Comprehensive energy system security domain boundary approximation method based on polynomial chaos expansion
The calculation of the security domain boundary of the integrated energy system is simplified through the polynomial chaos expansion method, solving the problems of large-scale computing and lack of universality of the existing methods, and achieving rapid and accurate security domain boundary construction and online monitoring in large-scale systems.
Patent Information
- Application Number
- CN202411010796.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-26
- Publication Date
- 2024-11-22
AI Technical Summary
The existing comprehensive energy system security domain boundary solution method is large in computing and lacks universality in large-scale systems, making it difficult to quickly build and online monitoring, especially in the case of load changes.
Using the method of polynomial chaotic expansion, by constructing a parametric optimization model and Lagrangian function, solving the first-order optimization conditions of KKT, obtaining polynomial approximation expressions, simplifying the calculation process of the boundaries of the security domain, and improving the accuracy through polynomial segmentation approximation .
The global accuracy of the security domain boundary under large-scale load changes is achieved, which reduces the computational volume, improves the system's situational awareness ability and security evaluation efficiency, and reduces the failure rate.
Smart Images

Figure CN119004780A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of static safety domains, and in particular to a method for approximating the boundary of a safety domain of an integrated energy system based on polynomial chaos expansion. Background Art
[0002] With the rapid development of gas turbines and power to gas (P2G) technology, the coupling between power grid and natural gas grid is becoming increasingly close. While promoting the flexibility of power grid and integrating higher penetration of renewable energy, it also brings potential operational risks to the integrated energy system (IES). In addition, the volatility and uncertainty of load and power supply further aggravate the security challenges of IES. As a system operation assessment method, security region (SR) has the functions of situational awareness, safety margin assessment and optimal control, and occupies an extremely important position in the field of IES security and stability analysis.
[0003] IES static security region refers to the set of all operating points that satisfy both the node energy flow balance equation and the static security constraints in the energy injection space. According to the different research objects of IES static safety and stability, the integrated energy system security region (IESSR) can be divided into: thermal security region (TSR), static voltage security region (VSR), pressure security region (PSR) and delivery security region (DSR). The current safe operation status of the system can be qualitatively evaluated by judging the relative position between the current operating point of the system and the security region boundary (SRB). Therefore, the search for the security region boundary is crucial to constructing the security region of the integrated energy system.
[0004] At present, the commonly used methods for solving the security region boundary can be divided into fitting method and function approximation method. The fitting method searches for a large number of security region boundary points through relevant algorithms, and then indirectly obtains the approximate expression of SRB based on the searched boundary points. Its calculation process is relatively complicated. In order to simplify the process of solving the security region boundary, the function approximation method no longer searches for the security region boundary points point by point, and directly uses nonlinear function expressions to approximate SRB. Both the fitting method and the function approximation method construct the integrated energy system security region boundary (IESSRB) by solving the function expression of SRB.
[0005] However, in practical applications, the above method has the following shortcomings:
[0006] ① In order to ensure the accuracy of the constructed SRB, the fitting method needs to search a large number of SR boundary points, which has a heavy computational burden; ② The function approximation method only constructs an approximate analytical expression of SRB for a specific research object, which lacks universality; ③ The continuous expansion of the IES scale will increase the dimension of the SRB criterion equation, thereby increasing the computational complexity of the above method, making it difficult to meet the needs of rapid construction of IESSRB. Therefore, it is of great significance to study how to establish an IESSRB general search model and realize accurate search of SRB according to actual IES needs for online monitoring and evaluation of IES security and stability.
[0007] References
[0008] [1]Gahleitner G, Hydrogen from renewable electricity: An international review of power-to-gas pilot plants for stationary applications[J]. International journal of hydrogen energy, 2013, 38(5):2039-2061.
[0009] [2]J Wu,J Yan,H Jia,et al.Integrated Energy Systems[J].AppliedEnergy,2016,167(apr.1):155-157.
[0010] [3]Fairley, Peter. The troubled link between gas and electricity grids[News][J]. IEEE Spectrum, 2016, 53(6):11-12. Summary of the invention
[0011] The present invention provides a safety domain boundary approximation method for an integrated energy system based on polynomial chaos expansion, which ensures the global accuracy of the approximate safety domain boundary under large-scale load changes, simplifies the calculation process of the safety domain boundary, and omits the search for safety domain boundary points; the approximation coefficient in the polynomial approximate expression of the safety domain boundary is obtained by one calculation, thereby quickly constructing the safety domain boundary, and then using the constructed safety domain boundary to conduct a safety assessment of the integrated energy system, thereby improving the situational awareness capability of the system, as described below for details:
[0012] A method for approximating the boundary of a safety region of an integrated energy system based on polynomial chaos expansion, the method comprising:
[0013] The state vector in the IESSR critical point search optimization model is approximated by the linear combination of polynomial function sequences, and the IESSRB parameterized optimization model is constructed.
[0014] The inequality constraints are transformed into equality constraints, the Lagrangian function of the IESSRB parameterized optimization model is constructed, the first-order partial derivative of the Lagrangian function with respect to the original dual variable is solved to make the result equal to 0, the KKT first-order optimization condition of IESSRB is constructed, and the parameterized nonlinear equation system model of IESSRB is obtained;
[0015] The KKT first-order optimization conditions of IESSRB are projected onto the polynomial basis function sequence, and a set of algebraic equations about the IESSRB polynomial approximation coefficients is constructed. The global approximate IESSRB polynomial analytical expression is obtained by solving the set of equations.
[0016] The curvature K of the IESSRB polynomial approximation analytical expression of the global approximation is calculated based on the algebraic equations of the polynomial approximation coefficients. max When the polynomial piecewise approximate expression of IESSRB is output, the construction of IESSRB is completed, and the overall safe operation area of IES is described. The system operation status, safety margin and adjustment direction are obtained by analyzing the relative position of the operating point and the safety domain.
[0017] Among them, when K ≥ K max When IESSRB is used, the variable parameter domain is segmented at the maximum curvature of IESSRB, and polynomial piecewise approximation of IESSRB is adopted until the polynomial piecewise approximation expressions of all IESSRB are obtained.
[0018] Among them, the IESSR critical point search optimization model is: TSR critical point search optimization model, VSR critical point search optimization model, PSR critical point search optimization model or DSR critical point search optimization model.
[0019] Among them, the IESSR parameterized optimization model is: TSRB parameterized optimization model, PSRB parameterized optimization model, or DSRB parameterized optimization model.
[0020] Furthermore, the TSR critical point search optimization model is:
[0021]
[0022] In the formula, y i Inject energy pattern y into nodes y=[y1,y2,…,y i,0 +Δy i ,y i+1 ,…,y j,0 +Δy j …,y m ] The TSR critical point corresponds to the energy injected into node i; P b,k-s is the active transmission amount of branch ks; is the active power transmission limit of branch ks; F TSRB,u-d (y)=0 is the implicit function expression of TSRB of branch ud.
[0023] Furthermore, the VSR critical point search optimization model is:
[0024]
[0025] In the formula, y i Inject energy pattern y into nodes y=[y1,y2,…,y i,0 +Δy i ,y i+1 ,…,y j,0 +Δy j …,y m ]The VSR critical point corresponds to the injected energy of node i; and is the voltage amplitude of node j V j The upper and lower limits of F VSRB,k (y)=0 is the implicit function expression of VSRB of node k.
[0026] Among them, the PSR critical point search optimization model is:
[0027]
[0028] In the formula, y i Inject energy pattern y into nodes y=[y1,y2,…,y i,0 +Δy i ,y i+1 ,…,y j,0 +Δy j …,y m]The PSR critical point corresponds to the injected energy of node i; and is the pressure π at node n n The upper and lower limits of F PSRB,m (y)=0 is the implicit function expression of PSRB of node m.
[0029] Wherein, the DSRB critical point search optimization model is:
[0030]
[0031] In the formula, y i Inject energy pattern y into nodes y=[y1,y2,…,y i,0 +Δy i ,y i+1 ,…,y j,0 +Δy j …,y m ] The DSR critical point corresponds to the energy injected into node i; F xy is the natural gas transmission flow of pipeline xy; is the natural gas transmission flow limit of pipeline xy; F DSRB,m-n (y)=0 is the implicit function expression of DSRB of node m.
[0032] Wherein, the TSRB parameterized optimization model is:
[0033]
[0034] Wherein, the VSRB parameterized optimization model is:
[0035]
[0036] Wherein, the PSRB parameterized optimization model is:
[0037]
[0038] Wherein, the DSRB parameterized optimization model is:
[0039]
[0040] The beneficial effects of the technical solution provided by the present invention are:
[0041] 1. This method can effectively avoid the complex iterative optimization process of solving the optimization problem, thereby reducing the amount of calculation of IES thermal security region boundary (TSRB), static voltage security region boundary (VSRB), pressure security region boundary (PSRB) and delivery security region boundary (DSRB), and realize the rapid construction of IESSRB;
[0042] 2. Compared with the existing OEF-based SRB construction method, this method not only greatly improves the construction efficiency, but also ensures the accuracy of the static safety and stability assessment of the system, thereby improving the security of the system and reducing the failure rate;
[0043] 3. This method can be applied to the SRB search of actual IES. Compared with the existing OEF-based SRB construction method, it can further improve the perception ability of large-scale IES stability situation. BRIEF DESCRIPTION OF THE DRAWINGS
[0044] Figure 1 It is a flow chart of a method for approximating the boundary of the safety region of an integrated energy system based on polynomial chaos expansion;
[0045] Figure 2 Schematic diagram of IESSRB using polynomial piecewise approximation in two-dimensional energy injection space;
[0046] Figure 3 This is the test system diagram for the IES E39-G20 node (IES Electricity39-Gas20);
[0047] Figure 4 It is a two-dimensional TSRB schematic diagram of the IES E39-G20 system with the coordinate axes as load nodes;
[0048] Figure 5 It is a two-dimensional VSRB schematic diagram of the IES E39-G20 system with the coordinate axes as load nodes;
[0049] Figure 6 It is a two-dimensional PSRB schematic diagram with the coordinate axes of the IES E39-G20 system as load nodes;
[0050] Figure 7 It is a two-dimensional DSRB schematic diagram of the IES E39-G20 system with the coordinate axes as load nodes. DETAILED DESCRIPTION
[0051] In order to make the objectives, technical solutions and advantages of the present invention more clear, the embodiments of the present invention are described in further detail below.
[0052] Example 1
[0053] According to the different research objects of IES static safety and stability, IES static safety domain can be divided into different types of security regions (SR). By constructing various types of security regions, effective evaluation and operation control of different types of IES security issues can be achieved. In view of the construction problem of IES thermal stability security region (TSR), static voltage security region (VSR), pressure security region (PSR) and delivery security region (DSR), an embodiment of the present invention provides a general search method for IES thermal stability security region boundary (TSRB), static voltage security region boundary (VSRB), pressure security region boundary (PSRB) and delivery security region boundary (DSRB) based on polynomial chaos expansion, such as Figure 1 As shown, for the convenience of description, the embodiment of the present invention collectively refers to TSRB, VSRB, PSRB and DSRB as security region boundaries (SRB), and the general search method includes the following steps:
[0054] 101: Input IES topology structure, system initial operation parameters, select the maximum load node i, variable parameter q, variable parameter domain Q and polynomial approximation order n representing IESSRB p ;
[0055] 102: Based on the generalized Galerkin method, construct a polynomial function sequence according to the variable parameter domain Q and the polynomial basis function sequence {φ j (q)};
[0056] 103: Using polynomial function sequences The linear combination of is used to approximate the state vector x' in the IESSR critical point search optimization model, and the IESSRB parameterized optimization model is constructed;
[0057] 104: Based on the primal-dual interior point method, the inequality constraints in the IESSRB parameterized optimization model are transformed into equality constraints, and the Lagrangian function of the IESSRB parameterized optimization model is constructed. The first-order partial derivative of the Lagrangian function with respect to the primal-dual variables is solved to make the result equal to 0, and the KKT first-order optimization condition of IESSRB is constructed, thereby obtaining the parameterized nonlinear equation system model of IESSRB;
[0058] 105: Construct the Galerkin projection equations and project the KKT first-order optimization conditions of IESSRB to the polynomial basis function sequence φ i (q) Construct a set of algebraic equations about the IESSRB polynomial approximation coefficients, and solve the set of equations to obtain the global approximate IESSRB polynomial analytical expression y i =y i (q);
[0059] 106: Set the maximum curvature threshold K max , calculate the global approximate IESSRB polynomial approximate analytical expression y according to the algebraic equation system of the polynomial approximation coefficients i =y i The curvature K of (q) is K≥K max When , execute step 107, otherwise execute step 108;
[0060] 107: Divide the variable parameter domain into segments at the maximum curvature value of IESSRB, adopt polynomial segmentation to approximate IESSRB, repeat steps 102-105, and obtain the polynomial segmentation approximate expression of IESSRB;
[0061] 108: Output the polynomial piecewise approximate expression of IESSRB to complete the construction of IESSRB;
[0062] 109: The constructed IESSRB is used to describe the overall safe operation area of IES. By analyzing the relative position of the operating point and the safety domain, a lot of safety information such as the system operation status, safety margin and adjustment direction can be obtained, which greatly improves the efficiency of IES safety assessment, reduces the amount of safety analysis calculations, and simplifies the solution of various optimization problems related to safety.
[0063] In summary, the embodiment of the present invention uses the polynomial chaos expansion method to quickly calculate the polynomial approximate expression of the IESSRB through the above steps 101 to 109, thereby realizing the accurate construction of the IESSRB, improving the efficiency of IES security assessment, and thus improving the security of the system and reducing the failure rate.
[0064] Example 2
[0065] The embodiment of the present invention takes TSR as an example. The following is a specific calculation formula: Figure 2 The scheme in Example 1 is further introduced, as described below:
[0066] 201: Select the node to inject energy y j ,y i As the horizontal and vertical axes of the energy injection space, respectively, given the node injection energy pattern y, the TSR critical point search optimization model is established. The specific process is as follows:
[0067] Where y=[y1,y2,…,y m ],m=N E +N G , N E is the total number of grid nodes in IES, N G is the total number of gas network nodes in IES.
[0068] For the actual power grid in IES, the thermal safety of the inter-regional interconnection section is the focus of the power grid dispatching center operators, and the interconnection section is usually composed of a few key branches. Therefore, the thermal safety of these key branches can be focused on to build the system's TSR. In view of this feature, the TSR critical point search optimization model for the thermal safety of a branch on the system's key section can be simplified as follows:
[0069]
[0070] In the formula, y i Inject energy pattern y into nodes y=[y1,y2,…,y m ] The TSR critical point corresponds to the energy injected into node i; P b,k-s is the active transmission amount of branch ks; is the active power transmission limit of branch ks; F TSRB,u-d (y) = 0 is the implicit function expression of TSRB of branch ud, which is composed of the node energy flow balance equation h(x, y) = 0 and the active power transmitted by branch ud |P b,u-d |Expressions that reach the limit Together, the node energy flow balance equation of IES can be specifically expressed as:
[0071]
[0072] Where P L,i , Q L,i are the active and reactive loads at node i respectively; P g,slack is the active power output of the gas turbine at the grid balancing node; P cp,i Active power consumed by the coupling equipment connecting the power grid and the gas grid; n E is the number of grid nodes; G ij , Bij are the conductance and susceptance of branch ij respectively; e i and f i are the real and imaginary parts of the voltage at node i, respectively; F sp,m is the gas supply volume of the gas source at node m; F L,m is the natural gas load at node m; F gas,m is the natural gas flow consumed by the gas turbine at node m; F P2G,m k is the natural gas flow input to the P2G device at the fixed pressure node of the gas network; mn is the pipeline constant; s mn The value reflects the flow direction of natural gas in the pipeline. When natural gas flows from node m to node n, s mn Take 1, otherwise take -1.
[0073] Active power transmitted by branch ud |P b,u-d | Reached limit The expression is:
[0074]
[0075] 202: Combine the PCE theory with the TSR critical point search optimization model shown in formula (1) to parameterize the TSR critical point search optimization model. Inject energy y into the node corresponding to the horizontal coordinate in the energy injection space j Consider it as a variable parameter q, determine its range of variation, that is, the variable parameter domain Q, and construct n according to the variable parameter domain p Order polynomial function sequence The objective function and state variables in the model are approximated by a linear combination of polynomial functions, so as to simulate the continuous change of the energy injected into the node corresponding to the TSR critical point on the TSRB, thereby constructing a TSRB parameterized optimization model;
[0076] Wherein, the step 202 includes:
[0077] 1) Construct a polynomial function based on a variable parameter domain, where Q = {y j |0≤y j ≤b}, the constructed polynomial function sequence and the polynomial basis function sequence {φ j (q)}.
[0078] 2) Using polynomial functions The linear combination of approximates the objective function y i 、Gas network node pressure π m , the natural gas flow rate F input by the P2G equipment to the gas grid P2G,m , the real and imaginary parts of the grid node voltage e i 、f i, active output of grid balancing node P g,slack and the reactive power output Q of the power grid generator G,i We can get:
[0079]
[0080] 3) The polynomial approximate variables in equation (4) are uniformly recorded as vector x'(q), expressed as:
[0081] x'(q)=[y i (q),π(q),F P2G (q),e(q),f(q),P g,slack (q),Q G (q)](5)
[0082] 4) In order to remove the absolute value of the inequality constraint in equation (1), the second equality constraint in equation (1) is transformed to obtain:
[0083]
[0084] 5) Substitute equation (4) into the changed TSR critical point search optimization model shown in equation (6) to construct the TSRB parameterized optimization model:
[0085]
[0086] 203: Based on the classical primal-dual interior point method, the slack variable u(q) is introduced to transform the inequality constraints in the TSRB parameterized optimization model into equality constraints, thereby constructing the TSRB parameterized Lagrangian function, solving the first-order partial derivatives of the Lagrangian function with respect to the primal-dual variables in the transformed TSRB parameterized optimization model, and constructing the parameterized KKT first-order optimization conditions of TSRB, thereby transforming the TSRB parameterized optimization model into a set of parameterized nonlinear algebraic equations.
[0087] Wherein, the step 203 includes:
[0088] 1) Introduce non-negative slack variables to transform the inequality constraints in equation (7) into equality constraints. In order to satisfy the non-negative condition of slack variables, introduce the logarithmic function of the slack variables u(q) to transform the objective function into a barrier function. After the transformation, the TSRB parameterized model is:
[0089]
[0090] Where μ is the barrier factor and v represents the number of branch thermal constraints.
[0091] 2) The parameterized Lagrangian function of TSRB constructed by equation (8) is:
[0092]
[0093] Where ξ(q) and λ(q) are Lagrange multiplier vectors of polynomial approximation, also called dual variable vectors; x'(q) and u(q) are primal variable vectors; T represents the transpose of the matrix; to simplify the expression, the primal dual variable vector of polynomial approximation is uniformly recorded as ω(q) = [x' T (q),u T (q),ξ T (q),λ T (q)] T .
[0094] 3) Since the first-order partial derivative of the Lagrangian function of equation (9) with respect to the vector ω(q) is 0, the first-order optimization condition of the parameterized KKT of TSRB is:
[0095]
[0096] In the formula, J FTSRB (x'(q)) is the implicit function F TSRB (y i The Jacobian matrix of (q),q) = 0; J Pb (x'(q)) is P b The Jacobian matrix of (x'(q),q); g is an r-dimensional column vector of all ones; U is a diagonal matrix of slack variables u(q).
[0097] For F in formula (10) TSRB (y i (q),q)=0, the node active power balance equation is f P , the dimension is d1; the node reactive power balance equation is f Q , the dimension is d2; the node flow balance equation is f F , the dimension is d3; the node voltage amplitude equation is f V 2, the dimension is d4; the equation corresponding to the active power transmission limit of branch ud is f s , then J FTSRB (x'(q)) T It can be expressed as:
[0098]
[0099] In the formula, It is a d1×(d2+d4-1)-dimensional matrix; It is a d1×(d3-1)-dimensional matrix; is a d1-dimensional column vector; It is a d2×(d2+d4-1)-dimensional matrix; It is a d2×(d4+1)-dimensional matrix; It is a d3×(d3-1) dimensional matrix; is a d3-dimensional column vector; is a d4×(d2+d4-1)-dimensional matrix. The specific expressions of each element in the above matrix are shown in equations (12)-(25). First-order partial derivative Both are d2+d4-1 dimensional row vectors. The expressions of the u-th element and the d-th element in the two row vectors are as shown in formula (26). The remaining elements are 0, and the first-order partial derivatives of the remaining variables in x'(q) are 0.
[0100] Assume that the grid balancing node is numbered a, the gas grid constant pressure node is numbered b, the gas turbine is connected to the gas grid node c and the grid node a, and the P2G device is connected to the gas grid node b and the grid node d.
[0101]
[0102]
[0103]
[0104]
[0105]
[0106] In the formula, t mn B is the compression ratio of the electric driven compressor k connected to the gas network pipeline mn; k 、z k is the design constant of compressor k.
[0107]
[0108]
[0109]
[0110]
[0111]
[0112]
[0113]
[0114] Where P g,a(q) is the active power generated by the gas-fired units connected to the grid balancing node; β G,c and γ G,c It is expressed as the fitting coefficient in the conversion relationship between natural gas and electric power of the gas generator set; f GHV Natural gas has a high calorific value.
[0115]
[0116]
[0117] In the formula, f LHV It is the low calorific value of natural gas.
[0118]
[0119] The branch transmission active power P of the polynomial approximation in the original branch thermal constraint b (x'(q),q) takes the first-order partial derivative of x'(q), and we get P b The Jacobian matrix J of (x'(q),q) Pb (x'(q)) T for:
[0120]
[0121] In formula (27) The specific expression is:
[0122]
[0123] 204: Project the parameterized nonlinear algebraic equations about TSRB shown in equation (10) into a set of polynomial basis functions {φ j (q)}, the Galekrin projection equation for constructing the TSRB polynomial approximation coefficients is:
[0124]
[0125] In the formula, c JFTSRB(x'),l is the matrix J FTSRB The l-order approximation coefficient of the vector x'(q) in (x'(q)); c ξ,v 、c λ,v are the v-order approximation coefficients of the vectors ξ(q) and λ(q) respectively; The matrices U -1 (q)'s l-order approximation coefficient; For the matrix The l-order approximation coefficient of the vector x'(q); c x',lis the l-order approximation coefficient of the vector x'(q); <·,·> represents the inner product calculation, the i-order polynomial function in the inner product space With the j-order polynomial basis function φ j (q) The inner product is:
[0126]
[0127] 205: Through the construction of the Galerkin projection equations for the TSRB polynomial approximation coefficients, the TSRB parameterized nonlinear equations can be transformed into algebraic equations, and then the polynomial approximation coefficients of TSRB can be obtained by using conventional methods for solving algebraic equations, such as the Newton-Raphson method. The TSRB polynomial approximation expression is thus constructed as:
[0128]
[0129] 206: Calculate the curvature K of the TSRB polynomial approximation expression as:
[0130]
[0131] 207: Determine the maximum curvature K of the approximate TSRB max Is it greater than the set curvature threshold? If K does not meet the convergence condition, at the maximum curvature K max The original variable parameter domain Q is segmented and then the polynomial segmentation is used to approximate TSRB, thereby obtaining the polynomial segmentation approximate expression of TSRB. The TSRB approximation process in the two-dimensional energy injection space is shown in the attached figure. Figure 2 shown.
[0132] In summary, the embodiment of the present invention not only simplifies the construction process of TSRB through the above steps 201-207, but also effectively avoids the complex iterative optimization process of the traditional optimal energy flow method, and significantly improves the accuracy of TSRB by using polynomial piecewise approximation TSRB, which has practical application value for IES thermal stability evaluation. By constructing TSRB, fast, accurate and reliable thermal stability information can be provided for operation and control personnel to improve the thermal stability safety of the power system in IES, which is conducive to improving the N-1 thermal stability safety monitoring and control capabilities of the power system.
[0133] Example 3
[0134] This embodiment takes VSR as an example. Figure 2 The scheme in Example 1 is further introduced, as described below:
[0135] 301: Select the node to inject energy y j ,y iAs the horizontal and vertical axes of the energy injection space, respectively, given the node injection energy pattern y, the VSR critical point search optimization model is established. The specific process is as follows:
[0136] Where y=[y1,y2,…,y m ],m=N E +N G (N E is the total number of grid nodes in IES, N G is the total number of gas network nodes in IES).
[0137] In the actual IES power grid, only a small number of node voltage amplitudes exceeding the limit will have a significant impact on the safety and stability of the power grid operation. Therefore, when constructing the VSR, the focus can be on building the VSR around these voltage safety dominant nodes that have a serious impact on the safety of system operation. The VSRB critical point search optimization model for a single voltage safety dominant node in the power system is obtained as follows:
[0138]
[0139] In the formula, y i Inject energy pattern y into nodes y=[y1,y2,…,y m ]The VSR critical point corresponds to the injected energy of node i; and is the voltage amplitude of node j V j The upper and lower limits of F VSRB,k (y) = 0 is the implicit function expression of VSRB of node k, which is composed of the node energy flow balance equation h(x, y) = 0 shown in formula (2) and the voltage amplitude V of node k k Reaching the limit The expressions together constitute .
[0140] Node k voltage amplitude V k Reaching the limit The expression is:
[0141]
[0142] 302: Combine the PCE theory with the VSR critical point search optimization model shown in formula (33) to parameterize the VSR critical point search optimization model. Inject energy y into the node corresponding to the horizontal coordinate in the energy injection space j Consider it as a variable parameter q, determine its range of variation, that is, the variable parameter domain Q, and construct n according to the variable parameter domain p Order polynomial function sequence As shown in formula (4), the objective function and state variables in the model are approximated by a linear combination of polynomial functions, thereby simulating the continuous change of the energy injected into the node corresponding to the VSR critical point on the VSRB. The VSRB parameterized optimization model is constructed as follows:
[0143]
[0144] 303: Based on the classical primal-dual interior point method, the slack variable u(q) is introduced to transform the inequality constraints in the VSRB parameterized optimization model into equality constraints, thereby constructing the VSRB parameterized Lagrangian function, solving the first-order partial derivatives of the Lagrangian function with respect to the primal-dual variables in the transformed VSRB parameterized optimization model, and constructing the parameterized KKT first-order optimization conditions of VSRB, thereby transforming the VSRB parameterized optimization model into a set of parameterized nonlinear algebraic equations.
[0145] Wherein, the step 303 includes:
[0146] 1) Introducing non-negative slack variables to transform the inequality constraints in equation (35) into equality constraints. In order to satisfy the non-negative condition of the slack variables, the logarithmic function of the slack variables u(q) is introduced to transform the objective function into a barrier function. After the transformation, the VSRB parameterized model is:
[0147]
[0148] 2) The parameterized Lagrangian function of TSRB is constructed by equation (36):
[0149]
[0150] Where ξ(q), λ(q) and η(q) are the Lagrange multiplier vectors of polynomial approximation, collectively referred to as the dual variable vector;
[0151] x'(q), l(q) and u(q) are the original variable vectors. To simplify the expression, the original dual variable vectors of the polynomial approximation are uniformly recorded as ω(q) = [x' T (q),l T (q),u T (q),ξ T (q),λ T (q),η T (q)] T .
[0152] 3) Since the first-order partial derivative of the Lagrangian function of equation (37) with respect to the vector ω(q) is 0, the first-order optimization condition of the parameterized KKT of VSRB is obtained as follows:
[0153]
[0154] In the formula, is the implicit function F VSRB (y i The Jacobian matrix of (q),q) = 0; J V (x'(q)) is the Jacobian matrix of V(x'(q),q); g is an r-dimensional column vector of all ones; L and U are diagonal matrices consisting of slack variables l(q) and u(q).
[0155] The specific expression of is shown in (11), where the dimensions of matrices N, H, A, T, U, S, L, K, G, D, C, B, E and F are the same as the dimensions of the matrices in step 203 in Example 2, and the expressions are shown in equations (12)-(25). The first-order partial derivative Both are d2+d4-1 dimensional row vectors. The expression of the i-th element in the two row vectors is shown in formula (39), the remaining elements are 0, and the first-order partial derivatives of the remaining variables in x'(q) are 0.
[0156]
[0157] The first-order partial derivative of the node voltage amplitude V(x'(q),q) approximated by the polynomial in the original node voltage amplitude constraint with respect to x'(q) is obtained, and the Jacobian matrix J of V(x'(q),q) is obtained. V (x'(q)) T for:
[0158]
[0159] In the formula, The specific expression is the same as formula (39).
[0160] 304: Project the parameterized nonlinear algebraic equations about VSRB shown in equation (38) into a set of polynomial basis functions {φ j (q)}, the Galekrin projection equation for constructing the VSRB polynomial approximation coefficients is:
[0161]
[0162] In the formula, For the matrix The l-order approximation coefficient of the vector x'(q); c ξ,v 、c λ,v and c η,v are the v-order approximation coefficients of the vectors ξ(q), λ(q) and η(q) respectively; The matrices L -1 (q), U-1 (q)'s l-order approximation coefficient; is the matrix J N The l-order approximation coefficient of the vector x'(q) in (x'(q)); c x',l is the l-th order approximation coefficient of the vector x'(q).
[0163] Steps 305-307 are the same as steps 205-207 and will not be repeated here.
[0164] In summary, the embodiment of the present invention not only simplifies the construction process of VSRB through the above steps 301-307, but also effectively avoids the complex iterative optimization process of the traditional optimal energy flow method, and significantly improves the accuracy of VSRB by using polynomial piecewise approximation VSRB, which has practical application value for IES voltage stability assessment. The constructed VSRB can be used to intuitively and qualitatively assess the voltage stability of the power system in IES, providing a scientific and reasonable basis for online voltage stability assessment and control.
[0165] Example 4
[0166] The embodiment of the present invention takes PSR as an example. Figure 2 The scheme in Example 1 is further introduced, as described below:
[0167] 401: Select the node to inject energy y j ,y i As the horizontal and vertical axes of the energy injection space, respectively, given the node injection energy pattern y, the PSR critical point search optimization model is established. The specific process is as follows:
[0168] Where y=[y1,y2,…,y m ],m=N E +N G (N E is the total number of grid nodes in IES, N G is the total number of gas network nodes in IES).
[0169] In the actual IES gas grid, only a small number of nodes with over-limit gas pressure will have a significant impact on the safety and stability of the gas grid operation. Therefore, when constructing the PSR, we can focus on these gas pressure safety dominant nodes that have a serious impact on the safety of system operation. The PSRB critical point search optimization model for a single pressure safety dominant node of the natural gas system is:
[0170]
[0171] In the formula, y i Inject energy pattern y into nodes y=[y1,y2,…,y i,0 +Δyi ,y i+1 ,…,y j,0 +Δy j …,y m ]The PSR critical point corresponds to the injected energy of node i; and is the pressure π at node n n The upper and lower limits of F PSRB,m (y) = 0 is the implicit function expression of PSRB of node m, which is composed of the node energy flow balance equation h(x, y) = 0 and the node m pressure π n Reaching the limit The expressions together constitute .
[0172] Pressure at node mπ m Reaching the limit The expression is:
[0173]
[0174] 402: Combine the PCE theory with the VSR critical point search optimization model shown in equation (42) to parameterize the PSR critical point search optimization model. Inject energy y into the node corresponding to the horizontal coordinate in the energy injection space j Consider it as a variable parameter q, determine its range of variation, that is, the variable parameter domain Q, and construct n according to the variable parameter domain p Order polynomial function sequence As shown in formula (4), the objective function and state variables in the model are approximated by a linear combination of polynomial functions, thereby simulating the continuous change of the energy injected into the node corresponding to the PSR critical point on the PSRB. The PSRB parameterized optimization model is constructed as follows:
[0175]
[0176] 403: Based on the classical primal-dual interior point method, the slack variable u(q) is introduced to transform the inequality constraints in the PSRB parameterized optimization model into equality constraints, thereby constructing the PSRB parameterized Lagrangian function, solving the first-order partial derivatives of the Lagrangian function with respect to the primal-dual variables in the transformed PSRB parameterized optimization model, and constructing the parameterized KKT first-order optimization conditions of PSRB, thereby transforming the PSRB parameterized optimization model into a set of parameterized nonlinear algebraic equations.
[0177] Wherein, step 403 includes:
[0178] 1) Introduce non-negative slack variables to transform the inequality constraints in equation (44) into equality constraints. In order to satisfy the non-negative condition of the slack variables, introduce the logarithmic function of the slack variables u(q) to transform the objective function into a barrier function. After the transformation, the PSRB parameterized model is:
[0179]
[0180] 2) The parameterized Lagrangian function of PSRB constructed by equation (45) is:
[0181]
[0182] In the formula, ξ(q), λ(q) and η(q) are the Lagrange multiplier vectors of polynomial approximation, collectively referred to as dual variable vectors; x'(q), l(q) and u(q) are the primal variable vectors. To simplify the expression, the primal dual variable vectors of polynomial approximation are uniformly recorded as ω(q) = [x' T (q),l T (q),u T (q),ξ T (q),λ T (q),η T (q)] T .
[0183] 3) Since the first-order partial derivative of the Lagrangian function of equation (46) with respect to the vector ω(q) is 0, the first-order optimization condition of the parameterized KKT of PSRB is:
[0184]
[0185] In the formula, is the implicit function F PSRB (y i The Jacobian matrix of (q),q) = 0; J π (x'(q)) is the Jacobian matrix of π(x'(q),q); g is an r-dimensional column vector of all 1s; L and U are diagonal matrices consisting of slack variables l(q) and u(q).
[0186] The specific expression of is shown in (11), where the dimensions of matrices N, H, A, T, U, S, L, K, G, D, C, B, E and F are the same as the dimensions of the matrices in step 203 in Example 2, and the expressions are shown in equations (12)-(25). The first-order partial derivative It is a d3-1 dimensional row vector. Except for the mth element which is 1, the remaining elements in the vector are 0, and the first-order partial derivatives of the remaining variables in x'(q) are 0.
[0187] The first-order partial derivative of the node voltage amplitude π(x'(q),q) approximated by the polynomial in the original node pressure constraint with respect to x'(q) is obtained, and the Jacobian matrix J of π(x'(q),q) is obtained. π (x'(q)) T for:
[0188]
[0189] In the formula, The remaining elements are 0.
[0190] 404: Project the parameterized nonlinear algebraic equations about PSRB shown in equation (47) into a set of polynomial basis functions {φ j (q)}, the Galekrin projection equation for constructing the PSRB polynomial approximation coefficients is:
[0191]
[0192] In the formula, For the matrix The l-order approximation coefficient of the vector x'(q); c ξ,v 、c λ,v and c η,v are the v-order approximation coefficients of the vectors ξ(q), λ(q) and η(q) respectively; The matrices L -1 (q), U -1 (q) is the l-order approximation coefficient; c Jπ(x'),l is the matrix J π The l-order approximation coefficient of the vector x'(q) in (x'(q)); c x',l is the l-th order approximation coefficient of the vector x'(q).
[0193] Steps 405-407 are the same as steps 305-307 and will not be described again here.
[0194] In summary, the embodiment of the present invention not only simplifies the construction process of PSRB through the above steps 401-407, but also effectively avoids the complex iterative optimization process of the traditional optimal energy flow method, and significantly improves the accuracy of PSRB by using polynomial piecewise approximation PSRB, which has practical application value for IES pressure stability assessment. By constructing the PSRB of the natural gas system in IES, the situational awareness of natural gas network security and IES security can be enhanced, thereby solving the security challenges caused by the close coupling of power and natural gas systems in IES.
[0195] Example 5
[0196] The embodiment of the present invention takes DSR as an example. Figure 2 The scheme in Example 1 is further introduced, as described below:
[0197] 501: Select the node to inject energy y j ,y iAs the horizontal and vertical axes of the energy injection space, respectively, given the node injection energy pattern y, the DSR critical point search optimization model is established. The specific process is as follows:
[0198] Where y=[y1,y2,…,y m ],m=N E +N G (N E is the total number of grid nodes in IES, N G is the total number of gas network nodes in IES).
[0199] In the actual IES gas network, only a small number of pipelines that exceed the limit of natural gas flow will have a significant impact on the safety and stability of the gas network operation. Therefore, when constructing the DSR, we can focus on these transmission safety-dominant pipelines that have a serious impact on the safety of system operation. The DSRB critical point search optimization model for a single transmission safety-dominant pipeline in the natural gas system is:
[0200]
[0201] In the formula, y i Inject energy pattern y into nodes y=[y1,y2,…,y i,0 +Δy i ,y i+1 ,…,y j,0 +Δy j …,y m ] The DSR critical point corresponds to the energy injected into node i; F xy is the natural gas transmission flow of pipeline xy; is the natural gas transmission flow limit of pipeline xy; F DSRB,m-n (y) = 0 is the implicit function expression of PSRB of node m, which is composed of the node energy flow balance equation h(x, y) = 0 shown in formula (2) and the natural gas flow rate transmitted by pipeline mn | F mn | Reached limit Sure.
[0202] The natural gas flow rate of pipeline mn has reached the limit The expression is:
[0203]
[0204] 502: Combine the PCE theory with the DSR critical point search optimization model shown in formula (50) to parameterize the DSR critical point search optimization model. Inject energy y into the node corresponding to the horizontal coordinate in the energy injection space j Consider it as a variable parameter q, determine its range of variation, that is, the variable parameter domain Q, and construct n according to the variable parameter domain p Order polynomial function sequence As shown in formula (4), the objective function and state variables in the model are approximated by a linear combination of polynomial functions, thereby simulating the continuous change of the energy injected into the node corresponding to the DSR critical point on the DSRB. The DSRB parameterized optimization model is constructed as follows:
[0205]
[0206] 503: Based on the classical primal-dual interior point method, the slack variable u(q) is introduced to transform the inequality constraints in the DSRB parameterized optimization model into equality constraints, thereby constructing the DSRB parameterized Lagrangian function, solving the first-order partial derivatives of the Lagrangian function with respect to the primal-dual variables in the transformed DSRB parameterized optimization model, and constructing the parameterized KKT first-order optimization conditions of DSRB, thereby transforming the DSRB parameterized optimization model into a set of parameterized nonlinear algebraic equations.
[0207] Wherein, the step 503 includes:
[0208] 1) Introduce non-negative slack variables to transform the inequality constraints in equation (52) into equality constraints. In order to satisfy the non-negative condition of the slack variables, introduce the logarithmic function of the slack variables u(q) to transform the objective function into a barrier function. After the transformation, the DSRB parameterized model is:
[0209]
[0210] 2) The parameterized Lagrangian function of DSRB is constructed by equation (53):
[0211]
[0212] In the formula, ξ(q) and λ(q) are Lagrange multiplier vectors of polynomial approximation, collectively referred to as dual variable vectors; x'(q) and u(q) are primal variable vectors. To simplify the expression, the primal dual variable vectors of polynomial approximation are uniformly recorded as ω(q) = [x' T (q),u T (q),ξ T (q),λ T (q)] T .
[0213] 3) Since the first-order partial derivative of the Lagrangian function of equation (54) with respect to the vector ω(q) is 0, the first-order optimization condition of the parameterized KKT of DSRB is:
[0214]
[0215] In the formula, is the implicit function F DSRB (y iThe Jacobian matrix of (q),q) = 0; J F (x'(q)) is the Jacobian matrix of F(x'(q),q); g is an r-dimensional column vector of all ones; L and U are diagonal matrices consisting of slack variables l(q) and u(q).
[0216] The specific expression of is shown in (11), where the dimensions of matrices N, H, A, T, U, S, L, K, G, D, C, B, E and F are the same as the dimensions of the matrices in step 203 in Example 2, and the expressions are shown in equations (12)-(25). The first-order partial derivative is a d3-1 dimensional row vector. The expressions of the mth and nth elements in the vector are shown in equation (56). The remaining elements are 0, and the first-order partial derivatives of the remaining variables in x'(q) are 0.
[0217]
[0218] The first-order partial derivative of the polynomial approximation of the pipeline transmission flow F(x'(q),q) with respect to x'(q) in the original pipeline transmission constraint is obtained, and the Jacobian matrix J of F(x'(q),q) is obtained. F (x'(q)) T for:
[0219]
[0220] In the formula, The result is the same as formula (56).
[0221] 504: Project the parameterized nonlinear algebraic equations about DSRB shown in equation (55) into a set of polynomial basis functions {φ j (q)}, the Galekrin projection equation for constructing the DSRB polynomial approximation coefficients is:
[0222]
[0223] In the formula, For the matrix The l-order approximation coefficient of the vector x'(q); c ξ,v 、c λ,v are the v-order approximation coefficients of the vectors ξ(q) and λ(q) respectively; The matrices U -1 (q)'s l-order approximation coefficient; is the matrix J F The l-order approximation coefficient of the vector x'(q) in (x'(q)); c x',l is the l-th order approximation coefficient of the vector x'(q); steps 505-507 are the same as steps 405-407, and will not be repeated here.
[0224] In summary, the embodiment of the present invention not only simplifies the construction process of DSRB through the above steps 501-507, but also effectively avoids the complex iterative optimization process of the traditional optimal energy flow method, and significantly improves the accuracy of DSRB by using polynomial piecewise approximation DSRB, which has practical application value for IES transmission security assessment. By constructing DSRB in IES, the monitoring, assessment and control capabilities of IES security can be improved.
[0225] Example 6
[0226] The following is a specific example. Figure 3 , Figure 4 , Figure 5 Table 1, Table 2, Table 3, Table 4, Table 5 and Table 6 verify the feasibility of the solutions in Examples 1, 2, 3, 4 and 5, as described below:
[0227] This embodiment uses the search for the two-dimensional thermal stability safety domain boundary TSRB, static voltage safety domain boundary VSRB, pressure safety domain boundary PSRB and transmission safety domain boundary DSRB of the IES E39-G20 node system as an example to verify the effectiveness of this method. The IES E39-G20 node test system is as follows: Figure 3 shown.
[0228] 1) Construction of TSRB
[0229] The thermal safety of the IES E39-G20 node system power grid branch 1-2 is studied, and its active power transmission limit The active load P6 of the grid node 6 is taken as a variable parameter, and the value range is P6∈[0,4.04]. 20 The TSRB of branch 1-2 is approximated by this method in the two-dimensional injection space of the coordinate axis.
[0230] The TSRB of branch 1-2 is approximated by a linear combination of i-th order polynomial functions about P6 as shown in equation (59).
[0231]
[0232] Polynomial function sequence constructed according to the generalized Galerkin method with P6 variation range and the polynomial basis function sequence {φ i (P6)} as shown in Table 1. Linear combination of Approximate the vector x' in equation (4) and construct the TSRB parameterized optimization model shown in equation (60).
[0233]
[0234] According to formula (60), a TSRB parameterized Lagrangian function similar to formula (10) is constructed. By solving the first-order partial derivative of vector ω(P6), the parameterized KKT first-order optimization condition of TSRB shown in formula (61) is constructed, and then the parameterized optimization model of TSRB (60) is transformed into a parameterized nonlinear equation group as shown in formula (61).
[0235]
[0236] Project equation (61) to the polynomial basis functions φ of each order i (P6), construct the projection equations of the polynomial approximation coefficients of VSRB as shown in equation (62), and solve equation (62) to obtain polynomial functions of different orders The approximation coefficient c i , the fourth-order polynomial approximate expression of TSRB is shown in formula (63).
[0237]
[0238]
[0239] Table 1 Polynomial functions used to approximate TSRB and polynomial basis functions {φ i (P6)}
[0240]
[0241] To verify the correctness of this method for searching TSRB, Figure 4 Further, the TSR boundary point is given by Figure 4 By comparison, it can be seen that for the same variable parameter domain, the approximate TSRB obtained by this method is almost consistent with the TSRB formed by the TSR boundary points. The mean, variance and maximum relative error of the polynomial approximate TSRB are 1.04%, 2.43% and 3.51%, respectively, which verifies the accuracy of this method in searching TSRB.
[0242] 2) Construction of VSRB
[0243] The voltage safety of grid node 25 in the IES E39-G20 test system is taken as the research object, and the upper and lower limits of the voltage amplitude are 1.05pu and 0.95pu respectively. 26 Gas network node 20 natural gas load F 20 Into the two-dimensional injection space of the coordinate axis, with P 26 is a variable parameter, using26 The VSRB of node 25 is approximated by a linear combination of i-order polynomial functions.
[0244] Setting P 26 The range of change, when When P 26 ∈[0,10.2]; when When P 26 ∈[0,6.65]. First to fourth order polynomials are used to approximate the upper and lower boundaries of VSR respectively, and the approximation coefficients of each order are shown in Table 2.
[0245] Table 2 Approximation coefficients of polynomial approximation expressions obtained by approximating VSRB with polynomials of different orders
[0246]
[0247] To verify the correctness of this method for searching TSRB, Figure 5 Further given the VSR boundary point, Figure 5 By comparison, it can be seen that when the polynomial approximation order N=1, the approximate boundary of VSR is a straight line, and the approximation effect is poor; as the polynomial approximation order increases, the approximate accuracy of VSRB increases significantly, and the approximate error mean, variance and maximum relative error gradually decrease; when N>3, the upper and lower boundaries of VSRB almost coincide with the VSRB formed by the VSR boundary points, which verifies the accuracy of this method in searching TSRB. Table 3 further gives the error of approximating two-dimensional VSRB with polynomial functions of different orders.
[0248] Table 3 Two-dimensional VSRB error approximated by polynomial functions of different orders
[0249]
[0250] 3) Construction of PSRB
[0251] Taking the pressure safety of gas network node 4 in the IES E39-G20 test system as the research object, the PSRB of node 4 is approximated in the two-dimensional injection space with P4 and F7 as the coordinate axes. The results are as follows: Figure 6 As shown, the approximate polynomial analytical expression of PSRB and the range of F7 are shown in Table 4.
[0252] Table 4 PSRB polynomial approximate expression in P4-F7 space
[0253]
[0254] Table 5 gives the mean relative error, variance and maximum relative error of the approximation of PSRB using the proposed PCE method. From the results in the table, it can be seen that the mean relative error, variance and maximum relative error of the approximation of PSRB using the third-order polynomial do not exceed 1.58%, 1.73% and 4.6% respectively, which can achieve accurate approximation of PSRB.
[0255] Table 5 Errors of polynomial function approximation of 2D PSRB
[0256]
[0257] 4) Construction of DSRB
[0258] The transmission safety of natural gas pipeline 3-4 in the IES E39-G20 test system is studied, and its transmission flow limit is 28Mm 3 / day. The natural gas load flow F7 of natural gas grid node 7 is taken as the independent variable, and the variation range F7∈[0,1.81]. In the two-dimensional injection space with P4 and F7 as the coordinate axes, the DSRB of pipeline 3-4 is approximated by a fourth-order polynomial. The results are as follows: Figure 7 As shown in (a), the detailed polynomial analytical expression is shown in equation (63).
[0259]
[0260] Taking the searched DSRB boundary points as the benchmark, the error comparison results show that the relative error mean of the DSRB approximated by the fourth-order polynomial is 6.84%, the variance is 115.62%, and the maximum relative error is 39.8%, and the approximation error is relatively large. In order to improve the approximate accuracy of DSRB, the order of the approximate polynomial can be increased, but this will lead to an increase in the dimension of the Galerkin projection equation group of the DSRB polynomial approximation coefficients, which increases the difficulty of solving the projection equation group. To solve this problem, this paper uses the fourth-order polynomial to approximate the maximum curvature K of DSRB. max =4.47 (F7 = 1.201pu) to partition the F7 variation range, and use 2nd and 3rd order polynomials to approximate the DSRB of F7∈[0,1.2) and F7∈[1.2,1.81] respectively. The results are as follows Figure 7 (b) shows that the mean, variance and maximum relative error of the segmented approximation error are 2.47%, 4.23% and 5.81% respectively. The error results show that the proposed PCE method improves the approximation accuracy of DSRB while reducing the computational complexity of the projection equations. The polynomial analytical expression of the segmented approximation DSRB is shown in Table 6.
[0261] Table 6 Piecewise approximate expressions of DSRB polynomials in P4-F7 space
[0262]
[0263] In summary, compared with the existing OEF-based SRB construction method, the PCE-based IESSRB approximation method proposed in the embodiment of the present invention can effectively improve the IES security situation awareness capability, making the IES security online monitoring, evaluation and optimization more scientific and efficient. The integrated energy system security domain boundary point search optimization model constructed by this method is equivalent to the traditional integrated energy system security domain boundary point search optimization model, and can realize the accurate search of the integrated energy system security domain boundary point. The proposed PCE-based approximation method can transform the original IESSR boundary point search optimization model into an IESSRB parameterized optimization model through parameterization, thereby converting the search of the original SR boundary point into the problem of solving the IESSRB polynomial approximation expression. Compared with the traditional method of constructing IESSRB, the PCE-based approximation method proposed in this article omits the search process of SR boundary points, and directly constructs IESSRB through the optimization model, thereby simplifying the calculation. The proposed KKT first-order optimization condition based on IESSRB parameterization and the generalized Galerkin method can transform the original IESSRB optimization problem into the calculation of the Galerkin projection equations about the polynomial approximation coefficients in the IESSRB polynomial approximation expression, reducing the difficulty of solving IESSRB. In order to further reduce the amount of calculation, the original variable parameter domain is segmented at the maximum curvature of the low-order polynomial approximation IESSRB, and then the polynomial segmented approximation IESSRB is used to ensure the approximation accuracy while reducing the dimension of the Galerkin projection equations.
[0264] Those skilled in the art will appreciate that the accompanying drawing is only a schematic diagram of a preferred embodiment, and the serial numbers of the embodiments of the present invention are only for description and do not represent the advantages or disadvantages of the embodiments.
[0265] 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, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims
1. A method for approximating the safety domain boundary of an integrated energy system based on polynomial chaos expansion, characterized in that: The method comprises: The state vector in the IESSR critical point search optimization model is approximated by the linear combination of polynomial function sequences, and the IESSRB parameterized optimization model is constructed. The inequality constraints are transformed into equality constraints, the Lagrangian function of the IESSRB parameterized optimization model is constructed, the first-order partial derivative of the Lagrangian function with respect to the original dual variable is solved to make the result equal to 0, the KKT first-order optimization condition of IESSRB is constructed, and the parameterized nonlinear equation system model of IESSRB is obtained; The KKT first-order optimization conditions of IESSRB are projected onto the polynomial basis function sequence, and a set of algebraic equations about the IESSRB polynomial approximation coefficients is constructed. The global approximate IESSRB polynomial analytical expression is obtained by solving the set of equations. The curvature K of the IESSRB polynomial approximation analytical expression of the global approximation is calculated based on the algebraic equations of the polynomial approximation coefficients. max When the polynomial piecewise approximate expression of IESSRB is output, the construction of IESSRB is completed, and the overall safe operation area of IES is described. The system operation status, safety margin and adjustment direction are obtained by analyzing the relative position of the operating point and the boundary of the safety domain.
2. According to claim 1, a method for approximating the boundary of the safety region of a comprehensive energy system based on polynomial chaos expansion is characterized in that: When K ≥ K max When IESSRB is used, the variable parameter domain is segmented at the maximum curvature of IESSRB, and polynomial piecewise approximation of IESSRB is adopted until the polynomial piecewise approximation expressions of all IESSRB are obtained.
3. According to claim 1, a method for approximating the safety domain boundary of a comprehensive energy system based on polynomial chaos expansion is characterized in that: The IESSR critical point search optimization model is: TSR critical point search optimization model, VSRB critical point search optimization model, PSRB critical point search optimization model or DSRB critical point search optimization model.
4. According to claim 1, a method for approximating the boundary of the safety region of a comprehensive energy system based on polynomial chaos expansion is characterized in that: The IESSR parameterized optimization model is: TSRB parameterized optimization model, PSRB parameterized optimization model, or DSRB parameterized optimization model.
5. According to claim 3, a method for approximating the boundary of the safety region of a comprehensive energy system based on polynomial chaos expansion is characterized in that: The TSR critical point search optimization model is: In the formula, y i Inject energy pattern y into nodes y=[y1,y2,…,y i,0 +Δy i ,y i+1 ,…,y j,0 +Δy j …,y m ] The TSR critical point corresponds to the energy injected into node i; P b,k-s is the active transmission amount of branch ks; is the active power transmission limit of branch ks; F TSRB,u-d (y)=0 is the implicit function expression of TSRB of branch ud.
6. According to claim 3, a method for approximating the boundary of the safety region of a comprehensive energy system based on polynomial chaos expansion is characterized in that: The VSRB critical point search optimization model is: In the formula, y i Inject energy pattern y into nodes y=[y1,y2,…,y i,0 +Δy i ,y i+1 ,…,y j,0 +Δy j …,y m ]The VSR critical point corresponds to the injected energy of node i; and is the voltage amplitude of node j V j The upper and lower limits of F VSRB,k (y)=0 is the implicit function expression of VSRB of node k.
7. The method for approximating the boundary of the safety region of a comprehensive energy system based on polynomial chaos expansion according to claim 3 is characterized in that: The PSRB critical point search optimization model is: In the formula, y i Inject energy pattern y into nodes y=[y1,y2,…,y i,0 +Δy i ,y i+1 ,…,y j,0 +Δy j …,y m ]The PSR critical point corresponds to the injected energy of node i; and is the pressure π at node n n The upper and lower limits of F PSRB,m (y)=0 is the implicit function expression of PSRB of node m.
8. The method for approximating the boundary of the safety region of a comprehensive energy system based on polynomial chaos expansion according to claim 3 is characterized in that: The DSRB critical point search optimization model is: In the formula, y i Inject energy pattern y into nodes y=[y1,y2,…,y i,0 +Δy i ,y i+1 ,…,y j,0 +Δy j …,y m ] The DSR critical point corresponds to the energy injected into node i; F xy is the natural gas transmission flow of pipeline xy; is the natural gas transmission flow limit of pipeline xy; F DSRB,m-n (y)=0 is the implicit function expression of DSRB of node m.
9. The method for approximating the safety domain boundary of a comprehensive energy system based on polynomial chaos expansion according to claim 4 is characterized in that: The TSRB parameterized optimization model is: The VSRB parameterized optimization model is:
10. The method for approximating the safety region boundary of a comprehensive energy system based on polynomial chaos expansion according to claim 4 is characterized in that: The PSRB parameterized optimization model is: The DSRB parameterized optimization model is:
Citation Information
Cited By
Improved polynomial chaos expansion based optimal scheduling method of cc-tscopf
CN122951234A
