Method and system for positioning gas pipeline leakage through combination of XGBoost algorithm and negative pressure waves

Abstract

The invention discloses a method and system for positioning gas pipeline leakage by combining an XGBoost algorithm and negative pressure waves, and belongs to the field of natural gas pipeline safety guarantee. The method comprises the following steps: 1) reading signals such as pressure and flow of each valve chamber and station yard through an SCADA system, and preprocessing the signals; 2) inputting the time sequence signal into a pre-trained XGBoost model, and predicting a leakage position and an aperture; 3) calculating the flow and pressure of the leakage point; 4) determining upstream and downstream on-way pressure drop of the leakage point; (5) the pipeline is segmented, and the gas flow rate and the pressure wave propagation speed are calculated segment by segment; and 6) calculating to obtain the signal time difference of the front and back stations, and further accurately positioning the position of the leakage point. Meanwhile, the device is provided with a leakage positioning system and can be used for detecting the pipeline leakage condition in real time and quickly positioning leakage points. Compared with a traditional method, the method has the advantages of accurate positioning, clear process, high operability and the like, and is suitable for being applied by technicians in the field.

Description

Technical Field

[0001] The present invention belongs to the field of natural gas pipeline safety assurance, and in particular relates to a method and system for locating gas pipeline leakage by combining an XGBoost algorithm with a negative pressure wave. Background Art

[0002] In the long-term operation of natural gas pipelines, there is a risk of pipeline failure caused by corrosion, welding defects and third-party intervention. As pipelines gradually enter the middle and late stages of service, leakage has become one of the most common types of failures in the operation of natural gas pipeline networks, and there is an urgent need to achieve efficient detection, accurate identification and real-time positioning of leaks.

[0003] At present, negative pressure wave detection technology is the main means of pipeline leakage monitoring, and its effectiveness is highly dependent on the accuracy of leakage signal acquisition. Traditional methods rely on SCADA systems or directly deployed sensors to capture leakage signals, but the inherent accuracy limitations of SCADA systems affect their performance in leak detection, resulting in insufficient accuracy of directly read leakage signals. In addition, due to the high compressibility of the transmission medium in natural gas pipelines, the pressure fluctuations inside the pipelines create additional interference to the identification of leakage signals, further increasing the error range of negative pressure wave positioning that relies solely on SCADA system time difference data.

[0004] In order to overcome the problem of accurate positioning in natural gas pipeline leakage detection, an effective solution is urgently needed to improve the detection accuracy on the basis of ensuring the signal acquisition of the SCADA system. At present, the traditional negative pressure wave detection method is difficult to meet the requirements of high-precision leak positioning due to the accuracy limitations of the SCADA system. At the same time, due to the compressibility of the natural gas medium and the interference of pressure fluctuations inside the pipeline, the negative pressure wave positioning technology that relies solely on SCADA time difference data often has large errors. Therefore, an intelligent detection model that can comprehensively utilize SCADA system signals and has strong anti-interference capabilities is needed to break through the accuracy bottleneck of the SCADA system, accurately analyze and predict the time series pressure drop rate signal, and ensure high-precision identification of the leak location and aperture. Summary of the invention

[0005] In view of the lack of a fast and accurate leak location method on existing long-distance gas pipelines, the present invention provides a method and system for locating gas pipeline leaks by combining the XGBoost algorithm and negative pressure waves. The leak location and leak aperture are predicted by a pre-trained XGBoost model as the initial value for leak location, and the negative pressure wave method is used to calculate them to obtain the accurate leak location.

[0006] According to a first aspect of the present application, a calculation method for locating gas pipeline leakage by combining an XGBoost algorithm with a negative pressure wave is provided, the method comprising:

[0007] Step 1: Collect pressure, flow and other signals from the SCADA system, and temporarily store and pre-process the signals to form a time series signal of the pressure drop rate;

[0008] Step 2: Input the time series signal into the pre-trained XGBoost model to predict the leakage position and leakage aperture value of the upstream valve chamber and the downstream valve chamber respectively, and take the arithmetic mean of the correlation values ​​of the upstream and downstream valve chambers as the predicted leakage aperture and position;

[0009] Step 3: Calculate the flow rate and pressure of the leakage point according to the leakage position and leakage aperture predicted by the XGBoost model. The specific calculation formula of the leakage flow rate of the leakage point is:

[0010] Q v =0.00078PD 2 (1)

[0011] Where Q v ——Gas leakage rate, m 3 / s; D——effective hole diameter, mm; P——pipeline operating pressure, MPa; According to the flow rate and location of the leakage point, the pressure of the front and rear stations and the flow rate of the pipeline section are read from the SCADA system at the same time, and the pressure of the leakage point is calculated from the upstream and downstream valve chambers of the leakage point as the starting point, and the arithmetic average of the upstream and downstream pressures is calculated as the calculated pressure of the leakage point. The specific calculation relationship is:

[0012]

[0013]

[0014]

[0015] Where p 0,1 ——Pressure of upstream calculated leakage point, MPa; p 0,2 ——Calculated pressure of downstream leakage point, MPa; p0——Calculated pressure of leakage point, MPa; ——Outlet pressure of upstream station, MPa; ——Inlet pressure of downstream station, MPa; L0——Distance from leakage location to upstream station, km; L——Length of pipe section, km;

[0016] Step 4: Calculate the pressure distribution along the leakage point according to the leakage flow and pressure of the leakage point, where the calculation relationship of the pressure along the leakage point is:

[0017]

[0018]

[0019] Where p x,1 ——Pressure along the pipe section upstream of the leakage point, MPa; p x,2 ——pressure along the pipe section downstream of the leakage point, MPa; x——location of each point along the line, km; ——Inlet pressure of downstream station, MPa;

[0020] Step 5: At the same time, the pipeline is divided into n sub-segments according to relevant factors such as the length of the pipeline segment and the height difference of the undulation, and the pipeline is numbered 1, 2, ..., i, ..., n from upstream to downstream; and the flow velocity and pressure wave propagation velocity of the gas in the pipeline are calculated section by section. According to the flow velocity and pressure wave propagation velocity along the way, the calculation relationship of the gas flow velocity is:

[0021]

[0022] Where u(x)——gas velocity, m / s; Q *,i ——gas mass flow rate, kg / s; i——pipeline section number; R——gas constant, J / (kg·K); T0——pipeline ground temperature, k; A——pipeline cross-sectional area, m 2 ; ——pressure at the starting point of the ith sub-pipe section, Pa; ——The pressure at the end point of the ith sub-pipeline section, Pa;

[0023] The calculation formula of pressure wave propagation velocity is:

[0024]

[0025] Where v i (x)——pressure wave propagation velocity, m / s; k——specific heat ratio; D——inner diameter of the pipe, m; E——elastic modulus of the pipe, Pa; e——pipe wall thickness, m.

[0026] Step 6: According to the flow velocity and pressure wave propagation velocity along the route, the time difference between the stations before and after the negative pressure wave propagates to the leakage point and the accurate leakage position are calculated. The calculation relationship for calculating the time difference between the stations before and after the negative pressure wave propagates to the leakage point is:

[0027]

[0028] Where Δt is the time difference, s;

[0029] Based on the time difference and the flow velocity and pressure wave propagation velocity, etc., the exact location of the leakage point can be calculated, and the calculation formula is as follows:

[0030]

[0031] Where L0′ is the new leakage position obtained after optimization; v s ,v x are the average pressure propagation velocities upstream and downstream, m / s, respectively. The calculation formula is:

[0032]

[0033]

[0034] Where i is the number of pipe sections divided in the upstream of the pipeline; n is the number of pipe sections divided in the downstream of the pipeline; based on the above formula, the accurate location of the gas pipeline negative pressure wave leakage can be obtained.

[0035] According to a second aspect of the present invention, a system for locating gas pipeline leaks by combining an XGBoost algorithm with a negative pressure wave is provided. The system comprises: a processor and a memory, wherein the memory stores computer program instructions and can temporarily store SCADA system imported data. When the computer program instructions are executed by the processor, the steps of a calculation method for locating gas pipeline leaks by combining an XGBoost algorithm with a negative pressure wave provided in the first aspect of the present application are implemented, specifically comprising the following modules:

[0036] 1) SCADA system data acquisition and temporary storage module, used to obtain data signals such as pressure, temperature and flow from the SCADA system and save the data;

[0037] 2) Data cleaning and preprocessing module, used to remove abnormal values ​​and missing values ​​contained in the signal, and use interpolation method to supplement the missing values, and calculate the pressure drop rate time series data;

[0038] 3) An XGBoost model calling module, used to import the pressure drop rate time series signal into the pre-trained XGBoost model to predict the leakage position and the initial value of the leakage aperture;

[0039] 4) A leakage position calculation module based on negative pressure waves: used to calculate the optimized leakage position according to the calculation method provided in the first aspect of the present invention, combined with the basic parameters of the pipeline and the data obtained by the SCADA system.

[0040] According to the above aspects and any possible implementation, an implementation is further provided, wherein the pre-training of the XGBoost model comprises:

[0041] The training data is divided into a training set and a test set, and the pressure drop rate time series data is passed to the input layer of the XGBoost model;

[0042] Performing data preprocessing according to the input layer of the XGBoost model to form time series samples, and extracting feature information of the sequence samples at different time scales;

[0043] The pre-constructed XGBoost model is trained according to the characteristic information of the pressure drop rate samples of the time series at different time scales to obtain the linear or nonlinear relationship between the time series of different pressure drop rate data and the leakage aperture and leakage position.

[0044] As described above and in any possible implementation, the present invention can achieve the following beneficial effects by adopting the above technical solutions:

[0045] 1) The present invention introduces a pre-trained XGBoost model to read the SCADA system time series to predict the leakage location and leakage aperture, which solves the problem that the leakage signal of the gas pipeline is difficult to capture and judge when a leakage occurs due to the compressibility of the transmission medium;

[0046] 2) The present invention predicts the leakage position and leakage aperture based on the XGBoost model, and performs an optimization calculation based on the negative pressure wave method on the predicted leakage position, thereby improving the leakage accuracy of the positioning method;

[0047] 3) The present invention provides a gas pipeline leakage location method based on the XGBoost model combined with negative pressure wave calculation, which can obtain data from the SCADA system and perform leakage location online. The calculation process is clear and the operability is strong, which is convenient for people in this technical field to use. It should be understood that the content described in the Summary of the Invention section is not intended to limit the key or important features of the embodiments of the present application, nor is it used to limit the scope of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 A flow chart of a calculation method for locating gas pipeline leaks by combining an XGBoost algorithm and negative pressure waves provided by the present invention.

[0049] Figure 2 A schematic diagram of the arrangement of valve chambers and leakage locations of a gas transmission trunk line according to an embodiment of the present invention.

[0050] Figure 3 A time series diagram of two pressure drop rates of an embodiment provided by the present invention.

[0051] Figure 4 A schematic diagram of the pressure along the route obtained by calculation according to an embodiment of the present invention.

[0052] Figure 5 This is a diagram illustrating the composition of a computing system for locating gas pipeline leaks by combining an XGBoost algorithm with negative pressure waves according to an embodiment of the present invention. DETAILED DESCRIPTION

[0053] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention is further described below in conjunction with the accompanying drawings, but the present invention is not limited to the following embodiments.

[0054] A calculation method for locating gas pipeline leaks by combining the XGBoost algorithm with negative pressure waves. First, data is obtained from the SCADA system and preprocessed; second, the data is input into the pre-trained XGBoost model to obtain the predicted leak location and leak aperture; finally, the accurate leak location and leak aperture are calculated by the negative pressure wave method. This method improves the safety of pipeline operation and the timeliness of emergency repairs. The specific positioning methods include:

[0055] 1) Read the pressure and flow data in the SCADA system and pre-process the data:

[0056] This part mainly includes the acquisition of SCADA system data, the calculation of the pressure drop rate data and the processing of outliers and missing values ​​in the SCADA system so that the data can be recognized by the pre-trained XGBoost model.

[0057] 2) Input the time series pressure drop rate signal into the pre-trained XGBoost model to obtain the predicted leak location and leak aperture:

[0058] Specifically, XGBoost (Extreme Gradient Boosting) is a machine learning model based on gradient boosting trees. It improves prediction accuracy by gradually building a collection of decision trees. Its advantages include high computational efficiency, control of overfitting, and excellent prediction performance for both scattered and continuous data.

[0059] The XGBoost model is pre-trained with existing controlled leakage simulation data, and has good performance in predicting the leakage diameter and leakage location based on the pipeline pressure drop rate.

[0060] 3) Based on the leak location and leak aperture predicted by the XGBoost model and the flow rate and other data obtained by the SCADA system, calculate the flow rate, pressure and other data of the pipeline leak point:

[0061] Specifically, the flow rate and pressure at the leakage point are calculated according to the predicted leakage position and leakage aperture, and the calculation formula is:

[0062] Q v =0.00078PD 2 (13)

[0063]

[0064]

[0065]

[0066] Where Q v ——Gas leakage rate, m 3 / s; D——effective hole diameter, mm; P——pipeline operating pressure, MPa;

[0067] p 0,1 ——Pressure of upstream calculated leakage point, MPa; p 0,2 ——Calculated pressure at downstream leakage point, MPa; p0——Calculated pressure at leakage point, MPa; ——Outlet pressure of upstream station, MPa; ——Inlet pressure of downstream station, MPa; L0——Distance from leakage location to upstream station, km; L——Length of pipe section, km;

[0068] 4) Based on the flow rate and pressure at the leak point and the pressure, flow rate and other data at the beginning and end of the pipeline, calculate the pressure distribution along the pipeline:

[0069] Specifically, the pressure distribution along the upstream and downstream pipelines of the leakage point in the pipeline is calculated based on the flow rate and pressure data of the leakage point. The calculation formula is:

[0070]

[0071]

[0072] Where p x,1 ——Pressure along the pipe section upstream of the leakage point, MPa; p x,2 ——pressure along the pipe section downstream of the leakage point, MPa; x——location of each point along the line, km; ——Inlet pressure of downstream station, MPa;

[0073] 5) Divide the pipeline into n sub-segments, and calculate the flow velocity and pressure wave propagation velocity of each sub-segment respectively. Specifically, the calculation relationship between the flow velocity and pressure wave propagation velocity of each sub-segment is as follows:

[0074]

[0075]

[0076] Where u(x)——gas velocity, m / s; Q *,i ——gas mass flow rate, kg / s; i——pipeline section number; R——gas constant, J / (kg·K); T0——pipeline ground temperature, k; A——pipeline cross-sectional area, m2 ; ——pressure at the starting point of the ith sub-pipe section, Pa; ——The pressure at the end point of the ith sub-pipeline section, Pa;

[0077] 6) According to the above method, the time difference of obtaining the leakage signal of the valve chambers or stations at both ends is calculated, and the exact location of the leakage is further calculated. The specific calculation relationship is as follows:

[0078]

[0079]

[0080] Where Δt is the time difference, s; L0′ is the new leakage position obtained after optimization; v s ,v x are the average pressure propagation velocities upstream and downstream, m / s, respectively, and the calculation formula is:

[0081]

[0082]

[0083] Where i is the number of pipe sections divided upstream of the pipeline; n is the number of pipe sections divided downstream of the pipeline.

[0084] Embodiment 1 Since it is difficult to directly obtain the real-time pressure change data of pipeline leakage and there is a lack of measured data for verification, this embodiment uses simulation data that does not participate in the XGBoost model training and adds white noise as an example to demonstrate the calculation method. Figure 2 As shown in the figure, a 200mm diameter leak occurs at a location 12.65km away from the upstream valve chamber on the pipe section from valve chamber 2 to valve chamber 3 in a pipeline model. The simulated pressure drop rate timing signals of the upstream valve chamber 2 and the downstream valve chamber 3 are calculated as follows: Figure 3 As shown, the time series signal of the pressure drop rate is input into the XGBoost model for prediction, and the predicted leakage position and leakage aperture are obtained. The prediction results are shown in Table 1 below:

[0085] Table 1 Predicted leakage position and leakage diameter and calculated average value

[0086]

[0087] According to the average prediction results, the flow rate at the leak and the pressure at the leak point are calculated. According to formula (13), combined with the pressure equivalent obtained by the SCADA system, the flow rate at the leak point is calculated to be 0.7943m 3 / s, and the flow rate at the leakage point is introduced into equations (14) to (16) to calculate the pressure at the leakage point to be 8.07MPa.

[0088] Based on the calculated flow rate and pressure at the leakage point, the pressure distribution along the upstream and downstream of the leakage point is calculated according to equations (17) and (18). The pressure distribution along the pipeline section after the leakage occurs is as follows: Figure 4 shown.

[0089] According to the calculation results of the pressure distribution along the pipe section and the temperature and other data obtained by the SCADA system, the pipe section is divided into fifty upstream and downstream sub-sections. The pipeline flow velocity and pressure wave propagation velocity of each sub-section are calculated according to formula (19) and formula (20). It is calculated that the gas flow velocity of each sub-section is in the range of 1.73 to 1.84 m / s, and the pressure wave propagation velocity of each sub-section is in the range of 464.98 to 465 m / s.

[0090] According to the flow velocity and pressure wave propagation speed of each pipe section, the time difference of the valve chambers at both ends receiving the leakage signal is calculated according to formula (21), and the time difference of the leakage signal is calculated to be 0.2644s.

[0091] According to formula (23) to formula (24), the pressure propagation speed upstream and downstream of the leakage point is calculated, which is 463.21 m / s for upstream leakage and 466.78 m / s for downstream leakage. By introducing the time difference of the leakage signal, the exact position of the leakage point can be calculated as 12.63 km according to formula (22).

[0092] The leakage position of the natural gas pipeline is calculated by the above method. Compared with the leakage positioning by the XGBoost model alone, the calculation method is used to further optimize the leakage positioning accuracy. It can be found that the leakage positioning accuracy is improved from 99.13% to 99.84%. Therefore, the use of this method can further improve the speed of natural gas pipeline emergency repair, thereby ensuring the operation and safety of the pipeline.

[0093] The above is an introduction to the method embodiment. The following is a further explanation of the scheme described in this application through an apparatus embodiment. Figure 5 A composition diagram of a computing system for locating gas pipeline leakage by combining an XGBoost algorithm with a negative pressure wave according to an embodiment of the present application is shown, Figure 5 As shown, the system includes: a processor and a memory, wherein the memory stores computer program instructions and can temporarily store SCADA system imported data, and when the computer program instructions are executed by the processor, the steps of the calculation method for combining the XGBoost algorithm and the negative pressure wave to locate the gas pipeline leakage provided in the first aspect of the present application are implemented, specifically including the following modules:

[0094] SCADA system data acquisition and temporary storage module, used to obtain data signals such as pressure, temperature and flow from the SCADA system and temporarily save the data;

[0095] The data cleaning and preprocessing module is used to remove abnormal values ​​and missing values ​​contained in the signal, and use the interpolation method to supplement the missing values, and calculate the pressure drop rate time series data;

[0096] An XGBoost model calling module is used to import the pressure drop rate time series signal into the pre-trained XGBoost model to predict the leakage position and the initial value of the leakage aperture;

[0097] The leakage position calculation module based on negative pressure waves is used to calculate the optimized leakage position according to the calculation method provided by the first aspect of the present invention, combined with the basic parameters of the pipeline and the data obtained by the SCADA system.

[0098] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for locating gas pipeline leakage by combining XGBoost algorithm and negative pressure wave, characterized in that: The following steps are involved: 1) Collect pressure, flow and other signals from the SCADA system, and temporarily store and pre-process the signals to form a time series signal of the pressure drop rate; 2) Input the time series signal into the pre-trained XGBoost model to predict the leakage position and leakage aperture value of the upstream valve chamber and the downstream valve chamber respectively. The predicted leakage aperture and position are the average values ​​of the upstream and downstream valve chambers. 3) According to the leakage position and leakage aperture predicted by the XGBoost model, the flow rate at the leakage point is calculated. The specific relationship is: Q v =0.00078PD 2 (1) Where Q v ——Gas leakage rate, m 3 / s; D——effective hole diameter, mm; P——pipeline operating pressure, MPa; p 0,1 ——Pressure upstream of the leakage point, MPa; p 0,2 ——pressure downstream of the leakage point, MPa; p0——calculated pressure at the leakage point, MPa; ——Outlet pressure of upstream station, MPa; ——Inlet pressure of downstream station, MPa; L0——Distance from leakage location to upstream station, km; L——Length of pipe section, km; 4) According to the leakage point flow and leakage point pressure, calculate the pressure distribution along the way, and the specific relationship is: Where p x,1 ——Pressure along the pipe section upstream of the leakage point, MPa; p x,2 ——pressure along the pipe section downstream of the leakage point, MPa; x——location of each point along the line, km; ——Inlet pressure of downstream station, MPa; 5) Divide the pipeline into n sub-segments according to the length and height difference of the pipeline, and calculate the flow velocity and pressure wave propagation velocity of the gas in the pipeline segment by segment. Where u(x)——gas velocity, m / s; Q *,i ——gas mass flow rate, kg / s; i——pipeline section number; R——gas constant, J / (kg·K); T0——pipeline ground temperature, k; A——pipeline cross-sectional area, m 2 ; ——pressure at the starting point of the ith sub-pipe section, Pa; ——The pressure at the end of the ith sub-pipeline, Pa; v i (x)——pressure wave propagation velocity, m / s; k——specific heat ratio; D——pipe inner diameter, m; E——pipe elastic modulus, Pa; e——pipe wall thickness, m; 6) According to the flow velocity and pressure wave propagation speed along the way, calculate the time difference between the front and rear stations, and get the accurate leakage location. The specific relationship is: Where Δt is the time difference, s; L0′ is the new leakage position obtained after optimization, km; v s , v x are the average pressure propagation velocities upstream and downstream, m / s, respectively. The calculation formula is: Where i is the number of pipe sections divided upstream of the pipeline; n is the number of pipe sections divided downstream of the pipeline.

2. The method for locating gas pipeline leakage by combining XGBoost algorithm and negative pressure wave as claimed in claim 1, characterized in that: Predict the time series data of pressure drop rate based on the pre-trained XGBoost model, including: The pre-trained XGBoost model is called, and the pressure drop rate time series data obtained from the SCADA system is used as the XGBoost model input value, and the predicted leakage location and predicted leakage aperture are obtained as output values ​​and output.

3. A system for locating gas pipeline leaks by combining XGBoost algorithm and negative pressure wave, the system comprising: A processor and a memory, wherein the memory stores computer program instructions and can temporarily store data imported from a SCADA system, and when the computer program instructions are executed by the processor, the steps of a method for locating gas pipeline leaks by combining an XGBoost algorithm with negative pressure waves provided in the first aspect of the present application are implemented.

Images