence: [20250526T140000Z, …, 20250526T140500Z]). ​​After receiving the data, node E-15 calculates the relative error: Take the maximum value max(E rel)=0.44%, compared with the permissible error threshold of 1.0% (according to IEC 60870-5-101 protocol settings), generating a "Verification Passed" verification result. For example, during pressure parameter compensation, node E-12 transmits [0.82, 0.81, 0.83] MPa (time window 14:00:00-14:02:30). Node E-15 receives the value and calculates: |0.81-0.82| / 0.82≈1.22%. This exceeds the threshold, triggering a "Verification Failed" status code (error code ERR-P-045). The system automatically initiates the secondary compensation protocol and re-requests E-12 to send the original ADC sample value (16-bit precision, serial number SAM-202505261400P). After three compensations, the error is still 1.15%. Finally, a synchronization consistency report (report number SYNC-202505261405) is generated, including the node IDs [E12, E15], the abnormal parameter [pressure], and the maximum error of 1.15%.

[0114] See also Figure 6 ,The edge computing verification module includes an integrity verification submodule, an anomaly location submodule, and a data repair submodule;

[0115] The integrity check submodule extracts the hash check values ​​of the flow, pressure, and turbidity parameters in the edge node based on the data synchronization consistency results, calculates the matching percentage between the check value and the standard value, and generates the integrity check rate;

[0116] The OPC UA server of node E-12 (ID: NODE_2025_12) obtains the SHA-256 hash value "a1b2c3d4" of the traffic parameter in the time window (14:00-14:05) on 2025-05-26T14:00:00Z. This is compared to the cloud-based standard value "a1b2c3d4" and the calculated match is 100%. The pressure parameter node hash "e5f6g7h8" matches the standard value "e5f6g7h9" in the first 32 characters, resulting in a total match length of 28 / 64 = 4. 3.75%. The turbidity parameter hash "i9j0k1l2" completely matches the standard value, generating a verification rate group of [100%, 43.75%, 100%]. For example, the traffic hash "m3n4o5p6" of node E-15 in the same time window matches the standard value "m3n4o5p7" by 56 / 64 bytes, with a matching degree of 87.5%. When the verification rate is lower than 90%, it is marked as abnormal. The pressure parameter hash matching rate of 82.3% for this node triggers an alert.

[0117] The anomaly location submodule calls the integrity check rate, locates the parameter type and node number whose check rate is lower than the industry standard threshold, calculates the distribution density of abnormal parameters, and generates an anomaly distribution map;

[0118] An industry standard threshold of 90% (based on the ISO / IEC 27001 information security standard) was set. The verification rate group [100%, 43.75%, 100%] was checked. A pressure parameter of 43.75% < 90% was marked as abnormal. Twenty nodes in the node cluster were counted for abnormalities: 12 nodes had abnormal pressure parameters, and 8 nodes had abnormal flow rates. The pressure parameter abnormality density was calculated as 12 / 20 = 60%, and the flow rate abnormality density was calculated as 8 / 20 = 40%. A two-dimensional coordinate map was generated with the node number [E-01 to E-20] on the X-axis and the parameter type [1: flow, 2: pressure, 3: turbidity] on the Y-axis. Outliers were marked as E-12 (2, 43.75%) and E-15 (1, 87.5%). For example, nodes E-08 to E-11 in region G-07 all showed abnormal pressure parameters, forming a clustered hotspot.

[0119] The data repair submodule performs incremental synchronization and redundancy verification of abnormal node data based on the abnormal distribution map, calculates the hash matching degree of the repaired data, and generates the water monitoring data adjustment results;

[0120] For the pressure parameter of node E-12, redundant data packets from 14:00:00 to 14:05:00 are obtained from neighboring nodes E-11 and E-13. Incremental synchronization is performed: E-11 transmits [0.82, 0.81, 0.83, 0.82, 0.82] MPa, and E-13 transmits [0.83, 0.82, 0.83, 0.84, 0.82] MPa. The median value sequence [0.82, 0.82, 0.83, 0.83, 0.82] is taken, and the new hash value "f7g8h9i0" is calculated. The match with the standard value "e5f6g7h9" is improved to 58 / 64 = 90.6%. During the secondary redundancy check, the same operating condition data for the six hours before 14:00:00 is retrieved from the historical database, and the weighted average is used to generate the repair value [0.821, 0.819, 0.825, 0.822, 0.818] MPa. The final hash match is 62 / 64 = 96.9%. The water service data adjustment result is generated, including the repair value sequence, new hash value, and a final verification rate of 96.9%, and is written to the OPC UA historical database of edge node E-12.

[0121] The digital twin-based smart water dynamic monitoring method includes the following steps:

[0122] S1: Real-time sampling data is acquired through pressure sensors and water quality detectors. A sliding window algorithm is used to calculate the standard deviation change rate of the data stream over multiple consecutive sampling periods. Based on the turbidity threshold and conductivity reference value in the water environment characteristic parameter library, the data fluctuation amplitude is compared with the preset range to generate a drift anomaly determination result.

[0123] S2: Extract model output variables based on the drift anomaly judgment results, and use the dynamic time warping algorithm to calculate the synchronization index between the variable fluctuation trajectory and the drift rate of the sampled data. When the synchronization index is lower than the set threshold, the model stability deviation result is generated;

[0124] S3: Call the attenuation coefficient in the model stability deviation result to perform double screening of the water service sampling data based on the change rate and standard deviation. When the data change rate exceeds the preset threshold and the standard deviation is higher than the historical mean multiple threshold, the abnormal item is marked and the data quality fluctuation assessment result is generated;

[0125] S4: Based on the distribution of abnormal items in the data quality fluctuation assessment results, the cosine similarity of the data sets between edge nodes is calculated. When the similarity is lower than the set threshold, a data compensation operation based on the Newton interpolation method is triggered to generate a data synchronization consistency result including the compensation value;

[0126] S5: Based on the node status code in the data synchronization consistency result, perform CRC32 check and timestamp continuity check on the edge computing node. When the check failure rate exceeds the set threshold or the timestamp interval exceeds the set threshold, generate a water monitoring data adjustment result containing the repair operation record.

[0127] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any modifications or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in the present invention should be included in the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be based on the scope of protection of the claims.

Claims

1. The digital twin-based smart water dynamic monitoring system is characterized by: The system comprises: The sensor drift monitoring module acquires real-time sampling data and model output data, detects data drift, calculates the standard deviation change rate of multiple sampling periods, determines data fluctuations based on water environment characteristics, and generates drift anomaly determination results; A model stability analysis module, based on the drift anomaly determination result, analyzes the stability of the model output variables, determines whether there is a synchronous change in combination with the drift rate of the sampled data, and generates a model stability deviation result; The data quality monitoring module analyzes the fluctuation trend of water service sampling data based on the stability deviation results of the model, determines whether it exceeds the quality range, uses the volatility and standard deviation methods to filter out abnormal data, and generates data quality fluctuation assessment results; The distributed data synchronization module obtains the data quality fluctuation assessment results, compares the data synchronization of the edge computing nodes, calculates the synchronization difference, and triggers the data compensation mechanism if the difference exceeds the threshold, performs correction, and generates a data synchronization consistency result; The edge computing verification module verifies the integrity of the edge computing node data based on the data synchronization consistency results, checks and compares whether the data meets the standards, and generates water monitoring data adjustment results.

2. The digital twin-based smart water affairs dynamic monitoring system according to claim 1 is characterized by: The drift anomaly judgment result specifically includes the anomaly threshold range, drift direction identifier, and confidence rating; the model stability deviation result includes fluctuation tolerance, output deviation, and attenuation coefficient; the data quality fluctuation assessment result includes the abnormal sampling point number, compliance rate index, and outlier distribution diagram; the data synchronization consistency result specifically includes the difference quantization matrix, compensation magnitude record, and node synchronization status code; the water monitoring data adjustment result includes the verification pass identifier, data correction value, and node data version number.

3. The digital twin-based smart water affairs dynamic monitoring system according to claim 1 is characterized by: The sensor drift monitoring module includes a data synchronization acquisition submodule, a dynamic index calculation submodule, and a drift threshold determination submodule; The data synchronization acquisition submodule obtains the real-time sampling data stream and the model output data stream, detects the numerical differences of the conductivity, flow, and pressure parameters in the two sets of data streams, extracts the data points whose difference values ​​exceed the allowable deviation of the industry standard within the continuous sampling period, and generates a drift characteristic signal; The dynamic index calculation submodule calls the drift characteristic signal, calculates the standard deviation of the temperature and turbidity parameters in adjacent sampling windows, obtains the relative change percentage of the standard deviation between windows, and generates the window standard deviation change rate; The drift threshold determination submodule compares the window standard deviation change rate with the dynamic threshold interval based on the turbidity reference range and pH reference range in the water environment characteristic parameter library. When the continuous window change rate exceeds the threshold, a drift abnormality determination result is generated.

4. The digital twin-based smart water affairs dynamic monitoring system according to claim 1 is characterized by: The model stability analysis module includes a variable fluctuation quantification submodule, a drift synchronization assessment submodule, and a stability deviation generation submodule; The variable fluctuation quantification submodule extracts the continuous sampling values ​​of the temperature, pressure, and flow variables in the model output based on the drift anomaly judgment result, calculates the absolute value of the standard deviation of each variable within the set time window, and generates the variable fluctuation amplitude; The drift synchronization evaluation submodule calls the variable fluctuation amplitude and the sampling data drift rate, using the formula: The deviation between the variable fluctuation and the drift rate is obtained by calculation, and the deviation is compared with the synchronization threshold to generate the synchronization deviation coefficient; Among them, S sync Represents the degree of synchronization deviation, ΔV i Represents the fluctuation range of the i-th variable, ΔD j represents the drift rate of the jth sampling parameter, n is the total number of model variables, and m is the number of sampling parameters; The stability deviation generation submodule calculates the sum of squares of the deviations of each variable and the reference value based on the synchronization deviation coefficient when the coefficient exceeds the stability judgment benchmark to generate a model stability deviation result.

5. The digital twin-based smart water affairs dynamic monitoring system according to claim 1 is characterized by: The data quality monitoring module includes a trend baseline construction submodule, an abnormal fluctuation detection submodule, and a quality assessment generation submodule; The trend baseline construction submodule obtains the continuous monitoring values ​​of temperature, turbidity and pH value in the water service sampling data based on the stability deviation results of the model, calculates the moving average of each parameter under standard working conditions, and generates a dynamic trend baseline; The abnormal fluctuation detection submodule calls the dynamic trend baseline and uses the formula: The calculation obtains the comprehensive fluctuation amount of each parameter deviating from the baseline in the current sampling window, compares the fluctuation amount with the quality fluctuation threshold, and generates the abnormal fluctuation index; Among them, Q anm represents the abnormal fluctuation, x k is the real-time value of the kth parameter, μ k is the dynamic trend baseline mean of the corresponding parameter, σ k is the standard deviation of real-time data, is the baseline standard deviation, p is the number of monitored parameters; The quality assessment generation submodule, based on the abnormal fluctuation index, screens the sampling points whose index exceeds the upper limit of the quality interval, calculates the proportion and duration of the exceeding points, and generates the data quality fluctuation assessment results.

6. The digital twin-based smart water affairs dynamic monitoring system according to claim 1 is characterized by: The distributed data synchronization module includes a difference quantization submodule, a compensation trigger submodule, and a synchronization verification submodule; The difference quantification submodule obtains the data quality fluctuation assessment results, extracts the synchronized timestamp data of the temperature, pressure, and flow parameters between edge nodes, calculates the absolute value of the standard deviation of the corresponding parameter values ​​of adjacent nodes, and generates a synchronization difference coefficient; The compensation trigger submodule calls the synchronization difference coefficient and compares each parameter coefficient with the industry standard synchronization threshold. When the number of parameters exceeding the threshold reaches a set ratio, the data compensation queue is activated and a compensation trigger instruction is generated. The synchronization verification submodule performs retransmission and verification of the missing data segments based on the compensation trigger instruction, calculates the maximum relative error of the parameter values ​​between the nodes after compensation, and generates a data synchronization consistency result.

7. The digital twin-based smart water affairs dynamic monitoring system according to claim 1 is characterized by: The edge computing verification module includes an integrity check submodule, an abnormality positioning submodule, and a data repair submodule; An integrity check submodule extracts hash check values ​​of the flow, pressure, and turbidity parameters in the edge node based on the data synchronization consistency result, calculates the matching percentage between the check value and the standard value, and generates an integrity check rate; The abnormality positioning submodule calls the integrity check rate, locates the parameter type and node number whose check rate is lower than the industry standard threshold, calculates the distribution density of abnormal parameters, and generates an abnormal distribution map; The data repair submodule performs incremental synchronization and redundancy check of abnormal node data according to the abnormal distribution map, calculates the hash matching degree of the repaired data, and generates the water monitoring data adjustment result.

8. The digital twin-based smart water affairs dynamic monitoring method is characterized by: The method is used in the digital twin-based smart water affairs dynamic monitoring system according to any one of claims 1 to 7, comprising the following steps: S1: Real-time sampling data is acquired through pressure sensors and water quality detectors. A sliding window algorithm is used to calculate the standard deviation change rate of the data stream over multiple consecutive sampling periods. Based on the turbidity threshold and conductivity reference value in the water environment characteristic parameter library, the data fluctuation amplitude is compared with the preset range to generate a drift anomaly determination result. S2: extracting the model output variables based on the drift anomaly determination result, and using the dynamic time warping algorithm to calculate the synchronization index between the variable fluctuation trajectory and the drift rate of the sampled data. When the synchronization index is lower than the set threshold, a model stability deviation result is generated; S3: Calling the attenuation coefficient in the stability deviation result of the model, performing double screening based on the standard deviation of the change rate on the water service sampling data, when the data change rate exceeds the preset threshold and the standard deviation is higher than the historical mean multiple threshold, marking the abnormal item and generating the data quality fluctuation assessment result; S4: Calculate the cosine similarity of the data sets between edge nodes based on the distribution of abnormal items in the data quality fluctuation assessment results. When the similarity is lower than a set threshold, trigger a data compensation operation based on the Newton interpolation method to generate a data synchronization consistency result including the compensation value. S5: Based on the node status code in the data synchronization consistency result, perform CRC32 check and timestamp continuity check on the edge computing node. When the check failure rate exceeds the set threshold or the timestamp interval exceeds the set threshold, generate a water monitoring data adjustment result containing a repair operation record.

Images