Load flexibility value evaluation method based on vertex search method

Abstract

The invention discloses a load flexibility value evaluation method based on a vertex search method, which comprises the following steps of: constructing an optimization problem, separating schedulable variables of each unit in a load side resource response main body from variables for providing flexibility to the outside, obtaining a constraint relationship among self flexibility variables of each unit, and then accessing a load side resource response main body; a new projection space convex polyhedron vertex is searched through outward expansion of an existing hyperplane, and approximate approximation of flexible external characteristic projection is obtained through loop iteration; and after the load side resource response main body is introduced, solving the optimal scheduling model of the main network to obtain the main network operation cost after the load side resource response main body is introduced, and quantitatively evaluating the operation cost change caused after the flexible resource represented by the adjustable load accesses the main network. According to the method, the flexibility boundary provided by the load side resource response main body to the main network can be accurately described, and the value generated by the flexibility after the load side resource response main body accesses the main network is effectively and quantitatively evaluated.

Description

Technical Field

[0001] The invention relates to the technical field of power systems, and in particular to a load flexibility value assessment method based on a vertex search method. Background Art

[0002] At present, a large number of dispersed and small-capacity adjustable load resources are aggregated into load-side resource response entities. Through the regulation of the energy management center, they participate in the operation of the power system as a main body. The interactive electric power capacity they can provide at the system coupling port is the interconnection line power.

[0003] The essence of obtaining the flexibility range provided by the load-side resource response subject to a single electrical node of the main grid is to project the high-dimensional flexibility space into the low-dimensional flexibility space. For this type of problem, the existing methods realize the dimensionality reduction projection of the feasible domain by constructing an optimization problem, which can be divided into the following three categories: geometric parameter approximation method, multi-parameter linear programming method and vertex search method. The geometric parameter approximation method pre-selects the geometric shape of the flexibility space, and then obtains the geometric parameters of the shape by solving the optimization problem. The result of this method is directly related to the pre-selected geometric shape, which has certain subjectivity and randomness. The accuracy needs to be improved when facing the problem of high-dimensional feasible domain; the multi-parameter linear programming method regards the coupling variables of the boundary nodes of the virtual power plant as variable planning parameters, and uses the tie line power as the planning parameter to determine the feasible domain of the DC tie line power of the interconnected power grid. However, for virtual power plants with many coupling variables, the computational burden of multi-parameter planning is heavy and the computational efficiency is low.

[0004] The vertex search method constructs an optimization problem and iteratively searches for the vertex of the flexible feasible domain to approximate the feasible domain. Compared with the first two methods, it has great advantages in calculation accuracy and speed, and is the most commonly used method in current research. Current research focuses on the operation strategy of the system after the flexible load is connected, and lacks a quantitative evaluation method for the value generated by different flexible loads. To this end, an effective value evaluation analysis model should be established to analyze the value of load flexibility and provide a basis for decision-making at all levels of government. Summary of the invention

[0005] In view of this, in order to accurately characterize the flexibility boundary provided by the load-side resource response subject to the main grid and quantitatively evaluate the value of the flexibility generated after the load-side resource response subject is connected to the main grid, the present invention provides a load flexibility value evaluation method based on the vertex search method to realize the projection of the high-dimensional flexibility space to the low-dimensional flexibility space.

[0006] The present invention discloses a load flexibility value evaluation method based on a vertex search method, which comprises:

[0007] Step 1: Construct an optimization problem, separate the dispatchable variables of each unit in the load-side resource response subject and the variables that provide flexibility to the outside, obtain the constraint relationship between the flexibility variables of each unit itself, and then connect them to the load-side resource response subject;

[0008] Step 2: Use the existing hyperplane to expand outward to find the vertices of the new projection space convex polyhedron, and obtain the approximate approximation of the projection of the flexibility external characteristics through cyclic iteration;

[0009] Step 3: After introducing the load-side resource response subject, solve the optimization scheduling model of the main grid to obtain the main grid operation cost after the introduction of the load-side resource response subject, and quantitatively evaluate the change in operation cost caused by the flexible resources represented by adjustable loads being connected to the main grid.

[0010] Furthermore, the step 1 comprises:

[0011] Taking DC power flow as an example, the node power constraints and transmission capacity constraints that should be satisfied are:

[0012] P m,i =P m,g -P m,l -P m,e (1)

[0013] P lmin ≤P n,l ≤P lmax (2)

[0014] Where P m,i represents the injected active power of the mth node inside the load-side resource response body; P m,g represents the active power provided by each generator set in the virtual power plant to the mth node; P m,l represents the active power consumed by the mth node load in the virtual power plant; P m,e represents the active power provided by the mth node in the virtual power plant to the main grid; P n,l represents the active power exchanged on the nth branch within the virtual power plant, P lmax and P lmin are the upper and lower limits of the exchange power;

[0015] For the load-side resource response entity, the internal operation constraints of each component unit are as follows:

[0016] f e (P q,o ,X q,h )=0 (3)

[0017] f ne (P q,o ,X q,h )≤0 (4)

[0018] Where P q,o represents the oth variable related to providing flexibility within the qth component unit of the load-side resource response subject; X q,h Then it is the hth variable in the internal variables of the qth component unit that is only related to its own operation; fe() and fne() represent the equality and inequality constraints of the internal operation of each component unit respectively;

[0019] Formulas (3) and (4) are finally transformed into the following constraints:

[0020] k e (P q,o )=0 (5)

[0021] k ne (P q,o )≤0 (6)

[0022] Where ke() and kne() represent the equality and inequality constraints between variables related to providing flexibility within each component unit, respectively.

[0023] Furthermore, the step 2 comprises:

[0024] Step 21: Initialize the projection plane, aggregate the dispatchable variables of each unit in the load-side resource response subject and the variables that provide flexibility to the outside into the output power of the load-side resource response subject node, that is, project the high-dimensional space onto the projection plane in the low-dimensional space to obtain the initial vertex set on the projection plane;

[0025] Step 22: The vertices in the vertex set are combined to form a plurality of high-dimensional initial hyperplanes, and the ideal low-dimensional projection surface is approximated by extrapolating the initial hyperplanes.

[0026] Furthermore, the step 21 comprises:

[0027] The objective function of the optimization problem is constructed as:

[0028] max P q,o (7)

[0029] min P q,o (8)

[0030] The constraints satisfied by the objective function are the internal operating constraints of each component unit;

[0031] By solving the optimization problem, we get the initial vertex set on the low-dimensional plane. The vertex set consists of vertices in each dimension. The vertex set is expressed as:

[0032] S0=[q max,1 ,q min,1 ,qmax,2 ,q min,2 ,...,q max,i ,q min,i ,] (9)

[0033] Among them, q max,i and q min,i They respectively represent the maximum and minimum projection points corresponding to the i-th dimension in the low-dimensional projection surface; S0 represents the set of initial vertices on the low-dimensional projection surface.

[0034] Further, the step 22 comprises:

[0035] Step 221: construct a hyperplane parallel to the initial hyperplane, under the constraint that the hyperplane has an intersection with the ideal projection plane, that is, ensure that there are feasible points on the extrapolated hyperplane, obtain the maximum distance of extrapolation, and the feasible point obtained under the maximum distance is the new vertex obtained by extrapolation;

[0036] Step 222: Update the existing vertex set with the obtained new vertex to obtain a new vertex set;

[0037] Step 223: Repeat the extrapolation process in the new vertex set, and obtain an approximate approximation of the ideal low-dimensional projection surface after a cycle.

[0038] Furthermore, the step 221 includes:

[0039] The mathematical form of any initial hyperplane is:

[0040] TQ= 1 (10)

[0041] Among them, T represents the normal vector of the hyperplane, and Q represents the variables of each dimension in the low-dimensional projection surface;

[0042] The extrapolated plane parallel to the initial hyperplane is constructed as:

[0043] TQ= d (11)

[0044] Among them, d is a parameter that characterizes the extrapolation distance;

[0045] The distance formula between the initial hyperplane and the constructed hyperplane is:

[0046]

[0047] Among them, d line is the distance between the initial hyperplane and the extrapolated hyperplane; ||T|| is the L2 norm of the normal vector;

[0048] Maximize d line Equivalent to maximizing the value of d, the constraints of the extrapolation process are internal operation constraints, and the objective function is:

[0049] max d=TQ (13)

[0050] After traversing all initial hyperplanes, new projection plane vertices are obtained.

[0051] Further, the step 222 includes:

[0052] The updated vertex set S is:

[0053] S1=S0+[q new1 ,q new2 , .....] (14)

[0054] After obtaining a new set of vertices, repeat the above steps and continue to extrapolate the new hyperplane to obtain new vertices.

[0055] Further, the step 223 includes:

[0056] After each extrapolation, it is necessary to judge whether the loop is over. If no new vertices are generated or the extrapolation distance of all extrapolated hyperplanes in this loop is less than the preset value, the loop ends and an approximation of the ideal low-dimensional projection surface is obtained.

[0057] Furthermore, in the step 21, the projection plane is initialized, that is, the maximum value point and the minimum value point of each flexibility variable are individually obtained;

[0058] The step of obtaining the maximum and minimum values ​​of each flexibility variable includes:

[0059] The scope of flexibility provided by the load-side resource response subject is obtained by solving the optimization problem. The objective function is:

[0060]

[0061] The constraints satisfied by this optimization problem are formulas (1), (2), (5), and (6). Solving this optimization problem yields the maximum flexibility that the virtual power plant can provide to the nodes of the main grid. The minimum flexibility is obtained by constructing a minimization problem.

[0062] Furthermore, the step 3 comprises:

[0063] After introducing the load-side resource response subject, the optimization scheduling model of the main network is:

[0064] minC(P ps ,P e )(16)

[0065] g ie (P ps ,P e )≤0 (17)

[0066] g ie (P ps ,P e )=0 (18)

[0067]

[0068] Where, the objective function is to minimize the operating cost of the main network; P ps and P e They represent variables related to the operation of the main grid itself and variables related to the flexibility provided by the load-side resource response subject to the main grid; constraints (17) and (18) represent the power flow constraints satisfied when the main grid is running; constraint (19) is the upper and lower limit constraints of the flexibility variable; the main grid operation cost after the introduction of the load-side resource response subject is obtained by solving the optimal scheduling model;

[0069] The flexibility value provided by the load-side resource response subject is regarded as the change in the main network operation cost caused by access, and the main network operation optimization problem is regarded as a linear programming. The optimal solution of the linear programming is the geometric intersection of multiple hyperplanes. When the resource limit changes, that is, some hyperplanes are translated, the intersection between the hyperplanes also changes accordingly.

[0070] When all planes are translated, the change of the optimal solution satisfies the superposition property, and the change of the final optimal solution and the change of the optimal value are:

[0071]

[0072] Δz=c B T B -1 Δb (30)

[0073] Among them, Δb1, Δb2, and Δb3 are the changes in the resource limits corresponding to the three intersecting planes, and B is the resource consumption matrix corresponding to the basis vectors.

[0074] Due to the adoption of the above-mentioned technical scheme, the present invention has the following advantages: various types of flexibility inside the load-side resource response subject are aggregated into the output power of the load-side resource response subject node, and its mathematical essence is the projection of the high-dimensional flexibility space into the low-dimensional flexibility space. Considering the internal network architecture of the load-side resource response subject, the vertex search method based on extrapolation is used to approximate the projection space by maximizing the Euclidean distance between hyperplanes. At the same time, based on the external characteristic equivalence method of the load-side resource response subject, a load flexibility resource value evaluation method is proposed to explain the changes in the main network operating cost and the optimal solution before and after the load-side resource response subject is connected from a geometric perspective. The proposed flexibility aggregation model can effectively reduce the operating cost of the main network, and the proposed load-side resource response subject value evaluation method based on the external characteristic equivalence can effectively evaluate the value generated by various types of flexible loads when the flexibility range changes slightly. BRIEF DESCRIPTION OF THE DRAWINGS

[0075] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the embodiments of the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0076] Figure 1 It is the projection diagram of the maximum value point of the embodiment of the present invention;

[0077] Figure 2 A schematic diagram of a hyperplane extrapolation process according to an embodiment of the present invention;

[0078] Figure 3 A flowchart of a flexibility characterization method based on vertex search according to an embodiment of the present invention;

[0079] Figure 4 A schematic diagram of the flexibility variation geometry of an embodiment of the present invention;

[0080] Figure 5 This is a schematic diagram of the optimal solution change of an embodiment of the present invention;

[0081] Figure 6 The present invention is a flowchart of a method for evaluating the value of load flexibility based on a vertex search method. DETAILED DESCRIPTION

[0082] The present invention is further described in conjunction with the accompanying drawings and embodiments, and the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by ordinary technicians in this field should fall within the scope of protection of the embodiments of the present invention.

[0083] See also Figure 6The present invention provides an embodiment of a load flexibility value evaluation method based on a vertex search method, which comprises:

[0084] Step 1: Construct an optimization problem, separate the dispatchable variables of each unit in the load-side resource response subject and the variables that provide flexibility to the outside, obtain the constraint relationship between the flexibility variables of each unit itself, and then connect them to the load-side resource response subject;

[0085] Step 2: Use the existing hyperplane to expand outward to find the vertices of the new projection space convex polyhedron, and obtain the approximate approximation of the projection of the flexibility external characteristics through cyclic iteration;

[0086] Step 3: After introducing the load-side resource response subject, solve the optimization scheduling model of the main grid to obtain the main grid operation cost after the introduction of the load-side resource response subject, and quantitatively evaluate the change in operation cost caused by the flexible resources represented by adjustable loads being connected to the main grid.

[0087] The step 1 comprises:

[0088] The distributed energy sources within the load-side resource response body can form a power system similar to the distribution network topology. Like the traditional power system, the virtual power plant also has similar constraints when operating. This application discusses the DC power flow, which should meet the node power constraints and transmission capacity constraints as shown below:

[0089] P m,i =P m,g -P m,l -P m,e (1)

[0090] P lmin ≤P n,l ≤P lmax (2)

[0091] Where P m,i represents the injected active power of the mth node inside the load-side resource response body; P m,g represents the active power provided by each generator set in the virtual power plant to the mth node; P m,l represents the active power consumed by the mth node load in the virtual power plant; P m,e represents the active power provided by the mth node in the virtual power plant to the main grid; P n,l represents the active power exchanged on the nth branch within the virtual power plant, P lmax and P lmin are the upper and lower limits of the exchange power.

[0092] For each component unit of the load-side resource response subject, such as controllable loads, they also need to meet their own operating constraints during operation. For example, when industrial high-energy-load equipment is in operation, there are constraints between its power consumption, operating temperature, floor space and other variables; when temperature-controlled loads are in operation, there are constraints between its power consumption, building temperature variation coefficient, temperature variation and other variables. The internal operating constraints of each component unit are as follows:

[0093] f e (P q,o ,X q,h )=0 (3)

[0094] f ne (P q,o ,X q,h )≤0 (4)

[0095] Among them, P q,o represents the oth variable related to providing flexibility within the qth component unit of the load-side resource response subject, such as active power, etc.; X q,h It is the hth variable among the internal variables of the qth component unit that is only related to its own operation, such as the temperature and floor area mentioned above; fe() and fne() respectively represent the equality and inequality constraints of the internal operation of each component unit.

[0096] When the load-side resource response body is working, among the variables of each internal component unit, only the variables related to providing flexibility are directly related to the subsequent optimization scheduling of the main network. For other variables, it is only necessary to ensure that they do not violate their own operating constraints during the optimization process. Therefore, it is only necessary to obtain the constraints between the flexibility variables. The mathematical essence of this problem is to project the high-dimensional flexibility space of the original variables of each component unit into the low-dimensional flexibility space of only the variables related to flexibility. Based on this idea, the constraints of formulas (3) and (4) can be finally transformed into the following constraints:

[0097] k e (P q,o )=0 (5)

[0098] k ne (P q,o )≤0 (6)

[0099] Among them, ke() and kne() represent the equality and inequality constraints between variables related to providing flexibility within each component unit.

[0100] The step 2 comprises:

[0101] Step 21: Initialize the projection plane:

[0102] The mathematical essence of the initialization process is to obtain the maximum and minimum points of each flexibility variable separately without considering the coupling relationship between the flexibility variables. This process can be achieved by constructing a relatively simple optimization problem, and its objective function is:

[0103] max P q,o (7)

[0104] min P q,o (8)

[0105] The satisfied constraints are the self-operation constraints, as shown in formulas (3) and (4).

[0106] By solving the above problem, we can get the initial vertex set on the low-dimensional plane (composed of vertices in each dimension), which can be expressed as follows:

[0107] S0=[q max,1 ,q min,1 ,q max,2 ,q min,2 ,...,q max,i ,q min,i ,] (9)

[0108] Among them, q max,i and q min,i They represent the maximum and minimum projection points of the i-th dimension in the low-dimensional projection surface respectively; S0 represents the set of initial vertices on the low-dimensional projection surface, such as Figure 1 shown.

[0109] Step 22: Extrapolation of the projection surface:

[0110] The vertices in the set S0 can be combined to form multiple high-dimensional initial hyperplanes. By extrapolating the hyperplane, the ideal low-dimensional projection surface can be continuously approached. The extrapolation process is also implemented by constructing an optimization problem. The core idea is: construct a hyperplane parallel to the initial hyperplane, and obtain the maximum distance of extrapolation under the constraint that the hyperplane has an intersection with the ideal projection surface (that is, to ensure that there are feasible points on the extrapolated hyperplane). The feasible point obtained under this distance is the new vertex obtained by extrapolation; after obtaining the new vertex, update the set S, and repeat the above extrapolation process in the new set. After a cycle, an approximate approximation of the ideal low-dimensional projection surface can be obtained, such as Figure 2 .

[0111] For any initial hyperplane, it can be expressed in the following mathematical form:

[0112] TQ= 1 (10)

[0113] Where T represents the normal vector of the hyperplane, and Q represents the variables of each dimension in the low-dimensional projection surface. The extrapolated plane parallel to it can be constructed as:

[0114] TQ= d (11)

[0115] Among them, d is a parameter that characterizes the extrapolation distance. In order to achieve sufficient extrapolation, the distance between the initial hyperplane and the constructed hyperplane needs to be as large as possible. The distance formula between the two hyperplanes is:

[0116]

[0117] Among them, d line is the distance between the initial hyperplane and the extrapolated hyperplane; ||T|| is the L2 norm of the normal vector. Since this process is extrapolation, d>1, and when the initial plane is selected, ||T|| is a constant, so maximizing d line This is equivalent to maximizing the value of d. Therefore, the constraints of the extrapolation process are equations (3) and (4), and the objective function is as follows:

[0118] max d=TQ (13)

[0119] After traversing the initial hyperplane, we can get new projection plane vertices that are no more than the number of hyperplanes, and then update the vertex set S as follows:

[0120] S1=S0+[q new1 ,q new2 , .....] (14)

[0121] After obtaining a new set of vertices, repeat the above steps and continue to extrapolate the new hyperplane to obtain new vertices.

[0122] Step 23: Loop end judgment:

[0123] After each extrapolation, it is necessary to determine whether the loop is over. If no new vertices are generated or the extrapolation distance of all extrapolated hyperplanes in this loop is less than a preset value, the loop ends. Figure 3 .

[0124] The flexibility provided by the load-side resource response subject to a single node of the main grid is the sum of the flexibility provided by each node within the virtual power plant to the main grid node. Therefore, by solving the optimization problem, the range of flexibility provided by the load-side resource response subject can be obtained. The objective function is as follows:

[0125]

[0126] The constraints satisfied by this optimization problem are formulas (1), (2), (5), and (6). This optimization problem can be solved to obtain the maximum flexibility that the virtual power plant can provide to the main grid node. Similarly, the minimum value of flexibility can be obtained by constructing a minimization problem.

[0127] The step 3 comprises:

[0128] The introduction of the load-side resource response subject enables the decentralized adjustable load resources to be effectively organized and managed before participating in the main grid dispatch. It directly provides the main grid with variable information related to flexibility regulation, effectively avoiding the information exchange barriers caused by information privacy between different energy entities. At the same time, the main grid can also avoid direct dispatch of the internal variables of numerous decentralized adjustable load resources when optimizing dispatch, which can greatly reduce the computational burden.

[0129] The load-side resource response subject directly provides flexibility to a single main grid node, which is equivalent to connecting a controllable load or generator set to the single node, which can flexibly adjust the power load according to the needs of the power market. For new power systems, the increasing proportion of renewable energy grid connection has brought considerable uncertainty and randomness to the power grid. The introduction of flexible resources such as decentralized adjustable load resources can mitigate this negative impact to a certain extent and improve the stability and security of the power system.

[0130] After introducing the load-side resource response subject, the optimization scheduling model of the main network is as follows:

[0131] min C(P ps ,P e ) (16)

[0132] g ie (P ps ,P e )≤0 (17)

[0133] g ie (P ps ,P e )=0 (18)

[0134]

[0135] Among them, the objective function is to minimize the operating cost of the main grid; Pps and Pe represent the variables related to the operation of the main grid itself and the variables related to the flexibility provided by the load-side resource response subject to the main grid; constraints (17) and (18) represent the flow constraints satisfied when the main grid is running; constraint (19) is the upper and lower limit constraints of the flexibility variable. By solving this optimization problem, the operating cost of the main grid after the introduction of the load-side resource response subject can be obtained.

[0136] After the load-side resource response subject is connected, it will provide flexibility services to the main grid, which can provide spare capacity for stabilizing grid frequency, load peak regulation, and renewable energy consumption, and can also reduce the grid operation cost to a certain extent. In order to quantitatively evaluate the changes in operation caused by the connection of flexible resources represented by adjustable loads to the main grid, this section proposes a flexibility value evaluation method based on external characteristic equivalence. Under the condition that the optimal basis remains unchanged, it explains the changes in the optimal solution after the connection of flexible resources from a geometric perspective. The flexibility value provided by the load-side resource response subject is regarded as the cost change caused by the connection.

[0137] The present invention discusses the main network operation optimization problem as a linear programming, and the standard form of linear programming is as follows:

[0138] max c T x(20)

[0139]

[0140] Among them, c T represents the value coefficient vector, b is the resource limit vector, A is the resource consumption matrix, and x is the decision variable vector.

[0141] If a set of basic variables is selected, the above problem can be transformed into the following form:

[0142] max c B T B -1 b+(c N T -c B T B -1 N)x N (twenty two)

[0143]

[0144] Where B and N represent the resource consumption matrices corresponding to basis vectors and non-basis vectors respectively; c B T and c N T Respectively represent the value coefficient vectors corresponding to basis vectors and non-basis vectors; B and x N represent the decision variables corresponding to basis vectors and non-basis vectors respectively.

[0145] The difference between whether the load-side resource response subject is connected or not is whether flexibility is provided to a single electrical node of the main grid. Mathematically, this difference is reflected in the modification of the resource restriction vector. Since the flexibility that a single load-side resource response subject can provide is limited, its impact on the resource restriction vector is small and will not affect the feasibility of the optimal basis. Therefore, it is believed that the optimal basis will not change before and after the load-side resource response subject is connected.

[0146] To ensure the optimality of the solution, the value of the non-basis vector is 0, so the optimal solution can be uniquely determined by the constraint equation. From a geometric point of view, the optimal solution is the geometric intersection of multiple hyperplanes. When the resource constraint changes, that is, some hyperplanes are translated, the intersection between the hyperplanes also changes accordingly. Figure 4 Taking the change of the intersection of three-dimensional planes as an example, the process is explained geometrically, such as Figure 5 .

[0147] When the resource limit corresponding to plane 3 changes by Δb3, the plane translation distance is:

[0148]

[0149] The distance the intersection point moves is:

[0150]

[0151] The cosine value in the above formula can be calculated using the following formula:

[0152]

[0153] Among them, θ represents the angle between the normal vector of the moving plane and the intersection line of the other two planes; the direction of the intersection line should be perpendicular to the normal vectors of planes 1 and 2; ||T3|| is the L2 norm of the normal vector of plane 3; ||T12|| is the L2 norm of the intersection line of planes 1 and 2.

[0154] It is worth noting that the components of each dimension of the line perpendicular to the normal vectors of planes 1 and 2 can be obtained from the algebraic co-factors of each dimension of plane 3 in the B matrix. Therefore, the value of the numerator of the above formula is the determinant value of the B matrix. Therefore, the distance moved by the intersection point can be finally expressed as follows:

[0155]

[0156] According to the previous analysis, the direction of the intersection change can be obtained by finding the algebraic co-factors of each dimension of plane 3 in the B matrix, so the final optimal solution change is:

[0157]

[0158] When all planes are translated, since the normal vectors of each plane remain unchanged, the change of the optimal solution satisfies the superposition property, and the final optimal solution and the change of the optimal value are as follows:

[0159]

[0160] Δz=cB T B -1 Δb (30)

[0161] The present application is further explained below in conjunction with a specific implementation plan:

[0162] 1.1 Accuracy evaluation of flexibility characterization methods

[0163] In order to evaluate the accuracy of the load flexibility characterization method based on vertex search proposed in this application, this section randomly samples from the load-side resource response subject flexibility range obtained by projection to verify the feasibility of the sample point.

[0164] The simulation samples the projection range 1000 times, and finally all sample points can meet the actual operation constraints. That is, under linear constraints, this method can completely restore the flexibility space.

[0165] Although this method can be very close to the actual projection space under linear constraints, it has limitations when facing nonlinear constraints. In the actual load-side resource response body, there are a large number of nonlinear constraints in addition to linear constraints, such as the introduction of binary variables by energy storage equipment and the non-convex characteristics brought about by the unit climbing constraints. Considering the time domain correlation of the unit, the constraint projection containing integer variables, and the high-dimensional flexibility space dimensionality reduction projection with uncertainty of new energy output will be the focus of subsequent research.

[0166] (II) Comparison of the flexibility provided by various types of flexible loads and their impact on the main grid operation cost

[0167] In order to compare the differences in flexibility that different flexible loads can provide within the load-side resource response entity, 5 temperature-controlled air-conditioning loads, 5 industrial high-energy loads and 5 V2G devices were connected to the load-side resource response entity, and the changes in the flexibility of the load-side resource response entity when the number of corresponding devices was reduced and the changes in the main network operating costs after connecting to the main network were compared.

[0168] It can be found that the reduction of V2G equipment and industrial high-energy-load equipment will cause the load-side resource response subject to lose some flexibility in absorbing the main grid power, but due to the decrease in internal load demand, it can provide additional flexibility to the main grid load, while the load-side resource response subject is the opposite; when providing power to the main grid, the reduction of temperature-controlled air-conditioning equipment has the greatest impact on flexibility, followed by V2G equipment, and in comparison, industrial high-energy-load equipment has the least impact on flexibility; when absorbing the main grid power, the impact of industrial high-energy-load equipment on flexibility is significantly reduced; when the number of equipment decreases, the flexibility shows a "saturation" trend. The reason is that when a single node of the load-side resource response subject provides flexibility to the main grid, it is limited by the power constraints of the branch connected to the node, and the power that the node can output or absorb is limited.

[0169] The reduction of V2G equipment and industrial high-energy-load equipment will reduce the main grid operation cost, while the reduction of temperature-controlled air-conditioning equipment will reduce the main grid operation cost. The reason is that the reduction of the former will increase the flexibility of the output power of the load-side resource response subject, and the main grid can perform scheduling optimization in a wider feasible domain, while the reduction of temperature-controlled air-conditioning equipment will reduce the main grid operation cost. The temperature-controlled air-conditioning equipment has the greatest impact on the operation cost, followed by V2G equipment, and the industrial high-energy-load equipment has the least impact.

[0170] (III) Validation of the value assessment method based on external characteristic equivalence

[0171] In order to verify the effectiveness of the flexible load value assessment method based on geometric derivation proposed in this invention, the three flexibility resource values ​​are ranked and the results are compared with the solution results of the optimization algorithm.

[0172] When the range of flexibility provided by the load-side resource response subject is small, the relative value of the main network cost change caused by unit equipment is ranked as follows:

[0173] Table 1 Comparison of value assessment methods based on external characteristic equivalence and optimization results

[0174]

[0175] It can be seen from Table 1 that the value brought by the flexibility provided by the temperature-controlled air conditioner is much higher than that of the other two types of equipment, and the geometry-based value evaluation method adopted by the present invention is basically consistent with the results of the optimization algorithm. This method has a more accurate effect when the flexibility variation range of the temperature-controlled air conditioner is small.

[0176] The present invention proposes a load flexibility model and a vertex search method based on extrapolation. In the flexibility model, each decentralized adjustable load unit participates in the main network scheduling by reporting its own flexibility variable information to the energy management system, avoiding the interaction between the main network and the operating variables inside the unit in the traditional scheduling process. This process greatly reduces the number of scheduling variables, so that the originally heavy computational burden can meet the actual needs of the project. When the main network is scheduled, it is necessary to obtain the coupling relationship between the flexibility variables. The mathematical essence of this problem is the projection of the high-dimensional flexibility space to the low-dimensional flexibility space, and the low-dimensional space is the coupling space between the variables related to flexibility. The introduction of the vertex search method based on extrapolation can accurately approximate the projection space. By maximizing the Euclidean distance of the hyperplane, the existing hyperplane is extrapolated to obtain new vertices. After the new vertices form a new hyperplane, they continue to be extrapolated, and so on and so forth until no new vertices appear or the extrapolation distance is less than the preset accuracy. A load flexibility value evaluation method based on external characteristic equivalence is proposed. Based on the optimal basis invariant assumption, the optimal solution of the main network operation and the optimal value of the operating cost after the load-side resource response subject is connected are analyzed and explained from a geometric perspective. The calculation examples show that the load-side resource response subject flexibility model proposed in the present invention can effectively reduce the operating cost of the main grid, and the proposed vertex search method based on extrapolation has a solution efficiency that can be applied in practice; and the flexible load value evaluation method based on external characteristic equivalence proposed in this application can effectively evaluate the value generated by various types of flexibility equipment when the flexibility capacity changes slightly.

[0177] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A load flexibility value assessment method based on vertex search method, characterized in that: include: Step 1: Construct an optimization problem, separate the dispatchable variables of each unit in the load-side resource response subject and the variables that provide flexibility to the outside, obtain the constraint relationship between the flexibility variables of each unit itself, and then connect them to the load-side resource response subject; Step 2: Use the existing hyperplane to expand outward to find the vertices of the new projection space convex polyhedron, and obtain the approximate approximation of the projection of the flexibility external characteristics through cyclic iteration; Step 3: After introducing the load-side resource response subject, solve the optimization scheduling model of the main grid to obtain the main grid operation cost after the introduction of the load-side resource response subject, and quantitatively evaluate the change in operation cost caused by the flexible resources represented by adjustable loads being connected to the main grid.

2. The method according to claim 1, characterized in that The step 1 comprises: Taking DC power flow as an example, the node power constraints and transmission capacity constraints that should be satisfied are: P m,i =P m,g -P m,l -P m,e (1) P lmin ≤P n,l ≤P lmax (2) Where P m,i represents the injected active power of the mth node inside the load-side resource response body; P m,g represents the active power provided by each generator set in the virtual power plant to the mth node; P m,l represents the active power consumed by the mth node load in the virtual power plant; P m,e represents the active power provided by the mth node in the virtual power plant to the main grid; P n,l represents the active power exchanged on the nth branch within the virtual power plant, P lmax and P lmin are the upper and lower limits of the exchange power; For the load-side resource response entity, the internal operation constraints of each component unit are as follows: f e (P q,o ,X q,h )=0 (3) f ne (P q,o ,X q,h )≤0 (4) Where P q,o represents the oth variable related to providing flexibility within the qth component unit of the load-side resource response subject; X q,h Then it is the hth variable in the internal variables of the qth component unit that is only related to its own operation; fe() and fne() represent the equality and inequality constraints of the internal operation of each component unit respectively; Formulas (3) and (4) are finally transformed into the following constraints: k e (P q,o )=0 (5) k ne (P q,o )≤0 (6) Where ke() and kne() represent the equality and inequality constraints between variables related to providing flexibility within each component unit.

3. The method according to claim 1, characterized in that The step 2 comprises: Step 21: Initialize the projection plane, aggregate the dispatchable variables of each unit in the load-side resource response subject and the variables that provide flexibility to the outside into the output power of the load-side resource response subject node, that is, project the high-dimensional space onto the projection plane in the low-dimensional space to obtain the initial vertex set on the projection plane; Step 22: The vertices in the vertex set are combined to form a plurality of high-dimensional initial hyperplanes, and the ideal low-dimensional projection surface is approximated by extrapolating the initial hyperplanes.

4. The method according to claim 3, characterized in that The step 21 comprises: The objective function of the optimization problem is constructed as: max P q,o (7)min P q,o (8) The constraints satisfied by the objective function are the internal operating constraints of each component unit; By solving the optimization problem, we get the initial vertex set on the low-dimensional plane. The vertex set consists of vertices in each dimension. The vertex set is expressed as: S0=[q max,1 ,q min,1 ,q max,2 ,q min,2 ,...,q max,i ,q min,i ,] (9) Among them, q max,i and q min,i They respectively represent the maximum and minimum projection points corresponding to the i-th dimension in the low-dimensional projection surface; S0 represents the set of initial vertices on the low-dimensional projection surface.

5. The method according to claim 3, characterized in that: The step 22 comprises: Step 221: construct a hyperplane parallel to the initial hyperplane, under the constraint that the hyperplane has an intersection with the ideal projection plane, that is, ensure that there are feasible points on the extrapolated hyperplane, obtain the maximum distance of extrapolation, and the feasible point obtained under the maximum distance is the new vertex obtained by extrapolation; Step 222: Update the existing vertex set with the obtained new vertex to obtain a new vertex set; Step 223: Repeat the extrapolation process in the new vertex set, and obtain an approximate approximation of the ideal low-dimensional projection surface after a cycle.

6. The method according to claim 5, characterized in that The step 221 includes: The mathematical form of any initial hyperplane is: TQ= 1 (10) Among them, T represents the normal vector of the hyperplane, and Q represents the variables of each dimension in the low-dimensional projection surface; The extrapolated plane parallel to the initial hyperplane is constructed as: TQ= d (11) Among them, d is a parameter that characterizes the extrapolation distance; The distance formula between the initial hyperplane and the constructed hyperplane is: Among them, d line is the distance between the initial hyperplane and the extrapolated hyperplane; ||T|| is the L2 norm of the normal vector; Maximize d line Equivalent to maximizing the value of d, the constraints of the extrapolation process are internal operation constraints, and the objective function is: max d=TQ (13) After traversing all initial hyperplanes, new projection plane vertices are obtained.

7. The method according to claim 6, characterized in that The step 222 comprises: The updated vertex set S is: S1=S0+[q new1 ,q new 2,.....] (14) Among them, q newi A collection of newly added vertices; After obtaining a new set of vertices, repeat the above steps and continue to extrapolate the new hyperplane to obtain new vertices.

8. The method according to claim 1, characterized in that: The step 223 includes: After each extrapolation, it is necessary to judge whether the loop is over. If no new vertices are generated or the extrapolation distance of all extrapolated hyperplanes in this loop is less than the preset value, the loop ends and an approximation of the ideal low-dimensional projection surface is obtained.

9. The method according to claim 1, characterized in that: In the step 21, the projection plane is initialized, that is, the maximum value point and the minimum value point of each flexibility variable are individually obtained; The step of obtaining the maximum and minimum values ​​of each flexibility variable includes: The scope of flexibility provided by the load-side resource response subject is obtained by solving the optimization problem. The objective function is: The constraints satisfied by this optimization problem are formulas (1), (2), (5), and (6). Solving this optimization problem yields the maximum flexibility that the virtual power plant can provide to the nodes of the main grid. The minimum flexibility is obtained by constructing a minimization problem.

10. The method according to claim 1, characterized in that The step 3 comprises: After introducing the load-side resource response subject, the optimization scheduling model of the main network is: minC(P ps, P e )(16) g ie (P ps, P e )≤0 (17) g ie (P ps ,P e )=0 (18) Where, the objective function is to minimize the operating cost of the main network; P ps and P e They represent variables related to the operation of the main grid itself and variables related to the flexibility provided by the load-side resource response subject to the main grid; constraints (17) and (18) represent the power flow constraints satisfied when the main grid is running; constraint (19) is the upper and lower limit constraints of the flexibility variable; the main grid operation cost after the introduction of the load-side resource response subject is obtained by solving the optimal scheduling model; The flexibility value provided by the load-side resource response subject is regarded as the change in the main network operation cost caused by access, and the main network operation optimization problem is regarded as a linear programming. The optimal solution of the linear programming is the geometric intersection of multiple hyperplanes. When the resource limit changes, that is, some hyperplanes are translated, the intersection between the hyperplanes also changes accordingly. When all planes are translated, the change of the optimal solution satisfies the superposition property, and the change of the final optimal solution and the change of the optimal value are: Δz=c B T B -1 Δb (30) Among them, Δx is the change of the optimal solution, Δz is the change of the optimal value, Δb1, Δb2, and Δb3 are the changes in the resource constraints corresponding to the three intersecting planes, B is the resource consumption matrix corresponding to the basis vector, and M ij is the algebraic cofactor of each dimension of the plane, b is the resource constraint vector in the optimization problem, Δb is the change of the resource constraint vector, c B T is the value coefficient vector corresponding to the basis vector.

Images