Method for calculating parameter changing domain of loads under a case that guarantees constant locational marginal price in electricity market

Abstract

The disclosure provides a method for calculating a parameter changing domain of loads under a case that guarantees a constant locational marginal price in an electricity market, which relates to the electricity market field of the power system. With the method in the disclosure, the clearing model on the locational marginal price in the general form is established, and the safe changing domain of the locational marginal price with respect to the loads may be derived and calculated based on the first-order KKT condition expansion of the clearing model on the locational marginal price in the general form. When the increment of the nodal loads is subordinate to the changing domain, the locational marginal price may remain unchanged. The parameter changing domain of loads in the power system may be used for the comprehensive evaluation of power market clearing results and assisting the operation of the power market.

Claims

1. A method for calculating a parameter changing domain of loads in a power system under a case that guarantees a constant locational marginal price in an electricity market, comprising: (1) establishing a clearing model on the constant locational marginal price in a general form, comprising: (1-1) establishing the clearing model on the constant locational marginal price as follows: $\underset{p_i}{\min }\underset{i=1}{\overset{N_g}{.Math.}}{c}_i{p}_i\mathrm{satisfying}\text{:}\underset{i=1}{\overset{N_g}{.Math.}}{p}_i=\underset{j=1}{\overset{N}{.Math.}}{D}_j{P}_i^{\min }\le p_i\le P_i^{\max },\forall i\in \underset{\_}{{𝒩}_g}-F_i^{\max }\le \underset{i=1}{\overset{N_g}{.Math.}}{S}_{l,i}{p}_i-\underset{j=1}{\overset{N}{.Math.}}{S}_{l,j}{p}_j\le F_i^{\max },\forall l\in \underset{\_}{ℒ},$ where, N.sub.g represents a number of generator nodes in the power system; represents a set of serial numbers of the generator nodes, {1, 2, . . . , N.sub.g}; N represents a total number of nodes in the power system; represents a set of serial numbers of branches in the power system, {1, 2, . . . , L}, L represents a total number of the branches in the power system; p.sub.l, iϵ represents a power variable of generator i; c.sub.i, iϵ represents a power cost coefficient of generator i; P.sub.i.sup.max represents an upper power limit of the generator node i; P.sub.i.sup.min represents a lower power limit of the generator node i; D.sub.j, jϵ{1, 2, . . . , N} represents a nodal load of the power system; F.sub.l.sup.max, lϵ represents a capacity of branch l; and S.sub.l,i, lϵ, iϵ{1, 2, . . . , N} represents a power transfer distribution factor; there is a following linear relationship between constant locational marginal prices and Lagrangian multipliers of constraints:
Λ=τ−S.sub.LN.sup.T(μ.sup.L,max−μ.sup.L,min), where, Λ represents a column vector including the constant locational marginal prices of the nodes in the power system in an order of serial numbers of the nodes; τ represents a Lagrangian multiplier for the power balance constraint $\underset{i=1}{\overset{Ng}{.Math.}}{p}_i=\underset{i=1}{\overset{N}{.Math.}}{D}_i;$ μ.sup.L,max represents a column vector including Lagrangian multipliers for the upper bound constraint $\underset{i=1}{\overset{Ng}{.Math.}}{S}_{l,i}{p}_i-\underset{i=1}{\overset{N}{.Math.}}{S}_{l,i}{D}_i\le F_l^{\max }$ in the power system in an order of serial numbers of the branches; μ.sup.L,min represents a column vector including Lagrangian multipliers for the lower bound constraint $-F_l^{\max }\le \underset{i=1}{\overset{Ng}{.Math.}}{S}_{l,i}{p}_i-\underset{i=1}{\overset{N}{.Math.}}{S}_{l,i}{D}_i$ in the power system in an order of serial numbers of the branches; S.sub.LN represents a power transfer distribution factor matrix; and T represents a matrix transpose; (1-2) letting a variable p′.sub.i=p.sub.i−P.sub.i.sup.min to transform a decision variable p.sub.i of the clearing model into a pure non-negative variable p′.sub.i, and introducing slack variables p.sub.i.sup.sl, f.sub.l.sup.sl,min, f.sub.l.sup.sl,max to transform the clearing model into a linear programming in the general form as follows: $\underset{p_i}{\min }\underset{i=1}{\overset{N_g}{.Math.}}{c}_i{p}_i^{\prime }\mathrm{satisfying}\text{:}\underset{i=1}{\overset{N_g}{.Math.}}{p}_i=\underset{i=1}{\overset{N}{.Math.}}{D}_i-\underset{i=1}{\overset{\mathrm{Ng}}{.Math.}}{P}_i^{\min }{p}_i^{\prime }+p_i^{\mathrm{sl}}=P_i^{\max }-P_i^{\min },\forall i\in \underset{\_}{{𝒩}_g}-F_i^{\max }\le \underset{i=1}{\overset{N_g}{.Math.}}{S}_{l,i}{p}_i^{\prime }+\underset{i=1}{\overset{N_g}{.Math.}}{S}_{l,i}{P}_i^{\min }-\underset{i=1}{\overset{N}{.Math.}}{S}_{l,j}{D}_i-f_l^{\mathrm{sl},\min },\forall l\in \underset{\_}{ℒ}\underset{i=1}{\overset{N_g}{.Math.}}{S}_{l,i}{p}_i^{\prime }+\underset{i=1}{\overset{N_g}{.Math.}}{S}_{l,i}{P}_i^{\min }-\underset{i=1}{\overset{N}{.Math.}}{S}_{l,j}{D}_i+f_l^{\mathrm{sl},\max }=F_l^{\max },\forall l\in \underset{\_}{ℒ}{p}_i^{\prime },p_i^{\mathrm{sl}},f_l^{\mathrm{sl},\min },f_l^{\mathrm{sl},\max }\ge 0;$ (1-3) simplifying the clearing model obtained in (1-2) into the linear programming in the general form, as follows: $\underset{x}{\min }.Math.c.Math.x\mathrm{satisfying}\text{:}A.Math.x=b,x\ge 0$ where, matrix A and vectors c, b correspond to parameters of the clearing model as follows: $A=\left[\begin{array}{cccc}{e}_G& & & \\ I_G& I_G& & \\ S_{\mathrm{LG}}& & -I_L& \\ S_{\mathrm{LG}}& & & I_L\end{array}\right],x=\left[\begin{array}{c}{p}^{\prime }\\ p^{\mathrm{sl}}\\ f^{\mathrm{sl},\min }\\ f^{\mathrm{sl},\max }\end{array}\right],b=]\begin{array}{c}{e}_{D}D-e_G{P}^{\min }\\ P^{\max }-P^{\min }\\ S_{LN}D-S_{LG}{P}^{\min }-F^{\max }\\ S_{LN}D-S_{LG}{P}^{\min }+F^{\max }\end{array}],c=\left[\begin{array}{cccc}{c}^T& 0& 0& 0\end{array}\right]$ where, e.sub.G is a matrix whose elements of dimension 1×N.sub.g are all 1; e.sub.D is a matrix whose elements of dimension 1×N are all 1; I.sub.G is a unit matrix with dimension N.sub.g×N.sub.g; I.sub.L is a unit matrix with dimension L×L; S.sub.LG is a sub-matrix formed by columns corresponding to the generator nodes; (2) deriving and calculating the parameter changing domain of loads under the case that guarantees the constant locational marginal price based on a first-order KKT (Karush-Kuhn-Tucker) condition of the clearing model in (1-3) in the general form, comprising: (2-1) deriving the first-order KKT condition in an incremental form as follows: $\{\begin{array}{c}A.Math.x^{*}=b\\ x^{*}\ge 0\\ A^T.Math.\omega +r=c^T,r^{*}\ge 0\\ r^T.Math.x^{*}=0\end{array},$ where, ω is a Lagrangian multiplier vector of the constraint condition A.Math.x=b; r is a Lagrangian multiplier vector of the constraint condition x≥0; c, b are independent variables in the KKT condition; ω, r, x* are dependent variables in the KKT condition; supposing that in a base state c=c.sub.0 and b=b.sub.0, the dependent variables in the KKT condition being ω=ω.sub.0, r=r.sub.0, x*=x*.sub.0; wherein in order to ensure that the dependent variables ω, r remain unchanged after the independent variable b is superimposed by Δb, it is necessary to ensure that when the independent variables become c=c.sub.0 and b=b.sub.0+Δb, the dependent variables in the KKT condition satisfy the following form: ω=ω.sub.0, r=r.sub.0, x*=x*.sub.0+Δx*; in the base state c=c.sub.0, b=b.sub.0, the KKT condition is as follows: $\{\begin{array}{c}A.Math.x_0^{*}=b_0\\ x_0^{*}\ge 0\\ A^T.Math.{\omega }_0+r_0=c^T,r\ge 0\\ r_0^T.Math.x_0^{*}=0\end{array},$ when the independent variables change in the base state, the KKT condition is as follows: $\{\begin{array}{c}A.Math.\left(x_0^{*}+\underset{\_}{\Delta x^{*}}\right)=b_0+\underset{\_}{\Delta b}\\ x_0^{*}+\underset{\_}{\Delta x^{*}}\ge 0\\ A^T.Math.{\omega }_0+r_0=c^T,r_0\ge 0\\ r_0^T\left(x_0^{*}+\underset{\_}{\Delta x^{*}}\right)=0\end{array},$ an expansion equation of the first-order KKT condition in the incremental form is derived from the above equation as follows: $\{\begin{array}{c}A.Math.\underset{\_}{\Delta x^{*}}=\underset{\_}{\Delta b}\\ r_0^T.Math.\underset{\_}{\Delta x^{*}}=0\\ x_0+\underset{\_}{\Delta x^{*}}\ge 0\end{array};$ (2-2) design a projection matrix to derive the parameter changing domain of loads under the case that guarantees the constant locational marginal price: defining the projection matrix P of an equation r.sub.0.sup.T.Math.Δx*=0 as follows:
P=I−r.sub.0.Math.(r.sub.0.sup.T.Math.r.sub.0).sup.−1.Math.r.sub.0.sup.T; the expansion equation of the first-order KKT condition in the incremental form is transformed into the parameter changing domain of loads under the case that guarantees the constant locational marginal price as follows:
S{A.Math.P.Math.Δy|x*+PΔy≥0, P=I−r.sub.0.Math.(r.sub.0.sup.T.Math.r.sub.0).sup.−1.Math.r.sub.0.sup.T, ΔyϵR.sup.n}; wherein an operation of the power system is controlled based on results derived from the steps (1) and (2).

Description

DETAILED DESCRIPTION

(1) The method for calculating the parameter changing domain of loads in the power system under the case that guarantees the constant locational marginal price in the electricity market, provided in the disclosure, may include the following.

(2) (1) A clearing model on the locational marginal price in a general form is established, which has the following specific process.

(3) (1-1) The clearing model on the locational marginal price is established as follows:

(4) $\underset{p_l}{\min }\underset{i=1}{\overset{\mathrm{Ng}}{.Math.}}{c}_i{p}_i\mathrm{satisfying}\text{:}\underset{i=1}{\overset{\mathrm{Ng}}{.Math.}}{p}_i=\underset{j=1}{\overset{N}{.Math.}}{D}_j{P}_i^{\min }\le p_i\le P_i^{\max },\forall i\in {𝒩}_g-F_l^{\max }\le \underset{i=1}{\overset{\mathrm{Ng}}{.Math.}}{S}_{i,l}{p}_i=\underset{j=1}{\overset{N}{.Math.}}{S}_{i,j}{D}_j\le F_l^{\max },\forall l\in ℒ,$
where,

(5) N.sub.g represents a number of generator nodes in the power system;

(6) represents a set of serial numbers of the generator nodes, {1, 2, . . . , N.sub.g};

(7) N represents a total number of nodes in the power system;

(8) represents a set of serial numbers of branches in the power system, {1, 2, . . . , L}, L represents a total number of the branches in the power system;

(9) p.sub.i, iϵ represents a power variable of generator i and c.sub.i, iϵ represents a power cost coefficient of generator i, which are declared and confirmed by the main body of each generator to the relevant power agencies.

(10) P.sub.i.sup.max represents an upper power limit of the generator node;

(11) P.sub.i.sup.min represents a lower power limit of the generator node;

(12) D.sub.j, jϵ{1, 2, . . . , N} represents a nodal load of the power system;

(13) F.sub.l.sup.max, lϵ represents a capacity of branch l; and

(14) S.sub.l,i, lϵ, iϵ{1, 2, . . . , N} represents a power transfer distribution factor, which is calculated by relevant power agencies and released upon application.

(15) There is the following linear relationship between locational marginal prices and Lagrangian multipliers of constraints:
Λ=τ−S.sub.LN.sup.T(μ.sup.L, max−μ.sup.L, min)
where,

(16) Λ represents a column vector including locational marginal prices of the nodes in the power system in an order of serial numbers of the nodes;

(17) τ represents a Lagrangian multiplier for the power balance constraint

(18) $\underset{i=1}{\overset{Ng}{.Math.}}{p}_i=\underset{i=1}{\overset{N}{.Math.}}{D}_i;$

(19) μ.sup.L, max represents a column vector including Lagrangian multipliers for the upper bound constraint

(20) $\underset{i=1}{\overset{Ng}{.Math.}}{S}_{l,i}{p}_i-\underset{i=1}{\overset{N}{.Math.}}{S}_{l,i}{D}_i\le F_l^{\max }$
in the power system in an order of serial numbers of the branches;

(21) μ.sup.L,min represents a column vector including Lagrangian multipliers for the lower bound constraint

(22) $-F_l^{\max }\le \underset{i=1}{\overset{Ng}{.Math.}}{S}_{l,i}{p}_l-\underset{i=1}{\overset{N}{.Math.}}{S}_{l,i}{D}_i$
in the power system in an order of serial numbers of the branches;

(23) S.sub.LN represents a power transfer distribution factor matrix; and

(24) T represents a matrix transpose.

(25) (1-2) Let a variable p′.sub.i=p.sub.i−P.sub.i.sup.min, to transform a decision variable p.sub.i of the clearing model into a pure non-negative variable p′.sub.i, and slack variables p.sub.i.sup.sl, f.sub.l.sup.sl,min, f.sub.l.sup.sl,max are introduced to transform the clearing model into a linear programming in a general form as follows:

(26) $\underset{p_i}{\min }\underset{i=1}{\overset{Ng}{.Math.}}{c}_i{p}_i^{\prime }$$\begin{array}{c}\begin{array}{cc}\mathrm{satisfying}:& \underset{j=1}{\overset{Ng}{.Math.}}{p}_i^{\prime }=\underset{i=1}{\overset{N}{.Math.}}{D}_i-\underset{i=1}{\overset{Ng}{.Math.}}{P}_i^{min}\end{array}\\ p_i^{\prime }+p_i^{sl}=P_i^{max}-P_i^{min},\forall i\in {𝒩}_g\\ -F_l^{max}=\underset{l=1}{\overset{Ng}{.Math.}}{S}_{l,i}{p}_i^{\prime }+\underset{l=1}{\overset{Ng}{.Math.}}{S}_{l,i}{P}_i^{min}-\underset{i=1}{\overset{N}{.Math.}}{S}_{l,i}{D}_i-f_l^{\mathrm{sl},min},\forall l\in ℒ\\ \underset{i=1}{\overset{Ng}{.Math.}}{S}_{l,l}{p}_i^{\prime }+\underset{l=1}{\overset{Ng}{.Math.}}{S}_{l,l}{P}_i^{min}-\underset{i=1}{\overset{N}{.Math.}}{S}_{l,i}{D}_i+f_l^{sl,max}=F_l^{max},\forall l\in ℒ\\ p_i^{\prime },p_i^{xl},f_l^{xl,min},f_l^{sl,max}\ge 0\end{array}.$

(27) (1-3) The clearing model obtained in (1-2) is simplified into the linear programming in the general form, as follows:

(28) $\underset{x}{\min }.Math.c.Math.x\mathrm{satisfying}\text{:}A.Math.x=bx\ge 0$
where,

(29) matrix A and vectors c, b correspond to parameters of the clearing model as follows:

(30) $A=\left[\begin{array}{cccc}{e}_G& & & \\ I_G& I_G& & \\ S_{\mathrm{LG}}& & -I_L& \\ S_{\mathrm{LG}}& & & I_L\end{array}\right],x=\left[\begin{array}{c}{p}^{\prime }\\ p^{\mathrm{sl}}\\ f^{\mathrm{sl},\min }\\ f^{\mathrm{sl},\max }\end{array}\right],b=]\begin{array}{c}{e}_{D}D-e_G{P}^{\min }\\ P^{\max }-P^{\min }\\ S_{LN}D-S_{LG}{P}^{\min }-F^{\max }\\ S_{LN}D-S_{LG}{P}^{\min }+F^{\max }\end{array}],c=\left[\begin{array}{cccc}{c}^T& 0& 0& 0\end{array}\right]$
where,

(31) e.sub.G is a matrix whose elements of dimension 1×N.sub.g are all 1;

(32) e.sub.D is a matrix whose elements of dimension 1×N are all 1;

(33) I.sub.G is a unit matrix with dimension N.sub.g×N.sub.g;

(34) I.sub.L is a unit matrix with dimension L×L;

(35) S.sub.LG is a sub-matrix formed by columns corresponding to the generator nodes.

(36) (2) The parameter changing domain of loads under the case that guarantees the constant locational marginal price is derived and calculated based on a first-order KKT (Karush-Kuhn-Tucker) condition of the clearing model in (1-3) in the general form, which may include the following.

(37) (2-1) The first-order KKT condition in an incremental form may be derived as follows:

(38) $\{\begin{array}{c}A.Math.x^{*}=b\\ x^{*}\ge 0\\ A^T.Math.\omega +r=c^T,r^{*}\ge 0\\ r^T.Math.x^{*}=0\end{array},$
where,

(39) ω is a Lagrangian multiplier vector of the constraint condition A.Math.x=b;

(40) r is a Lagrangian multiplier vector of the constraint condition x≥0;

(41) c, b are independent variables in the KKT condition;

(42) ω, r, x* are dependent variables in the KKT condition.

(43) According to the definition of the locational marginal price in the power system, it may be seen from (1-1) that the relationship between the locational marginal price and the nodal load is equivalent to the relationship between the dependent variables ω, r and the independent variable b in the KKT condition. Therefore, “when the nodal load changes, the locational marginal price remains unchanged” is equivalent to “when the independent variable b changes, the dependent variables ω, r remain unchanged”.

(44) It is supposed that in a base state c=c.sub.0 and b=b.sub.0, the dependent variables in the KKT condition may be ω=ω.sub.0, r=r.sub.0, x*=x.sub.0. In order to ensure that the dependent variables ω, r remain unchanged after the independent variable b is superimposed by Δb, it is necessary to ensure that when the independent variables become c=c.sub.0 and b=b.sub.0+Δb, the dependent variables in the KKT condition satisfy the following form: ω=ω.sub.0, r=r.sub.0, x-=x*.sub.0+Δx*.

(45) Therefore, in the base state c=c.sub.0, b=b.sub.0, the KKT condition is as follows:

(46) 0$\{\begin{array}{c}A.Math.x_0^{*}=b_0\\ x_0^{*}\ge 0\\ A^T.Math.{\omega }_0+r_0=c^T,r\ge 0\\ r_0^T.Math.x_0^{*}=0\end{array}.$

(47) When the independent variables change in the base state, that is, c=c.sub.0 and b=b.sub.0+Δb, the KKT condition is as follows:

(48) $\{\begin{array}{c}A.Math.\left(x_0^{*}+\Delta x^{*}\right)=b_0+\Delta b\\ x_0^{*}+\Delta x^{*}\ge 0\\ A^T.Math.{\omega }_0+r_0=c^T,r_0\ge 0\\ r_0^T\left(x_0^{*}+\Delta x^{*}\right)=0\end{array}.$

(49) The expansion equation of the first-order KKT condition in the incremental form may be derived from the above equation as follows:

(50) $\{\begin{array}{c}A.Math.\Delta x^{*}=\Delta b\\ r_0^T.Math.\Delta x^{*}=0\\ x_0+\Delta x^{*}\ge 0\end{array}.$

(51) (2-2) A projection matrix is designed to derive the parameter changing domain of loads under the case that guarantees the constant locational marginal price:

(52) The projection matrix P of an equation r.sub.0.sup.T.Math.Δx*=0 is defined as follows:
P=I−r.sub.0.Math.(r.sub.0.sup.T.Math.r.sub.0).sup.−1.Math.r.sub.0.sup.T

(53) The expansion equation of the first-order KKT condition in the incremental form may be transformed into the parameter changing domain of loads under the case that guarantees the constant locational marginal price as follows:
S{A.Math.P.Math.Δy|x*+PΔy≥0,P=I−r.sub.0.Math.(r.sub.0.sup.T.Math.r.sub.0).sup.−1.Math.r.sub.0.sup.T,ΔtϵR.sup.n}.

(54) The method for calculating the parameter changing domain of loads in the power system under the case that guarantees the constant locational marginal price in the electricity market, provided in the disclosure, may have the following advantages.