Energy pipeline leakage intelligent detection method and system

Abstract

This invention discloses an intelligent detection method and system for energy pipeline network leakage. It includes: (1) acquiring historical time-series data of instantaneous and cumulative flow of network sources and users over a time period T, forming an original data sample D; (2) preprocessing the original data sample D to obtain a processed sample CD; (3) calculating the working fluid imbalance CD1 based on the processed sample CD, and statistically analyzing its mean µ and variance σ²; (4) determining the significance level α, and using the µ test to determine the mean µ1 and µ2 of the working fluid imbalance, respectively; (5) calculating the probability density CP1 of pipeline network leakage and the probability density CP2 of pipeline network non-leakage; (6) combining CP1 and CP2 as a sample, including time-series correlation features, inputting it into a support vector machine model to obtain the pipeline network leakage status judgment result. This invention combines statistical probability with the support vector machine algorithm to eliminate fluctuation errors in flow modeling, achieving high accuracy and a low misjudgment rate.

Description

Technical Field

[0001] This invention belongs to the field of pipeline leakage detection, and relates to an intelligent detection method and system for energy pipeline leakage. Background Technology

[0002] In various scenarios such as steel, power, and industrial parks, there are numerous energy pipeline networks (including steam, water, compressed air, etc.). Many of these networks were constructed in phases, resulting in highly complex structures and numerous, scattered user points. During operation, these complex networks inevitably encounter various problems such as pipe aging, material corrosion, weld damage, stress changes, and human sabotage, leading to pipeline leaks, energy waste, economic losses, environmental pollution, and even safety accidents.

[0003] Because pipeline networks cover a large area and leaks can cause serious damage, leak detection technology has always been a key research area both domestically and internationally. Currently, the mainstream detection methods are divided into two categories: direct methods and indirect methods.

[0004] Direct methods include acoustic detection, temperature detection, gas detection, and pressure testing. These methods detect and judge the effects of pipeline leaks on the pipe wall and surrounding medium, such as sound waves or vibrations at specific frequencies, abnormal temperature changes, tracer gas diffusion, and pressure surges. However, these methods are costly to implement, have poor real-time performance, are easily affected by the environment, and rely heavily on human experience and subjective judgment. They also lack systematic analytical capabilities, and the detection results are difficult to quantify.

[0005] Indirect methods include statistical detection, neural network methods, model recognition, and quality balance methods. These methods rely on a large amount of historical operational data to extract relevant feature data. Based on the differences between data before and after a pipeline leak, they establish corresponding black-box, gray-box, and white-box models for pipeline leak analysis using statistical methods or underlying mechanisms to determine the leak situation. These methods do not require additional implementation costs, can achieve 24 / 7 online monitoring, and can be continuously iterated and updated over time. However, these methods can introduce fluctuation errors during modeling.

[0006] Therefore, if a smart detection method and system for energy pipeline leaks can be developed to solve the above problems, it will have great application value. Summary of the Invention

[0007] To overcome the above-mentioned defects, the purpose of this invention is to provide an intelligent detection method and system for energy pipeline network leaks.

[0008] To achieve the above objectives, the present invention provides an intelligent detection method for energy pipeline network leaks, comprising the following steps: (1) Obtain historical time-series data of instantaneous and cumulative traffic of network sources and users within time period T to form the original data sample D; (2) The original data sample D is preprocessed to obtain the processed sample CD; (3) Calculate the unbalanced amount of working fluid CD1 based on the processed sample CD, and calculate its mean µ and variance σ². (4) Based on the ratio of pipeline leakage time to pipeline operation time during the sampling history, the significance level α is determined, and the mean values ​​of working fluid imbalance, µ1 and µ2, are determined by the µ test method for pipeline leakage and non-leakage, respectively. (5) The mean, µ1, µ2 and variance σ of the working fluid imbalance were obtained from the calculations for the pipeline network under leak and non-leak conditions. 2 Calculate the probability density CP1 of pipeline leakage and the probability density CP2 of pipeline non-leakage; (6) Combine CP1 and CP2 as a sample, including time series correlation features, and input them into the support vector machine model to obtain the pipeline leakage status judgment result.

[0009] Furthermore, the preprocessing in step (2) includes: calculating the moving average of the traffic of the network source and each user within a period of t2 using instantaneous traffic and cumulative traffic respectively, and removing gross errors from the traffic difference between the two according to the Raida criterion.

[0010] Furthermore, in step (2), the data preprocessing will filter out the outage data in the pipeline network, calculate the moving average of the instantaneous flow of the network source and each user at any time as the average from any time to t2 before any time, and calculate the moving average of the cumulative flow of the network source and each user at any time as the cumulative flow of any time minus the cumulative flow of t2 before any time.

[0011] Further, in step (2), the data preprocessing removes gross errors in the difference between the instantaneous flow rate and the cumulative flow rate within the period t2 caused by sensor failure or other reasons in the original sample D; according to the Raida criterion, the standard deviation is calculated and processed from the difference in the flow rate moving average; data with error values ​​greater than the standard deviation are discarded and replaced with data from the previous moment.

[0012] Furthermore, in step (4), the significance level α is the ratio of pipeline leakage time to pipeline operation time during the sampling history period, calculated by the statistical measure. and rejection domain Calculate the average working fluid imbalance values ​​µ1 and µ2 for pipeline leakage and non-leakage conditions, respectively.

[0013] Furthermore, in step (5), the probability density CP1 is obtained through the formula... The probability density CP2 is calculated using the formula... calculate.

[0014] Furthermore, in step (1), the time period T is set to more than half a year to ensure full coverage of all working conditions, and the data acquisition interval t1 is set to 1-5 minutes to ensure the accuracy of data calculation.

[0015] Furthermore, the period t2 in step (2) is 4-20 times the number taking interval t1.

[0016] To achieve the above objectives, the present invention provides an intelligent detection system for energy pipeline leaks, comprising: The data acquisition module is used to acquire historical time-series data of instantaneous and cumulative traffic of network sources and users within a time period T, forming the original data sample D; The data preprocessing module preprocesses the original data sample D to obtain the processed sample CD. The working fluid imbalance calculation module calculates the working fluid imbalance CD1 based on the processed sample CD. The statistical parameter estimation module is used to calculate the average value μ and variance σ2 of the working fluid imbalance CD1, and determine the significance level α based on the ratio of pipeline leakage time to total operating time during historical operation. The mean values ​​μ1 and μ2 of the working fluid imbalance when the pipeline is in a leaking state and a non-leaking state are estimated respectively by the μ test method. The probability density calculation module calculates the mean, µ1, µ2, and variance σ of the working fluid imbalance under both leaking and non-leaking pipeline conditions. 2 Calculate the probability density CP1 of pipeline leakage and the probability density CP2 of pipeline non-leakage; The leakage detection module is used to concatenate CP1 and CP2 and combine their time series correlation features to form an input feature vector, which is then input into a trained support vector machine model to output the pipeline leakage status detection result at the current moment.

[0017] The present invention has the following beneficial effects: (1) This invention is an online detection method based on statistical probability and support vector machine algorithm, which does not require additional hardware equipment and construction.

[0018] (2) This invention analyzes and preprocesses instantaneous flow and cumulative flow to filter out error data, improve the accuracy of data used in modeling, and enhance model precision.

[0019] (3) This invention combines statistical probability with support vector machine algorithm to eliminate fluctuation error in flow modeling. It establishes a support vector machine judgment model of pipeline leakage status based on the probability of pipeline leakage and non-leakage and its time series correlation characteristics. The model has high accuracy and low misjudgment rate. Attached Figure Description

[0020] Figure 1 This is a flowchart of the industrial user branch pipe network leakage detection method of the present invention.

[0021] Figure 2 This is a schematic diagram of a branch pipe network for industrial users.

[0022] Figure 3 The graph shows the 20-minute average change curves of the instantaneous flow rate and cumulative flow rate of the main pipe.

[0023] Figure 4 This is a comparison chart of the pipeline leakage status model assessment results and the actual pipeline leakage status. Detailed Implementation

[0024] The embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0025] In the description of this invention, it should be understood that the terms "center", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention.

[0026] The terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this invention, unless otherwise stated, "a plurality of" means two or more.

[0027] In the description of this invention, it should be noted that, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "joining" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal communication between two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0028] This invention relates to a method for detecting leaks in industrial user branch pipe networks based on statistical probability and support vector machines, such as... Figure 1 As shown, it includes the following steps: (1) Obtain historical operating data of instantaneous and cumulative flow for network sources and users with a period of T from the pipeline database, with a data collection interval of t1, to obtain the original data sample D; the time period T is set to more than half a year to ensure full coverage of operating conditions, and the data collection interval t1 is set to 1-5 minutes to ensure the accuracy of data calculation. Among them, the network source and users can be any branch pipe segment, which must have flow measurement points at the inlet and all outlets. (2) Perform data preprocessing on the original data sample D, including: The moving average of the traffic of the network source and each user within a period of t2 is calculated using instantaneous traffic and cumulative traffic respectively. Gross errors are removed from the difference between the two traffic values ​​according to the Raida criterion. The processed sample is denoted as CD. According to the Raida criterion, when removing gross errors from the original sample D, the standard deviation of the traffic of the network source and all users must be calculated simultaneously, and the data at any point in the calculated traffic value where a gross error occurs must be discarded.

[0029] Data preprocessing will filter out outage data in the pipeline network, calculate the moving average of the instantaneous flow of the network source and each user at any time as the average from any time to t2 times before any time, and calculate the moving average of the cumulative flow of the network source and each user at any time as the cumulative flow of any time minus the cumulative flow of t2 times before any time.

[0030] Data preprocessing removes gross errors in the moving average difference between instantaneous and cumulative flow rates over period t2 in the original sample D, caused by sensor malfunctions or other reasons. According to the Raida criterion, the standard deviation is calculated from the difference in the moving average flow rate. Data with errors greater than the standard deviation are discarded and replaced with data from the previous time step.

[0031] To ensure calculation accuracy, the period t2 is 4-20 times the data taking interval t1.

[0032] (3) Calculate the network working fluid imbalance CD1 for the flow of network sources and users in sample CD; (4) Calculate the average value µ and variance σ of the unbalanced working fluid CD1. 2 The imbalance of the working fluid in the pipeline follows a normal distribution regardless of whether there is a leak or no leak, and the variance is σ. 2 .

[0033] (5) Analyze the ratio of pipeline leakage time to pipeline operation time during the sampling history period, determine the significance level α, and determine the mean values ​​of working fluid imbalance (µ1 and µ2) under pipeline leakage and non-leakage conditions respectively using the µ test method; the significance level α is the ratio of pipeline leakage time to pipeline operation time during the sampling history period, and is determined by calculating the statistical statistic. and rejection domain The average values ​​of the working fluid imbalance, µ1 and µ2, can be calculated separately for pipeline leakage and non-leakage conditions.

[0034] (6) The mean, µ1, µ2 and variance σ of the working fluid imbalance were obtained from the calculations for pipeline leakage and non-leakage conditions. 2 Calculate the probability density CP1 for pipeline leakage and the probability density CP2 for pipeline without leakage, given the unbalanced working fluid quantity CD1. The probability density CP1 is calculated using the formula... The probability density CP2 is calculated using the formula... calculate.

[0035] (7) CP1 and CP2 are concatenated and combined as samples, including time series correlation features, and input into the support vector machine model to obtain the pipeline leakage status judgment result. CP1 and CP2 are concatenated as a matrix. The support vector machine model inputs parameters CP1 and CP2, as well as the autocorrelation and partial correlation features between the current value and its past values. The time span is t2, including t, t-1, t-2, t-3, t-4, ..., t-t2+1. There are 2×t2 input variables. The output parameter is the pipeline leakage status at the current moment. At the same time, the support vector machine model is trained to continuously adjust the parameters with the goal of reducing the judgment residual, so as to obtain a satisfactory judgment model.

[0036] The intelligent energy pipeline leakage detection system based on statistical probability and support vector machine of the present invention includes: The data acquisition module is used to acquire historical time-series data of instantaneous and cumulative traffic of network sources and users within a time period T, forming the original data sample D; The data preprocessing module preprocesses the original data sample D to obtain the processed sample CD. The working fluid imbalance calculation module calculates the working fluid imbalance CD1 based on the processed sample CD. The statistical parameter estimation module is used to calculate the average value μ and variance σ2 of the working fluid imbalance CD1, and determine the significance level α based on the ratio of pipeline leakage time to total operating time during historical operation. The mean values ​​μ1 and μ2 of the working fluid imbalance when the pipeline is in a leaking state and a non-leaking state are estimated respectively by the μ test method. The probability density calculation module calculates the mean, µ1, µ2, and variance σ of the working fluid imbalance under both leaking and non-leaking pipeline conditions. 2 Calculate the probability density CP1 of pipeline leakage and the probability density CP2 of pipeline non-leakage; The leakage detection module is used to concatenate CP1 and CP2 and combine their time series correlation features to form an input feature vector, which is then input into a trained support vector machine model to output the pipeline leakage status detection result at the current moment.

[0037] Example This embodiment is a flowchart illustrating the method for detecting leaks in industrial user branch pipe networks.

[0038] The framework of this invention mainly consists of core steps such as historical data sampling, data preprocessing, calculation of normal distribution, hypothesis testing, determination of pipeline leakage probability density, determination of pipeline non-leakage probability density, support vector machine model training, and pipeline leakage status judgment. Detailed procedures are as follows: Figure 1 As shown. Taking a certain pipeline system as an example, the pipeline structure is as follows. Figure 2 As shown, the specific operation steps are as follows: Historical time-series data of instantaneous and cumulative flow of network sources and users from 00:00 on July 1, 2023 to 00:00 on December 31, 2023 were collected from the pipeline network database. The data collection interval was 1 minute. After removing outage data, the 20-minute moving average sequence of instantaneous and cumulative flow of the main pipeline was calculated as follows: Figure 3 As shown. The data is preprocessed, and the standard deviation range of the flow difference between the two is calculated according to the Raida criterion. Gross errors are filtered out, and the working fluid imbalance of the pipeline network, as well as the mean µ and variance σ of the working fluid imbalance array, are further calculated. 2 Based on historical data and relevant records, the leakage time in the pipeline network during the sampling period accounted for 0.647% of the pipeline network operating time, with a significance level of α of 0.647%. The statistical statistic was calculated... and rejection domain Obtain the average working fluid imbalance values ​​µ1 and µ2 under pipeline leakage and no leakage conditions, and process the working fluid imbalance array through... Calculate the pipeline leakage probability density at the corresponding time using the formula. Calculate the probability density of no leakage in the pipeline network at the corresponding time, and then convert the probability density of leakage in the pipeline network... P 1 and the probability density of no leakage in the pipeline network P A support vector machine model is built using the autocorrelation and partial correlation features between 2 and its past values ​​as input parameters, where the time span is 20, and includes... P 1(t), P 1(t-1), P 1(t-2), P 1(t-3), ... P 1(t-19) P 2(t), P 2(t-1), P 2(t-2), P 2(t-3), ... P 2(t-19), and divide the data into datasets, one part as training data and the other part as test data. Use the training data to train the pipeline leakage state support vector machine model, and output the pipeline leakage state at the current time.

[0039] Figure 4 The data compares the pipeline leakage status assessment results with the actual results over a period of time (from 00:00 on December 23, 2023 to 00:00 on December 24, 2023). The absolute percentage error of the model's assessment values ​​for the test set is 3.275%, indicating that the model error is very small and meets the needs of practical engineering.

[0040] The present invention has been described in detail above with reference to the accompanying drawings. However, the present invention is not limited to the embodiments described above. Within the scope of knowledge possessed by those skilled in the art, various changes can be made without departing from the spirit of the present invention. Many other changes and modifications made without departing from the concept and scope of the present invention should be considered within the scope of protection of the present invention.

[0041] 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 variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A smart detection method for leaks in energy pipeline networks, characterized in that, Includes the following steps: (1) Obtain historical time-series data of instantaneous and cumulative traffic of network sources and users within time period T to form the original data sample D; (2) The original data sample D is preprocessed to obtain the processed sample CD; (3) Calculate the unbalanced amount of working fluid CD1 based on the processed sample CD, and calculate its mean µ and variance σ². (4) Based on the ratio of pipeline leakage time to pipeline operation time during the sampling history, the significance level α is determined, and the mean values ​​of working fluid imbalance, µ1 and µ2, are determined by the µ test method for pipeline leakage and non-leakage, respectively. (5) The mean, µ1, µ2 and variance σ of the working fluid imbalance were obtained from the calculations for the pipeline network under leak and non-leak conditions. 2 Calculate the probability density CP1 of pipeline leakage and the probability density CP2 of pipeline non-leakage; (6) Combine CP1 and CP2 as a sample, including time series correlation features, and input them into the support vector machine model to obtain the pipeline leakage status judgment result.

2. The intelligent detection method for energy pipeline network leakage according to claim 1, characterized in that, The preprocessing in step (2) includes: calculating the moving average of the traffic of the network source and each user within a period of t2 using instantaneous traffic and cumulative traffic respectively, and removing gross errors from the traffic difference between the two according to the Raida criterion.

3. The intelligent detection method for energy pipeline network leakage according to claim 2, characterized in that, In step (2), the data preprocessing will filter out the outage data in the pipeline network, calculate the moving average of the instantaneous flow of the network source and each user at any time as the average from any time to t2 before any time, and calculate the moving average of the cumulative flow of the network source and each user at any time as the cumulative flow of any time minus the cumulative flow of t2 before any time.

4. The intelligent detection method for energy pipeline network leakage according to claim 2, characterized in that, In step (2), the data preprocessing removes gross errors in the difference between the instantaneous flow rate and the cumulative flow rate within the period t2 caused by sensor failure or other reasons in the original sample D; according to the Laida criterion, the standard deviation is calculated and processed from the difference in the flow rate moving average; data with error values ​​greater than the standard deviation are discarded and replaced with data from the previous moment.

5. The intelligent detection method for energy pipeline network leakage according to claim 1, characterized in that, In step (4), the significance level α is the ratio of pipeline leakage time to pipeline operation time during the sampling history period, calculated by the statistical measure. and rejection domain Calculate the average working fluid imbalance values ​​µ1 and µ2 for pipeline leakage and non-leakage conditions, respectively.

6. The intelligent detection method for energy pipeline network leakage according to claim 1, characterized in that, In step (5), the probability density CP1 is obtained through the formula The probability density CP2 is calculated using the formula... calculate.

7. The intelligent detection method for energy pipeline network leakage according to claim 1, characterized in that, In step (1), the time period T is set to ensure full coverage of all working conditions and is set to more than half a year. The data acquisition interval t1 is set to ensure the accuracy of data calculation and is set to 1-5 minutes.

8. The intelligent detection method for energy pipeline network leakage according to claim 2, characterized in that, In step (2), the period t2 is 4-20 times the number taking interval t1.

9. An intelligent detection system for leaks in energy pipeline networks, characterized in that, include: The data acquisition module is used to acquire historical time-series data of instantaneous and cumulative traffic of network sources and users within a time period T, forming the original data sample D; The data preprocessing module preprocesses the original data sample D to obtain the processed sample CD. The working fluid imbalance calculation module calculates the working fluid imbalance CD1 based on the processed sample CD. The statistical parameter estimation module is used to calculate the average value μ and variance σ2 of the working fluid imbalance CD1, and determine the significance level α based on the ratio of pipeline leakage time to total operating time during historical operation. The mean values ​​μ1 and μ2 of the working fluid imbalance when the pipeline is in a leaking state and a non-leaking state are estimated respectively by the μ test method. The probability density calculation module calculates the mean, µ1, µ2, and variance σ of the working fluid imbalance under both leaking and non-leaking pipeline conditions. 2 Calculate the probability density CP1 of pipeline leakage and the probability density CP2 of pipeline non-leakage; The leakage detection module is used to concatenate CP1 and CP2 and combine their time series correlation features to form an input feature vector, which is then input into a trained support vector machine model to output the pipeline leakage status detection result at the current moment.

Images