Air compression station pipeline leakage detection method based on dynamic pressure and multi-branch pipe pressure

Abstract

The invention discloses an air compression station pipeline leakage detection method based on dynamic pressure and multi-branch-pipe pressure, and belongs to the technical field of air compression station intelligent control and industrial internet of things. The unified pressure reference is corrected through the air state, and the air density and the dynamic viscosity are obtained; on-way resistance and local resistance of the pipeline are dynamically calculated based on the fluid mechanics principle, and a normal pressure drop prediction model is constructed in combination with a support vector regression or multi-layer perceptron algorithm; through deviation analysis of real-time pressure drop and predicted pressure drop and multi-branch pipe topology cooperation, accurate judgment and positioning of leakage are realized; and the recursive least square method is adopted to adaptively update the parameters of the resistance model to cope with the influences of pipeline aging, working condition change and the like, and the problems that a traditional detection method is single in dimension, static in model, lack in positioning capability and insufficient in adaptability are solved.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of intelligent control of air compression stations and industrial Internet of Things, and particularly relates to an air compression station pipeline leakage detection method based on dynamic pressure and multi-branch pipe pressure. BACKGROUND

[0002] In the industrial production scene of an air compression station, compressed air is delivered to various air-consuming terminals through a complex pipeline network, and pipeline leakage is one of the core causes of energy waste and supply pressure fluctuation.

[0003] The existing technology has the following key defects in detecting pipeline leakage:

[0004] 1. Single detection dimension: only relying on the static comparison of the outlet pressure of the station house mother pipe or the end pressure of a single branch pipe, the difference between system overall pressure fluctuation and local leakage cannot be distinguished, and the misjudgment rate is more than 40%;

[0005] 2. Static resistance model: fixed resistance parameters in the design stage are used, and the dynamic influence of factors such as pipeline aging, dust accumulation and scaling on resistance is not considered, resulting in a leakage determination deviation of more than 25%;

[0006] 3. Lack of positioning ability: lacking the cooperative analysis logic of multi-branch pipe pressure data, the leakage branch pipe or even the leakage point cannot be accurately positioned, and the operation and maintenance efficiency is low;

[0007] 4. Insufficient adaptability: facing changes in working conditions (such as air compressor start-stop and sudden change in air consumption), the model parameters cannot be updated in real time, and the long-term detection accuracy continues to decline. SUMMARY

[0008] To solve the problems raised in the background art, the application provides an air compression station pipeline leakage detection method based on dynamic pressure and multi-branch pipe pressure, to solve the problems of single detection dimension, static resistance model, lack of positioning ability and insufficient adaptability of traditional air compression station.

[0009] To achieve the above purpose, the application provides the following technical scheme:

[0010] The air compression station pipeline leakage detection method based on dynamic pressure and multi-branch pipe pressure comprises the following steps:

[0011] S1: data acquisition; real-time acquisition of the gauge pressure P start (t) of the outlet of the station house mother pipe, the environment temperature T ambient (t), the atmospheric pressure P atm , the end gauge pressure P end,j (t) of each branch pipe, and the real-time volume flow Q j (t), wherein j=1, 2, …, n, and n is the total number of branch pipes;

[0012] S2: Air state correction; convert the gage pressure P start (t) to absolute pressure P end,j (t) to absolute pressure P abs,start (t) and P abs,j (t), and combine the ideal gas state equation and T ambient (t) to calculate the air density , while obtaining the air dynamic viscosity μ(t) in real time;

[0013] S3: Dynamic calculation of pipeline resistance; based on the specifications and measured parameters of each branch pipe, local resistance coefficient ξ j , air density , and air dynamic viscosity μ(t), calculate the Reynolds number and the frictional resistance coefficient, and calculate the total resistance of the branch pipe according to the frictional resistance coefficient and the local resistance coefficient ;

[0014] S4: Normal pressure drop benchmark modeling; collect Q j (t k ), ρ(t k ), and △P hist,J (t k ) data under the preset group of normal operating conditions, where is the sampling time, , use support vector regression SVR or multilayer perceptron MLP algorithm, take Q j (t k ) and ρ(t k ) as input, as output, minimize the mean square error MSE to complete training, and obtain the normal pressure drop prediction model of the branch pipe j;

[0015] S5: Leak detection and positioning; by calculating the deviation between the prediction result of the pressure drop prediction model and the real-time pressure drop at the current time, make a logical judgment of the deviation for the leakage;

[0016] S6: Dynamic model updating; to cope with the resistance parameter drift caused by pipeline aging and dust accumulation, use recursive least squares RLS to realize adaptive updating of the total resistance .

[0017] Compared with the prior art, the beneficial effects of the present application are:

[0018] This application constructs a complete technical system encompassing data acquisition, air condition correction, dynamic calculation of pipeline resistance, normal pressure drop benchmark modeling, leak detection and location, and dynamic model updating. Through an innovative architecture that utilizes multi-dimensional data input for dynamic pressure monitoring, multi-branch collaborative analysis, and adaptive updating of the resistance model, it achieves accurate detection and second-level location of pipeline leaks, solving the problems of traditional air compressor stations, such as limited detection dimensions, static resistance models, lack of location capabilities, and insufficient adaptability. Attached Figure Description

[0019] Figure 1 This is a flowchart illustrating the process of this application. Detailed Implementation

[0020] To facilitate understanding of the technical content of this invention by those skilled in the art, the invention will be further described in detail below with reference to the accompanying drawings and specific examples. It should be understood that the specific examples described herein are merely illustrative and not intended to limit the scope of the invention.

[0021] Example 1

[0022] like Figure 1 As shown, the method for detecting pipeline leaks in an air compressor station based on dynamic pressure and multi-branch pressure includes the following steps:

[0023] S1: Data Acquisition; Real-time acquisition of 7 types of industrial field data, including:

[0024] Station building main pipe outlet: gauge pressure P start (t), unit bar; ambient temperature T ambient (t), unit: °C; atmospheric pressure P atm The unit is bar;

[0025] Branch pipe j: End gauge pressure P end,j (t), unit bar; real-time volumetric flow rate Q j (t), unit Nm 3 / min; j=1,2,…,n, where n is the total number of branches;

[0026] Air compressor cluster: Operating status (start / stop signals, frequency, load rate); Discharge pressure P exh (t).

[0027] S2: Air State Correction; To eliminate the interference of temperature and pressure on the physical properties of air, a three-dimensional state correction is performed:

[0028] S2.1: Pressure Standardization: Standardize the gauge pressure P of the main pipe and each branch pipe. start (t), P end,j (t) is converted to absolute pressure P abs,start (t) and P abs,j(t), specifically:

[0029] Absolute pressure of main pipe: ;

[0030] Absolute pressure of branch pipe: ;

[0031] S2.2: Air density calculation: Based on the ideal gas law (PV=nRT), combined with the thermodynamic temperature T(t)=273.15+T ambient (t), calculate air density ;

[0032] Where R is the gas constant, and the air density in the pipe is the same, so the density calculation formulas for the main pipe and the branch pipe are equivalent. R=287J / (mol·K) is the air gas constant.

[0033] S2.3: Aerodynamic viscosity correction: Obtain the aerodynamic viscosity μ(t) in real time according to the Sutherland formula or industrial calibration table, in units of... Its nonlinear relationship with T(t) is as follows:

[0034] ;

[0035] In the formula The viscosity at T0 = 273.15 K. S = 110.4K is Sutherland's constant.

[0036] S3: Dynamic calculation of pipeline resistance; for a single branch pipe, based on the principles of fluid mechanics, the friction resistance and local resistance are dynamically calculated parameter by parameter, including:

[0037] S3.1: Branch pipe parameter pre-configuration: Enter design / calibration parameters including pipe diameter d j Unit: mm; Length: L j Unit: m; Local drag coefficient ξ j (Edges, valves, tees, etc.); Pipe wall roughness k j The unit is mm. Use the default value or measure it using a laser thickness gauge.

[0038] S3.2: Cross-sectional area of ​​the branch pipe Flow rate ;

[0039] Reynolds number ,when When the temperature is less than 2300, it is considered a laminar flow condition. A value ≥2300 indicates a turbulent operating condition.

[0040] S3.3: When the flow is judged to be laminar: friction coefficient (Direct derivation of Poiseuille's law);

[0041] When the flow is determined to be turbulent: the Colebrook formula is used for iterative solution (applicable to turbulent flow in rough pipes):

[0042] ;

[0043] The formula converges rapidly using the Newton-Raphson iteration method, with the initial value taken as... Iteration accuracy ;

[0044] S3.4: Total Resistance The resistance model is expressed as:

[0045] ;

[0046] In the formula m y Let be the number of local resistance elements in the y-th segment of the branch pipe. Let be the local resistance coefficient of the z-th element.

[0047] S4: During the system's stable operation window (which must meet the following conditions: ① no start / stop of the air compressor / load rate fluctuation <5%; ② no manual valve operation; ③ continuous duration ≥30 minutes), construct a nonlinear mapping model of "flow rate-density-pressure drop":

[0048] S4.1: Collect at least 100 sets of Q data under normal operating conditions. j (t k ), ρ(t) k ) and △P hist,J (t k ) data, among which Sampling time, ;

[0049] S4.2: Employ Support Vector Regression (SVR) or Multilayer Perceptron (MLP) algorithms, using Q... j (t k ) and ρ(t k ) is the input. As the output, the mean square error (MSE) is minimized to complete the training and obtain the normal pressure drop prediction model for branch j.

[0050] ;

[0051] in To predict pressure drop, As basis functions, For model weights, if SVR is used, a quadratic polynomial kernel is selected; if MLP is used, the ReLU activation function is selected to ensure fitting accuracy.

[0052] S5: Leak Detection and Location; Intelligent leak detection and location are achieved through pressure drop deviation analysis and multi-branch topology collaboration. Specifically:

[0053] Real-time pressure drop calculation: Pressure drop at the current moment ;

[0054] Normal pressure drop prediction: Substitute the current Q j (t), From the normal pressure drop prediction model, the predicted pressure drop is obtained. ;

[0055] Pressure drop deviation quantification: ;

[0056] The specific logic for determining leakage is as follows:

[0057] Set statistical threshold , Set a duration threshold for the standard deviation of the pressure drop deviation during the steady-state period. ,

[0058] like And duration ≥ If so, it is determined that there is a leak in branch pipe j;

[0059] Multi-branch coordinated positioning includes:

[0060] The system combines a pre-stored pipeline topology diagram containing the hierarchy of main pipes and branches, and the connection relationships between branches, to make a judgment. If two or more adjacent branches trigger a leak simultaneously, the leak is primarily identified as a leak in the main pipe section. If only a single branch triggers the leak, the location is determined to be the section from the beginning to the end of that branch.

[0061] S6: Dynamic model update; To address the drift in resistance parameters caused by pipe aging and ash accumulation, the Recursive Least Squares (RLS) method is used to realize the total resistance. The adaptive update is as follows:

[0062] Data filtering: Extract 50 sets of stable operating condition data (air compressor load rate fluctuation <3%, no valve action) from real-time data every hour.

[0063] Parameter update: For the observed values, For the parameters to be estimated, the prediction error is minimized using the RLS algorithm to update the parameters of the resistance model;

[0064] Adaptive update cycle: If a branch pipe has been leaked more than 5 times in the past 24 hours, the update cycle will be automatically shortened to 30 minutes (the default cycle is 2 hours).

[0065] In this embodiment, the beneficial effects that this application can achieve are as follows:

[0066] Breakthrough in accuracy: Dynamic pressure monitoring + multi-branch collaborative analysis eliminates interference from system pressure fluctuations, and the accuracy of leak detection is ≥95% (30%+ higher than traditional methods).

[0067] Enhanced adaptability: The resistance model dynamically tracks changes in operating conditions such as pipeline aging and dust accumulation, with a long-term detection deviation rate of ≤5% (compared to ≥25% for traditional fixed models).

[0068] Intelligent positioning: Multi-branch topology collaborative positioning, leakage branch positioning error <5m, supports rapid operation and maintenance, and reduces air compressor station energy consumption loss by 15%~25%.

[0069] Those skilled in the art will recognize that the embodiments described herein are intended to help the reader understand the principles of this application, and should be understood as not limiting the scope of protection of this application to such specific statements and embodiments. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

Claims

1. A method for detecting pipeline leakage in air compressor stations based on dynamic pressure and multi-branch pipe pressure, characterized in that, Includes the following steps: S1: Data Acquisition; Real-time acquisition of gauge pressure P at the outlet of the main pipe of the station building start (t), ambient temperature T ambient (t), atmospheric pressure P atm The gauge pressure P at the end of each branch pipe end,j (t), real-time volumetric flow rate Q j (t), where j=1,2,…,n, and n is the total number of branches; S2: Air condition correction; adjusts the gauge pressure P of the main pipe and each branch pipe. start (t), P end,j (t) is converted to absolute pressure P abs,start (t) and P abs,j (t), and combined with the ideal gas law and T ambient (t) Calculate air density Meanwhile, the aerodynamic viscosity μ(t) is acquired in real time; S3: Dynamic calculation of pipeline resistance; based on the specifications and measurement parameters of each branch pipe, and the local resistance coefficient ξ. j air density Given the aerodynamic viscosity μ(t), calculate the Reynolds number and determine the friction coefficient. Based on the friction coefficient and local drag coefficient, calculate the total resistance of the branch pipe. ; S4: Normal pressure drop baseline modeling; Q is collected under normal operating conditions of the preset group. j (t k ), ρ(t) k ) and △P hist,J (t k ) data, among which Sampling time, Support Vector Regression (SVR) or Multilayer Perceptron (MLP) algorithms are used, with Q... j (t k ) and ρ(t k ) is the input. As the output, the mean square error (MSE) is minimized to complete the training and obtain the normal pressure drop prediction model for branch j. S5: Leak detection and location; By calculating the deviation between the prediction results of the pressure drop prediction model and the real-time pressure drop at the current moment, a logical judgment of leakage is made based on the deviation. S6: Dynamic model update; To address the drift in resistance parameters caused by pipe aging and ash accumulation, the Recursive Least Squares (RLS) method is used to realize the total resistance. Adaptive updates.

2. The method for detecting pipeline leakage in an air compressor station based on dynamic pressure and multi-branch pipe pressure as described in claim 1, characterized in that, In S2, the calculation method for gauge pressure conversion is as follows: Absolute pressure of main pipe: ; Absolute pressure of branch pipe: .

3. The method for detecting pipeline leakage in an air compressor station based on dynamic pressure and multi-branch pipe pressure as described in claim 2, characterized in that, In S2, the air density of the main pipe and each branch pipe The specific calculation method is as follows: ； Where R is the gas constant, and T(t) = 273.15 + T ambient (t).

4. The method for detecting pipeline leakage in an air compressor station based on dynamic pressure and multi-branch pipe pressure as described in claim 3, characterized in that, In S2, the aerodynamic viscosity μ(t) is obtained in real time according to the Sutherland formula or industrial calibration table, and its nonlinear relationship with T(t) is as follows: ； In the formula The viscosity at T0 = 273.15 K. S = 110.4K is Sutherland's constant.

5. The method for detecting pipeline leakage in an air compressor station based on dynamic pressure and multi-branch pipe pressure as described in claim 4, characterized in that, In S3, the specifications and measurement parameters include pipe diameter d. j Length L j Local drag coefficient ξ j Pipe wall roughness k j The specific method for calculating the Reynolds number is as follows: Cross-sectional area of ​​branch pipe Flow rate ; Then Reynolds number ,when When the temperature is less than 2300, it is considered a laminar flow condition. A value ≥2300 indicates a turbulent operating condition.

6. The method for detecting pipeline leakage in an air compressor station based on dynamic pressure and multi-branch pipe pressure as described in claim 5, characterized in that, When the condition is determined to be laminar flow: Friction coefficient ; When the flow is determined to be turbulent: the Colebrook formula is used for iterative solution. ； The formula converges rapidly using the Newton-Raphson iteration method, with the initial value taken as... Iteration accuracy .

7. The method for detecting pipeline leakage in an air compressor station based on dynamic pressure and multi-branch pipe pressure as described in claim 6, characterized in that, In S3, the total resistance The resistance model is expressed as: ； In the formula m y Let be the number of local resistance elements in the y-th segment of the branch pipe. Let be the local resistance coefficient of the z-th element.

8. The method for detecting pipeline leakage in an air compressor station based on dynamic pressure and multi-branch pipe pressure according to claim 7, characterized in that, In S4, the normal pressure drop prediction model is expressed as: ； in To predict pressure drop, As basis functions, For model weights, if SVR is used, a quadratic polynomial kernel is selected; if MLP is used, the ReLU activation function is selected to ensure fitting accuracy.

9. The method for detecting pipeline leakage in an air compressor station based on dynamic pressure and multi-branch pipe pressure as described in claim 8, characterized in that, S5 achieves intelligent leak detection and location through pressure drop deviation analysis and multi-branch topology collaboration. Specifically: Pressure drop deviation analysis includes: Real-time pressure drop calculation: Pressure drop at the current moment ; Normal pressure drop prediction: Substitute the current Q j (t), From the normal pressure drop prediction model, the predicted pressure drop is obtained. ; Pressure drop deviation quantification: ; The specific logic for determining leakage is as follows: Set statistical threshold , Set a duration threshold for the standard deviation of the pressure drop deviation during the steady-state period. , like And duration ≥ If so, it is determined that there is a leak in branch pipe j; Multi-branch coordinated positioning includes: The system combines a pre-stored pipeline topology diagram containing the hierarchy of main pipes and branches, and the connection relationships between branches, to make a judgment. If two or more adjacent branches trigger a leak simultaneously, the leak is primarily identified as a leak in the main pipe section. If only a single branch triggers the leak, the location is determined to be the section from the beginning to the end of that branch.

10. The method for detecting pipeline leakage in an air compressor station based on dynamic pressure and multi-branch pipe pressure according to claim 9, characterized in that, S6 specifically refers to: Data filtering: Extract preset groups of stable operating condition data from real-time data every hour; Parameter update: For the observed values, For the parameters to be estimated, the prediction error is minimized using the RLS algorithm to update the parameters of the resistance model; Adaptive update cycle: If the number of leak detections in a branch pipe within a preset hour is greater than the preset number, the update cycle will be shortened.

Images