CN-122020927-A - Natural gas pipeline network layered decoupling optimization method integrating self-adaptive trust domain

CN122020927ACN 122020927 ACN122020927 ACN 122020927ACN-122020927-A

Abstract

The invention discloses a natural gas pipeline network layered decoupling optimization method integrating a self-adaptive trust domain, and aims to solve the problems of insufficient optimization accuracy and practicality caused by simplified physical characteristic description, incomplete constraint consideration and low solving efficiency of the conventional method. The method establishes a layered decoupling architecture by constructing a hydraulic initialization, a hydraulic scheduling decision and a physical state recovery model so as to separate complex physical state calculation and economic scheduling decision. And a layered decoupling self-adaptive solving algorithm is adopted to coordinate the solving of each model, and the damping factor and the trust domain radius are self-adaptively adjusted according to the physical consistency residual error, so that the solving stability and efficiency are improved. According to the invention, the iteration step length is dynamically controlled by introducing the self-adaptive trust domain mechanism, so that the defect of difficult solving of the high-dimensional nonlinear model is effectively overcome, and a pipe network operation scheme with economy, physical consistency and calculation robustness is obtained.

Inventors

LIANG GUANGCHUAN
QIN CAN
ZHOU JUN
PENG JINGHONG
LIU SHITAO
CHEN HAOBIN

Assignees

西南石油大学

Dates

Publication Date: 20260512
Application Date: 20260201

Claims (6)

1. The natural gas pipeline network layered decoupling optimization method integrating the self-adaptive trust domain is characterized by comprising the following steps of: The method comprises the following steps of S1, constructing a hydraulic initialization model, a hydraulic scheduling decision model and a physical state recovery model according to natural gas pipe network structure data, wherein the hydraulic initialization model is used for obtaining a topological feasible hydraulic approximate initial solution, the hydraulic scheduling decision model is used for solving an economic hydraulic scheduling scheme, and the physical state recovery model is used for solving a high-fidelity physical equation and providing accurate hydraulics and thermodynamic states; S2, performing system initialization and hot start steps, namely giving initial values to characteristic parameters, solving the hydraulic initialization model to obtain initial hydraulic state solution, and calling the physical state recovery model to calculate theoretical compression factors and friction coefficients as the characteristic parameters of primary iteration; the characteristic parameters comprise pipeline average temperature, pipeline compression factor and pipeline friction coefficient; S3, executing a main iteration loop of a hierarchical decoupling self-adaptive solving algorithm, solving the hydraulic scheduling decision model by utilizing known characteristic parameters and trust domain radius in the kth iteration to obtain a hydraulic scheduling scheme, inputting the hydraulic scheduling scheme into the physical state recovery model, calculating a characteristic parameter theoretical value, calculating a physical consistency residual error, and generating a characteristic parameter actual value and trust domain radius of the next iteration based on the physical consistency residual error self-adaptive adjustment damping factor and trust domain radius; s4, judging whether the convergence condition is met, calculating the relative change of the decision variable and the updating residual error of the physical parameter, and turning to the step S5 when the convergence condition is met, otherwise, enabling the iteration times k=k+1, and returning to the step S3; and S5, outputting the current hydraulic scheduling scheme as a natural gas pipeline network operation optimization scheme, wherein the optimization scheme comprises the pressure of each node, the temperature of each node, the flow rate of each pipeline, the rotating speed of each compressor, the starting number of each compressor and the power of each compressor.
2. The method for optimizing the hierarchical decoupling of a natural gas pipeline network fused with an adaptive trust domain according to claim 1, wherein in step S1, the objective function of the hydraulic initialization model is a weighted sum of minimized pipeline pressure square difference and relaxation variable, and the expression is: In the formula, The objective function value of the hydraulic initialization model is set; is the theoretical pressure squared difference of forward flow; Is the theoretical pressure squared difference of the reverse flow; A relaxation variable that is forward flow; A relaxation variable that is counter-current; penalty coefficients for relaxation variables; Is a gas station set; Is a pipeline set; Constraint conditions of the hydraulic initialization model comprise node flow balance constraint, internal gas station constraint, compressor operation boundary constraint and relaxed pipeline hydraulic constraint; The node flow balance constraint expression is: In the formula, Is an element Said elements being a pipe and a compressor; Is a node Is provided; Is a node Is set to the required flow rate; a set of all elements; is a node set; the relaxed pipeline hydraulic constraint expression is: In the formula, Is a node Is the pressure square of (2); Is a node Is the pressure square of (2); The pressure drop equation coefficient of the pipeline; an average temperature constant value for the initial pipe; is an initial compression factor constant value; is the constant value of the initial friction coefficient; The value of 1 is used as a pipeline flow direction variable to represent that natural gas is fed from a node Flow direction node A value of 0 indicates opposite; Is a large M value; maximum flow for the pipe.
3. The method for hierarchical decoupling optimization of a natural gas pipeline network fused with an adaptive trust domain according to claim 1, wherein in step S1, the hydraulic scheduling decision model is a mixed integer secondary constraint programming problem, and the expression is as follows, with the objective of minimizing the total running cost of the pipeline network under given characteristic parameters: In the formula, The objective function value of the hydraulic scheduling decision model; is the first First air compressing station Power of the table compressor; is the electricity price coefficient; penalty coefficients for pressure energy loss; Is a gas station set; the number of compressors in the air compressing station is the number; Constraint conditions of the hydraulic scheduling decision model comprise node flow balance constraint, gas station internal constraint, compressor performance constraint, pipeline hydraulic constraint and dynamic trust domain constraint; in the hydraulic constraint of the pipeline, the average temperature, the compression factor and the friction coefficient of the pipeline are based on the fixed value updated in the previous iteration, and the hydraulic constraint expression of the pipeline is as follows: In the formula, The average temperature of the pipeline is fixed; is a fixed compression factor; Is a fixed friction coefficient; The pressure drop equation coefficient of the pipeline; a variable for the flow direction of the pipeline; Is a large M value; maximum flow for the pipe; The internal gas station constraint comprises a gas station state constraint, a gas station flow coupling constraint and a gas station pressure coupling constraint; The compressor performance constraints include a compressor speed constraint, a compressor power constraint, and a compressor outlet temperature constraint.
4. The method for hierarchical decoupling optimization of a natural gas pipeline network with fusion of adaptive trust domains according to claim 1, wherein in step S1, the physical state recovery model is an unconstrained optimization problem, and the expression is as follows, with the objective of minimizing the sum of squares of residuals of thermodynamic constraint equations: In the formula, Is a set of all thermodynamic constraints; is the first Residual errors of thermodynamic constraint equations; Is a thermodynamic state variable vector; The thermodynamic constraint equation comprises a pipeline temperature drop constraint equation, a pipeline average temperature constraint equation and a node thermodynamic mixing constraint equation; Based on the temperature distribution obtained by solving, the state equation expression of the physical state recovery model is as follows: In the formula, Is a pipeline Is a compression factor of (2); Is the average pressure of the pipeline; Is the average temperature of the pipeline; A calculation function for a state equation; The friction coefficient calculation equation expression of the physical state recovery model is as follows: In the formula, Is a pipeline Friction coefficient of (a); Is the pipeline flow; is the natural gas kinematic viscosity; Is the diameter of the pipeline; absolute roughness of the pipeline; A function is calculated for the friction coefficient.
5. The method for optimizing the hierarchical decoupling of the natural gas pipeline network fused with the self-adaptive trust domain according to claim 1, wherein in step S3, the specific flow of the hierarchical decoupling self-adaptive solving algorithm is as follows: S301, initializing characteristic parameters and trust domain radius of the 1 st iteration, setting initial compression factors Coefficient of initial friction Average temperature with initial pipeline Setting initial trust domain radius to empirical constant value The value range is 0.1-0.3, and the iteration times are set And maximum number of iterations ; S302, solving the hydraulic scheduling decision model and utilizing the first step Characteristic parameters updated by multiple iterations Trust domain radius Solving the hydraulic scheduling decision model and outputting the first Node pressure for a number of iterations Flow rate of pipeline Pipeline flow direction variable ; S303, calculating the node pressure and the average pipeline pressure, and squaring according to the node pressure Calculating node pressure The expression is: In the formula, Is the first Node of secondary iteration Pressure, MPa, according to node pressure And node pressure Calculating the average pressure of the pipeline The expression is: In the formula, Is the first Pipeline for multiple iterations Average pressure, MPa; s304, solving the physical state recovery model according to the node pressure Node pressure Average pressure of pipeline Flow rate of pipeline Variable in direction of flow of pipeline Solving the physical state recovery model, and outputting the node temperature Node temperature Average temperature with pipeline According to the average temperature of the pipeline Average pressure with the pipeline Calculating compression factor Coefficient of friction ; S305, calculating local physical residual errors; The local physical residual Is a pipeline Is expressed as: In the formula, Is the first Pipeline for multiple iterations Local physical residual errors; is the first Theoretical pipeline average temperature of the second iteration; is the first The average temperature of the fixed pipeline in the iteration is calculated; is the first Theoretical compression factor of the second iteration; is the first A fixed compression factor for the second iteration; is the first Theoretical friction coefficient of the secondary iteration; is the first Fixed friction coefficient of the second iteration; s306, adaptively adjusting damping factors and trust domain radiuses; the damping factor From the local physical residual Dynamic adjustment, the expression is: In the formula, Is a pipeline First, the Damping factor of the second iteration; The value range is 0.1-0.3 for the minimum damping factor; is an superparameter, and the value range is 2-5; and carrying out damping update on the average temperature, the compression factor and the friction coefficient of the pipeline, wherein the expression is as follows: Calculating physical consistency residuals The expression is: In the formula, Is the first Physical consistency residual error of the secondary iteration; Is the total number of the pipelines; The trust domain radius According to the physical consistency residual error Self-adaptive adjustment, the expression is: In the formula, Is the first Trust domain radius for the next iteration; The value range is 0.01-0.1 for a preset physical consistency residual error threshold; The value range of the expansion factor is 1.2-2.0; the value range of the shrinkage factor is 0.5-0.9; is the maximum trust domain radius; Is the minimum trust domain radius; S307, carrying out iterative convergence judgment on the current result, and calculating the relative change of the decision variable and the update residual error of the physical parameter, wherein the judgment expression of the convergence condition is as follows: In the formula, Is the first Total relative change in the number of iterations; is the first Iterating the objective function value of the hydraulic scheduling decision model for the second time; is the first Iterating the objective function value of the hydraulic scheduling decision model for the second time; is the first Decision variable vectors of the secondary iteration, including node pressure, pipeline flow and pipeline flow direction; is the first Decision variable vectors of the secondary iteration; is the first Theoretical compression factor vector of the secondary iteration; is the first Theoretical compression factor vector of the secondary iteration; is the first Theoretical friction coefficient vector of the secondary iteration; is the first Theoretical friction coefficient vector of the secondary iteration; Is the L2 norm; When (when) When the iteration is stopped, the running scheme of the hydraulic scheduling decision model is output as the natural gas pipe network running optimization scheme, or when the iteration times are counted Up to the maximum number of iterations When the iteration is forcedly terminated and the current solution is output, when And is also provided with Time, order The process returns to step S302.
6. The method for optimizing hierarchical decoupling of natural gas networks incorporating adaptive trust domains according to claim 1, wherein in step S3, the dynamic trust domain constraints include node pressure trust domain constraints and pipeline traffic trust domain constraints; The node pressure trust domain constraint expression is: In the formula, For the node of the current iteration Pressure squaring; is the first Node of secondary iteration Pressure squaring; is the first Trust domain radius for the next iteration; is a node set; The pipeline flow trust domain constraint forward flow trust domain constraint, the reverse flow trust domain constraint and the flow component constraint, and the forward flow trust domain constraint expression is: In the formula, The forward flow of the pipeline in the current iteration is obtained; is the first The pipeline forward flow of the secondary iteration; the reverse traffic trust domain constraint expression is: In the formula, Reversing the flow for the pipeline of the current iteration; is the first Pipeline reverse flow of the second iteration; the flow component constraint expression is: In the formula, Is a pipeline Is a net flow of (c).

Description

Natural gas pipeline network layered decoupling optimization method integrating self-adaptive trust domain Technical Field The invention relates to the technical field of natural gas pipeline network operation and control, in particular to a natural gas pipeline network layered decoupling optimization method integrating a self-adaptive trust domain. Background The natural gas network serves as a core infrastructure connecting the gas source with the end user, and its operating state is directly related to the safety and economy of the energy supply. With the increasing expansion of the natural gas pipeline network scale, the operation condition of the pipeline network becomes more complex, and extremely high requirements are put on the accuracy, the calculation efficiency and the robustness of the operation optimization technology. The existing natural gas pipe network operation optimization method is mainly divided into two types, namely a linearization or convex relaxation method of a simplified physical model, and a complex nonlinear physical parameter (such as a compression factor, a pipeline friction coefficient, a gas temperature and the like) in the pipe transportation process is assumed to be a constant or a simplified function, so that a mixed integer nonlinear programming problem is converted into a form which is easy to solve. Although the method has high calculation efficiency, the dynamic characteristics of gas physical properties along with pressure and temperature change under high-pressure transmission are ignored, so that the deviation between the hydraulic/thermal state of the optimization scheme and the actual physical law is large, namely the physical consistency is poor, and the scheme is often not feasible in practice. The other is an overall optimization method of directly coupling the high-precision physical equation. Although the method ensures the physical precision, extremely strong nonlinearity and non-convexity are introduced, so that the optimization model is large in scale and complex in solving space. In practical application, the direct solution of the model often faces the problems of excessively long calculation time, easy sinking of local optimum and even incapability of convergence, and the real-time scheduling requirement of a pipe network is difficult to meet. Therefore, a new optimization method is needed, not only can the hydraulic thermodynamic state of the pipe network be accurately described by using a high-fidelity physical model, but also the solving difficulty can be reduced by an effective decoupling strategy and a convergence control mechanism, so that a natural gas pipe network operation scheduling scheme with economy, physical consistency and calculation robustness is obtained. Disclosure of Invention Aiming at the problem that the operation optimization of a natural gas pipe network in the prior art is difficult to consider both the accuracy of a physical model and the solution convergence stability, the invention provides a natural gas pipe network layering decoupling optimization method fused with a self-adaptive trust domain. The invention aims to overcome the defects of large calculated amount and difficult convergence caused by excessively simplifying physical parameters or directly coupling a high-dimensional nonlinear physical equation when a complex pipe network is processed by the traditional optimization method. By constructing a layered decoupling architecture, complex physical state calculation and economic dispatch decision are separated, and an adaptive trust domain mechanism is introduced to dynamically adjust a variable trust domain and an iteration step length, so that a pipe network operation scheme meeting high-fidelity physical constraint and having economic optimality is obtained. The invention discloses a natural gas pipeline network layering decoupling optimization method integrating a self-adaptive trust domain, which is characterized by comprising the following steps: The method comprises the following steps of S1, constructing a hydraulic initialization model, a hydraulic scheduling decision model and a physical state recovery model according to natural gas pipe network structure data, wherein the hydraulic initialization model is used for obtaining a topological feasible hydraulic approximate initial solution, the hydraulic scheduling decision model is used for solving an economic hydraulic scheduling scheme, and the physical state recovery model is used for solving a high-fidelity physical equation and providing accurate hydraulics and thermodynamic states; S2, performing system initialization and hot start steps, namely giving initial values to characteristic parameters, solving the hydraulic initialization model to obtain initial hydraulic state solution, and calling the physical state recovery model to calculate theoretical compression factors and friction coefficients as the characteristic parameters of primary iteration; th