A method for identifying plug flow in a pipeline based on a box-cox transformation

Abstract

The application discloses a pipeline plug flow identification method based on BOX-COX transformation, comprising the following steps: obtaining a plug flow pressure signal in a pipeline and pre-processing the plug flow pressure signal; analyzing time domain characteristics of the plug flow pressure signal and calculating pipeline characteristics in the time domain, wherein the pipeline characteristics include kurtosis, shape factor, root mean square, peak value, mean value, pulse factor and margin factor; fusing the pipeline characteristics by using a factor analysis method; and using BOX-COX transformation to Gaussianize the pipeline characteristics, using a 3sigma function to determine upper and lower boundary lines of the plug flow in the pipeline, and judging whether the plug flow appears in the pipeline or not. The application can diagnose the liquid accumulation state in the pipeline under the condition that the pipeline transportation is normally working, without stopping the system for detection. Meanwhile, the application can avoid artificial and objective plug flow stage division, and has the characteristics of simplicity, objectivity and convenience.

Description

Technical Field

[0001] This invention belongs to the field of pipeline transportation technology and relates to the problem of slug flow identification in pipes, specifically a method for identifying slug flow in pipes based on BOX-COX transformation. Background Technology

[0002] Subsea pipeline transportation technology is widely used in marine engineering. Underwater gas pipeline transportation is crucial for energy transmission, directly impacting the safety and reliability of the system. When compressed air, natural gas, and hydrogen enter the underwater pipeline after pressure regulation and drying at a ground station, the temperature decreases with increasing water depth. Heat exchange between the pipe wall and seawater causes the gas temperature inside the pipe to gradually drop. Upon reaching the pressure dew point, water vapor condenses and precipitates, forming liquid inside the pipeline due to gravity, especially in low-lying areas. This transforms the flow inside the pipeline from a single-phase gaseous flow into a complex and potentially hazardous two-phase gas-liquid flow.

[0003] Due to the turbulent characteristics of fluid flow and the complex interphase interactions of gas-liquid two-phase flow, various complex changes in flow patterns and fluctuations in fluid pressure occur, leading to a series of hazards. First, liquid accumulation inevitably reduces the pipeline's flow area, decreases transport efficiency, and increases pressure drop and energy consumption. Second, the presence of water exacerbates corrosion within the gas pipeline and poses a risk of water hammer. Simultaneously, pressure and flow fluctuations induce pipeline vibration and surge in pressure regulating equipment, which in turn affects the two-phase flow within the pipe, further exacerbating pressure and flow fluctuations. Failure to clean the accumulated liquid will severely impact the safe and efficient operation of the underwater gas transmission system, potentially even causing accidents. Therefore, it is essential to remove accumulated liquid from gas pipelines through cleaning operations. Actual cleaning operations rely primarily on experience or are conducted periodically, but indiscriminate and frequent cleaning not only increases system operating costs but also increases the frequency and duration of system shutdowns. The best solution to this problem is to accurately predict the state of the accumulated liquid before it causes deterioration or failure in the system's gas transmission performance, thereby determining the cleaning time and plan.

[0004] As is well known, temperature decreases with increasing water depth. Moisture in compressed gas condenses and accumulates at the lowest point of the pipeline, forming slug flow and affecting system operation. Slug flow has three main causes: laminar flow, gas expansion, and topographical factors. For slug flow identification, indirect measurement methods are more practical in engineering and establishing a relationship between slug flow and flow parameters provides a more meaningful explanation of the mechanism. Indirect measurement methods mainly include three processes: characteristic signal acquisition, feature extraction, and identification. Pressure fluctuation signals generated by changes in slug flow can be analyzed through remote measurement to identify slug flow within the pipeline. Summary of the Invention

[0005] To address the problems existing in the prior art, this invention discloses a method for identifying slug flow in a pipeline based on BOX-COX transformation, specifically including the following steps:

[0006] Acquire the slug flow pressure signal inside the pipeline and preprocess it;

[0007] The time-domain characteristics of the slug flow pressure signal are analyzed, and the pipe characteristics in the time domain are calculated, including kurtosis, shape factor, root mean square, peak value, mean value, impulse factor, and margin factor.

[0008] Factor analysis was used to integrate pipeline characteristics;

[0009] By utilizing the Gaussianization of pipe characteristics through BOX-COX transformation, the upper and lower boundary lines of slug flow in the pipe are determined by the 3sigma function to determine whether slug flow occurs in the pipe.

[0010] When using factor analysis to fuse pipeline characteristics:

[0011] Let P be the pipe characteristic that characterizes slug flow. j , Each P j Contains a unique factor U j (i = 1, L, r), each P j Given F1, F2, L, F7 (r≥7) and U j (i = 1, ..., r) can be linearly represented as:

[0012]

[0013]

[0014] In the formula, i = 1, 2, ... r; j = 1, 2, ..., 7, a ij λ represents the factor loading coefficients. j and r j They are respectively the Fth j The corresponding eigenvalues ​​and variance contribution rates, f ij Let g be the coefficient of the decision matrix for the i-th feature and the j-th feature. i The weights of feature i;

[0015] The merged pipeline characteristics P j Perform a box-cox conversion:

[0016] Pipe characteristics P j The one-step increment is expressed as:

[0017] ΔP j (t)=P j (t)-P j(t-1) (12)

[0018] Pipe characteristic increment ΔP j The normalized formula for (t) is:

[0019]

[0020] The normalized pipeline characteristic increment, ΔP jmax and ΔP jmin These are the maximum and minimum values ​​of the pipeline characteristic increment;

[0021] The Gaussianization derivation is as follows:

[0022]

[0023] This is the Gaussianized result of the normalized pipeline characteristic increment, where 'a' is a positive number used to ensure... If positive, ξ can be solved using the maximum likelihood estimation method during the Gaussianization of pipeline characteristic increments, i.e.,

[0024]

[0025] Where ξ is lnL max (ξ) is the solution value corresponding to the maximum value;

[0026] After Gaussian processing, the pipeline characteristic increments Follows a normal distribution N(μ,σ) 2 ), while in the normal distribution μ and σ 2 The mean and standard deviation are given, and the maximum likelihood function is used to estimate the solution.

[0027]

[0028] When using the 3sigma function to determine the upper and lower boundary lines of slug flow in a pipe, and to determine whether slug flow has occurred in the pipe:

[0029] Based on the properties of the normal distribution, the upper and lower boundary lines of slug flow in a pipe can be represented as:

[0030]

[0031] Set a trigger mechanism so that the first point exceeding the boundary line is automatically identified as the occurrence of a slug flow:

[0032]

[0033] The specific method for determining the occurrence of slug flow in the pipeline based on the set upper and lower boundary lines is as follows: Increment the Gaussianized pipeline characteristic... The first point beyond the upper and lower boundary lines is considered the occurrence of slug flow.

[0034] By employing the above technical solution, this invention provides a method for identifying slug flow in pipelines based on Box-Cox transform. This method involves arranging multiple pressure sensors along the pipeline to obtain pressure signals indicating the onset of slug flow. The liquid accumulation volume is selected as the abscissa (ml), and pressure fluctuation as the ordinate (kPa) to obtain a time-domain characteristic graph of the pressure signal. Factor analysis is used to analyze seven features and obtain their weighting factors, which are then fused. The fused features are Gaussianized using Box-Cox transform, and 3 sigma is set as the upper and lower bound thresholds. A triggering mechanism is established, and the first point exceeding the threshold is identified as the occurrence of slug flow. This invention can diagnose the liquid accumulation state in the pipeline while it is operating normally without stopping the system for detection. Furthermore, this invention avoids the need for manual, objective slug flow stage division, and is simple, objective, and convenient. Attached Figure Description

[0035] To more clearly illustrate the technical solutions in the embodiments of this application 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 some embodiments recorded in this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0036] Figure 1 This is a diagram illustrating the process of liquid accumulation inside a pipe, provided by the present invention.

[0037] Figure 2 This is a flowchart of the pipe slug flow identification method based on BOX-COX transformation provided by the present invention;

[0038] Figure 3 This is a feature signal calculation diagram obtained in the embodiments of the present invention;

[0039] Figure 4 is an adaptive recognition diagram of Gaussianization features in an embodiment of the present invention. Detailed Implementation

[0040] To make the technical solutions and advantages of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention:

[0041] This embodiment presents a method for identifying slug flow in a pipeline based on the BOX-COX transformation, with appended... Figure 1 A diagram illustrating the process of liquid accumulation inside a pipe; Figure 2 The flowchart of the pipe slug flow identification method based on BOX-COX transformation includes the following steps:

[0042] Step 1: Acquisition and preprocessing of pressure wave signals;

[0043] Multiple pressure sensors are arranged on the pipeline, with the arrangement based on integer multiples of the pipe diameter ratio. The required number of sensors is determined by the pipe length and the spacing between the sensors. Liquid tends to accumulate in low-lying areas of the pipeline; pressure sensors are placed on both sides of these low-lying areas. After the sensors are set up, pressure signals are tested and acquired using LabVIEW. When there is no liquid in the pipe, the pressure signal is stable with no significant fluctuations. Subsequently, pressure signals can be acquired under conditions of liquid accumulation and slug flow.

[0044] Specifically, the compressed air for the experiment was supplied to a 1.5 m³ AtlasCopco air tank via an AtlasCopco air compressor, and an AtlasCopco purifier was used to filter the compressed air entering the pipeline. Pressure was regulated using an IR3020-03 pressure reducing valve with a range of 0.01–0.8 MPa and a sensitivity of 0.2%, and flow rate was measured using a FESTO-565406 sensor (range 0–1000 L / min) with an accuracy of ±0.3%. The intake velocity was regulated by an AS4000-03 sensor with a range of 0–1670 L / min. Pressure measurement was performed using an MPM489 sensor (range 0–1.6 MPa) with an accuracy of ±0.5%.

[0045] Step 2: Obtain the signal characteristics that characterize slug flow;

[0046] The pressure signal of the slug flow in the pipe from zero to its origin was acquired, and seven characteristics of the time-domain signal were calculated, as shown in the attached figure. Figure 3 As shown, seven time-domain characteristics of slug flow are calculated. The horizontal axis represents the volume of the accumulated fluid (ml), and the vertical axis represents the pressure value (kPa).

[0047] Step 3: Feature fusion;

[0048] The features calculated in step 2 are then fused. Factor analysis is used to determine the weights of each feature. The selected features are imported into MINITAB software for further processing to obtain the weights of the fused features.

[0049] Step 4: Gaussianize the features using the Box-Cox transform;

[0050] The features fused in step 3 are transformed using a Box-Cox transformation to conform to a Gaussian distribution. Gaussianization of the fused features is then programmed using MATLAB software, with a threshold of 3 sigma set for adaptive slug flow identification, as shown in the attached figure. Figure 4a As shown in -4e. The threshold is set as follows:

[0051]

[0052] Step 5: Set up a trigger mechanism so that the first point exceeding the threshold is automatically identified as the occurrence of slug flow.

[0053]

[0054] After extracting features from the pressure fluctuation signals in this experiment, the signals are fused and then subjected to BOX-COX transformation. With the threshold set, when the pressure signal of the slug flow exceeds the threshold, it indicates that a slug flow has occurred in the pipeline.

[0055] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A method for identifying plug flow in a pipeline based on a BOX-COX transformation, characterized in that Comprising: acquiring and preprocessing the plug flow pressure signal in the pipeline; analyzing the time domain characteristics of the plug flow pressure signal to calculate the pipeline characteristics in the time domain, wherein the pipeline characteristics include kurtosis, shape factor, root mean square, peak value, mean value, pulse factor and margin factor; fusing the pipeline characteristics by using factor analysis method; gaussing the pipeline characteristics by using BOX-COX transformation, determining the upper and lower boundary lines of the plug flow in the pipeline by using 3 sigma function, and judging whether the plug flow appears in the pipeline; The fused pipe characteristics P j Perform a BOX-COX transformation: Piping characteristics P j One step increment is represented as: ΔP j (t) = P j (t) - P j (t - 1) Pipe property increment ΔP j The normalized calculation of (t) is: ΔP is the normalized pipe property increment jmax and ΔP jmin are the maximum and minimum values of the pipe property increment; The Gaussianization derivation is as follows: is the result of the Gaussianization of the normalized pipe characteristics increment, where a is a positive number used to ensure is positive, in the process of Gaussianization of the pipe characteristics increment, ξ can be solved by the maximum likelihood estimation method, that is, where ξ is lnL max (ξ) is the solution value corresponding to the maximum After Gaussian processing, the pipeline characteristic increments Follows a normal distribution N(μ,σ) 2 ), while in the normal distribution μ and σ 2 The mean and standard deviation are given, and the maximum likelihood function is used to estimate the solution. when the 3 sigma function is used to determine the upper and lower boundary lines of the plug flow in the pipeline to judge whether the plug flow appears in the pipeline: according to the properties of normal distribution, the upper and lower boundary lines of the plug flow in the pipeline are expressed as: setting a trigger mechanism, the first point exceeding the boundary line is automatically identified as the appearance of the plug flow: The specific method for determining the occurrence of the plug flow in the pipeline according to the set upper and lower boundary lines is: taking the Gaussized pipeline characteristic increment The first point exceeding the upper and lower boundary lines is determined as the occurrence of the plug flow.

2. The method of claim 1, wherein the BOX-COX transformation is used to identify the plug flow in the pipeline. when the factor analysis method is used to fuse the pipeline characteristics: Let P represent the pipe characteristics of the plug flow j , Each P j contains a unique factor U j (i = 1,..., r), each P j is linearly represented by F1, F2,..., F7 (r > 7) and U j (i = 1,..., r) In the formula, i = 1, 2, ..., r; j = 1, 2, ..., 7, a ij λ represents the factor loading coefficients. j and r j They are respectively the Fth j The corresponding eigenvalues ​​and variance contribution rates, f ij Let g be the coefficient of the decision matrix for the i-th feature and the j-th feature. i The weights are for feature i.

Images