Pipe network leakage early warning method based on online prediction and data fusion

Abstract

The present application relates to the technical field of pipeline network leakage early warning, and particularly relates to a pipeline network leakage early warning method based on online prediction and data fusion, comprising the following steps: a simulation data generation module, which generates simulation data by using a simulation model and historical data of a pipeline system; a monitoring data acquisition and transmission module, which is responsible for collecting real-time monitoring data of the pipeline system, and the real-time monitoring data is collected by a sensor device in real time; a real-time online prediction and data fusion module, which predicts future pipeline network operation conditions; a leakage early warning module, which sends an alarm when the prediction result shows that there is a leakage condition, and sends the alarm information to relevant personnel; and a feedback control module, which automatically adjusts the control strategy of the pipeline system according to the prediction result and the leakage early warning information. The present application predicts the pipeline network operation conditions in real time, discovers potential problems and abnormal conditions in a timely manner, and reduces the occurrence of operation accidents.

Description

Technical Field

[0001] This invention relates to the field of pipeline network leakage early warning technology, and in particular to a pipeline network leakage early warning method based on online prediction and data fusion. Background Technology

[0002] Pipeline transportation systems, as a fundamental infrastructure of modern cities, play a vital role in water supply, gas supply, heating, and other essential services. With urban development and population growth, the demand for these services continues to increase. To meet these demands, pipeline networks are becoming increasingly complex, and their coverage is constantly expanding.

[0003] However, at the same time, pipeline leaks have become an increasingly serious problem. Due to various reasons, such as aging, material fatigue, external impact, construction errors, or natural disasters, pipelines may develop cracks or breakages. Such leaks not only lead to the waste of precious resources but can also pollute the surrounding environment, such as seeping into the soil or water sources. More seriously, leaks of natural gas or chemicals can lead to explosions, fires, or the release of toxic gases, posing a serious threat to human life.

[0004] While some traditional detection and maintenance methods exist, such as periodic inspections, physical testing, or acoustic detection, these methods often fail to detect small or initial leaks in real time. Furthermore, due to the vastness and complexity of piping networks, regular physical inspections are time-consuming, labor-intensive, and not always effective. Therefore, for operators, detecting leaks and taking swift action in the shortest possible time has become a pressing issue.

[0005] In recent years, the rapid development of sensing, communication, and data analytics technologies has provided new possibilities for real-time monitoring and analysis of pipeline networks. By installing various sensors to monitor pipeline flow, pressure, temperature, and other key parameters in real time, and transmitting this data to a central data processing center, the operational status of the entire system can be analyzed in real time. However, extracting meaningful information from massive amounts of data and predicting and warning of potential leaks remains a technical challenge.

[0006] Therefore, proposing a method that can predict pipeline operation in real time and accurately warn of leaks can not only ensure the efficient use of resources and reduce environmental risks, but also has important practical significance and broad application prospects. Summary of the Invention

[0007] To achieve the above objectives, this invention provides a pipeline leakage early warning method based on online prediction and data fusion.

[0008] A pipeline leakage early warning method based on online prediction and data fusion includes the following steps:

[0009] S1: Simulation data generation module, which uses the simulation model of the pipeline system and historical data to generate simulation data to assist in prediction and analysis. The simulation data includes simulations of flow rate, pressure parameters and abnormal conditions under different operating conditions.

[0010] S2: Monitoring data acquisition and transmission module, responsible for collecting real-time monitoring data of the pipeline system, including flow, pressure and water quality parameters. The real-time monitoring data is collected in real time through sensor devices and transmitted to the data processing center through the communication system.

[0011] S3: Real-time online prediction and data fusion module. This module uses real-time monitoring data and simulation data, combined with prediction algorithms and data mining technology, to make real-time online predictions of pipeline network operation. By fusing and analyzing real-time monitoring data with historical simulation data, it predicts the future operation of the pipeline network.

[0012] S4: Leakage warning module. Based on the prediction results and characteristic threshold settings, this leakage warning module provides early warning of pipeline leaks. When the prediction results indicate that a leak exists, the leakage warning module issues an alarm and sends the alarm information to relevant personnel.

[0013] S5: Feedback control module, which automatically adjusts the control strategy of the pipeline system based on the prediction results and leakage early warning information.

[0014] Furthermore, S1 specifically includes:

[0015] S11: Select a finite difference model and collect historical data, including flow rate, pressure, and water quality;

[0016] S12: Perform data cleaning, delete or correct abnormal, incomplete or inaccurate data, and analyze the statistical characteristics of historical data;

[0017] S13: Model parameterization, using historical data to parameterize the model, including parameters such as pipe roughness, connection point characteristics, and valve operating characteristics;

[0018] S14: Simulate different operating conditions. Use the model to simulate various operating conditions. Based on the simulated operating conditions, the model will output predicted flow and pressure data.

[0019] S15: Verification and calibration. Compare the simulated data with historical data. If there are differences between the simulated data and historical data, adjust or calibrate the model until the simulated data matches the actual situation.

[0020] Furthermore, the sensor device in S2 includes:

[0021] Flow sensors are used to measure the flow rate of fluid through pipes;

[0022] Pressure sensors are used to monitor the pressure inside pipes;

[0023] Water quality sensors: pH sensor, turbidity sensor, conductivity sensor;

[0024] Leak detection sensor.

[0025] Furthermore, S3 specifically includes:

[0026] S31: Data acquisition and processing. Collect real-time monitoring data of the pipeline network, including flow, pressure, and temperature parameters. Clean, denoise, and detect anomalies in the collected data to ensure the accuracy and reliability of the data. Preprocess the monitoring data and simulation data, including normalization and standardization.

[0027] S32: Feature extraction and selection. Based on the characteristics of the pipeline network and operational requirements, features are selected for extraction. Features include time features, frequency domain features, and statistical features. A feature fingerprint database corresponding to the operating status is established, including feature thresholds corresponding to normal and leakage states. Feature selection algorithms are used to select features that have a significant impact on the prediction model. Feature selection algorithms include correlation analysis and information gain.

[0028] S33: Predictive model establishment. Based on the collected data and features, a predictive model is established using a support vector machine model. Historical data is used for model training, and model parameters are optimized and adjusted to improve the accuracy and stability of the prediction. The layout of pressure monitoring points is optimized.

[0029] Furthermore, the optimized arrangement of S33 specifically includes:

[0030] A prediction confidence level is defined to assess the reliability of pressure, flow rate, and temperature predictions. The prediction confidence level for the i-th point is defined as follows:

[0031]

[0032] For all selected time points, The pressure predicted for point i, For the actual pressure at point i, the closer the confidence level is to 1, the closer the predicted value is to the actual value.

[0033] Select a set of monitoring points , To determine the number of monitoring points, machine learning algorithms, such as support vector machines or artificial neural networks, are used to optimize the objective function TRAC, resulting in an optimized set of pressure monitoring points. .

[0034] Furthermore, S3 also includes:

[0035] Online prediction and real-time updates utilize established prediction models to forecast the real-time status of the pipeline network, obtaining prediction results for the network's operation over a future period. As new monitoring data continuously arrives, the prediction model is updated back to S33 to maintain its accuracy and adaptability.

[0036] Data mining and model optimization utilize data mining techniques to analyze and mine prediction results, discover potential patterns and correlations, and optimize and improve prediction models based on the mining results.

[0037] Furthermore, S4 specifically includes:

[0038] S41: Determine the early warning threshold for the test point Select a set of test points near the potential hazard area. , The number of test points;

[0039] S42: Assume the initial pipeline network is a complete, leak-free pipeline, and utilize the pressure data from monitoring points. And predict the pressure data at the test points. Compare the simulated data at the test points. and forecast data Define pressure deviation:

[0040] ;

[0041] S43: Imposing various types of defects on the simulation model The simulation yielded pressure data and pressure feature fingerprint database of the defective pipeline network under different operating conditions. Using pressure data from monitoring points Predict and obtain pressure data at test points of defective structures Compare the simulated data at the test points. and forecast data Define the pressure deviation of a defective pipeline network:

[0042] ;

[0043] S44: Obtain the relative pressure deviation between the defective and complete pipe networks. :

[0044] And establish a pressure feature fingerprint database based on the defect level Di. The quantitative relationship expression is as follows At this time, the warning threshold .

[0045] Furthermore, the control strategy in S5 includes adjusting valve opening and regulating pump station operation to ensure the stable operation of the pipeline network.

[0046] The beneficial effects of this invention are:

[0047] This invention provides real-time online prediction of pipeline network operation, timely detection of potential problems and anomalies, and reduction of operational accidents; data fusion analysis improves the accuracy and reliability of predictions; and automatic adjustment of control strategies enables dynamic regulation of pipeline network pressure, thereby improving the operational efficiency and stability of the pipeline network. Attached Figure Description

[0048] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only for this invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0049] Figure 1 This is a schematic diagram of the system structure according to an embodiment of the present invention. Detailed Implementation

[0050] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments.

[0051] It should be noted that, unless otherwise defined, the technical or scientific terms used in this invention should have the ordinary meaning understood by one of ordinary skill in the art to which this invention pertains. The terms "first," "second," and similar terms used in this invention do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Terms such as "comprising" or "including" mean that the element or object preceding the word encompasses the elements or objects listed following the word and their equivalents, without excluding other elements or objects. Terms such as "connected" or "linked" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. Terms such as "upper," "lower," "left," and "right" are used only to indicate relative positional relationships; when the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0052] like Figure 1 As shown, the pipeline leakage early warning method based on online prediction and data fusion includes the following steps:

[0053] S1: Simulation data generation module, which uses the simulation model of the pipeline system and historical data to generate simulation data to assist in prediction and analysis. The simulation data includes simulations of flow rate, pressure parameters and abnormal conditions under different operating conditions.

[0054] S2: Monitoring data acquisition and transmission module, responsible for collecting real-time monitoring data of the pipeline system, including flow, pressure and water quality parameters. The real-time monitoring data is collected in real time through sensor devices and transmitted to the data processing center through the communication system.

[0055] S3: Real-time online prediction and data fusion module. This module uses real-time monitoring data and simulation data, combined with prediction algorithms and data mining technology, to make real-time online predictions of pipeline network operation. By fusing and analyzing real-time monitoring data with historical simulation data, it predicts the future operation of the pipeline network.

[0056] S4: Leakage warning module. Based on the prediction results and characteristic threshold settings, this leakage warning module provides early warning of pipeline leaks. When the prediction results indicate that a leak exists, the leakage warning module issues an alarm and sends the alarm information to relevant personnel.

[0057] S5: Feedback control module, which automatically adjusts the control strategy of the pipeline system based on the prediction results and leakage early warning information.

[0058] S1 specifically includes:

[0059] S11: Select a finite difference model and collect historical data, including flow rate, pressure, and water quality;

[0060] S12: Perform data cleaning, delete or correct abnormal, incomplete or inaccurate data, and analyze the statistical characteristics of historical data;

[0061] S13: Model parameterization, using historical data to parameterize the model, including parameters such as pipe roughness, connection point characteristics, and valve operating characteristics;

[0062] S14: Simulate different operating conditions. Use the model to simulate various operating conditions. Based on the simulated operating conditions, the model will output predicted flow and pressure data.

[0063] S15: Verification and calibration. Compare the simulated data with historical data. If there are differences between the simulated data and historical data, adjust or calibrate the model until the simulated data matches the actual situation.

[0064] The sensor devices in S2 include:

[0065] Flow sensors are used to measure the flow rate of fluid through pipes;

[0066] Pressure sensors are used to monitor the pressure inside pipes;

[0067] Water quality sensors: pH sensor, turbidity sensor, conductivity sensor;

[0068] Leak detection sensor;

[0069] These sensors can help detect leaks or other pipeline anomalies, as leaks or malfunctions can cause changes in acceleration or vibration.

[0070] S3 specifically includes:

[0071] S31: Data acquisition and processing. Collect real-time monitoring data of the pipeline network, including flow, pressure, and temperature parameters. Clean, denoise, and detect anomalies in the collected data to ensure the accuracy and reliability of the data. Preprocess the monitoring data and simulation data, including normalization and standardization.

[0072] S32: Feature extraction and selection. Based on the characteristics of the pipeline network and operational requirements, features are selected for extraction. Features include time features, frequency domain features, and statistical features. A feature fingerprint database corresponding to the operating status is established, including feature thresholds corresponding to normal and leakage states. Feature selection algorithms are used to select features that have a significant impact on the prediction model. Feature selection algorithms include correlation analysis and information gain.

[0073] S33: Predictive model establishment. Based on the collected data and features, a predictive model is established using a support vector machine model. The model is trained using historical data, and the model parameters are optimized and adjusted to improve the accuracy and stability of the prediction. The layout of pressure (and flow, temperature) monitoring points is optimized.

[0074] The optimized layout of S33 specifically includes:

[0075] Prediction Confidence Ratio (TRAC) is defined to assess the reliability of pressure, flow rate, and temperature predictions. The prediction confidence ratio for the i-th point is defined as follows:

[0076]

[0077] For all selected time points, The pressure predicted for point i, For the actual pressure at point i, the closer the confidence level is to 1, the closer the predicted value is to the actual value.

[0078] Select a set of monitoring points , To determine the number of monitoring points, machine learning algorithms, such as support vector machines or artificial neural networks, are used to optimize the objective function TRAC, resulting in an optimized set of pressure monitoring points. .

[0079] S3 also includes:

[0080] Online prediction and real-time updates utilize established prediction models to forecast the real-time status of the pipeline network, obtaining prediction results for the network's operation over a future period. As new monitoring data continuously arrives, the prediction model is updated back to S33 to maintain its accuracy and adaptability.

[0081] Data mining and model optimization utilize data mining techniques to analyze and mine prediction results, discover potential patterns and correlations, and optimize and improve prediction models based on the mining results to enhance the accuracy and stability of predictions.

[0082] S4 specifically includes:

[0083] S41: Determine the early warning threshold for the test point Select a set of test points near potentially hazardous areas (such as valves or switches). , The number of test points;

[0084] S42: Assuming the initial pipeline network is a complete and leak-free pipeline, utilize the pressure (or pressure gradient, flow rate, temperature) data from monitoring points. And predict the pressure (or pressure gradient, flow rate, temperature) data at the test point. Compare the simulated data at the test points. and forecast data Define pressure (or pressure gradient, flow rate, temperature) deviation:

[0085] ;

[0086] S43: Imposing various types of defects on numerical models Simulations were used to obtain pressure (or pressure gradient, flow rate, temperature) data and pressure (or pressure gradient, flow rate, temperature) feature fingerprint databases of defective pipeline networks under different operating conditions. Using pressure (or pressure gradient, flow rate, temperature) data from monitoring points Predict and obtain pressure (or pressure gradient, flow rate, temperature) data at test points in the defective structure. Compare the simulated data at the test points. and forecast data Define the pressure (or pressure gradient, flow rate, temperature) deviation of a defective pipeline network:

[0087] ;

[0088] S44: Obtain the relative pressure (or pressure gradient, flow rate, temperature) deviation between the defective and complete pipe networks. And establish a defect degree Di-pressure feature fingerprint database. The quantitative relationship expression is as follows At this time, the warning threshold .

[0089] The control strategies in S5 include adjusting valve opening and regulating pump station operation to ensure the stable operation of the pipeline network.

[0090] Those skilled in the art should understand that the discussion of any of the above embodiments is merely exemplary and is not intended to imply that the scope of the invention is limited to these examples; within the framework of the invention, the technical features of the above embodiments or different embodiments can also be combined, the steps can be implemented in any order, and there are many other variations of the different aspects of the invention as described above, which are not provided in detail for the sake of brevity.

[0091] This invention is intended to cover all such substitutions, modifications, and variations falling within the broad scope of the claims. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A method for early warning of pipe network leakage based on online prediction and data fusion, characterized in that, The method comprises the following steps: S1: The simulation data generation module generates simulation data for assisting in prediction and analysis by using a simulation model of the pipeline system and first historical data, the simulation data including flow, pressure parameters and simulation of abnormal conditions under different working conditions; S2: The monitoring data acquisition and transmission module collects real-time monitoring data of the pipeline system, including flow, pressure and water quality parameters, the real-time monitoring data being collected by a sensor device in real time and transmitted to a data processing center through a communication system; S3: The real-time online prediction and data fusion module predicts the operation of the pipeline network in real time by using real-time monitoring data and simulation data, combining prediction algorithms and data mining techniques, and by fusing and analyzing the real-time monitoring data and the simulation data to predict the future operation of the pipeline network; specifically comprising: S31: Data acquisition and processing, collecting real-time monitoring data of the pipeline network, including flow, pressure and temperature parameters, cleaning, denoising and anomaly detection of the collected data to ensure the accuracy and reliability of the data, and pre-processing the monitoring data and the simulation data, including normalization and standardization; S32: Feature extraction and selection, selecting features for extraction according to the characteristics and operation requirements of the pipeline network, the features including time features, frequency domain features and statistical features, establishing a feature fingerprint library corresponding to the operation state, including feature thresholds corresponding to normal state and leakage state, and selecting features that have an important influence on the prediction model by using a feature selection algorithm, the feature selection algorithm including correlation analysis and information gain; S33: Prediction model establishment, establishing a prediction model according to the collected data and features, using a support vector machine model, training the prediction model by using second historical data, and optimizing and adjusting the parameters of the prediction model to improve the accuracy and stability of the prediction, and optimizing the arrangement of pressure monitoring points; S4: The leakage warning module warns of pipeline leakage based on the prediction results and feature threshold settings, and sends an alarm to relevant personnel when the prediction results show that there is a leakage; S5: The feedback control module automatically adjusts the control strategy of the pipeline system according to the prediction results and the leakage warning information.

2. The method of online prediction and data fusion based early warning of pipe network leakage according to claim 1, characterized in that, The S1 specifically comprises: S11: Selecting a finite difference model and collecting first historical data, including flow, pressure and water quality; S12: Data cleaning, deleting or correcting abnormal, incomplete or inaccurate data, and analyzing the statistical characteristics of the first historical data; S13: Parameterizing the simulation model by using the first historical data, the parameters including the roughness of the pipeline, the characteristics of the connection points and the operating characteristics of the valves; S14: Simulating different working conditions by using the simulation model, the simulation model outputting simulation data of flow and pressure based on the simulated working conditions; S15: Verification and correction, comparing the simulation data with the first historical data, adjusting or correcting the simulation model if there is a difference between the simulation data and the first historical data, until the simulation data matches the actual situation.

3. The method of online prediction and data fusion based early warning of pipe network leakage according to claim 2, characterized in that, The sensor device in the S2 comprises: a flow sensor for measuring the flow of fluid through the pipeline; Pressure sensor for monitoring the pressure in the pipeline; Water quality sensors: pH sensor, turbidity sensor, conductivity sensor.

4. The method of online prediction and data fusion based early warning of pipe network leakage according to claim 1, characterized in that, The optimization arrangement of S33 specifically includes: A prediction confidence is defined to conduct a credibility assessment to determine the credibility of pressure, flow rate, and temperature prediction, and the prediction confidence of the i-th point is defined as: ； Ppred(i) for all selected time points, Ppred(i) for the i-th point, Ptrue(i) for the i-th point, the closer the confidence is to 1, the closer the predicted value is to the true value; Selecting a monitoring point set , The number of monitoring points is monitored, and a machine learning algorithm is used to optimize the target function TRAC to obtain an optimized pressure monitoring point .

5. The method of online prediction and data fusion based early warning of pipe network leakage according to claim 4, characterized in that, S3 further includes: Online prediction and real-time updating, using the established prediction model to predict the real-time pipeline state, obtaining the prediction result of the pipeline operation in the future period of time, and returning to S33 to update the prediction model as new monitoring data continuously arrives, so as to maintain the accuracy and adaptability of the model; Data mining and model optimization, using data mining technology to analyze and mine the prediction result, discovering potential rules and correlations, and optimizing and improving the prediction model according to the mining result.

6. The method of online prediction and data fusion based early warning of pipe network leakage according to claim 5, characterized in that, S4 specifically includes: S41: determining a test point early warning threshold , selecting a test point set near the potential danger area , is the number of test points; S42: Set initial pipe network as complete no-leak pipe, use monitoring point pressure data and predict test point pressure data , compare test point simulation data and prediction data , define pressure deviation: ； S43: Apply various types of defects to the simulation model , simulate defect pipe network pressure data and pressure characteristic fingerprint library under different working conditions , use monitoring point pressure data , predict the test point pressure data of the defect structure , compare the simulation data of the test point and the prediction data , define the pressure deviation of the pipe network containing defects: ； S44: Obtain the relative pressure deviation of the defective and complete pipe networks : ; and establishing a degree of defect Di - the pressure characteristic fingerprint library The quantitative relationship expression is At this time, the early warning threshold .

7. The method of online prediction and data fusion based early warning of pipe network leakage according to claim 6, characterized in that, The control strategy in S5 includes adjusting the valve opening and adjusting the pump station operation to ensure the stable operation of the pipeline network.

Images