Novel distribution network multi-virtual power plant distributed resource aggregation and optimization operation method

Abstract

The invention discloses a novel distribution network multi-virtual power plant distributed resource aggregation and optimization operation method. The method comprises the following steps: 1, solving a virtual power plant feasible region based on a vertex enumeration method; 2, realizing multi-virtual power plant cooperative operation optimization based on a binary consistency algorithm; 3, based on the virtual power plant feasible region solving method based on the vertex enumeration method and the multi-virtual power plant cooperative operation optimization method based on the binary consistency algorithm, constructing a resource aggregation and aggregate optimization operation framework process; in the process, firstly, distributed resources are aggregated through a vertex enumeration method to obtain a virtual power plant feasible region; and carrying out collaborative operation optimization on the multiple virtual power plants by utilizing a binary consistency algorithm. By means of the method, efficient aggregation and optimized operation of distributed resources are achieved, the operation efficiency and economical efficiency of the power distribution network are improved, and the method has important economic and social benefits.

Description

Technical Field

[0001] The present invention relates to the technical field of power system dispatching and optimization, and in particular to a method for aggregating and optimizing distributed resources of multiple virtual power plants in a novel power distribution network. Background Art

[0002] With the rapid development of distributed energy and the gradual opening of the power market, virtual power plants (VPPs) have attracted widespread attention as a new type of energy management and scheduling model. Virtual power plants use advanced information and communication technologies to organically aggregate distributed power sources, energy storage systems, electric vehicles and other resources to achieve coordinated and optimized scheduling of these resources. However, existing virtual power plant optimization scheduling methods often ignore the synergy between resources, especially in the application of new distribution networks, and therefore fail to fully tap the potential of virtual power plants. Summary of the invention

[0003] The purpose of the present invention is to provide a novel method for aggregating and optimizing the distributed resources of multiple virtual power plants in a distribution network, thereby achieving efficient utilization of resources and improving the operating efficiency and economy of the distribution network by establishing an aggregation model of distributed resources and optimizing the scheduling strategy.

[0004] To achieve the above object, the present invention provides the following technical solutions:

[0005] Step 1: Solve the feasible region of the virtual power plant based on the vertex enumeration method;

[0006] 1.1) Classify distributed resources, including distributed power sources, energy storage systems, electric vehicles and controllable loads;

[0007] 1.2) Considering the physical characteristics and operation constraints of resources, mathematical models of various types of resources are established; the models include wind power generation model, photovoltaic power generation model, energy storage equipment model, and electric vehicle model;

[0008] The wind power generation model is constructed as follows:

[0009] Wind power generation uses a doubly-fed induction generator, and the rotor is excited by AC voltage. When the external wind speed and wind direction change, the rotor winding changes accordingly, and the stator winding remains unchanged. The generator can output AC power with a constant frequency. When the external wind speed is just enough to drive the wind turbine to start generating electricity, the wind speed is defined as the cut-in wind speed; when the wind speed cannot make the wind turbine run to generate electricity, it is defined as the cut-out speed; when the external wind speed makes the wind turbine reach the nominal power, the wind speed is called the nominal wind speed. The expected output power changes with wind force as shown below:

[0010]

[0011] Where: P per , P rat Respectively represent the predicted output of the fan and the rated output of the fan; ν per Represents the predicted wind speed value; ν in , ν rat , ν out They represent the cut-in wind speed, nominal wind speed and cut-out wind speed respectively;

[0012] In the wind turbine, the variation law of wind energy parameters is described by the Weibull distribution function, and the probability distribution density function is as follows:

[0013]

[0014] Where: h,j Indicates the current wind speed at a certain height; k h,j is the Wilbull shape parameter; c h,j is the distribution parameter;

[0015] After obtaining the scale and shape parameters, the wind speed series is fitted, and the Weibull cumulative probability distribution, the expected value of the wind speed Weibull and the variance are expressed as follows:

[0016]

[0017]

[0018]

[0019] Where: Γ is the gamma function, which represents the mathematical expectation operator. The smaller the variance, the more stable the wind speed, and the smaller the randomness of the wind turbine, the more accurate the prediction.

[0020] The construction method of photovoltaic power generation model is:

[0021] When calculating the power generated by a photovoltaic power generation system, the illumination time and light intensity are used as the determining factors of the power generation, which are expressed as follows:

[0022] P ill =S ill l illt l mpt l inv (1-0.005T a -0.0125) (6)

[0023] Where: S ill It indicates the actual area of ​​the solar panel that can receive sunlight when it is put into operation; illt represents the light intensity at time t; l mpt is the tracking efficiency of the maximum power point; l invis the power generation efficiency of the inverter in the photovoltaic power generation system; T a is the ambient temperature.

[0024] The energy storage equipment model includes a battery model and a thermal tank model:

[0025] Build a battery model:

[0026] In the virtual power plant, a circuit model that can effectively reflect the electrical characteristics of the battery port is used as the research model of the battery. The equivalent circuit is represented by a modified battery model, which consists of an ideal voltage source, an internal resistance, and a parallel RC reactance. In daily scheduling, the factors that affect the working of the battery include the battery capacity C cell , State of Charge SOC cell , charge and discharge capacity and discharge depth, the mathematical model is expressed as:

[0027]

[0028]

[0029] Where: Ccell represents the remaining energy in the battery; They represent the charging and discharging power of the battery at time t respectively; Represent the charging and discharging efficiency coefficients respectively.

[0030] Build the hot tank model:

[0031] The heat storage model of the heat storage device at time t is as follows:

[0032]

[0033] Where: SOC tst (t) represents the residual heat in the heat storage tank at time t; Indicates the heat storage tank's charge and power at time t; Indicates the thermal efficiency coefficient of the heat storage tank.

[0034] The electric vehicle model is constructed as follows:

[0035] Fluctuation of electric vehicle load Taking into account the randomness of travel patterns, Monte Carlo sampling is used to sample the corresponding functions of mileage, travel time and return time, simulating the random travel patterns of electric vehicles. The model is established by taking the i-th electric vehicle as an example:

[0036]

[0037]

[0038] S socev,i (t) = S socev,i(t-1)+η ev P ev,i (t) (12)

[0039] S socev,min ≤S socev (t)≤S socev,max (13)

[0040] 0.948I i ≤S socev,l,i ≤I i (14)

[0041]

[0042]

[0043] Where: P ev,in (t),P ev,out (t) respectively represent the charging and discharging power of the i-th electric vehicle at time t; Respectively represent the maximum and minimum charging power of electric vehicles; They represent the maximum and minimum discharge power of electric vehicles respectively; U ev,in , U ev,out It is a binary variable, which is 1 when the electric vehicle is involved in charging and discharging, otherwise it is 0; S socev,i (t), S socev,i (t-1), S socev,min , S socev,max are the battery capacity constraints of the i-th electric vehicle at time t and time t-1, as well as the minimum and maximum values ​​of the battery capacity of the electric vehicle; I i is the battery capacity of the i-th electric car; η ev is the charge and discharge efficiency; equations (10)-(14) are the constraints of the i-th electric vehicle; equation (15) is the total power of the electric vehicle; and equation (16) is the fluctuation range of the total power of the electric vehicle.

[0044] 1.3) Determine the objective function of vertex enumeration and set constraints;

[0045] 1.4) Construct the optimization problem M0, where the constraints include fully controlled power source, semi-controlled power source and energy storage resource constraints;

[0046] Step 2: Realize the coordinated operation optimization of multiple virtual power plants based on the binary consensus algorithm;

[0047] With the goal of minimizing the overall operating cost and maximizing the reliability of the distribution network, an optimal dispatching model is established:

[0048] 2.1) The economic optimization model of multiple VPPs takes the optimal total cost of each aggregated VPP as the objective function, including active and reactive costs. The active cost adopts a piecewise quadratic function aggregation cost model, and the reactive cost includes fixed and opportunity costs. At the same time, relevant constraints are given, including power balance constraints, power increment constraints, and node voltage constraints;

[0049] The specific method is:

[0050] The objective function of the economic optimization model of multiple VPPs is to optimize the total cost of each aggregated VPP. The total cost of VPPi includes the active cost. Reactive power cost Right now:

[0051]

[0052] Where: I is the number of VPPs included in the distribution network;

[0053] The active cost adopts a piecewise quadratic function aggregation cost model, which is the superposition of the costs of various resources within the VPP; the reactive cost includes fixed cost and opportunity cost. The fixed cost is the unit time cost converted based on the planned service life of the equipment, and the opportunity cost is the part of the cost lost by the VPP in active power generation due to providing reactive services. Since various power generation resources within the VPP have corresponding power constraints during actual operation, when the reactive output reaches a certain level, it will inevitably affect the active power generation income. The specific expression is:

[0054]

[0055]

[0056] Where: They are the costs of FCP, HCP and energy storage resources; and is the cost coefficient of FCPm; and is the cost coefficient of HCPn; and is the cost coefficient of energy storage resource l; α i,n , β i,n and γ i,n They are the secondary, primary and constant cost coefficients when VPPi outputs active power in n sections respectively; They are the active powers of FCPm, HCPn and energy storage resource l respectively;

[0057] Constraints

[0058]

[0059]

[0060]

[0061] Where: P Pr and Q Pr are the total active and reactive power requirements of the system respectively; P Loss and Q Loss They are the total active and reactive losses of the system respectively; ΔP i and ΔQ i are the active and reactive power increments of VPP respectively; U j is the voltage at node j; U min and U max are the minimum and maximum values ​​of the node voltage respectively; R ij and X ij are the resistance and reactance of line ij respectively; and are the injected active and reactive powers of node i respectively: N is the total number of nodes.

[0062] 2.2) Each VPP in the active distribution network is abstracted as a VPP node to interact with the communication network information and form a communication network topology;

[0063] The adjacency matrix A represents the association between nodes. A is a 0-1 matrix with diagonal elements of 0 and off-diagonal elements of a. ij . b i is the degree of the middle node i, then B = diag(b1, b2, ..., b N ), the Laplace matrix is ​​L = BA = [l ij ],in

[0064]

[0065] Set two consistency variables λ and w. The state variables λ and w of each node are iteratively updated and finally reach consistency. For node i, let λ(i, k) and w(i, k) represent the binary consistency variables of the kth iteration. When any two nodes i and j in the system satisfy |λ i,k -λ j,k |→0 and |w i,k -w j,k |→0, that is, the system converges when the state variables of all nodes are consistent within the convergence condition range;

[0066] Next, the objective function can be transformed into:

[0067]

[0068] By taking partial derivatives of the above formula, we get the equivalent optimal condition:

[0069]

[0070] Where: and They are respectively the slight increase rates of active and reactive costs: i and w i They are the consistency variables of VPPi; and They are the network loss slight increase rate of active network loss of node i to node active injection, active network loss to node reactive injection, reactive network loss to node active injection and reactive network loss to node reactive injection. The specific expression is:

[0071]

[0072]

[0073] According to the above equations, the binary consistency variable expression can be obtained by simultaneous transformation:

[0074]

[0075] The update rule of the binary consensus algorithm is expressed as:

[0076]

[0077] Where: i,k+1 and w i,k+1 is the binary consistency variable after the k+1th iteration of VPPi; I D I is an autonomous node; G is the dominant node; μ p and μ Q are active and reactive power regulation coefficients respectively; d ij is the state transfer matrix coefficient, which is determined by the topology of the communication network; ΔP Pr and ΔQ Pr are active and reactive power deviations respectively. ij , ΔP Pr and ΔQ Pr The expressions are:

[0078]

[0079]

[0080] Where: l ij,k is the k-th iteration l ij ;

[0081] Finally, according to equations (17) and (23), the power update rule of each VPP node is obtained as follows:

[0082]

[0083] Where: P i,k and Q i,k are the output active and reactive powers of VPPi at the kth iteration, respectively;

[0084] 2.3) Set the constraints of the optimization model, including the power constraints and voltage constraints of the resources; the specific method is:

[0085] 2.3.1) The power limit constraint is processed through the power correction process: when the power exceeds the limit, in order not to change the regulation capability property represented by the over-limit VPP, that is, the power factor, the actual output power of the VPP is set according to the VFR boundary value under the same power factor, and the node is removed from the communication network topology, and the network topology matrix is ​​updated; when multiple VPPs exceed the power limit, the VPP with the largest degree of over-limit is selected, and the above PC process is implemented for the VPP, and then the consistency calculation is re-performed, and the power over-limit check is completed again until the operating points of all VPPs are within the VFR boundary;

[0086] 2.3.2) The voltage correction for the over-limit voltage is realized by the correction equation of the binary consistency variable:

[0087] In order to characterize the voltage change and degree of voltage over-limit during the voltage over-limit process of consistency iteration, it is necessary to clarify the impact of the injected power change of each node on the voltage. Therefore, voltage sensitivity is selected to form the basis for solving the voltage over-limit problem.

[0088]

[0089] Where: ΔU w is the voltage change value of node m: B i,m is the intersection of the line number set from node i to the power inflow node of the distribution network and the line number set from node m to the power inflow node of the distribution network;

[0090] From the voltage sensitivity, we can know that the voltage over-limit problem in the set scenario is essentially that the power output of the VPP node is too high or too low, which makes it unable to support the corresponding voltage level, and then causes the power to exceed the specified range and the voltage to exceed the limit. Therefore, the voltage over-limit problem is solved by adjusting the external output power characteristics of each VPP node;

[0091] For voltage exceeding the upper limit: the node voltage is affected by the output power of each VPP node. It can be seen from formula (31) that the calculation of the output power of each VPP node is directly related to the consistency variable. By adding the correction equation of the binary consistency variable, the update law of the binary consistency variable is corrected to solve the voltage exceeding the limit problem. When the voltage constraint is ignored for optimization calculation, if only one VPP node voltage exceeds the upper limit globally, the economic optimum of the VPP node when it is kept at the upper limit voltage is the global economic optimum considering the voltage constraint. Therefore, the power output characteristics of the VPP node with the most serious voltage exceeding the limit are adjusted first.

[0092] The voltage inequality constraint of node m, where the voltage exceeds the upper limit most seriously, is written as an equality constraint:

[0093] U m =U max (33)

[0094] Increase the Lagrange day The corresponding Lagrange constant calculation formulas expressed by equations (23) and (24) are:

[0095]

[0096] Where: and The calculation formula is:

[0097]

[0098] Based on formula (34), the k-th iteration correction equation of the binary consistency variable can be set as:

[0099]

[0100] Where: Δλ i,k and Δw i,k are the consistency variable correction values ​​of consistency variables λ and w of VPP node i at the kth iteration respectively; and is the voltage correction value set based on formula (35), and its expression is:

[0101]

[0102] Where: k λ and k w All are voltage over-limit adjustment coefficients; k am The number of iterations to pass the out-of-limit value to the out-of-limit node.

[0103] At this point, the consistency variable correction equation given by equation (36) can be used to adjust the node power output by correcting the consistency variable to solve the node voltage over-limit problem;

[0104] For the case where the node voltage exceeds the lower limit, the formula is:

[0105]

[0106]

[0107] Step 3: Based on the above-mentioned feasible domain solution method of virtual power plants based on vertex enumeration method and the collaborative operation optimization method of multiple virtual power plants based on binary consistency algorithm, a resource aggregation and aggregate optimization operation framework process is constructed; in this process, distributed resources are first aggregated through vertex enumeration method to obtain the feasible domain of virtual power plants; then the binary consistency algorithm is used to optimize the collaborative operation of multiple virtual power plants, and power constraints and voltage constraints are continuously processed during the optimization process to ensure stable operation of the system.

[0108] Compared with the prior art, the present invention has the following beneficial effects:

[0109] 1. The present invention improves resource utilization through accurate modeling. By carrying out detailed mathematical modeling of resources such as distributed power sources, energy storage systems, electric vehicles and controllable loads, the physical characteristics and operation constraints of resources are fully considered, and the accuracy and reliability of the resource model are improved.

[0110] 2. The present invention takes the lowest overall operating cost of the distribution network as the goal, establishes a multi-objective optimization model, comprehensively considers factors such as power purchase cost, power sales revenue, ancillary service revenue and operation and maintenance cost, and optimizes the objective function. The trading mechanism for virtual power plants to participate in the power market is designed, including trading strategies for the day-ahead market, intraday market and real-time market. Through reasonable quotation and bidding strategies, the power purchase cost is reduced and the power sales revenue is increased.

[0111] 3. The present invention takes into account the reliability indicators of the system, such as power supply reliability and power quality, in the optimization model, and improves the overall reliability of the system by optimizing the scheduling strategy. Through the prediction model and scenario generation technology, the uncertainty of resource output and load demand is effectively handled, and the risk resistance of the system is enhanced.

[0112] 4. The present invention organically integrates renewable energy such as photovoltaic and wind power with resources such as energy storage systems and controllable loads through an efficient resource aggregation model, thereby improving the absorption capacity and utilization rate of renewable energy. BRIEF DESCRIPTION OF THE DRAWINGS

[0113] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. Among them:

[0114] Figure 1 It is a flow chart of the optimization operation framework of the aggregate based on the binary consensus algorithm;

[0115] Figure 2 It is a power topology and communication topology diagram of the active power distribution network system in Example 1;

[0116] Figure 3 This is a schematic diagram of multi-period operation of resource aggregation and aggregate optimization in Example 1;

[0117] Figure 4 are the VFR, operating point and operating point iteration path of each VPP in Example 1;

[0118] Figure 5 It is the iterative convergence process of each VPP in Example 1;

[0119] Figure 6 This is the iterative convergence process of the total output power of the system in Example 1;

[0120] Figure 7 It is the iterative convergence process of each VPP in Example 2;

[0121] Figure 8 This is the VPP operating point correction process of Example 2;

[0122] Fig. 9 This is the maximum over-limit node voltage correction process of embodiment 2. DETAILED DESCRIPTION

[0123] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.

[0124] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0125] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0126] In order to verify the feasibility of the patented algorithm, a 13-node active distribution network system including 7 9-node VPPs was selected for case testing. All case tests were based on the MATLAB-2022A platform, Intel(R)Core(TM)i5-12400 CPU and commercial solver Cplex-12.10.0. For the example scenario designed in this article, the following assumptions are made: 1) Only the reactive output of the VPP is considered as the reactive support source of the voltage, and no additional reactive compensation equipment is considered. 2) The decision variables of the binary consistency algorithm for solving the optimization problem are the power output of each VPP and the PCC of the distribution network.

[0127] The power topology diagram and communication topology of the active distribution network system used in this patent are as follows Figure 2 The communication topology node where VPP1 is located is selected as the dominant node.

[0128] This paper sets two groups of system power demand values ​​for example verification to reflect the feasibility of the proposed resource aggregation and aggregate optimization operation framework process under different power requirements. The time interval of each group of examples is 15 minutes. Figure 3 As shown. The initial state of the system in Example N2 is based on the running results of the previous example. This paper verifies the feasibility of the algorithm and framework under positive power demand through Example N1, and verifies the feasibility of the power limit processing solution and the voltage limit correction equation through Example N2.

[0129] At the same time, a centralized optimization control example is set to verify the accuracy of the results, and a control example of the traditional one-element consistency algorithm is set to verify the advantages of the binary consistency algorithm proposed in this paper. The system-related parameters are set as follows: the confidence level of the distributed power uncertainty probability model is 0.95; Δk in the vertex enumeration method is π / 100; the power regulation coefficient μ P and μ Q is 0.25; the convergence termination condition is 0.001; the voltage over-limit adjustment coefficient k λ and k w Both are 0.23.

[0130] Example 1

[0131] This section will analyze the comparison between the proposed method and the centralized optimization and traditional consistency algorithms in turn.

[0132] The vertex enumeration method is used to solve all the VFRs of the seven VPPs at the next moment (09:00) within 67.952s after 08:45, and the VFRs are used to characterize the active-reactive regulation capability of distributed resources during the period of 09:00-09:15. Figure 4 The 157 iterations of the binary consensus algorithm convergence process and the VPP operation points after iterative convergence are given. Finally, the aggregate optimization operation solution is completed within 311.587ms after 09:00. It can be seen that the framework process proposed in this paper can realize the effective control of multiple VPPs, and make full use of their active-reactive regulation capabilities, realize the solution of the active-reactive economic optimization problem of the power grid, and improve the operational flexibility of the power grid.

[0133] The active power deviation of the active distribution network output calculated by the proposed method is 0.0520%, the reactive power deviation is 0.0507%, and the cost deviation is 0.0042%. The calculation error with centralized optimization mainly comes from the VFR boundary error, the aggregation cost model error, and the error caused by the voltage correction process. The iterative convergence process of binary consistency variables, active-reactive power, and corresponding costs is shown in Figure 5 As shown. Figure 6 After the last iteration, the difference between the solid line and the dotted line is the network loss, of which the active network loss accounts for 3.7488% and the reactive network loss accounts for 5.7292%.

[0134] The control group of the traditional consistency algorithm needs to calculate and iterate two consistency variables separately, which greatly increases the calculation time. Due to the active-reactive coupling problem, it is impossible to set interactive iterations between the convergence of the two consistency variables, so the final convergence result may be a local optimum, not a global optimum.

[0135] Example 2

[0136] All VFR calculations in case N2 were completed within 76.105s in the previous period (09:00-09:15). The iterative convergence process of binary consistency variables, active-reactive power, and corresponding costs is shown in Figure 7 The binary consistency variables converged for the first time after 305.7ms iterations, but VPP1, VPP3, and VPP6 all had power limit violations, and active distribution network nodes 6, 12, and 13 all had voltage limit violations.

[0137] In this regard, according to the power over-limit processing principle, VPP3 (5 nodes) with the largest power over-limit amplitude is operated at a fixed power factor angle of 0.24, and the node is removed from the communication network topology. The new round of iterations converged after the 647.1ms iteration of the total iteration process, but VPP1 and VPP6 still had power over-limits. Using the same method as above, power over-limit adjustments were made to VPP1 and VPP6 after the 647.1 and 954.8ms iterations, respectively. Finally, convergence was completed after the 1331.9ms iteration, at which time all VPPs in the system had no power over-limits. See the power operating point adjustment process. Figure 8 After 954.8ms iteration, the voltage still has the problem of exceeding the upper limit. The binary consistency variable correction equation is used to correct VPP7 (13 nodes) with the largest degree of exceeding the limit, and finally convergence is achieved after the 1665.4ms iteration. The voltage change during the voltage correction process is shown in Fig. 9 . In summary, this paper fully utilizes the active-reactive regulation capability of distributed resources to achieve voltage support at each node and improve the flexibility of distributed resources in participating in power grid operation. The deviation of active power output of the active distribution network calculated by the proposed method is 0.8367%, the deviation of reactive power is 0.8367%, and the deviation of cost is 0.1513%. The proportion of active network loss is 4.5477%, and the proportion of reactive network loss is 7.7660%.

[0138] The specific implementation methods of the present invention are not exhaustive and do not constitute a limitation on the scope of protection of the claims. Based on the inspiration gained from the embodiments of the present invention, those skilled in the art can think of other substantially equivalent alternatives without creative work, all of which are within the scope of protection of the present invention.

Claims

1. A novel method for aggregation and optimization of distributed resources of multiple virtual power plants in a distribution network, characterized in that: The steps are: Step 1: Solve the feasible region of the virtual power plant based on the vertex enumeration method; 1.1) Classify distributed resources, including distributed power sources, energy storage systems, electric vehicles and controllable loads; 1.2) Considering the physical characteristics and operation constraints of resources, mathematical models of various types of resources are established; the models include wind power generation model, photovoltaic power generation model, energy storage equipment model, and electric vehicle model; 1.3) Determine the objective function of vertex enumeration and set constraints; 1.4) Construct the optimization problem M0, where the constraints include fully controlled power source, semi-controlled power source and energy storage resource constraints; Step 2: Realize the coordinated operation optimization of multiple virtual power plants based on the binary consensus algorithm; With the goal of minimizing the overall operating cost and maximizing the reliability of the distribution network, an optimal dispatching model is established: 2.1) The economic optimization model of multiple VPPs takes the optimal total cost of each aggregated VPP as the objective function, including active and reactive costs. The active cost adopts a piecewise quadratic function aggregation cost model, and the reactive cost includes fixed and opportunity costs. At the same time, relevant constraints are given, including power balance constraints, power increment constraints, and node voltage constraints; 2.2) Each VPP in the active distribution network is abstracted as a VPP node to interact with the communication network information and form a communication network topology; 2.3) Set the constraints of the optimization model, including power constraints and voltage constraints of resources; Step 3: Based on the above-mentioned feasible domain solution method of virtual power plants based on vertex enumeration method and the collaborative operation optimization method of multiple virtual power plants based on binary consistency algorithm, a resource aggregation and aggregate optimization operation framework process is constructed; in this process, distributed resources are first aggregated through vertex enumeration method to obtain the feasible domain of virtual power plants; then the binary consistency algorithm is used to optimize the collaborative operation of multiple virtual power plants, and power constraints and voltage constraints are continuously processed during the optimization process to ensure stable operation of the system.

2. According to claim 1, a novel method for distributed resource aggregation and optimization operation of multiple virtual power plants in a distribution network is characterized in that: In step 1, the method for constructing the wind power generation model is: Wind power generation uses a doubly-fed induction generator, and the rotor is excited by AC voltage. When the external wind speed and wind direction change, the rotor winding changes accordingly, and the stator winding remains unchanged. The generator can output AC power with a constant frequency. When the external wind speed is just enough to drive the wind turbine to start generating electricity, the wind speed is defined as the cut-in wind speed; when the wind speed cannot make the wind turbine run to generate electricity, it is defined as the cut-out speed; when the external wind speed makes the wind turbine reach the nominal power, the wind speed is called the nominal wind speed. The expected output power changes with wind force as shown below: Where: P per , P rat They represent the predicted output of the fan and the rated output of the fan respectively; v per Indicates the predicted wind speed value; v in 、v rat 、v out They represent the cut-in wind speed, nominal wind speed and cut-out wind speed respectively; In the wind turbine, the variation law of wind energy parameters is described by the Weibull distribution function, and the probability distribution density function is as follows: Where: v h，j Indicates the current wind speed at a certain height; k h，j is the Wilbull shape parameter; c h，j is the distribution parameter; After obtaining the scale and shape parameters, the wind speed series is fitted, and the Weibull cumulative probability distribution, the expected value of the wind speed Weibull and the variance are expressed as follows: Where: Γ is the gamma function, which represents the mathematical expectation operator. The smaller the variance, the more stable the wind speed, and the smaller the randomness of the wind turbine, the more accurate the prediction.

3. According to claim 1, a novel method for distributed resource aggregation and optimization operation of multiple virtual power plants in a distribution network is characterized in that: In the step 1, the photovoltaic power generation model is constructed by: When calculating the power generated by a photovoltaic power generation system, the illumination time and light intensity are used as the determining factors of the power generation, which are expressed as follows: P ill =S ill l illt l mpt l inv (1-0.005T a -0.0125) (6) Where: S ill It indicates the actual area of ​​the solar panel that can receive sunlight when it is put into operation; illt represents the light intensity at time t; l mpt is the tracking efficiency of the maximum power point; l inv is the power generation efficiency of the inverter in the photovoltaic power generation system; T a is the ambient temperature.

4. According to claim 1, a novel method for distributed resource aggregation and optimization operation of multiple virtual power plants in a distribution network is characterized in that: In step 1, the energy storage device model includes a battery model and a hot tank model: Build a battery model: In the virtual power plant, a circuit model that can effectively reflect the electrical characteristics of the battery port is used as the research model of the battery. The equivalent circuit is represented by a modified battery model, which consists of an ideal voltage source, an internal resistance, and a parallel RC reactance. In daily scheduling, the factors that affect the working of the battery include the battery capacity C cell , State of Charge SOC cell , charge and discharge capacity and discharge depth, the mathematical model is expressed as: Where: Ccell represents the remaining energy in the battery; They represent the charging and discharging power of the battery at time t respectively; Respectively represent the charging and discharging efficiency coefficients; Build the hot tank model: The heat storage model of the heat storage device at time t is as follows: Where: SOC tst (t) represents the residual heat in the heat storage tank at time t; Indicates the heat storage tank's charge and power at time t; Indicates the thermal efficiency coefficient of the heat storage tank.

5. According to claim 1, a novel method for distributed resource aggregation and optimized operation of multiple virtual power plants in a distribution network is characterized in that: In the step 1, the method for constructing the electric vehicle model is: Fluctuation of electric vehicle load Taking into account the randomness of travel patterns, Monte Carlo sampling is used to sample the corresponding functions of mileage, travel time and return time, simulating the random travel patterns of electric vehicles. The model is established by taking the i-th electric vehicle as an example: S socev，i (t)S socev，i (t-1)+η ev P ev，i (t) (12) S socev,min ≤S socew (t)≤S socev,max (13) 0.948I i ≤S socev，l，i ≤I i (14) Where: P ev，in (t),P ev，out (t) respectively represent the charging and discharging power of the i-th electric vehicle at time t; Respectively represent the maximum and minimum charging power of electric vehicles; They represent the maximum and minimum discharge power of electric vehicles respectively; U ev，in , U ev，out It is a binary variable, which is 1 when the electric vehicle is involved in charging and discharging, otherwise it is 0; S socev，i (t), S socev，i (t-1), S socev，min , S socev，max are the battery capacity constraints of the i-th electric vehicle at time t and time t-1, as well as the minimum and maximum values ​​of the battery capacity of the electric vehicle; I i is the battery capacity of the i-th electric car; η ev is the charge and discharge efficiency; equations (10)-(14) are the constraints of the i-th electric vehicle; equation (15) is the total power of the electric vehicle; and equation (16) is the fluctuation range of the total power of the electric vehicle.

6. A novel method for aggregation and optimization of distributed resources of multiple virtual power plants in a distribution network according to claim 1, characterized in that: In the above 2.1), the specific method is: The objective function of the economic optimization model of multiple VPPs is to optimize the total cost of each aggregated VPP. The total cost of VPPi includes the active cost. Reactive power cost Right now: Where: I is the number of VPPs included in the distribution network; The active cost adopts a piecewise quadratic function aggregation cost model, which is the superposition of the costs of various resources within the VPP; the reactive cost includes fixed cost and opportunity cost. The fixed cost is the unit time cost converted based on the planned service life of the equipment, and the opportunity cost is the part of the cost lost by the VPP in active power generation due to providing reactive services. Since various power generation resources within the VPP have corresponding power constraints during actual operation, when the reactive output reaches a certain level, it will inevitably affect the active power generation income. The specific expression is: Where: They are the costs of FCP, HCP and energy storage resources; and is the cost coefficient of FCPm; and is the cost coefficient of HCPn; and is the cost coefficient of energy storage resource l; α i，n , β i，n and γ i，n They are the secondary, primary and constant cost coefficients when VPPi outputs active power in n sections respectively; P l E They are the active powers of FCPm, HCPn and energy storage resource l respectively; Constraints Where: P pr and Q Pr are the total active and reactive power requirements of the system respectively; P Loss and Q Loss They are the total active and reactive losses of the system respectively; ΔP i and ΔQ i are the active and reactive power increments of VPP respectively; U j is the voltage at node j; U min and U max are the minimum and maximum values ​​of the node voltage respectively; R ij and X ij are the resistance and reactance of line ij respectively; P i N and are the injected active and reactive powers of node i respectively: N is the total number of nodes.

7. A novel method for aggregation and optimization of distributed resources of multiple virtual power plants in a distribution network according to claim 1, characterized in that: In the above 2.2), the specific method is: The adjacency matrix A represents the association between nodes. A is a 0-1 matrix with diagonal elements of 0 and off-diagonal elements of a. ij ; b i is the degree of the middle node i, then B = diag(b1, b2, ...., b N ), the Laplace matrix is ​​L = BA = [l ij ],in Set two consistency variables λ and w. The state variables λ and w of each node are iteratively updated and finally reach consistency. For node i, let λ(i, k) and w(i, k) represent the binary consistency variables of the kth iteration. When any two nodes i and j in the system satisfy |λ i，k -λ j，k |→0 and |w i，k -w j，k |→0, that is, the system converges when the state variables of all nodes are consistent within the convergence condition range; Next, the objective function can be transformed into: By taking partial derivatives of the above formula, we get the equivalent optimal condition: Where: and They are respectively the cost increase rates of active power and reactive power: i and w i They are the consistency variables of VPPi; and They are the network loss slight increase rate of active network loss of node i to node active injection, active network loss to node reactive injection, reactive network loss to node active injection and reactive network loss to node reactive injection. The specific expression is: According to the above equations, the binary consistency variable expression can be obtained by simultaneous transformation: The update rule of the binary consensus algorithm is expressed as: Where: i，k+1 and w i，k+1 is the binary consistency variable after the k+1th iteration of VPPi; I D I is an autonomous node; G is the dominant node; μ P and μ Q are active and reactive power regulation coefficients respectively; d ij is the state transfer matrix coefficient, which is determined by the topology of the communication network; ΔP pr and ΔQ Pr are active and reactive power deviations respectively; d ij , ΔP pr and ΔQ pr The expressions are: Where: l ij，k is the k-th iteration l ij ; Finally, according to equations (17) and (23), the power update rule of each VPP node is obtained as follows: Where: P i，k and Q i，k are the output active and reactive powers of VPPi at the kth iteration, respectively; 8. A novel method for aggregation and optimization of distributed resources of multiple virtual power plants in a distribution network according to claim 1, characterized in that: In the above 2.3), the specific method is: 2.3.1) The power limit constraint is processed through the power correction process: when the power exceeds the limit, in order not to change the regulation capability property represented by the over-limit VPP, that is, the power factor, the actual output power of the VPP is set according to the VFR boundary value under the same power factor, and the node is removed from the communication network topology, and the network topology matrix is ​​updated; when multiple VPPs exceed the power limit, the VPP with the largest degree of over-limit is selected, and the above PC process is implemented for the VPP, and then the consistency calculation is re-performed, and the power over-limit check is completed again until the operating points of all VPPs are within the VFR boundary; 2.3.2) The voltage correction for the over-limit voltage is realized by the correction equation of the binary consistency variable: In order to characterize the voltage change and degree of voltage over-limit during the voltage over-limit process of consistency iteration, it is necessary to clarify the impact of the injected power change of each node on the voltage. Therefore, voltage sensitivity is selected to form the basis for solving the voltage over-limit problem. Where: ΔU w is the voltage change value of node m: B i，m is the intersection of the line number set from node i to the power inflow node of the distribution network and the line number set from node m to the power inflow node of the distribution network; From the voltage sensitivity, we can know that the voltage over-limit problem in the set scenario is essentially that the power output of the VPP node is too high or too low, which makes it unable to support the corresponding voltage level, and then causes the power to exceed the specified range and the voltage to exceed the limit. Therefore, the voltage over-limit problem is solved by adjusting the external output power characteristics of each VPP node; For voltage exceeding the upper limit: the node voltage is affected by the output power of each VPP node. From formula (31), it can be seen that the calculation of the output power of each VPP node is directly related to the consistency variable; By adding the correction equation of the binary consistency variable and modifying the update rule of the binary consistency variable to solve the voltage over-limit problem, when the voltage constraint is ignored for optimization calculation, if only one VPP node has a voltage over-limit globally, then the economic optimum of the VPP node when it is kept at the upper limit voltage is the global economic optimum considering the voltage constraint, so the power output characteristics of the VPP node with the most serious voltage over-limit are adjusted first; The voltage U of the node m where the voltage exceeds the upper limit the most seriously m Inequality constraints are written as equality constraints: IN m =U max (33) Increase the Lagrange day The corresponding Lagrange constant calculation formulas expressed by equations (23) and (24) are: Where: and The calculation formula is: Based on formula (34), the k-th iteration correction equation of the binary consistency variable can be set as: Where: Δλ i,k and Δw i，k are the consistency variable correction values ​​of the consistency variables λ and w of the kth iteration of VPP node i, respectively; and is the voltage correction value set based on formula (35), and its expression is: Where: k λ and k w All are voltage over-limit adjustment coefficients; k am The number of iterations for passing the out-of-limit value to the out-of-limit node; At this point, the consistency variable correction equation given by equation (36) can be used to adjust the node power output by correcting the consistency variable to solve the node voltage over-limit problem; For the case where the node voltage exceeds the lower limit, the formula is: Where: and are the voltage correction values ​​based on formula (37) respectively.

Images